
Les outils d'IA en ligne transforment rapidement l'ingénierie mécanique en augmentant les capacités humaines en matière de conception et d'analyse, fabricationet la maintenance. Ces systèmes d'IA peuvent traiter de grandes quantités de données, identifier des modèles complexes et générer des solutions nouvelles beaucoup plus rapidement que les méthodes traditionnelles. Par exemple, l'IA peut vous aider à optimiser les conceptions en termes de performance et de fabricabilité, à accélérer les simulations complexes, à prédire les propriétés des matériaux et à automatiser un large éventail de tâches analytiques.
Les invites fournies ci-dessous aideront par exemple à la conception générative, à l'accélération des simulations (FEA/CFD), à la maintenance prédictive où l'IA analyse les données des capteurs des machines pour prévoir les défaillances potentielles, ce qui permet un entretien proactif et minimise les temps d'arrêt, à la sélection des matériaux et à bien d'autres choses encore.
- Cette page est spécifique à un domaine. Si nécessaire, vous pouvez effectuer une recherche complète sur tous les domaines et tous les critères dans notre >. Répertoire d'invites AI <, dédié à conception de produits et innovation.
- Compte tenu des ressources et du temps disponibles sur le serveur, les invites elles-mêmes sont réservées aux membres enregistrés et ne sont pas visibles ci-dessous si vous n'êtes pas connecté. Vous pouvez vous inscrire, 100% gratuit :
Adhésion requise
Vous devez être membre du site pour accéder à ce contenu.
- Analyse des causes profondes
- Ingénierie mécanique
Invitation à l'IA à Diagramme en arête de poisson Facteurs d'échec
- Amélioration continue, Conception pour la fabrication (DfM), Analyse des défaillances, Production allégée, Génie mécanique, Amélioration des processus, Gestion de la qualité, Analyse des causes profondes, Six Sigma
Aide à structurer un diagramme en arête de poisson (Ishikawa) pour la défaillance d'un composant mécanique en suggérant des catégories de facteurs contributifs potentiels (par exemple, l'homme, la machine, le matériel, la méthode, l'environnement et les mesures) et des questions spécifiques à poser pour chaque catégorie en fonction de la description de la défaillance. Cette invite facilite l'analyse systématique des causes profondes. Le résultat est un schéma au format markdown.
Sortie :
- Markdown
- ne nécessite pas d'Internet en direct
- Champs : {composant_that_failed} {failure_mode_description} {operating_conditions_at_failure_text}
Act as a Root Cause Analysis (RCA) Facilitator.
Your TASK is to help structure a Fishbone (Ishikawa) Diagram to investigate the root cause of a failure involving `{component_that_failed}`.
The described failure is: `{failure_mode_description}`.
The failure occurred under these conditions: `{operating_conditions_at_failure_text}`.
You should propose key questions for standard Fishbone categories tailored to this mechanical failure context.
**FISHBONE DIAGRAM STRUCTURE INPUTS (MUST be Markdown format):**
**Problem Statement (Head of the Fish):** Failure of `{component_that_failed}`: `{failure_mode_description}`
**Main Bones (Categories) and Potential Contributing Factor Questions:**
**1. Machine (Equipment / Technology)**
* Was the `{component_that_failed}` the correct type/model/specification for the application?
* Was the equipment where `{component_that_failed}` is installed operating correctly before/during the failure? (e.g.
speed
load
pressure
temperature within design limits described in `{operating_conditions_at_failure_text}`?)
* Had there been any recent maintenance
repair
or modification to the machine or `{component_that_failed}`? Were procedures followed?
* Was auxiliary equipment (e.g.
cooling
lubrication
power supply) functioning correctly?
* Is there a history of similar failures with this machine or other similar machines?
* Could any tooling
fixtures
or associated parts have contributed to the failure of `{component_that_failed}`?
**2. Method (Process / Procedure)**
* Were correct operating procedures being followed when the failure occurred
considering `{operating_conditions_at_failure_text}`?
* Were installation or assembly procedures for `{component_that_failed}` followed correctly?
* Were maintenance procedures adequate and followed correctly for `{component_that_failed}` and related systems?
* Were there any recent changes in operating procedures
set-points
or work instructions?
* Was the system being operated outside of its design intent or capacity?
* Could any testing or quality control procedures related to `{component_that_failed}` have missed a defect?
**3. Material (Includes Raw Materials
Consumables
`{component_that_failed}` itself)**
* Was the `{component_that_failed}` made from the specified material? Was material certification available/correct?
* Could there have been a defect in the material of `{component_that_failed}` (e.g.
inclusions
porosity
incorrect heat treatment
flaws)?
* If consumables are involved (e.g.
lubricants
hydraulic fluids
coolants)
were they the correct type
clean
and at the correct level/condition?
* Has the `{component_that_failed}` been exposed to any corrosive or degrading substances not accounted for in its design?
* Could there have been issues with material handling or storage of `{component_that_failed}` before installation?
**4. Manpower (People / Personnel)**
* Was the operator/maintenance personnel adequately trained and qualified for the task they were performing related to `{component_that_failed}` or its system?
* Was there sufficient experience or supervision?
* Could human error (e.g.
misjudgment
incorrect assembly
misreading instructions
fatigue) have contributed?
* Were personnel following safety procedures? Were they rushed or under stress?
* Was there clear communication regarding operational or maintenance status?
**5. Measurement (Inspection / Instrumentation)**
* Were measuring instruments or sensors used to monitor `{operating_conditions_at_failure_text}` (e.g.
temperature
pressure
vibration
current) calibrated and functioning correctly?
* Were any warning signs or abnormal readings from instruments ignored or misinterpreted prior to the failure of `{component_that_failed}`?
* Were quality control checks or inspections of `{component_that_failed}` (pre-installation or during service) performed correctly and were the criteria appropriate?
* Could there be inaccuracies in the data used to assess the condition of `{component_that_failed}`?
**6. Environment (Operating Conditions / Surroundings)**
* Were the environmental conditions (temperature
humidity
cleanliness
vibration from external sources) as described in `{operating_conditions_at_failure_text}` within design limits for `{component_that_failed}`?
* Could any unusual environmental factors (e.g.
sudden impact
flooding
power surge
foreign object ingress) have contributed?
* Was the `{component_that_failed}` properly protected from the operating environment?
* Could long-term environmental exposure (e.g.
corrosion
UV degradation) have weakened `{component_that_failed}`?
**Instructions for User**: Use these questions as starting points to brainstorm specific potential causes under each category for the failure of `{component_that_failed}`. Further drill down with 'Why?' for each identified cause.
- Idéal pour : Aider les ingénieurs en mécanique à enquêter systématiquement sur les défaillances des composants en leur fournissant des questions personnalisées pour chaque catégorie d'un diagramme en arête de poisson (Ishikawa).
