Equipment State Monitoring via Operation-Pattern-Independent Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional equipment state monitoring methods using shared models struggle to accurately determine equipment states due to differences in operation environments, leading to reduced determination accuracy.
Innovation Solution
An equipment state monitor device that extracts feature value data from operation data, converts it into a feature value space independent of operation patterns, and determines the equipment state based on a comparison with a feature value distribution, eliminating the need for selecting a general-purpose learning model and performing additional learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a general-purpose shared model is selected and additional learning is performed using target device data, then the model can be adapted to the target device, but determination accuracy is reduced due to operation environment differences between training data and target device data
Solution Approach 1:
The patent introduces a feature value space conversion mechanism as an intermediary between the shared model and target device data. This conversion process transforms feature values from different operation environments into a unified feature value space, allowing the model to generalize across environments without requiring additional learning. The conversion acts as a mediator that eliminates the negative impact of operation environment differences while preserving the adaptability benefits of the shared model approach.
Solution Approach 2:
The patent applies parameter changes by converting feature values using different conversion formulas based on the operation pattern. Instead of retraining the model with target device data, the system changes the parameter representation (feature values) through mathematical conversion. This allows the same shared model to accurately determine equipment states across different operation environments by transforming the input parameters into a consistent feature value space.
2Measurement precision
If multiple shared models are prepared for different operation patterns to maintain determination accuracy, then accuracy for each operation pattern is preserved, but device complexity increases
Solution Approach 1:
The patent makes a single shared model universal across multiple operation patterns by introducing a feature value conversion mechanism. Instead of maintaining separate models for different operation patterns, the system uses one shared model that can handle all operation patterns through feature value conversion. This single model performs multiple functions (handling different operation patterns) while maintaining determination accuracy, thereby reducing device complexity.
Solution Approach 2:
The patent uses parameter changes (feature value conversion) to enable a single model to adapt to different operation patterns. By converting feature values based on the detected operation pattern, the system allows one model to function effectively across multiple scenarios without requiring multiple specialized models, thus simplifying the overall system architecture.
3Reliability
If additional learning is performed using target device data, then the model becomes specialized for the target device, but the learning process requires time and computational resources
Solution Approach 1:
The patent performs preliminary action by pre-converting feature values based on operation patterns before they are used for equipment state determination. The conversion formulas are prepared in advance for different operation patterns, allowing the system to quickly transform incoming data without requiring time-consuming additional learning processes. This preliminary preparation of conversion mechanisms eliminates the need for time-intensive model retraining while maintaining high reliability.
Solution Approach 2:
The patent uses copying by creating converted copies of feature values in a unified feature value space rather than modifying the original data or retraining models. Instead of performing additional learning that would take time, the system creates transformed copies of the input data that are compatible with the shared model, enabling rapid deployment without learning delays.
Data Source
AI summary
An equipment state monitor device includes: a feature value extraction unit that extracts feature value data regarding a state of equipment that is an object to be monitored, from operation data indicating the state of the equipment; a data conversion unit that converts the feature value data extracted from the operation data into determination data in a feature value space that does not depend on an operation pattern of the equipment in an operation environment in which the state of the equipment is monitored; a determination unit that determines the state of the equipment on the basis of a result of comparison between the determination data and the feature value distribution indicating a determination range in the feature value space; and an output unit that outputs a result of determining the state of the equipment.