Hot Rolling Line Material Prediction Using Classified Regression Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hot rolling line control systems face high computational load and reduced prediction accuracy when predicting materials for multiple steel grades, as they rely on collecting and regressing data close to the input values, leading to decreased accuracy for data far from the collected range and excessive data storage requirements.
Innovation Solution
A hot rolling line control system with a classification criteria creation and material model regression unit that classifies and regresses operation and material measurement data to create a material model, using a classification-property learning approach to predict materials while managing calculation load, by focusing on the gradient between data and explanatory factors rather than data proximity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data close to the input value is collected and regressed for material prediction, then prediction accuracy for nearby data is improved, but prediction accuracy for data far from the collected range decreases and calculation load increases
Solution Approach 1:
The patent segments the continuous data space into discrete classification categories based on the gradient between explanatory factors and data. By dividing the prediction space into distinct classes and creating separate regression models for each class, the system achieves accurate predictions across the entire data range without requiring excessive computational resources for any single prediction task.
Solution Approach 2:
The patent transforms the prediction approach by changing from direct value-based proximity matching to gradient-based classification. Instead of measuring similarity based on absolute data values, the system uses the gradient (rate of change) between explanatory factors and target data as the classification parameter, which enables more efficient and accurate predictions across diverse data ranges.
2Measurement precision
If data close to the input value is collected for regression, then local prediction accuracy is improved, but data storage requirements and complexity increase
Solution Approach 1:
The patent segments the data storage requirement into classification categories rather than storing all possible data points. By organizing data into discrete gradient-based classes, the system reduces storage complexity while maintaining prediction accuracy, as each class contains representative data that captures the essential characteristics of that gradient range.
Solution Approach 2:
The patent extracts the essential feature (gradient between explanatory factors and data) from the complex data set and uses this extracted feature for classification. This extraction approach simplifies the data storage structure by focusing on the most discriminative characteristic rather than storing and processing all raw data dimensions.
3Measurement precision
If classification criteria are created and data is classified before regression, then prediction accuracy for multiple steel grades is improved, but system complexity increases
Solution Approach 1:
The patent changes the classification parameter from traditional value-based thresholds to gradient-based categorization. This parameter transformation simplifies the classification logic by using a single, meaningful gradient metric rather than multiple complex thresholds, thereby improving multi-grade prediction accuracy while minimizing the increase in system complexity.
Solution Approach 2:
The patent implements a dynamic classification system where the classification criteria are based on the gradient relationship between explanatory factors and target data. This dynamic approach allows the classification structure to adapt to different data patterns and steel grades automatically, improving prediction accuracy across multiple grades without requiring rigid, pre-defined complex classification rules.
Data Source
AI summary
A hot rolling line control system includes a rolling condition setting unit, an operation data collection unit that collects rolling conditions and operation data of a line during rolling, an operation data storage unit, a material measurement data storage unit that stores material actual measurement data obtained by measuring a material of a rolled steel sheet, a material prediction unit that predicts material of rolled steel sheet, and a material prediction data storage unit that stores material prediction data in the material prediction unit, and the material prediction unit includes a classification criteria creation and material model regression unit that creates classification criteria using operation data and the material actual measurement data, classifies the operation data and the material actual measurement data according to the created classification criteria, and regresses the classified operation data and material actual measurement data to create a material model for each classification.