- Analyse des causes profondes
- Ingénierie mécanique
Invitation à l'IA à Protocole des 5 raisons pour l'anomalie d'un processus
- 5 pourquoi, Amélioration continue, Analyse des défaillances, Production allégée, Génie mécanique, Techniques de résolution de problèmes, Amélioration des processus, Gestion de la qualité, Analyse des causes profondes
Guide l'utilisateur dans une analyse structurée des 5 causes fondamentales d'une anomalie du processus de fabrication dans le domaine de l'ingénierie mécanique. Cette invite permet d'aller à la source de la cause fondamentale en demandant itérativement pourquoi en fonction du problème initial et du contexte du processus. Le résultat est un parcours de questionnement structuré basé sur du texte.
Sortie :
- Texte
- ne nécessite pas d'Internet en direct
- Champs : {initial_problem_statement_text} {nom_du_processus_et_contexte}
Act as a Quality Engineering Coach facilitating a "5 Whys" Root Cause Analysis.
Your TASK is to guide the user through the 5 Whys methodology to find the potential root cause of the `{initial_problem_statement_text}` within the `{process_name_and_context}`.
You will provide a structured questioning pathway. For each 'Why?'
you will prompt the user for an answer
and then formulate the next 'Why?' based on a hypothetical (but plausible for mechanical engineering) user response. The user would then answer your 'Why?' in a real scenario.
Since this is not interactive
generate a plausible chain of 5 Whys and answers to illustrate the process
and then provide a template for the user to fill.
**ILLUSTRATIVE 5 WHYS EXAMPLE (AI Generated Plausible Chain):**
**Process & Context**: `{process_name_and_context}` (e.g.
CNC Machining of aluminum brackets)
**Initial Problem**: `{initial_problem_statement_text}` (e.g.
High scrap rate (20%) due to oversized holes on Part #XYZ)
**1. Why is there a high scrap rate (20%) due to oversized holes on Part #XYZ?**
* *AI's Assumed Plausible User Answer*: The holes are being drilled too large by the CNC machine.
**2. Why are the holes being drilled too large by the CNC machine?**
* *AI's Assumed Plausible User Answer*: The drill bit is wobbling excessively during operation.
**3. Why is the drill bit wobbling excessively?**
* *AI's Assumed Plausible User Answer*: The drill bit is not securely held in the tool holder collet.
**4. Why is the drill bit not securely held in the tool holder collet?**
* *AI's Assumed Plausible User Answer*: The collet is worn beyond its service limit and has lost its clamping force.
**5. Why is the collet worn beyond its service limit and not replaced?**
* *AI's Assumed Plausible User Answer*: The preventive maintenance schedule for checking/replacing collets based on usage hours was not being consistently followed due to a lack of clear responsibility assigned for this specific task.
* **Potential Root Cause Reached**: Lack of clear responsibility and adherence to preventive maintenance for tooling components.
**TEMPLATE FOR USER TO CONDUCT THEIR OWN 5 WHYS:**
*(Please use this template to conduct your own analysis. For each 'Why?'
record your team's answer
then formulate the next 'Why?' based on that answer.)*
**Process & Context**: `{process_name_and_context}`
**Initial Problem Statement**: `{initial_problem_statement_text}`
**1. Why? (Why is/did [`{initial_problem_statement_text}`] occur?)**
* **Answer 1**: _________________________________________________________
**2. Why? (Why did [Answer 1] occur?)**
* **Answer 2**: _________________________________________________________
**3. Why? (Why did [Answer 2] occur?)**
* **Answer 3**: _________________________________________________________
**4. Why? (Why did [Answer 3] occur?)**
* **Answer 4**: _________________________________________________________
**5. Why? (Why did [Answer 4] occur?)**
* **Answer 5**: _________________________________________________________
* **(Continue if needed - '5' is a guideline
not a strict limit. Stop when you reach an actionable root cause
often related to a process
system
or policy.)**
**Potential Root Cause(s) Identified**: _________________________________________________________
**Recommended Corrective Actions to Address Root Cause(s)**: _________________________________
**IMPORTANT**: The key is to avoid jumping to conclusions and to base each 'Why?' on the factual answer to the previous question. The goal is to find systemic causes
not just to assign blame.
- Idéal pour : Guider les ingénieurs en mécanique à travers la méthodologie des 5 raisons pour découvrir systématiquement les raisons fondamentales des anomalies ou des défaillances du processus de fabrication.
- Analyse des causes profondes
- Ingénierie mécanique
Invitation à l'IA à Analyse de l'arbre des défaillances Configuration de l'événement supérieur
- Conception pour la fabrication (DfM), Validation de la conception, Analyse des défaillances, Analyse de l'arbre des défaillances (ADF), Génie mécanique, Gestion de la qualité, Analyse des risques, Gestion des risques, Safety
Aide à lancer une analyse de l'arbre des défaillances (FTA) en définissant l'événement indésirable le plus important et en suggérant des défaillances de sous-systèmes contribuant immédiatement ou des événements de base pour un système mécanique décrit. Cette invite constitue un point de départ pour une évaluation quantitative ou qualitative détaillée des risques. Le résultat est un schéma d'arborescence au format markdown.
Sortie :
- Markdown
- ne nécessite pas d'Internet en direct
- Champs : {system_description_text} {undesired_top_event_failure_description} {key_subsystems_or_components_list_csv}
Act as a System Safety Engineer specializing in Fault Tree Analysis (FTA).
Your TASK is to help set up the initial levels of a Fault Tree for the `{system_description_text}`.
The TOP EVENT (the main undesired failure) is: `{undesired_top_event_failure_description}`.
Consider the `{key_subsystems_or_components_list_csv}` (CSV: 'Subsystem_Or_Component_Name
Brief_Function') as potential contributors.
You should propose immediate contributing events (intermediate events or basic events) and the logical gates (AND
OR) that connect them to the Top Event or to each other at the first couple of levels.
**FAULT TREE ANALYSIS - INITIAL STRUCTURE (MUST be Markdown format):**
**System Under Analysis**: `{system_description_text}`
**Top Undesired Event**: `{undesired_top_event_failure_description}`
**Level 0: Top Event**
```mermaid
graph TD
TE(`{undesired_top_event_failure_description}`")
```
**Level 1: Immediate Contributing Events / Sub-System Failures**
* **Guidance**: Think about the major ways the Top Event could occur. These could be failures of major subsystems listed in `{key_subsystems_or_components_list_csv}` or general failure categories. Determine if these immediate causes need to ALL occur (AND gate) or if ANY ONE of them occurring is sufficient (OR gate) to cause the Top Event.
**Proposed Structure (Example - AI to generate based on inputs):**
* *If the Top Event can be caused by failure of Subsystem A OR Subsystem B OR an External Event:*
```mermaid
graph TD
TE("`{undesired_top_event_failure_description}`") -->|OR Gate G1| IE1("Failure of [Subsystem A Name from CSV]")
TE -->|OR Gate G1| IE2("Failure of [Subsystem B Name from CSV]")
TE -->|OR Gate G1| IE3("Relevant External Event Causing Failure
e.g.
Power Loss")
```
* *If the Top Event occurs only if Component X AND Component Y fail simultaneously:*
```mermaid
graph TD
TE("`{undesired_top_event_failure_description}`") -->|AND Gate G2| BE1("Failure of [Component X Name from CSV]")
TE -->|AND Gate G2| BE2("Failure of [Component Y Name from CSV]")
```
**Level 2: Further Breakdown of Level 1 Intermediate Events (Illustrative for one branch)**
* **Guidance**: Take ONE of the Intermediate Events (IE) from Level 1 and break it down further. Identify how that specific subsystem or intermediate event could fail.
