Machinery Failure Prediction Horizon Learning From Run-to-Failure Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting machinery failures rely on manual selection of prediction horizons, which can be subjective and inaccurate, leading to reduced model quality and increased costs due to unnecessary maintenance and potential breakdowns.
Innovation Solution
A computer-implemented method that iteratively evaluates condition monitoring data to determine candidate cut-off values by calculating distances between relative frequency distributions of positive and negative sub-sequences, training a failure prediction model to automatically learn a prediction horizon, and selecting the best model for accurate failure prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If manual selection of prediction horizon is used, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The system enables the prediction model to automatically determine the optimal prediction horizon by iteratively evaluating candidate cut-off values and selecting the one that maximizes model performance metrics, eliminating the need for manual intervention while achieving high precision through data-driven optimization
Solution Approach 2:
The system systematically varies the prediction horizon parameter across multiple candidate values, evaluates each candidate's performance using iterative training and validation, and selects the optimal parameter setting based on quantitative metrics, thereby transforming subjective manual selection into objective automated optimization
2Device complexity
If manual selection of prediction horizon is used, then device complexity is reduced, but reliability deteriorates
Solution Approach 1:
The system performs preliminary iterative evaluation of multiple candidate cut-off values and prediction horizon settings before final model deployment, pre-identifying the optimal configuration through systematic exploration of parameter space, which ensures high reliability from the outset while maintaining manageable complexity through structured preprocessing
Solution Approach 2:
The system implements feedback loops where model performance metrics from iterative training cycles inform the selection of subsequent candidate cut-off values, continuously refining the prediction horizon determination based on actual model behavior and performance data, thereby enhancing reliability through self-correcting optimization
3Measurement precision
If iterative evaluation of candidate cut-off values is performed, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The system evaluates a finite set of discrete candidate cut-off values rather than exhaustively searching the entire parameter space, performing partial evaluation of promising candidates identified through preliminary analysis, which achieves sufficient precision for practical applications while significantly reducing computational time compared to exhaustive search methods
Solution Approach 2:
The system segments the prediction horizon determination process into distinct iterative phases, evaluating candidate cut-off values in structured batches and using results from earlier iterations to guide subsequent evaluations, which organizes the computational workload efficiently and reduces overall processing time while maintaining precision
Data Source
AI summary
The present disclosure relates to computer-implemented methods, software, and systems for predicting failure event occurrence for a machine asset. Run-to-failure sequences of time series data that include an occurrence of a failure event for the machine asset are received. One or more candidate cut-off values are determined based on iterative evaluation of a plurality of potential cut-off points. A candidate cut-off value is identified as substantially corresponding to a local peak point for calculated distances between relative frequency distributions of positive and negative sub-sequences. A failure prediction model is iteratively trained to iteratively extract sets of relevant features to determine a prediction horizon for an occurrence of the failure event for the machine asset. A candidate cut-off value associated with a model of highest quality from a set of failure prediction models determined during the iterations is selected to determine the prediction horizon for the machine asset.