Machine Tool Chip Imaging for Early Machining Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine tools struggle to detect abnormalities in machining processes effectively, leading to potential tool damage and workpiece defects, as they rely solely on machining load monitoring, which may not capture all anomalies, and existing methods like grinding wheel wear detection are limited in scope.
Innovation Solution
A control device and program that photographs chips produced during machining, determines a reference model of abnormal chips based on pre-alarm stop images, and judges abnormality occurrence by comparing subsequent chip images to this model, allowing for earlier detection of machining issues before alarm stop.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine tools rely solely on machining load monitoring to detect abnormalities, then the monitoring system remains simple, but abnormality detection precision is insufficient and cannot capture all anomalies
Solution Approach 1:
The patent combines multiple detection methods into a unified monitoring system: machining load monitoring, chip image capture, and AI-based chip morphology analysis are integrated to work together. This merging allows the system to leverage the strengths of each method while compensating for their individual limitations, thereby improving overall abnormality detection precision without requiring completely separate systems
Solution Approach 2:
The patent introduces chip images as an intermediary element between the machining process and the detection system. By capturing and analyzing chip morphology through imaging, the system gains an additional dimension of information about machining conditions. The AI model uses these chip images as intermediate data to detect abnormalities that machining load monitoring alone cannot identify
2Reliability
If machine tools wait for alarm stop to detect abnormalities, then the control logic remains simple, but tool damage and workpiece defects occur before detection
Solution Approach 1:
The patent implements preliminary action by continuously capturing chip images during the machining process and using an AI model to predict potential abnormalities before they cause alarm stops or damage. The system performs preliminary analysis of chip morphology trends to identify early signs of problems, allowing preventive measures to be taken before critical failures occur
Solution Approach 2:
The patent establishes a feedback loop where chip image analysis results are continuously fed back to adjust the detection threshold and alert operators. When the AI model detects chip morphology changes indicating potential abnormalities, it provides feedback that triggers early warnings, allowing the system to respond before alarm stops occur. This feedback mechanism enables dynamic adjustment of detection sensitivity based on real-time machining conditions
3Loss of substance
If conventional methods detect abnormalities only at alarm stop, then the detection system remains simple, but the number of workpieces discarded due to abnormality increases
Solution Approach 1:
The patent implements self-service by enabling the system to automatically detect, analyze, and respond to machining abnormalities without requiring constant human intervention. The AI model autonomously analyzes chip images, identifies abnormal patterns, and triggers appropriate responses, allowing the system to serve itself in monitoring and detecting issues that would otherwise require manual inspection and lead to increased scrap
Data Source
AI summary
A control device, which controls a machine tool capable of alarm stop based on abnormality in machining load, includes: a photographing unit which photographs chips produced as a result of machining of a workpiece; a reference model determination unit which determines in advance a reference model of chips for determining as abnormality in machining, based on images of chips photographed within a predetermined period before alarm stop, in response to an occurrence of alarm stop; and a judgment unit which judges as abnormality occurrence in machining, in a case of a degree of similarity in chips photographed at a predetermined timing in machining later, relative to a reference model of chips that was determined.