* **Example (Continuing from OR Gate G1
focusing on IE1 'Failure of Subsystem A'):**
* *If 'Failure of Subsystem A' can be caused by 'Component A1 Failure' OR 'Component A2 Failure':*
```mermaid
graph TD
TE("`{undesired_top_event_failure_description}`") -->|OR Gate G1| IE1("Failure of [Subsystem A Name]")
TE -->|OR Gate G1| IE2("Failure of [Subsystem B Name]")
TE -->|OR Gate G1| IE3("External Event")
IE1 -->|OR Gate G1A| BE_A1("Failure of [Component A1 of Subsystem A]")
IE1 -->|OR Gate G1A| BE_A2("Failure of [Component A2 of Subsystem A]")
```
* The events BE_A1
BE_A2 would be ""Basic Events"" if they represent the limit of resolution (e.g.
a specific part failing
human error
software glitch) for this initial setup
or they could be further developed Intermediate Events.
**Key Considerations for Further Development by User:**
* **Basic Events**: These are typically failures of individual components
human errors
or external events that require no further decomposition. Their probabilities of occurrence are often estimated from historical data
handbook data
or expert judgment.
* **Gate Logic**: Carefully determine if contributing events need an AND gate (all must occur) or an OR gate (any one can cause the higher-level event).
* **Mutual Exclusivity**: Assume events are independent unless otherwise specified.
* **Data Requirements**: For a quantitative FTA
failure probabilities for all basic events are needed.
* **Common Cause Failures**: Consider if a single event could cause multiple basic events to fail simultaneously (this adds complexity beyond this initial setup but is important for full FTA).
**AI's Proposed Initial Breakdown (specific to your inputs):**
*(The AI should now provide a concrete proposed Mermaid diagram snippet for Level 0 and Level 1
and one branch of Level 2
based on the user's specific `{system_description_text}`
`{undesired_top_event_failure_description}`
and `{key_subsystems_or_components_list_csv}`. It should make reasonable assumptions about how these subsystems might contribute to the top event
stating the gate logic clearly.)*
```mermaid
graph TD
TE("`{undesired_top_event_failure_description}`")
// AI will populate the connections and Level 1 / Level 2 events here
// Example: If key_subsystems_or_components_list_csv includes 'Hydraulic_Pump
Provides_Pressure' and 'Control_Valve
Directs_Flow'
// and top event is 'System_Fails_to_Actuate'
// TE -->|OR Gate G_Main| Pump_Failure("Hydraulic Pump Fails")
// TE -->|OR Gate G_Main| Valve_Failure("Control Valve Fails")
// TE -->|OR Gate G_Main| Electrical_Failure("Control System Electrical Failure")
// Pump_Failure -->|OR Gate G_Pump| Motor_Fails("Pump Motor Fails (Basic Event)")
// Pump_Failure -->|OR Gate G_Pump| Pump_Internal_Leak("Pump Internal Leakage (Basic Event)")
```
**IMPORTANT**: This prompt generates a STARTING POINT for an FTA. A complete FTA is a detailed and iterative process. The Mermaid syntax is provided to suggest a visual structure; the user would use FTA software or draw this out. The AI's main role here is to structure the initial decomposition logically."
- Idéal pour : Aider les ingénieurs en mécanique à commencer une analyse de l'arbre des défaillances (FTA) en définissant l'événement principal et en suggérant les branches initiales et les portes logiques pour les défaillances contribuant au sous-système.
- Analyse des causes profondes
- Ingénierie mécanique
Invitation à l'IA à Comparative RCA for Repetitive Failures
- Amélioration continue, Action corrective, Conception pour la fabrication (DfM), Analyse des défaillances, Production allégée, Génie mécanique, Amélioration des processus, Gestion de la qualité, Analyse des causes profondes
Analyzes textual descriptions from multiple incident reports of a repetitive failure in a mechanical system. This prompt aims to identify common patterns potential shared root causes and any differentiating factors across incidents helping to solve persistent issues. The output is a markdown formatted comparative analysis.
Sortie :
- Markdown
- ne nécessite pas d'Internet en direct
- Fields: {system_or_component_name} {common_failure_description} {multiple_failure_incident_reports_text}
Act as a Senior Reliability Engineer conducting a Root Cause Analysis (RCA) on REPETITIVE failures.
Your TASK is to analyze the information provided in `{multiple_failure_incident_reports_text}` concerning recurring instances of '`{common_failure_description}`' affecting the `{system_or_component_name}`.
The goal is to identify common patterns
potential shared root causes
and any significant differentiating factors or unique conditions across the incidents.
The `{multiple_failure_incident_reports_text}` is a single block of text where each incident report is clearly demarcated (e.g.
by '---INCIDENT REPORT X START---' and '---INCIDENT REPORT X END---'
or user ensures separation). Each report may contain details like date
operator
specific symptoms
environmental conditions
immediate actions taken
and initial findings.
**COMPARATIVE ROOT CAUSE ANALYSIS REPORT (MUST be Markdown format):**
**1. Overview of Repetitive Failure:**
* **System/Component**: `{system_or_component_name}`
* **Common Failure Mode**: `{common_failure_description}`
* **Number of Incident Reports Analyzed**: [AI to count based on demarcations in `{multiple_failure_incident_reports_text}`]
**2. Data Extraction and Tabulation (Conceptual - AI to perform this internally):**
* For each incident report
extract key information such as:
* Incident ID/Date
* Specific symptoms observed (beyond the `{common_failure_description}`)
* Operating conditions at time of failure (load
speed
temperature
etc.)
* Environmental conditions
* Maintenance history just prior
* Operator actions or comments
* Any parts replaced or immediate fixes tried.
* *(AI should internally process this information to find patterns. A table won't be in the final output unless it's a summary table
but the AI's logic should be based on this kind of structured comparison.)*
**3. Identification of Common Patterns and Themes Across Incidents:**
* **Symptomology**: Are there consistent preceding symptoms or secondary effects noted across multiple reports before or during the `{common_failure_description}`?
* **Operating Conditions**: Do failures tend to occur under specific loads
speeds
temperatures
or during particular phases of operation (startup
shutdown
steady-state)?
* **Environmental Factors**: Is there a correlation with specific environmental conditions (e.g.
high humidity
dusty environment
specific time of day/year)?
* **Maintenance Activities**: Do failures cluster after certain maintenance activities
or if maintenance is overdue?
* **Component Batch/Supplier (if mentioned in reports)**: Is there any indication of issues related to specific batches or suppliers of the `{system_or_component_name}` or its sub-parts?
* **Human Factors**: Any patterns related to operator experience
shift changes
or specific procedures being followed/not followed?
**4. Identification of Differentiating Factors and Unique Conditions:**
* Are there any incidents that stand out as different in terms
of symptoms
conditions
or severity?
* What unique factors were present in these outlier incidents?
