Motor Abnormality Detection via Data-Mechanical Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting and diagnosing mechanical equipment states using sensor data are inadequate in detecting mechanical abnormalities such as secular changes and oscillations in motor-driven mechanisms, as they rely solely on statistical methods and cannot effectively determine abnormalities across the entire mechanical system.
Innovation Solution
An abnormality determining system that acquires time-series data on inputs and outputs of a motor-driven mechanism, using machine learning techniques like the Hotelling T2 method to detect data abnormalities and separately determine mechanical abnormalities based on the acquisition aspects of the data, such as frequency and combination, to identify secular changes and oscillations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If statistical methods are used for predicting and diagnosing mechanical equipment states, then the analysis can be performed on sensor data, but the detection of mechanical abnormalities such as secular changes and oscillations is inadequate
Solution Approach 1:
The patent segments the analysis into two distinct parts: data abnormality detection (using statistical methods like Hotelling's T2 method) and mechanical abnormality determination (based on acquisition aspects). This segmentation allows each method to focus on its strength, with statistical methods handling data patterns and acquisition aspect analysis handling mechanical fault identification, thereby resolving the contradiction between measurement precision and reliability.
Solution Approach 2:
The patent introduces 'acquisition aspect' as an intermediary concept that bridges sensor data and mechanical abnormality determination. By analyzing how data is acquired (frequency, combination, timing) rather than just the data values themselves, the system can reliably determine mechanical abnormalities while maintaining the benefit of statistical data analysis.
2Ease of operation
If statistical methods alone are used for data analysis, then the processing is straightforward, but the determination of mechanical abnormalities across the entire mechanical system cannot be effectively performed
Solution Approach 1:
The patent divides the diagnostic process into two segments: a simple statistical data abnormality detection part (maintaining ease of operation) and a mechanical abnormality determination part based on acquisition aspects (improving detection capability). This segmentation allows the system to remain easy to operate while significantly improving mechanical abnormality detection across the entire system.
Solution Approach 2:
The patent adds a new dimension of analysis by considering 'acquisition aspects' (how, when, and how frequently data is acquired) in addition to the traditional data value analysis. This dimensional expansion enables comprehensive mechanical system abnormality detection without complicating the basic data processing operations.
3Device complexity
If only data values are analyzed without considering acquisition aspects, then the analysis is simpler, but mechanical abnormalities such as secular changes and oscillations cannot be detected
Solution Approach 1:
The patent segments the analysis complexity: data value analysis remains simple (using established statistical methods), while acquisition aspect analysis handles the detection of mechanical abnormalities. This segmentation increases detection reliability without substantially increasing overall system complexity, as the two analyses are performed in parallel independent tracks.
Solution Approach 2:
The patent uses 'acquisition aspect' as an intermediary that translates complex mechanical system behavior into analyzable patterns. By mediating between raw sensor data and mechanical abnormality determination through acquisition characteristics, the system achieves high reliability detection without requiring complex analysis methods.
Data Source
AI summary
An abnormality determining apparatus for a motor driven mechanism includes circuitry. The circuitry is configured to acquire time-series data with respect to an input to and an output from a motor which drives the motor driven mechanism, detect data abnormality in the time-series data, and determine, based on the data abnormality, whether mechanical abnormality in the motor driven mechanism occurs.