Cold Rolling Condition Prediction for Stable High-Speed Mills
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in cold rolling mills is maintaining productivity and stability while rolling difficult-to-roll materials with high loads, as changes in lubrication state, roll temperature, and thermal expansion affect rolling conditions, leading to variability in operating speed and potential issues like chattering and sheet thickness variations.
Innovation Solution
A cold rolling mill rolling condition calculation method that uses a prediction model trained with past rolling performance data to estimate and adjust rolling constraints, ensuring stable operation by inputting multi-dimensional data and changing target rolling conditions to meet predetermined criteria, thereby stabilizing the rolling process and maintaining productivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the rolling speed is increased to improve productivity, then the operating speed and productivity are improved, but the rolling stability deteriorates due to changes in lubrication state, thermal expansion, and temperature rise
Solution Approach 1:
The system performs preliminary learning of past operating conditions using a neural network to predict future rolling constraints before actual rolling occurs. This allows the rolling speed to be pre-adjusted based on predicted conditions, preventing stability issues before they arise while maintaining high productivity.
Solution Approach 2:
The system continuously monitors rolling performance data and feeds it back to the neural network model, which updates its predictions and adjusts rolling conditions in real-time. This closed-loop feedback mechanism maintains rolling stability even at high speeds by dynamically compensating for lubrication state changes, thermal expansion, and temperature rise.
2Reliability
If the rolling speed is reduced to maintain rolling stability, then the rolling stability is improved, but the productivity deteriorates
Solution Approach 1:
The system dynamically adjusts the rolling speed based on real-time predictions from the neural network model. Rather than maintaining a fixed conservative speed, the system optimizes speed dynamically according to actual rolling conditions, achieving both high productivity and rolling stability through adaptive control.
3Reliability
If the pass schedule and rolling speed are manually set to avoid facility constraints, then the facility constraint is satisfied, but the productivity varies depending on operator experience
Solution Approach 1:
The system performs self-service by automatically learning from past operating conditions and independently determining optimal rolling speeds and pass schedules. The neural network model eliminates dependence on operator experience by autonomously predicting rolling constraints and adjusting parameters to satisfy facility constraints while maximizing productivity consistently.
4Productivity
If the rolling speed is increased for difficult-to-roll materials, then the productivity is improved, but the rolling constraint condition deteriorates due to high load and thermal effects
Solution Approach 1:
The system performs preliminary learning and prediction of rolling constraints specific to difficult-to-roll materials before processing. By analyzing past performance data for similar high-load materials, the system pre-determines optimal rolling speeds that maintain manufacturing precision while maximizing productivity, avoiding the need to reduce speed due to thermal effects and high load.
Data Source
AI summary
A cold rolling mill rolling condition calculation method includes: an estimation step of estimating a rolling constraint condition with respect to a target steady rolling condition of a roll target material, by inputting second multi-dimensional data to a prediction model, the prediction model having been trained with explanatory variable and response variable, the explanatory variable being first multi-dimensional data generated based on non-steady rolling performance data, among past rolling performance in rolling a roll material by a cold rolling mill, and the response variable being steady rolling performance data and rolling constraint condition data during steady rolling, and the second multi-dimensional data having been generated based on non-steady rolling performance data of the roll target material; and a change step of changing the target steady rolling condition so that the estimated rolling constraint condition satisfies a predetermined condition.