Tooling Machine Image Monitoring for Real-Time Tool Life Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting the remaining useful life of work tools in manufacturing processes are time-consuming, costly, and impractical for real-time mass production, particularly for technologies involving multiple tools and work pieces, and are often impractical for real-time monitoring, lacking a quick and cost-sensitive solution.
Innovation Solution
A computer-implemented method using a local binary pattern (LBP) algorithm to optimize images captured by an image capturing device, followed by a machine-learned computing model trained with histogram data to predict the remaining useful life of work tools, providing real-time monitoring and replacement indications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional prediction models are used for each work piece quality individually, then measurement precision is improved, but productivity deteriorates due to slow and lengthy model modification processes
Solution Approach 1:
The patent combines multiple individual prediction models into a single unified predictive model that can handle multiple work piece quality parameters simultaneously. This integration allows the system to predict various quality metrics (e.g., surface roughness, dimensional accuracy) in one go, eliminating the need to modify separate models for each quality parameter and thereby significantly improving productivity while maintaining prediction accuracy.
Solution Approach 2:
The predictive model is designed with multi-functionality to accommodate different work piece quality parameters and various tool types. By making the model universal, it can adapt to predict multiple quality aspects without requiring individual model modifications, thus resolving the contradiction between maintaining precision and improving modification speed.
2Reliability
If complex manufacturing methods are used for tool life prediction, then reliability is improved, but ease of manufacture deteriorates making them impractical for real-time mass production
Solution Approach 1:
The patent replaces complex mechanical and manual monitoring methods with an automated image processing and machine learning-based prediction system. By using digital image capture, LBP algorithm processing, and neural network models, the system achieves reliable tool life prediction without the complexity of physical measurements and manual analysis, making it suitable for real-time mass production environments.
Solution Approach 2:
The system enables self-service monitoring where the machine automatically captures images, processes them through the LBP algorithm, and uses the machine-learned model to predict tool life without requiring external intervention or complex manual procedures. This automation simplifies implementation while maintaining high reliability in predicting tool remaining useful life.
3Measurement precision
If traditional monitoring systems are deployed for multiple tools and work pieces, then measurement precision is improved, but device complexity increases making them costly and impractical
Solution Approach 1:
The patent develops a universal image processing and prediction system that can monitor multiple tools and work pieces using the same hardware setup and software algorithm. The LBP algorithm and machine-learned model are designed to be tool-agnostic and work piece-agnostic, allowing a single system configuration to serve multiple functions without increasing device complexity or cost.
Solution Approach 2:
The system segments the monitoring task into distinct modular components: image capture, LBP algorithm processing, histogram generation, and machine-learned prediction. This segmentation allows each component to be independently optimized and reused across different tools and work pieces, reducing overall system complexity while maintaining measurement precision.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
The present disclosure provides a computer implemented method of operating a tooling machine. The method comprising: providing a work tool (12) and a work piece (14), providing an image capturing device (16), starting a tooling process and engaging the work tool (12) with the work piece (14), capturing a basic image (18) of a surface of the work piece (14) by the image capturing device (16), the image capturing device (16) transmitting the basic image (18 ) to a processing unit (22) of a computer (20), the processing unit (22) creating an optimized image (24), wherein creating the optimized image (24) comprises processing the basic image (18) by using an image optimization algorithm via the processing unit (22). It is further proposed creating a histogram (26) of the optimized image (24), using data of the histogram (26) as one or more indicators (28), utilizing the one or more indicators (28) as an input data for a machine-learned model (30), and utilizing an output data of the machine-learned model (30) for providing an indication for a need of a replacement of the work tool (10) or for stopping the tooling machine. Furthermore, a tooling machine (10) using such method is disclosed.