Equipment Anomaly Detection Using Normal-State AI Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in equipment anomaly detection lies in the difficulty of collecting sufficient anomaly data, leading to unbalanced training datasets and poor prediction performance of AI models, especially for electrical or mechanical equipment, due to the scarcity of aging and anomaly data.
Innovation Solution
The proposed solution involves acquiring and using large amounts of normal operation data to train machine learning models, combining time-domain and frequency-domain signals or image and image frequency-domain features, without requiring anomaly data, and implementing edge computing for real-time anomaly detection using a processor and data acquisition device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If anomaly data is collected for training AI models, then the model can detect equipment anomalies, but it is extremely difficult to obtain sufficient aging and anomaly data
Solution Approach 1:
Instead of training the model to detect anomalies directly using anomaly data, the patent inverts the approach by training the model to recognize normal operation patterns. The anomaly detection is then achieved by identifying deviations from these learned normal patterns, eliminating the need for scarce anomaly data while maintaining detection capability
Solution Approach 2:
The patent creates a virtual representation of normal equipment operation through comprehensive data collection during normal states. This virtual model serves as a reference copy that the system compares against real-time operation to detect anomalies, allowing detection without requiring physical anomaly data
2Measurement precision
If large amounts of normal and anomaly data are collected for AI model training, then the prediction performance improves, but the training data becomes unbalanced due to scarcity of anomaly data
Solution Approach 1:
The patent extracts and removes the problematic anomaly data requirement from the training process. By focusing training exclusively on normal operation data and using pattern deviation for detection, the method eliminates data imbalance while maintaining high prediction performance through the learned normal operation baseline
3Ease of manufacture
If AI models are trained with unbalanced data, then the model can be trained, but the prediction performance for detecting equipment anomalies decreases
Solution Approach 1:
The patent inverts the detection logic: instead of teaching the model what anomalies look like, it teaches the model what normal operation looks like and detects anomalies as deviations from this normal pattern, achieving high detection accuracy without unbalanced training data
Data Source
AI summary
A method and an apparatus for equipment anomaly detection are provided. In the method, multiple signals of an equipment during normal operation or appearance images of the equipment when an appearance is not damaged are acquired in advance by using a data acquisition device to train a machine learning model stored in a storage device. A real-time signal of the equipment during a current operation or a current image of the appearance of the equipment is acquired by using the data acquisition device, and input to the trained machine learning model to output a detection result indicating a current operation state of the equipment or a current state of the appearance of the equipment.