Industrial Equipment Behavior Modeling for Predictive Maintenance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methodologies for supervising industrial equipment reliability, such as predictive maintenance, physical models, and bench tests, are limited in simulating the impact of manufacturing, maintenance, and usage on equipment behavior, leading to inaccurate predictions and insufficient proactive maintenance.
Innovation Solution
A digital system that generates a specific behavior model for industrial equipment series, correlating manufacturing, maintenance, and usage logs to simulate equipment behavior over its service life, optimizing maintenance decisions and extending service life.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If predictive maintenance, physical models, and bench tests are used to supervise equipment reliability, then maintenance decisions can be made based on observed data and physical principles, but the ability to simulate the impact of manufacturing, maintenance, and usage on equipment behavior is limited, leading to inaccurate predictions
Solution Approach 1:
The patent creates a digital copy of the physical equipment through a behavior model that replicates equipment responses to various inputs. This digital twin approach allows virtual simulation of manufacturing, maintenance, and usage scenarios without affecting the actual equipment, enabling accurate prediction of equipment behavior and reliability under different conditions.
Solution Approach 2:
The system performs preliminary simulations of various manufacturing, maintenance, and usage scenarios before actual equipment operations. By pre-evaluating different decisions and their potential impacts on equipment behavior, the system enables proactive optimization of maintenance schedules and operational parameters to maximize reliability.
2Reliability
If comprehensive manufacturing, maintenance, and usage logs are collected and processed, then equipment behavior can be accurately simulated and predicted, but the system complexity and data processing requirements increase significantly
Solution Approach 1:
The patent extracts only the critical parameters and features from comprehensive manufacturing, maintenance, and usage logs that have the most significant impact on equipment behavior. This selective extraction reduces data dimensionality and processing complexity while maintaining the accuracy needed for reliable predictions, focusing computational resources on the most influential factors.
Solution Approach 2:
The behavior model serves as an intermediary layer between raw comprehensive logs and final reliability predictions. This intermediary processing layer transforms complex multi-source data into meaningful equipment behavior patterns, simplifying the overall system architecture while preserving the ability to handle comprehensive operational data.
3Productivity
If maintenance schedules are optimized based on simulated equipment behavior, then maintenance costs can be reduced and service life extended, but the initial model training and validation require significant time and computational resources
Solution Approach 1:
The patent implements a phased model development approach where an initial simplified behavior model is trained on representative subsets of data to achieve functional capability quickly. This partial implementation allows early optimization of maintenance schedules with acceptable accuracy, while more comprehensive model training continues in parallel to improve precision over time without delaying initial benefits.
Data Source
AI summary
The present invention relates to a system (1) for supervision of the operation and maintenance of an item of equipment (2) in a facility (3), in which the equipment (2) is operated up to maintenance (6) to be performed, with at least one subsequent projected maintenance (7), inducing a manufacturing and maintenance log (9), a usage log (11), a log (120) of statuses (13). A correlation (14) is determined between causes and consequences of aging of the equipment (2), by characterizing tasks (90) and conditions (110) impacting the status (13).For other equipment, data corresponding to said correlation (14) is recovered and extracted, in order to train a virtual model (16).The day before the maintenance (6) to be carried out, based on the logs (9, 11), tasks (90) of the maintenance (6) to be carried out and projected conditions (110) of a scenario (8), said model (16) generates a projected status (130) compared to a minimal operating status (17) for said equipment (2).