* Could these differences point to multiple root causes or aggravating factors for the `{common_failure_description}`?
**5. Hypothesis Generation for Potential Shared Root Cause(s):**
Based on the common patterns
propose 2-3 primary hypotheses for the underlying root cause(s) of the repetitive '`{common_failure_description}`'. For each hypothesis:
* **Hypothesis Statement**: (e.g.
'Material fatigue due to cyclic loading under X condition'
'Inadequate lubrication leading to premature wear'
'Sensor malfunction providing incorrect feedback to control system').
* **Supporting Evidence from Reports**: Briefly list the common patterns from section 3 that support this hypothesis.
**6. Recommended Next Steps for Investigation / Verification:**
* What specific data collection
tests
or analyses should be performed to confirm or refute the proposed hypotheses? Examples:
* `Detailed metallurgical analysis of failed components from multiple incidents.`
* `Targeted inspection of [specific sub-component] across all similar units.`
* `Review of design specifications vs. actual operating conditions.`
* `Interviews with operators and maintenance staff involved in the incidents.`
* `Monitoring specific parameters (e.g.
vibration
temperature) that might be precursors.`
**7. Interim Containment or Mitigation Actions (if obvious from analysis):**
* Are there any immediate actions that could be taken to potentially reduce the frequency or severity of the failures while the full RCA is ongoing
based on the patterns identified?
**IMPORTANT**: The analysis should focus on synthesizing information from MULTIPLE reports to find trends that might not be obvious from a single incident. The AI should clearly articulate the logic connecting observed patterns to potential root causes.
- Best for: Assisting mechanical engineers in diagnosing recurrent system failures by comparatively analyzing multiple incident reports to identify common patterns and potential shared root causes.
- Optimisation de la conception expérimentale
- Ingénierie mécanique
Invitation à l'IA à Critique du plan DOE pour une expérience factorielle
- Génie mécanique, Optimisation des processus, Contrôle de qualité, Gestion de la qualité, Statistical Analysis, Tests statistiques
Critique une proposition de plan d'expériences (DOE) pour une expérience factorielle en suggérant des améliorations pour la sélection des facteurs, l'adéquation du niveau de confusion et la puissance statistique. Cette invite aide les ingénieurs en mécanique à optimiser leurs plans d'expériences pour plus de robustesse et d'efficacité. Le résultat est une critique au format markdown.
Sortie :
- Markdown
- ne nécessite pas d'Internet en direct
- Champs : {experimental_objective_text} {facteurs_et_niveaux_json} {proposed_experimental_runs_table_csv} {response_variables_list_csv}
Act as a Statistical Consultant specializing in Design of Experiments (DOE) for engineering applications.
Your TASK is to critique the proposed DOE plan for a factorial experiment
based on the following inputs:
* `{experimental_objective_text}`: Clear statement of what the experiment aims to achieve (e.g.
'To determine the main effects and two-factor interactions of cutting speed
feed rate
and depth of cut on surface roughness and tool wear in milling 6061 Aluminum.').
* `{factors_and_levels_json}`: A JSON string defining factors and their levels (e.g.
`{"CuttingSpeed_m_min": [100
150
200]
"FeedRate_mm_rev": [0.1
0.2]
"DepthOfCut_mm": [0.5
1.0]}`). The actual JSON will be standard.
* `{proposed_experimental_runs_table_csv}`: A CSV string of the proposed experimental runs
showing combinations of factor levels (e.g.
'Run
CuttingSpeed
FeedRate
DepthOfCut
...'). If it's a standard design (e.g.
full factorial
fractional factorial)
this might be implied or the user might just state the design type.
* `{response_variables_list_csv}`: CSV string listing the output variables to be measured (e.g.
'SurfaceRoughness_Ra_microns
ToolWear_VB_mm').
**CRITIQUE OF DOE PLAN (MUST be Markdown format):**
**1. Alignment with Objective:**
* **Assessment**: Does the selection of factors
levels
and responses in `{factors_and_levels_json}` and `{response_variables_list_csv}` directly support achieving the `{experimental_objective_text}`?
* **Recommendations**: [e.g.
'The objective mentions interactions
ensure the design specified in `{proposed_experimental_runs_table_csv}` allows estimation of these (e.g.
full factorial or appropriate fractional factorial).' or 'Consider if [Additional Factor] might be relevant to the objective.']
**2. Factor Selection and Levels:**
* **Assessment**: Are the factors in `{factors_and_levels_json}` truly independent and controllable? Are the chosen levels appropriate (e.g.
spanning a reasonable range
not too close
not too far apart to cause process instability)? Are there enough levels to detect non-linearity if expected (more than 2 for a factor)?
* **Recommendations**: [e.g.
'For Factor X
the levels [L1
L2] are very close; consider widening the range if feasible to better observe its effect.' or 'If quadratic effects are suspected for Factor Y
three levels would be necessary.']
**3. Experimental Design Choice (based on `{proposed_experimental_runs_table_csv}` or implied design):**
* **Assessment**:
* **Type of Design**: (e.g.
Full factorial
Fractional factorial
other). Is it clearly stated or inferable?
* **Resolution (for fractional factorials)**: If fractional
what is its resolution and what interactions are confounded? Is this acceptable given the `{experimental_objective_text}`?
* **Number of Runs**: Is it practical? Is it sufficient for the effects being estimated?
* **Randomization**: Is randomization of run order planned? (CRITICAL - should be mentioned as essential).
* **Replication**: Are replications planned
especially at center points (if any) or for key runs
to estimate pure error?
* **Recommendations**: [e.g.
'The proposed fractional factorial design (if identified) confounds main effects with two-factor interactions involving Factor Z. If Factor Z interactions are critical
a higher resolution design or full factorial is needed.' or 'Strongly recommend randomizing the run order to mitigate effects of lurking variables.' or 'Include 3-5 replications at the center point (if applicable) to check for curvature and get a robust estimate of error.']
**4. Response Variables:**
* **Assessment**: Are the `{response_variables_list_csv}` well-defined
measurable
and relevant to the objective? Is the measurement system capable (repeatable and reproducible)? (Latter is an assumption
but good to mention).
* **Recommendations**: [e.g.
'Ensure a consistent measurement protocol for [Response Y].']
**5. Statistical Power and Analysis:**
* **Assessment**: While a full power analysis is complex
comment qualitatively if the design seems underpowered for detecting effects of practical importance
especially if there are few runs or high expected variability.
* **Recommendations**: [e.g.
'With only N runs
detecting small but significant interactions might be challenging. Consider if effect sizes are expected to be large.' or 'Plan for ANOVA and regression analysis. Check model assumptions (normality
constant variance of residuals) post-experiment.']
**Summary of Key Recommendations:**
* [List the top 3-4 most critical suggestions for improving the DOE plan.]
**IMPORTANT**: The critique should be constructive and provide actionable advice. If the `{proposed_experimental_runs_table_csv}` is not detailed
critique based on the likely design implied by factors/levels and objective.
- Meilleur pour : Critiquer les plans d'expériences (DOE) pour les expériences factorielles en ingénierie mécanique afin d'améliorer la pertinence du niveau de sélection des facteurs et la validité statistique.
