Drive System Predictive Maintenance for Part-Level Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques struggle to accurately identify parts in an industrial machine's drive system that are approaching an abnormal state, making it difficult to perform maintenance before a failure occurs.
Innovation Solution
A predictive maintenance device and system that utilizes a machine learning-trained determination model to analyze state data from an industrial machine's drive system, determining the possibility of abnormality for each part and identifying which parts require maintenance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional threshold-based abnormality detection is used, then the detection method is simple, but the accuracy of identifying parts approaching abnormal state is insufficient
Solution Approach 1:
The patent replaces the conventional mechanical threshold-based detection system with a machine learning-based determination model. The processor inputs state data (vibration, temperature, current) into the determination model, which outputs abnormality probabilities for each part. This substitution enables accurate identification of parts approaching abnormal states without requiring complex manual threshold setting for each component.
Solution Approach 2:
The determination model acts as an intermediary between the raw state data from sensors and the abnormality detection result. It processes the multi-parameter state data (vibration acceleration, temperature, current) and transforms it into meaningful abnormality probabilities for each part, bridging the gap between simple sensor readings and accurate diagnostic information.
2Measurement precision
If multiple parts in the drive system are monitored individually, then the identification accuracy improves, but the analysis complexity increases
Solution Approach 1:
The patent merges the monitoring of multiple parts (motor, speed reducer, timing belt) into a single unified determination model. The model simultaneously processes state data from all parts and outputs abnormality probabilities for each component in one operation. This combining approach achieves accurate identification of specific problematic parts while avoiding the complexity of separate analysis systems for each component.
Solution Approach 2:
The determination model is designed with multi-functionality to handle various types of state data (vibration, temperature, current) and evaluate multiple parts simultaneously. This universal model can identify abnormalities in any part of the drive system without requiring separate specialized models, reducing overall system complexity while maintaining high identification accuracy.
Data Source
AI summary
A predictive maintenance device for an industrial machine includes a determination model and a processor. The determination model has been trained by machine learning to output a possibility of an abnormality for each of a plurality of components by inputting state data. The state data indicates a state of a drive system of the industrial machine. The plurality of components are included in the drive system. The plurality of components are connected to each other so as to operate in conjunction with each other. The processor acquires the state data. The processor acquires the possibility of the abnormality in each of the plurality of components from the state data by using the determination model. The processor determines the component to be maintained from the plurality of components based on the possibility of the abnormality.