Device State Detection Using Segmented Pattern Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for automatically recognizing the status of devices, especially in error-prone and noisy measurement data, often lead to incorrect pattern recognition due to measurement noise and statistical inaccuracies, resulting in unreliable device status detection.
Innovation Solution
A system that includes a signal detection device and an analysis unit capable of independently searching for and comparing pattern sections in measurement data, using dynamic time normalization and multiple position recording to accurately determine the device's state by verifying the presence of predetermined patterns, thereby reducing false recognition rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If pattern recognition is performed on measurement data using conventional methods, then device status detection can be automated, but false recognition rates increase due to measurement noise and statistical inaccuracies
Solution Approach 1:
The predetermined pattern is divided into at least two pattern sections that are searched for independently in the measurement data. This segmentation allows the system to verify the presence of complete patterns while filtering out false positives caused by noise, as isolated noise segments are unlikely to form complete sequential patterns.
Solution Approach 2:
The system uses the detected positions of pattern sections and their order to verify pattern presence, creating a feedback mechanism that confirms pattern recognition accuracy. The sequential verification of pattern sections provides feedback that reduces false recognitions by requiring multiple conditions to be met simultaneously.
2Device complexity
If the entire predetermined pattern is searched for as a single unit in measurement data, then pattern detection is simpler, but false recognitions increase in noisy data
Solution Approach 1:
The pattern is segmented into multiple sections that are searched independently. While this increases computational steps, each individual search is simpler and more robust to noise, as the system can identify and discard incomplete or noisy pattern sections before confirming overall pattern presence.
Solution Approach 2:
The system performs multiple partial searches for pattern sections rather than a single complete pattern search. This excessive action of searching for sections separately and then verifying their sequence provides more reliable detection by requiring multiple independent confirmations, thereby reducing false positives.
3Productivity
If measurement data is analyzed without dividing patterns into sections, then analysis process is faster, but false positives occur more frequently
Solution Approach 1:
By segmenting the pattern into sections that can be searched independently and in parallel, the system maintains analytical speed while improving reliability. The segmented approach allows for efficient processing of each section while the sequential verification ensures accurate overall pattern detection.
Solution Approach 2:
The system performs multiple partial comparisons of pattern sections against measurement data rather than a single complete pattern match. This excessive partial action increases computational effort but significantly reduces false positives by requiring multiple independent confirmations of pattern presence.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention relates to a system (1) for the automatic detection of a state of a device (2), comprising: a signal acquisition means (10) for acquiring defective measurement data (D) of a physical variable characterising the device (2); and an analysis unit (11) for identifying a specified pattern (M) in the measurement data (D) acquired by the signal acquisition means (10). The analysis unit (11) is designed: - to compare at least two different pattern sections (M1-M4) of the specified pattern (M) separately from each other with the measurement data (D); - on the basis of the respective comparison, to determine at least one position of each of the pattern sections (M1-M4) in the measurement data (D); - on the basis of the positions determined and the order of the positions of the pattern sections (M1-M4), to detect the specified pattern (M) at one or more positions in the measurement data (D) and, - on the basis of the one or more positions of the specified pattern (M), to determine the state of the device (2).