- Optimisation de la conception expérimentale
- Ingénierie mécanique
Invitation à l'IA à Suggestion du groupe de contrôle pour l'essai des matériaux
- Matériels, Génie mécanique, Propriétés mécaniques, Suivi des performances, Assurance qualité, Contrôle de qualité, Statistical Analysis, Traitement de surface, Méthodes d'essai
Suggère des groupes de contrôle appropriés et des mesures de référence pour une étude expérimentale sur un nouveau matériau ou un traitement de surface dans une application mécanique, afin de garantir des comparaisons valables et des conclusions fiables. Cette invite aide les ingénieurs à concevoir des protocoles d'essai des matériaux plus robustes. Le résultat est une recommandation textuelle.
Sortie :
- Texte
- ne nécessite pas d'Internet en direct
- Champs : {test_material_or_treatment_description} {experimental_conditions_text} {performance_metrics_to_be_measured_list_csv}
Act as an Experimental Design Specialist in Materials Science and Engineering.
Your TASK is to recommend appropriate control groups and baseline measurements for an experimental study involving `{test_material_or_treatment_description}` under `{experimental_conditions_text}`
where `{performance_metrics_to_be_measured_list_csv}` (CSV: 'Metric_Name
Units') are the key outputs.
The goal is to ensure that any observed changes in performance can be confidently attributed to the `{test_material_or_treatment_description}`.
**RECOMMENDATIONS FOR CONTROL GROUPS AND BASELINE MEASUREMENTS:**
**1. Understanding the Core Investigation:**
* The primary goal is to evaluate the effect of `{test_material_or_treatment_description}`.
* The `{experimental_conditions_text}` (e.g.
'High-temperature tensile testing at 600°C'
'Cyclic fatigue testing under 200 MPa load for 10^6 cycles'
'Wear testing against a steel counterface with 10N load for 5 hours') define the environment.
* The `{performance_metrics_to_be_measured_list_csv}` (e.g.
'Ultimate_Tensile_Strength_MPa
Elongation_Percent'
'Fatigue_Life_Cycles'
'Wear_Rate_mm3_Nm') are the indicators of performance.
**2. Recommended Control Group(s):**
* **A. Untreated/Standard Material Control:**
* **Description**: Samples made from the SAME BASE MATERIAL as the `{test_material_or_treatment_description}` but WITHOUT the specific new material feature or treatment being tested. If the test involves a new alloy
the control might be the conventional alloy it aims to replace or a version of the new alloy without a critical processing step.
* **Justification**: This is the MOST CRITICAL control. It allows for direct comparison to determine if the `{test_material_or_treatment_description}` provides any benefit (or detriment) over the standard or untreated state.
* **Processing**: These control samples should
as much as possible
undergo all other processing steps (e.g.
heat treatments
machining) that the test samples experience
EXCEPT for the specific treatment/feature being evaluated.
* **B. (Optional
if applicable) Benchmark/Reference Material Control:**
* **Description**: Samples made from a well-characterized
industry-standard benchmark material that is commonly used in similar applications or for which extensive performance data exists.
* **Justification**: This allows comparison against a known quantity and can help validate the testing procedure if the benchmark material behaves as expected. It also positions the performance of the `{test_material_or_treatment_description}` within the broader field.
* **C. (Optional
if treatment involves application) Placebo/Sham Treatment Control:**
* **Description**: If the treatment involves a complex application process (e.g.
a coating applied via a specific sequence of steps
some of which might independently affect the material)
a sham control experiences all application steps EXCEPT the active treatment ingredient/process.
* **Justification**: Helps to isolate the effect of the active treatment component from the effects of the application process itself.
**3. Baseline Measurements (Pre-Test Characterization):**
* For ALL samples (both test and control groups)
consider performing and recording the following baseline measurements BEFORE subjecting them to the main `{experimental_conditions_text}`:
* **Initial Microstructure Analysis**: (e.g.
Optical microscopy
SEM) To document the starting state
grain size
presence of defects
or treatment-induced surface changes.
* **Initial Hardness Testing**: A quick way to check for consistency or initial effects of a surface treatment.
* **Precise Dimensional Measurements**: Especially important for wear or deformation studies.
* **Surface Roughness**: If surface properties are critical or affected by the treatment.
* **Compositional Analysis (Spot Checks)**: To verify material or coating composition if it's a key variable.
* **Justification**: Baseline data helps confirm initial sample consistency
can reveal pre-existing flaws
and provides a reference point for assessing changes after testing.
**4. Experimental Considerations:**
* **Sample Size**: Ensure a sufficient number of samples in each group (test and control) for statistical validity.
* **Randomization**: If there are variations in the testing apparatus or over time
randomize the testing order of samples from different groups.
* **Identical Test Conditions**: CRITICAL - All groups (test and control) MUST be subjected to the EXACT SAME `{experimental_conditions_text}` and measurement procedures for the `{performance_metrics_to_be_measured_list_csv}`.
**Summary**: By including these control groups and baseline measurements
the experiment will be better able to isolate the true effect of the `{test_material_or_treatment_description}` and produce more reliable and defensible conclusions.
- Idéal pour : Guider les ingénieurs en mécanique dans la sélection de groupes de contrôle appropriés et de mesures de référence pour les essais de matériaux, afin de garantir la validité expérimentale et l'interprétation fiable des résultats.
- Optimisation de la conception expérimentale
- Ingénierie mécanique
Invitation à l'IA à Placement des capteurs pour les essais de vibration
- Génie mécanique, Assurance qualité, Contrôle de qualité, Sensors, Traitement du signal, Ingénierie des structures, Méthodes d'essai, Analyse des vibrations
Recommande les types de capteurs optimaux et les stratégies de placement pour les essais de vibration sur une structure mécanique afin de capturer les modes pertinents et de garantir la qualité des données sur la base de la description de la structure et des objectifs de l'essai. Cette invite aide à planifier une analyse modale expérimentale efficace ou une surveillance des vibrations. La sortie est une recommandation au format markdown.
Sortie :
- Markdown
- ne nécessite pas d'Internet en direct
- Champs : {structure_description_and_material} {texte_de_l'essai_de_vibration} {frequency_range_of_interest_text} (en anglais) {liste_des_types_de_capteurs_disponibles_csv}
Act as a Vibration Testing and Modal Analysis Expert.
Your TASK is to recommend optimal sensor types and placement strategies for vibration testing on the `{structure_description_and_material}`.
The recommendations should align with the `{vibration_test_objectives_text}`
consider the `{frequency_range_of_interest_text}`
and select from the `{available_sensor_types_list_csv}` (CSV: 'SensorType
KeySpecification_e.g._Sensitivity_FrequencyRange_Weight').
**SENSOR PLACEMENT AND TYPE RECOMMENDATIONS (MUST be Markdown format):**
**1. Analysis of Inputs:**
* **Structure**: `{structure_description_and_material}` (e.g.
'Cantilevered steel beam
1m long
5cm x 1cm cross-section'
'Aluminum plate
50cm x 50cm x 0.5cm
simply supported on four edges'
'Complex welded frame assembly').
