Movable Equipment Anomaly Detection Using Linked Periodic Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing abnormality detection methods for movable apparatuses face challenges in accurately detecting abnormalities due to individual differences in data, lack of failure data, and difficulty in extracting features from non-periodic data, especially in industrial machinery, leading to inefficiencies in failure detection.
Innovation Solution
An abnormality detection apparatus that converts non-periodic operation-related data into periodic data, generates reference data during stable operation periods, and performs change detection processing using linked data to identify abnormalities and estimate failure timing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-periodic operation-related data is directly analyzed, then the original data characteristics are preserved, but the detection accuracy is insufficient due to difficulty in extracting features
Solution Approach 1:
The patent applies periodic action by converting non-periodic operation-related data into periodic data through temporal linking. Multiple pieces of operation data collected at different time points are linked together to form periodic sequences, enabling the application of spectral analysis and other periodic signal processing methods to detect abnormalities that would be difficult to identify in non-periodic data.
2Reliability
If reference data is generated during stable operation periods, then the detection baseline is improved, but the processing time to accumulate sufficient data is increased
Solution Approach 1:
The patent applies preliminary action by proactively collecting and storing operation-related data during stable operation periods before abnormalities occur. This reference data is accumulated and processed in advance to establish a baseline for comparison, enabling faster and more reliable abnormality detection when issues arise without requiring extensive real-time data accumulation.
3Measurement precision
If multiple pieces of operation-related data are linked to form periodic data, then the feature extraction capability is improved, but the computational load is increased
Solution Approach 1:
The patent applies segmentation by dividing operation-related data into multiple discrete pieces collected at different time points, then selectively linking specific segments to form periodic sequences. This segmented approach allows for targeted feature extraction from relevant data portions rather than processing entire datasets, reducing computational power consumption while maintaining detection accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An abnormality detection apparatus comprising a target data generation unit configured to generate, based on operation-related data 310 resulting from an operation of a movable apparatus, a plurality of target data 320 that are temporally separated, and a detection processing execution unit configured to execute change detection processing on linked data 330 obtained by linking the plurality of target data 320 with reference data 342.