Rolling Load Prediction Using Temperature Proxy Inputs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current rolling load prediction models struggle to accurately predict rolling loads due to the difficulty in measuring steel temperature and limited operation record data, resulting in suboptimal rolling quality and efficiency.
Innovation Solution
Incorporating temperature-related factors, such as skid rail temperatures and steel slab temperatures, into neural network models using deep learning to improve prediction accuracy, allowing for more precise input variable selection and enhanced model training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional rolling load prediction models are used without temperature data, then the model complexity remains low, but the prediction accuracy deteriorates due to inability to capture temperature-related rolling load variations
Solution Approach 1:
The patent applies preliminary action by measuring and recording temperature data (steel slab temperature, skid rail temperature, ambient temperature) in advance during the rolling process. These temperature parameters are captured before being input to the neural network model, allowing the model to utilize temperature information for more accurate rolling load prediction without requiring complex real-time temperature measurement systems during the actual prediction phase.
Solution Approach 2:
The patent uses a neural network model as an intermediary that processes temperature data and other rolling parameters to predict rolling load. The neural network acts as a mediator between the measured temperature parameters and the final rolling load prediction, transforming multiple input parameters including temperature into accurate load predictions without requiring direct complex physical relationships to be modeled.
2Measurement precision
If temperature measurement systems are added to capture steel temperature, then prediction accuracy improves, but the difficulty of detecting and measuring increases due to the challenging nature of steel temperature measurement
Solution Approach 1:
The patent uses skid rail temperature as an intermediary parameter to indirectly estimate steel temperature. Instead of directly measuring the difficult-to-access steel temperature, the system measures the skid rail temperature which is in thermal contact with the steel slab, and uses this as a proxy input variable in the neural network model to capture temperature-related effects on rolling load.
Solution Approach 2:
The patent replaces direct mechanical/physical temperature measurement of steel (which is difficult due to high temperature and moving conditions) with indirect thermal measurement through skid rails and other accessible components. The neural network model substitutes direct steel temperature input with multiple indirect temperature parameters that are easier to measure and still capture the essential temperature effects.
3Reliability
If more operation record data including temperature is collected for training, then the neural network training quality improves, but the data collection complexity and time required increase
Solution Approach 1:
The patent applies universality by using a multi-functional data collection system that simultaneously captures multiple parameters (rolling speed, force, temperature, ambient conditions) during normal rolling operations. This approach allows the system to collect comprehensive training data for the neural network without requiring separate dedicated measurement campaigns, as the same sensors serve multiple data collection purposes during routine production.
Data Source
AI summary
A rolling load prediction method predicts a rolling load of a rolling mill for rolling steel and includes predicting the rolling load of the rolling mill in a case where the steel is rolled under an operating condition for prediction, by inputting the operating condition for prediction into a rolling load prediction model that has been trained with operation record data including at least a factor related to a temperature of the steel as an input variable and an actual value of the rolling load of the rolling mill as an output variable.