* **Objectives**: `{vibration_test_objectives_text}` (e.g.
'Identify first 5 natural frequencies and mode shapes'
'Monitor operational vibration levels at critical bearing locations'
'Assess damping effectiveness of a new treatment').
* **Frequency Range**: `{frequency_range_of_interest_text}` (e.g.
'0-500 Hz'
'Up to 2 kHz').
* **Available Sensors**: `{available_sensor_types_list_csv}` (e.g.
'Accelerometer_A
100mV/g
1-5000Hz
10grams'
'Displacement_Sensor_B
non-contact_eddy
DC-1000Hz
50grams_probe').
**2. Recommended Sensor Type(s):**
* **Selection Rationale**: Based on the `{vibration_test_objectives_text}`
`{frequency_range_of_interest_text}`
and characteristics of the `{structure_description_and_material}` (e.g.
size
stiffness
expected displacement/acceleration levels).
* **Primary Choice(s) from `{available_sensor_types_list_csv}`**:
* [Sensor Type 1]: Justify why it's suitable (e.g.
'Accelerometer_A is suitable due to its wide frequency range covering the interest area
good sensitivity
and relatively low mass which minimizes mass loading on lighter structures.').
* [Sensor Type 2 (if needed or alternative)]: Justify.
* **Considerations**:
* **Mass Loading**: Ensure sensor mass is significantly less than the dynamic mass of the structure at the attachment point (typically <10%).
* **Dynamic Range**: Sensor must handle expected vibration amplitudes without clipping or poor signal-to-noise.
* **Environmental Conditions**: Temperature
humidity (if relevant).
**3. Sensor Placement Strategy:**
* **Goal**: To adequately capture the modes of interest (if modal analysis) or critical operational responses.
* **General Principles**:
* Place sensors where significant motion is expected for the modes of interest.
* Avoid placing sensors at nodal points/lines of key modes if those modes are to be measured.
* Ensure good mechanical coupling between sensor and structure (e.g.
stud mount
rigid adhesive
magnetic base on suitable surface).
* Consider sensor orientation to capture motion in relevant directions (uniaxial
triaxial sensors).
* **Specific Recommendations for `{structure_description_and_material}`**:
* **Driving Point (for modal testing with shaker)**: Place a sensor near the excitation point to measure input (if not using an impedance head).
* **Response Points (General Grid/Targeted)**:
* If identifying mode shapes: Distribute sensors across the structure to provide sufficient spatial resolution. A preliminary Finite Element Analysis (FEA) model
if available
can guide optimal placement by showing high displacement areas for different modes.
* If simple structure (e.g.
beam
plate): Suggest a grid or specific points (e.g.
'For the cantilever beam
place sensors at L/4
L/2
3L/4
and L from the fixed end to capture bending modes.').
* If monitoring specific locations (from `{vibration_test_objectives_text}`
e.g.
'bearing housings'): Prioritize these locations.
* **Number of Sensors**: Based on objectives
complexity of modes
and available channels. Suggest a minimum or typical number.
* **Pre-Test Checks**:
* Perform a "tap test" or preliminary sweep to ensure sensors are working and capturing signals as expected.
**4. Data Acquisition Considerations (Briefly):**
* Ensure sampling rate is adequate (at least 2.56 times the max `{frequency_range_of_interest_text}`
ideally 5-10 times for better time domain representation).
* Anti-aliasing filters are essential.
* Cable routing to minimize noise.
**IMPORTANT**: These are general guidelines. The optimal setup can depend on subtle details of the structure and test. If a pre-test FEA modal analysis is feasible for the user
it's highly recommended for guiding sensor placement for complex mode shapes.
- Idéal pour : Guider les ingénieurs en mécanique dans le choix et le placement des capteurs pour les essais de vibration, afin de garantir une acquisition optimale des données pour l'analyse modale ou la surveillance de l'état de santé.
- Optimisation de la conception expérimentale
- Ingénierie mécanique
Invitation à l'IA à Optimisation des variables du protocole de test d'usure
- Science des matériaux, Génie mécanique, Optimisation des processus, Contrôle de qualité, Statistical Analysis, Méthodes d'essai, Tribologie, Technologie portable
Analyse un protocole d'essai d'usure pour un composant mécanique en suggérant des moyens de réduire le nombre de variables ou d'améliorer le contrôle des paramètres afin d'isoler des effets spécifiques et d'améliorer la répétabilité et la fiabilité des essais. Cette invite permet d'affiner les montages expérimentaux pour les études tribologiques. Le résultat est une liste de recommandations au format markdown.
Sortie :
- Markdown
- ne nécessite pas d'Internet en direct
- Champs : {wear_test_protocol_description_text} {component_material_and_counterface_material} (matériau du composant et de la contreface) {key_variables_being_investigated_list_csv}
Act as a Tribology Specialist with expertise in wear testing methodologies.
Your TASK is to analyze the provided `{wear_test_protocol_description_text}` for testing a component made of `{component_material_and_counterface_material}` (specify both
e.g.
'Component: Bearing Steel
Counterface: Stainless Steel 304'). The aim is to suggest improvements for reducing uncontrolled variables
isolating the effects of `{key_variables_being_investigated_list_csv}` (CSV: 'Variable_Name
Range_or_Levels')
and enhancing overall test repeatability and reliability.
**RECOMMENDATIONS FOR WEAR TESTING PROTOCOL OPTIMIZATION (MUST be Markdown format):**
**1. Review of Current Protocol and Objectives:**
* **Understanding the Protocol**: Briefly summarize the core elements of the `{wear_test_protocol_description_text}` (e.g.
'Pin-on-disk test
10N load
0.5 m/s sliding speed
1000m distance
ambient temperature
dry contact').
* **Investigated Variables**: Clarify the specific variables from `{key_variables_being_investigated_list_csv}` that the protocol aims to study (e.g.
'Effect of Load (5N
10N
15N)'
'Effect of Lubricant Type (Oil A
Oil B
Dry)').
* **Materials**: `{component_material_and_counterface_material}`.
**2. Identification of Potential Uncontrolled or Confounding Variables:**
Based on the protocol description
identify factors that might not be explicitly controlled or could interfere with isolating the effects of the `{key_variables_being_investigated_list_csv}`. Examples:
* **Environmental Factors**:
* Temperature fluctuations (ambient vs. localized heating due to friction).
* Humidity variations.
* Contamination (dust
debris from previous tests).
* **Specimen Preparation Inconsistencies**:
* Surface finish variations (initial roughness of component and counterface).
* Cleaning procedures before test.
* Specimen alignment and clamping.
* **Test Rig / Operational Factors**:
* Actual load application (static vs. dynamic components
precise load control).
* Speed fluctuations.
* Vibration from the test rig or surroundings.
* Wear debris accumulation or removal during the test.
* **Measurement Inconsistencies (for wear quantification)**:
* Method of wear measurement (mass loss
profilometry
wear scar dimensions) and its precision/repeatability.
* Timing of measurements.
**3. Recommendations for Improving Control and Isolation of Variables:**
* **For each identified potential issue
suggest specific improvements:**
* **Environmental Control**:
* `e.g.
