Machine Tool Diagnostic Model Relearning Based on Tool Usage Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machining abnormality diagnosis systems using machine learning face challenges in determining the necessity of relearning the diagnostic model, particularly in distinguishing between misdiagnosis caused by threshold issues and model deficiencies, leading to over-detection and increased tool replacement costs.
Innovation Solution
A method and device that determine the necessity of relearning a diagnostic model by analyzing the frequency distribution of tool usage data, specifically using a logarithmic normal distribution to assess whether the model accurately represents tool wear, and optionally removing samples to account for sudden abnormalities, thereby reducing unnecessary relearning and tool replacement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the threshold is adjusted to reduce missing (false-negative), then over-detection (false-positive) increases, leading to increased tool replacement costs
Solution Approach 1:
The patent changes the parameter being analyzed from simple threshold adjustment to the distribution characteristics of tool usage data. By examining whether the data follows a logarithmic normal distribution, the system can identify appropriate thresholds without increasing over-detection, thus resolving the contradiction between reliability and loss.
2Measurement precision
If relearning of the diagnostic model is performed frequently to improve diagnosis accuracy, then computational load and time consumption increase
Solution Approach 1:
The patent performs preliminary analysis of the tool usage data distribution before deciding on relearning. By checking whether the data follows a logarithmic normal distribution in advance, the system can determine if the current model is still valid, avoiding unnecessary relearning operations and thus reducing time loss while maintaining diagnosis accuracy.
Solution Approach 2:
The patent implements a feedback mechanism where the distribution characteristics of tool usage data are continuously monitored. When the data deviates from the expected logarithmic normal distribution, this triggers model relearning, creating a closed-loop system that maintains accuracy without excessive relearning operations.
3Loss of substance
If the threshold is decreased to reduce over-detection, then missing (false-negative) increases, leading to machine abnormalities and product defects
Solution Approach 1:
Instead of simply adjusting the threshold parameter, the patent changes to analyzing the distribution parameters of tool usage data. By identifying whether data follows a logarithmic normal distribution, the system can set optimal thresholds that prevent both over-detection and missing, thus resolving the contradiction between cost and reliability.
Data Source
AI summary
A relearning necessity determination method is provided for determining a necessity of relearning of a learned diagnostic model in a machine tool including a machining abnormality diagnosing unit. The machining abnormality diagnosing unit determines normal or abnormality of machining using the diagnostic model generated through machine learning. The method includes storing a cumulative cutting time or a cumulative cutting distance of a tool mounted to the machine tool as a tool usage, storing the tool usage when the machining abnormality diagnosing unit diagnoses the machining as machining abnormality, and determining the necessity of the relearning of the diagnostic model based on a frequency distribution of the tool usage stored in the storing of the tool usage.