Learning Model for Press Compensation Without Setup Trial and Error
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In press systems, determining the optimal die height for processing various materials is often time-consuming and requires trial and error, as it is influenced by multiple parameters such as plate thickness, hardness, and temperature, making it inefficient to adjust compensation amounts for accurate control variable settings.
Innovation Solution
A learning device that uses a model to generate a compensation amount for command values in a control system, allowing for the evaluation and adjustment of control variables based on specific parameters of the material being processed, such as material characteristics and physical properties, without the need for repeated actual operations, using data-driven control methods like VRFT, FRIT, or ERIT.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If trial and error method is used to determine optimal die height for each plate thickness, then appropriate control variable settings can be achieved, but huge amount of time is required
Solution Approach 1:
The system performs preliminary learning by collecting operation data and generating a learned model in advance, storing the relationship between plate thickness and optimal die height compensation amounts. When actual processing is needed, the pre-learned model directly provides the compensation amount without requiring trial and error adjustments, thus resolving the contradiction between achieving accurate control variable settings and reducing adjustment time.
2Reliability
If trial and error is performed for various objects to be processed to confirm compensation amount appropriateness, then appropriate compensation can be verified, but huge amount of time is required
Solution Approach 1:
The system creates a virtual copy of the learning process by generating a learned model that simulates the relationship between material parameters and optimal compensation amounts. Instead of performing actual trial and error on physical objects, the system uses the learned model to predict appropriate compensation amounts, thereby verifying reliability without consuming excessive time on actual processing trials.
Solution Approach 2:
The system incorporates feedback mechanisms by collecting operation data from actual processing results and using it to generate and refine the learned model. The model learns from past operations and continuously improves its predictions, allowing the system to verify compensation amount appropriateness efficiently while maintaining high reliability through data-driven validation.
3Manufacturing precision
If multiple parameters (hardness, temperature, material) are considered for compensation, then comprehensive control accuracy is achieved, but learning complexity increases
Solution Approach 1:
The system segments the learning process by first collecting operation data that includes multiple parameters (plate thickness, hardness, temperature, material type), then using machine learning algorithms to automatically generate a comprehensive learned model. The model internally handles the complexity of multiple parameters through data-driven pattern recognition, allowing comprehensive control accuracy to be achieved without manually increasing system complexity.
Data Source
AI summary
This learning device provides a learned model to an adjuster containing a learned model learned to output a predetermined compensation amount to a controller, in a control system including the controller outputting a command value obtained by compensating a target value based on a compensation amount and a control object controlled to process an object to be processed. The learning device includes: an evaluation part obtaining operation data including the target value, command value and control variable and evaluates the quality of the control variable; a learning part generating candidate compensation amounts based on the operation data, and learning, as teacher data, the generated candidate compensation amount and the specific parameter of the object, and generating a learned model; and a setting part providing the learned model to the adjuster if the evaluated quality is within an allowable range.