Machine Tool Diagnosis Using Context-Based Processing Interval Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems that use vibration sensors to detect abnormalities in machine tools face challenges in accurately identifying the processing interval, as noise from tool replacement or position changes can obscure the actual processing time, and internal states with movable and immovable parts can cause false abnormalities, making it difficult to determine tool abnormalities accurately.
Innovation Solution
A diagnosis device with a first acquiring unit for context information, a second acquiring unit for detection information, an extracting unit for feature information, a selecting unit for reference feature information, a calculating unit for likelihood of the processing interval, and an estimating unit to determine the actual processing interval based on the comparison of feature information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a vibration sensor is used to detect abnormalities during cutting, then abnormality detection capability is improved, but noise from non-processing intervals (tool replacement, position changes) mixes into the detection data, worsening measurement precision
Solution Approach 1:
The patent segments the detection information into multiple intervals based on context information (processing intervals and non-processing intervals). By dividing the continuous detection data into discrete segments corresponding to different operational phases, the system can analyze only the relevant processing intervals for abnormality detection, excluding noise from tool replacement and position changes.
Solution Approach 2:
The patent extracts feature information specifically from processing intervals by using context information to identify and isolate these segments from the overall detection data. This extraction process removes harmful non-processing interval data, allowing abnormality determination to be based solely on relevant cutting operation data.
2Quantity of substance
If detection information from all intervals is used for abnormality determination, then more data is available for analysis, but false abnormalities are caused by noise during non-processing intervals, worsening reliability
Solution Approach 1:
The patent converts the harmful effect of having abundant detection information that includes noise into a benefit by using context information to intelligently select and process only the useful portions. The large volume of detection data becomes advantageous because the system can identify and utilize the relevant processing interval data while discarding non-processing interval noise, turning data quantity into diagnostic quality.
3Duration of action of stationary object
If vibration detection is performed during all operations including tool replacement and position changes, then continuous monitoring is achieved, but false abnormalities are generated from immovable part actions, worsening measurement precision
Solution Approach 1:
The patent performs preliminary classification of detection intervals using context information before conducting abnormality analysis. By预先 identifying processing intervals versus non-processing intervals (tool replacement, position changes), the system prepares the data in advance, ensuring that only relevant processing interval data is subjected to abnormality determination algorithms, thus preventing false positives from immovable part actions.
Data Source
AI summary
A device includes: a first acquiring unit to acquire context information corresponding to running operation among pieces of context information; a second acquiring unit to acquire detection information output from a detecting unit detecting a physical quantity of a target device; an extracting unit to extract, from the detection information, feature information indicating a feature of the detection information in an interval including a specific operation interval of the target device; a selecting unit to select reference feature information used as reference based on the feature information, and sequentially select pieces of target feature information; a calculating unit to calculate a likelihood of a process interval based on a comparison between the reference feature information and each piece of target feature information; a determining unit to determine whether the target feature information corresponding to the likelihood is included in the process interval based on the likelihood; and an estimating unit to estimate the process interval based on a determination result.