Consider conducting tests in a temperature and humidity controlled chamber if feasible
or at least monitor and record ambient conditions.`
* `e.g.
Implement strict cleaning protocols for the test chamber and specimens between runs.`
* **Specimen Preparation Standardization**:
* `e.g.
Define and adhere to a specific surface preparation procedure (e.g.
grinding
polishing to a consistent Ra value). Verify roughness before each test.`
* `e.g.
Use standardized cleaning solvents and drying methods.`
* `e.g.
Develop a fixture or procedure for consistent alignment.`
* **Test Rig Calibration and Monitoring**:
* `e.g.
Regularly calibrate load cells
speed sensors
and environmental sensors.`
* `e.g.
Monitor key parameters like load
speed
and friction coefficient IN-SITU if possible.`
* **Wear Debris Management**:
* `e.g.
Decide on a strategy: either allow debris to accumulate naturally (if studying three-body wear is intended) or implement a method for controlled removal (e.g.
periodic cleaning
inert gas flow) if two-body abrasion is the focus. Document the choice.`
* **Standardized Wear Measurement**:
* `e.g.
Clearly define the wear measurement technique
including specific locations for profilometry scans or number of mass measurements. Calibrate measurement instruments.`
* **Isolating `{key_variables_being_investigated_list_csv}`**:
* `e.g.
When investigating 'Load'
ensure ALL other parameters (speed
environment
lubricant if any
material batch
surface prep) are kept as constant as possible across different load levels.`
* `e.g.
Use a full factorial or well-designed fractional factorial approach if multiple variables from the list are changed simultaneously to understand interactions.`
**4. Enhancing Repeatability and Reliability:**
* **Replicates**: `Perform multiple test runs (e.g.
3-5 replicates) for each unique test condition to assess variability and calculate confidence intervals.`
* **Randomization**: `Randomize the order of test runs for different conditions to minimize systematic errors related to time or drift.`
* **Reference Runs**: `Periodically run a test with a standard reference material pair under fixed conditions to check for drift in the test rig performance.`
**IMPORTANT**: The goal is to make the experimental results more attributable to the `{key_variables_being_investigated_list_csv}` by minimizing other sources of variation.
- Idéal pour : Améliorer la fiabilité et la précision des protocoles de test d'usure en identifiant et en contrôlant les variables, ce qui permet de mieux isoler les effets étudiés dans les études tribologiques.
- Modélisation prédictive
- Ingénierie mécanique
Invitation à l'IA à Stratégie du modèle de prédiction des propriétés des matériaux
- Machine Learning, Matériels, Génie mécanique, Propriétés mécaniques, Algorithmes de maintenance prédictive, Optimisation des processus, Gestion de la qualité, Statistical Analysis
Décrit une stratégie de développement d'un modèle prédictif pour une propriété de matériau spécifique basée sur des paramètres de composition et de traitement à l'aide d'un ensemble de données décrit. Cette invite aide les ingénieurs en mécanique à initier une conception ou une sélection de matériaux basée sur des données. Le résultat est un document de stratégie formaté en markdown.
Sortie :
- Markdown
- ne nécessite pas d'Internet en direct
- Champs : {target_material_property_to_predict} {discription_des_caractéristiques_des_entrées_disponibles} {texte_de_la_taille_et_des_caractéristiques du jeu de données}
Act as a Materials Informatics Specialist.
Your TASK is to outline a strategy for developing a predictive model for `{target_material_property_to_predict}` (e.g.
'Tensile Strength in MPa'
'Hardness in HRC'
'Fatigue Life in cycles').
The model will be based on input features described in `{available_input_features_csv_description}` (CSV string detailing feature names
types (numeric/categorical)
and example ranges
e.g.
'FeatureName
DataType
ExampleRange
CarbonContent_Numeric_0.1-1.0%
HeatTreatmentTemp_Numeric_800-1200C
AlloyingElementX_Categorical_Present/Absent').
Consider the `{dataset_size_and_characteristics_text}` (e.g.
'Approximately 500 data points
some missing values
potential outliers noted in preliminary analysis
data from various literature sources').
**PREDICTIVE MODEL DEVELOPMENT STRATEGY (MUST be Markdown format):**
**1. Project Goal:**
* To develop a predictive model for `{target_material_property_to_predict}` using the features outlined in `{available_input_features_csv_description}`.
* Potential use cases: Material screening
alloy design optimization
property estimation where experimental data is scarce.
**2. Data Preprocessing and Exploration (Key Steps):**
* **2.1. Data Loading and Initial Inspection**:
* Load the dataset.
* Verify feature names
data types against `{available_input_features_csv_description}`.
* Initial check for gross errors or inconsistencies.
* **2.2. Handling Missing Values**: Based on `{dataset_size_and_characteristics_text}` if it mentions missing data.
* Strategy: (e.g.
Imputation using mean/median/mode
K-Nearest Neighbors imputation
or model-based imputation if complex. Justify choice).
* Consider adding an indicator column for imputed values.
* **2.3. Outlier Detection and Treatment**: Based on `{dataset_size_and_characteristics_text}` if it mentions outliers.
* Strategy: (e.g.
Z-score
IQR method
Isolation Forest). Decide whether to cap
transform
or remove outliers
and justify.
* **2.4. Feature Engineering (if applicable)**:
* Transformations: (e.g.
Logarithmic
square root transformations for skewed data; Polynomial features if non-linear relationships are suspected).
* Categorical Encoding: For features identified as categorical in `{available_input_features_csv_description}` (e.g.
One-Hot Encoding
Label Encoding).
* Interaction Terms: Consider creating interaction terms between key features if domain knowledge suggests their importance.
* **2.5. Feature Scaling/Normalization**:
* Strategy: (e.g.
StandardScaler
MinMaxScaler) especially important for distance-based algorithms or neural networks.
* **2.6. Exploratory Data Analysis (EDA)**:
* Visualize distributions of individual features and the `{target_material_property_to_predict}`.
* Plot relationships between input features and the target property (scatter plots
correlation matrix).
* Identify potential correlations or patterns.
**3. Model Selection and Training:**
* **3.1. Splitting the Dataset**:
* Training set (e.g.
70-80%)
Validation set (e.g.
10-15%)
Test set (e.g.
10-15%). Stratified splitting if the target variable is highly imbalanced or for classification tasks (though this is regression).
* **3.2. Candidate Model Algorithms (Suggest 2-3 to explore)**:
* **Baseline Model**: Simple Linear Regression or a decision tree regressor.
* **More Complex Models**:
* Ensemble methods (Random Forest Regressor
Gradient Boosting Regressor like XGBoost
LightGBM) - often perform well on tabular data.
* Support Vector Regression (SVR).
* Neural Networks (Multilayer Perceptron - MLP) - consider if dataset is large enough and complex non-linearities are expected.
* Justify choices based on `{dataset_size_and_characteristics_text}` and nature of features.
* **3.3. Hyperparameter Tuning**:
* Strategy: (e.g.
GridSearchCV
RandomizedSearchCV
Bayesian Optimization) using the validation set.
* **3.4. Training**: Train candidate models on the training set using optimized hyperparameters.
