Machine Tool Anomaly Detection by Workpiece-Specific ML Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting abnormalities in mechanical processing equipment are uncertain and inaccurate due to reliance on engineer qualifications and unified thresholds, which fail to account for variations in processing different workpieces.
Innovation Solution
A method utilizing machine learning models tailored to specific workpiece types, processed parts, and processing stages to analyze processing data, separating data based on workpiece type, removing irrelevant periods, and applying machine learning classification to enhance detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If unified upper and lower thresholds of processing data set are used for abnormality detection, then the detection method is simple to implement, but the detection accuracy deteriorates due to inability to account for variations in processing different workpieces
Solution Approach 1:
The patent segments the processing data into multiple groups based on workpiece types, processing conditions, and equipment states. Instead of using a single unified threshold for all operations, the system creates separate threshold ranges for different segmentation categories (e.g., different workpiece types, processing speeds, feed rates). This allows the detection system to adapt to variations in processing conditions while maintaining systematic organization and manageable complexity.
2Ease of operation
If maintenance engineers evaluate abnormalities based on their qualifications, then the evaluation process is simple, but the reliability deteriorates due to great deal of uncertainty in evaluation
Solution Approach 1:
The patent implements a feedback mechanism where the system continuously monitors processing data, compares it against dynamically generated threshold ranges, and provides real-time feedback on equipment status. The system automatically adjusts threshold ranges based on historical data and actual equipment responses, creating a closed-loop feedback system that improves reliability without requiring subjective engineer evaluation.
Solution Approach 2:
The system performs self-diagnosis by automatically analyzing processing data against established threshold ranges and generating abnormality detection results without requiring manual engineer intervention. The system serves itself by autonomously identifying abnormalities, reducing reliance on engineer qualifications while maintaining operational simplicity through automated alerting and reporting.
3Productivity
If processing data is analyzed without separating by workpiece type, then the analysis process is fast and simple, but the detection precision deteriorates due to challenge in accurately detecting abnormalities for different workpiece types
Solution Approach 1:
The patent applies preliminary action by pre-segmenting processing data according to workpiece types and processing conditions before performing abnormality detection. The system预先 (in advance) categorizes data into different groups based on workpiece characteristics, allowing subsequent analysis to focus on specific segments. This preliminary segmentation enables both efficient processing (by focusing analysis on relevant data subsets) and high precision (by using type-specific threshold ranges).
Data Source
AI summary
A method, system, and storage medium for detecting abnormality of mechanical processing equipment. The method includes obtaining processing data of the mechanical processing equipment during machining of a workpiece, obtaining a type of the workpiece of the processing data, separating the processing data based on the type of machined workpiece, selecting a machine learning model corresponding to the workpiece type of the separated processing data to process the separated processing data to obtain an abnormality detection result. With the method for detecting abnormality of mechanical processing equipment disclosed in the present disclosure, abnormality of the mechanical processing equipment can be accurately detected, thereby improving the yield rate of the processed workpiece.