Abnormality Prediction Model Using Probability Density Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing abnormality prediction methods for devices, such as gas engines, face challenges in accurately predicting issues on newly introduced or unfailed devices due to individual differences among devices of the same model, leading to reduced prediction accuracy.
Innovation Solution
A prediction system that converts operation data into probability densities using the variational Bayesian method, allowing for the creation of a prediction model that accounts for and excludes individual differences among devices, enabling accurate abnormality prediction by learning from corrected data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a prediction model is created using data collected when an abnormality has occurred in the device, then the prediction model reflects intrinsic properties of the device, but this method cannot be used when abnormality prediction is performed on a newly introduced device or on a device in which an abnormality does not occur
Solution Approach 1:
The patent creates a prediction model by copying data from reference devices (devices of the same model where abnormalities have occurred) and transforming it into a form that can be applied to target devices. The learning data is constructed by combining operation data from multiple reference devices, effectively creating a representative model that can be copied and applied to new devices without requiring abnormality data from each individual device.
2Adaptability or versatility
If a prediction model is created using data collected from other devices of the same model in which an abnormality has occurred, then the model can be applied to newly introduced devices, but there is an individual difference among the devices leading to reduced prediction accuracy
Solution Approach 1:
The patent merges operation data from multiple reference devices to create a comprehensive learning data set. By combining data from multiple sources and transforming it into probability density representations, the system creates a prediction model that captures common patterns across devices while filtering out individual differences, thereby improving prediction accuracy for target devices.
Solution Approach 2:
The patent transforms operation data into probability density parameters, changing the representation from raw operational values to statistical distributions. This parameter transformation allows the system to capture the essence of device behavior while eliminating individual device variations, enabling accurate prediction across different devices of the same model.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
Provided is a prediction system that predicts whether a prescribed event will occur in a device, without being affected by differences among individual devices. The prediction system comprises: a data acquisition unit which acquires operation data representing the operation status of a device; a probability density estimation unit which estimates the probability density of the operation data; and an abnormality prediction unit which predicts whether an abnormality will occur in the device on the basis of the probability density estimation results of the operation data and a prediction model.