**4. Model Evaluation:**
* **4.1. Performance Metrics (for regression)**:
* Primary metrics: Root Mean Squared Error (RMSE)
Mean Absolute Error (MAE).
* Secondary metrics: R-squared (Coefficient of Determination)
Mean Absolute Percentage Error (MAPE).
* **4.2. Evaluation on Test Set**: Assess final model performance on the unseen test set to estimate generalization ability.
* **4.3. Residual Analysis**: Plot residuals to check for patterns
homoscedasticity
and normality.
* **4.4. Feature Importance Analysis**: (e.g.
from tree-based models) to understand which features are most influential.
**5. Iteration and Refinement:**
* Based on evaluation
iterate on feature engineering
model selection
or hyperparameter tuning.
* Consider ensemble stacking if individual models show complementary strengths.
**6. Deployment Considerations (Briefly
if applicable):**
* How will the model be used? API
embedded system
standalone tool?
* Monitoring model performance over time if new data comes in.
**IMPORTANT**: This strategy should be comprehensive yet adaptable. The specific choices for methods will depend on the detailed exploration of the data. Emphasize the iterative nature of model development.
- Idéal pour : Guider les ingénieurs en mécanique tout au long du processus de développement d'un modèle prédictif basé sur des données pour les propriétés des matériaux, du traitement initial des données à l'évaluation du modèle.
- Modélisation prédictive
- Ingénierie mécanique
Invitation à l'IA à Identification des principaux paramètres d'entrée du modèle RUL
- Corrosion, Conception pour la fabrication (DfM), Analyse des modes de défaillance et de leurs effets (AMDEC), Génie mécanique, Algorithmes de maintenance prédictive, Optimisation des processus, Sensors, Analyse des vibrations
Identifie et répertorie les paramètres d'entrée clés et les types de données de capteur les plus pertinents pour développer un modèle prédictif de durée de vie utile restante (RUL) pour un type spécifique d'équipement mécanique rotatif. Cette invite aide à sélectionner les données appropriées pour le pronostic. La sortie est une liste au format CSV.
Sortie :
- CSV
- ne nécessite pas d'Internet en direct
- Champs : {equipment_type_and_function} {known_failure_modes_list_csv} {available_sensor_data_streams_description_text}
Act as a Prognostics and Health Management (PHM) Specialist.
Your TASK is to identify key input parameters and sensor data types that would be MOST RELEVANT for developing a Remaining Useful Life (RUL) predictive model for `{equipment_type_and_function}`.
Consider the `{known_failure_modes_list_csv}` (CSV: 'FailureMode_ID
Description
AffectedComponents') and the `{available_sensor_data_streams_description_text}` (e.g.
'Vibration (accelerometers on bearings
10kHz sampling)
Temperature (thermocouples on casing
motor winding
1Hz)
Oil pressure (transducer
10Hz)
Rotational speed (encoder
1kHz)
Acoustic emission (sensor on housing
1MHz range)
Load current (motor control unit
50Hz)').
**KEY INPUT PARAMETERS FOR RUL MODEL (MUST be CSV format):**
**CSV Header**: `Parameter_Rank
Parameter_Name_Or_Feature
Sensor_Or_Data_Source
Relevance_To_Failure_Modes
Potential_Feature_Engineering_Notes`
**Analysis Logic to Generate Rows:**
1. **Understand Equipment and Failure Modes**:
* Analyze `{equipment_type_and_function}` (e.g.
'Centrifugal Pump for corrosive fluids'
'High-speed gearbox for wind turbine'
'Aircraft engine bearing assembly').
* Review each failure mode in `{known_failure_modes_list_csv}`. For each mode
think about what physical parameters would change as degradation progresses towards that failure.
2. **Map Sensor Data to Degradation Indicators**:
* For each sensor stream in `{available_sensor_data_streams_description_text}`
assess its potential to capture indicators of the identified failure modes.
3. **Prioritize Parameters**: Rank parameters based on their likely sensitivity to degradation and relevance to the most critical or common failure modes.
4. **Suggest Feature Engineering**: For raw sensor data
suggest derived features that are often more informative for RUL models.
**Example Parameters to Consider (AI to generate specific to inputs):**
* **Vibration-based features**:
* RMS
Kurtosis
Crest Factor
Skewness (overall or in specific frequency bands).
* Spectral power in bands around bearing defect frequencies
gear mesh frequencies
unbalance frequencies.
* Cepstral analysis features for identifying harmonics/sidebands.
* **Temperature-based features**:
* Absolute temperature levels.
* Temperature trends/rate of change.
* Temperature differentials between components.
* **Process/Operational Parameters**:
* Load levels
torque
current drawn (can indicate increased friction or effort).
* Pressure drops
flow rates (for pumps
hydraulic systems).
* Speed variations.
* **Oil/Lubricant Related (if applicable & sensors exist)**:
* Particle count
viscosity
water content (if online sensors
otherwise lab data).
* **Acoustic Emission features**:
* Hit counts
energy levels
peak amplitudes (good for crack detection
early wear).
* **Usage/Cycle Counters**:
* Number of starts/stops
operating hours
load cycles.
**Output CSV Content Example (Conceptual - AI generates actuals):**
`1
Vibration RMS (Bearing Housing Z-axis)
Accelerometer
Bearing wear
Spalling
Calculate rolling window RMS
Trend analysis`
`2
Oil Temperature Trend
Thermocouple (Oil Sump)
Lubricant degradation
Overheating
Slope of temperature over time
Threshold alerts`
`3
Spectral Peak at 2x RPM
Accelerometer (Motor Shaft)
Misalignment
Unbalance
FFT analysis
Monitor amplitude growth`
`4
Motor Current Mean
Motor Control Unit
Increased friction
Winding fault
Filter out transients
Compare to baseline`
`5
Acoustic Emission Hit Rate
AE Sensor
Crack initiation
Severe wear
Denoise signal
Detect sudden increases`
**IMPORTANT**: The list should be prioritized
with the most promising parameters ranked higher. The 'Relevance_To_Failure_Modes' column should specifically link the parameter to failure modes from `{known_failure_modes_list_csv}` if possible. 'Potential_Feature_Engineering_Notes' gives hints for data processing.
- Idéal pour : Aider les ingénieurs mécaniciens à sélectionner les données de capteurs et les paramètres opérationnels les plus pertinents pour développer des modèles prédictifs de durée de vie utile restante (RUL) pour les équipements rotatifs.
Sommes-nous en train de supposer que l'IA peut toujours générer les meilleurs messages en génie mécanique ? Comment sont-elles générées ?
L'IA va-t-elle rendre les ingénieurs humains superflus ?
Articles Similaires
Ordre de modification technique (ECO) : meilleures pratiques pour minimiser les perturbations et les coûts
Du laboratoire au marché : le rôle de la production pilote
Plus de 45 astuces de sciences cognitives pour les jeux et le marketing : psychologiques et d'engagement
Plus de 45 astuces de sciences cognitives pour les jeux et le marketing : psychologiques et d'engagement
Dernières publications et brevets sur les zéolites
Dernières publications et brevets sur les structures organométalliques (MOF)