Fault Prediction System Using Dynamic Reference Time Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fault prediction systems face limitations in accurately predicting fault occurrences due to variations in the periods between symptom onset and actual fault events, leading to reduced prediction precision.
Innovation Solution
A fault predicting system that adjusts reference time points and prediction target periods based on predetermined length settings, using machine learning to generate a fault prediction model with feature quantities from operation log information and fault index values, allowing for more precise symptom sensing and prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If operation logs are partially retrieved with uniform temporal differences prior to fault occurrences, then the fault prediction system can be simplified, but the prediction precision deteriorates because the operation logs do not necessarily include symptoms of faults
Solution Approach 1:
The patent applies dynamics by making the log retrieval period variable rather than uniform. The control unit dynamically adjusts the retrieval period based on the specific fault type and its characteristics, allowing the system to adapt to different fault patterns while maintaining prediction precision without excessive complexity
Solution Approach 2:
The patent changes the temporal parameter (retrieval period length) based on fault type. By adjusting this parameter dynamically according to fault characteristics, the system retrieves logs that actually contain fault symptoms, thereby improving prediction precision while managing system complexity through parameter optimization
2Measurement precision
If the retrieval period is extended to ensure inclusion of fault symptoms, then the prediction precision improves, but the data processing time and computational load increase
Solution Approach 1:
The patent optimizes the retrieval period parameter to be just long enough to capture fault symptoms for each specific fault type. This prevents unnecessary extension of the retrieval period, thereby maintaining prediction precision while minimizing data processing time and computational load
3Measurement precision
If different retrieval periods are used for different fault types, then the prediction precision improves, but the system complexity increases due to need for fault-type-specific parameters
Solution Approach 1:
The patent creates a universal control unit that handles multiple fault types by selecting appropriate retrieval periods based on fault type identification. This multi-functional approach allows the system to manage different retrieval periods through a single centralized component, improving prediction precision while containing system complexity through unified control logic
Data Source
AI summary
A reference time point which is equivalent to a fault-prediction-execution time point is sequentially changed by setting, as a symptom sensing period, a period from a time point which is a predetermined first period-length before the reference time point until the reference time point, and by setting, as a prediction target period, a period from the reference time point until a time point which is a predetermined second period-length after the reference time point if a fault does not occur in the period from the reference time point until the time point which is the second period-length after the reference time point, and a period from the reference time point until a fault-occurrence time point if a fault occurs in the period from the reference time point until the time point which is the second period-length after the reference time point, and machine learning is performed by using, as an explanatory variable, a feature quantity based on the operation log information of the symptom sensing period, and using, as a response variable, a fault index value based on whether or not there is a particular-event occurrence in the operation log information corresponding to a fault occurrence in the fault record information of the prediction target period, and a period-length from the reference time point until the fault-occurrence time point.