Wire-Cut EDM Consumable Life Prediction Using Machining Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional simulations for wire-cut electrical discharge machining do not accurately account for the varying degree of consumable deterioration due to machining conditions, leading to errors in estimating remaining life and reducing machining efficiency and accuracy.
Innovation Solution
A machine learning device and prediction system that obtain input data on machining conditions and consumables' deterioration levels to generate a trained model, which predicts the degree of consumable deterioration before machining, using supervised learning to improve accuracy and reduce worker load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional simulations are used to estimate remaining life, then the process is simple, but the prediction accuracy of consumable deterioration is poor
Solution Approach 1:
The patent replaces conventional simulation methods with a machine learning model that uses supervised learning to predict consumable deterioration. The system collects actual machining data including machining conditions (electric discharge voltage, fluid pressure, plate thickness) and consumable usage data, then trains a prediction model that accurately estimates remaining life without relying on simplified conventional simulations.
Solution Approach 2:
The system enables self-service by automatically collecting machining data and consumable information, training the prediction model autonomously, and providing accurate remaining life predictions without requiring manual intervention or complex conventional simulation processes. The machine learning model serves itself by continuously learning from actual operational data.
2Extent of automation
If conventional simulation methods are used, then worker load is high due to manual determination, but automation is low
Solution Approach 1:
The system automatically collects machining data from the wire-cut electrical discharge machine, gathers consumable information, trains the prediction model, and provides remaining life predictions without manual worker intervention. This self-service approach eliminates the need for workers to manually determine deterioration rates while providing accurate automated predictions.
Solution Approach 2:
The system implements feedback by continuously collecting actual machining data and consumable usage information, using this feedback to train and improve the prediction model. The model learns from real operational data and provides accurate predictions, creating a closed-loop system that reduces worker load while increasing automation.
3Productivity
If machining conditions are not considered in simulation, then the process is simple, but machining efficiency and accuracy deteriorate
Solution Approach 1:
The patent incorporates machining condition parameters (electric discharge voltage, machining fluid type, fluid pressure, workpiece plate thickness) as input features for the prediction model. By considering these parameter changes and their impact on consumable deterioration, the system provides accurate remaining life predictions that enable timely consumable replacement, thereby maintaining both machining efficiency and accuracy.
Data Source
AI summary
A machine learning device includes an input data acquisition unit which acquires input data containing a machining condition for any wire-cut electrical discharge machining applied to any workpiece by any wire-cut electrical discharge machining machine and consumables information including the degree of degradation of at least one of an electrode wire, ion exchange resin, a power supply die, and an electrode wire guide roller before wire-cut electrical discharge machining. The device also includes a label acquisition unit which acquires label data indicating the degree of degradation of at least one of the electrode wire, the ion exchange resin, the power supply die, and the electrode wire guide roller after the wire-cut electrical discharge machining under the machining condition contained in the input data, and a learning unit which uses the input data and the label data to execute supervised learning, thereby generating a learned model.