Rolling Mill Setup Values That Capture Operator Adjustments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining mill setup values in rolling mills struggle to reflect immediate adjustments made by operators through manual intervention, leading to delays in incorporating these changes into subsequent mill setups. Additionally, the precision of these methods depends on the accuracy of mathematical models, which can be inconsistent.
Innovation Solution
A method using trained models that incorporate relations between previous and current mill setup values, allowing for the determination of setup conditions that reflect operator-driven manual adjustments. This involves a two-step process using first and second trained models based on the appropriateness of using previous setup values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If mill setup values are determined using trained neural networks based on large volume of past records, then automation is improved, but the ability to immediately reflect operator's manual adjustments is worsened
Solution Approach 1:
The patent segments the mill setup determination into two distinct components: (1) automatic mill setup values calculated by trained neural networks based on past records, and (2) manual adjustment values input by operators. These two components are processed separately and then combined, allowing each to function independently without interfering with the other's timing or accuracy.
Solution Approach 2:
The patent introduces an intermediary mechanism that receives both the automatic mill setup values from the neural network and the manual adjustment values from the operator, then combines them to produce the final mill setup values. This intermediary process enables the immediate reflection of manual adjustments while preserving the automated calculation foundation.
2Manufacturing precision
If optimization is performed using sequential quadratic programming on a mathematical model, then manufacturing precision is improved, but the method becomes sensitive to model precision and requires more complex calculations
Solution Approach 1:
The patent uses trained neural networks that have been trained on historical mill setup data and actual operator adjustments. The neural network creates a learned copy of the relationship between manufacturing conditions and optimal setup values, replacing the need for complex real-time mathematical modeling while capturing the essential patterns from past performance data.
Solution Approach 2:
The system uses past mill setup records and operator adjustment patterns to automatically improve its own accuracy over time. The neural network learns from historical data and continuously refines its predictions, enabling the system to self-improve without requiring external recalibration or complex model adjustments.
3Stability of the object's composition
If mill setup values are determined solely by automated systems, then consistency is improved, but adaptability to operator expertise and manual insights is worsened
Solution Approach 1:
The patent merges two different sources of mill setup determination: the automated neural network calculations based on historical data patterns, and the operator's manual adjustments based on experience and real-time observations. By combining these two approaches, the system achieves both the consistency of automated processing and the adaptability of human expertise.
Solution Approach 2:
The system incorporates feedback from operator manual adjustments into the neural network training process. By learning from actual operator modifications to automated recommendations, the system adapts to incorporate human expertise and preferences, improving both consistency and adaptability over time.
Data Source
AI summary
A set condition determining method for manufacturing facilities includes: inputting, into a trained model, a manufacturing condition for a target product and a setup condition that is for a product manufactured in same manufacturing facilities before manufacture of the target product and that reflects setup condition modification by an operator's manual manipulation; and obtaining a setup condition for the target product. The trained model has been trained with input being: manufacturing conditions for the target product; and setup conditions that are for the product manufactured in the same manufacturing facilities before the manufacture of the target product and that reflect setup condition modification by an operator's manual manipulation, and output being setup conditions for the target product.