Steelworks Process Prediction Using Regression Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting and controlling steelworks processes, such as converter and electric arc furnace processes, face inefficiencies due to reliance on expensive and inaccurate measurement techniques, which fail to account for multiple influencing variables and system changes, leading to suboptimal production outcomes.
Innovation Solution
A method utilizing regression analysis and classification methods, specifically Support Vector Machines (SVM), to monitor and relate multiple input variables to target variables, enabling precise prediction and control of steelworks processes by preprocessing and dynamically adapting to changing conditions, incorporating both static and dynamic input variables.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sub-lances are used to determine temperature and chemical composition of the melt, then measurement precision is improved, but device complexity and maintenance costs increase
Solution Approach 1:
The patent replaces the mechanical sub-lance measurement system with a combination of acoustic sensors, optical sensors, and computational models. Acoustic emissions from bubble formation and optical measurements of flame characteristics are used to infer melt temperature and composition without physical contact, thereby eliminating the complexity and maintenance burden of sub-lances while maintaining measurement precision.
Solution Approach 2:
The patent introduces intermediate measurement parameters (acoustic emissions, optical flame characteristics, exhaust gas composition) that can be measured non-invasively and are correlated with the desired melt properties through regression models. These intermediates serve as proxies that avoid direct contact with the molten metal while still providing the necessary information for process control.
2Measurement precision
If multiple input variables are monitored using regression analysis, then prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The patent employs a multi-functional monitoring system where a single integrated platform collects and processes multiple types of data (acoustic, optical, thermal, chemical) simultaneously. The regression analysis framework is designed to handle various input variables (exhaust gas composition, temperature profiles, acoustic emissions) within a unified computational model, avoiding the need for separate specialized systems for each measurement type.
Solution Approach 2:
The system uses readily available process data from existing sensors and measurement devices, combining them with additional non-invasive measurements. The regression models are trained on historical process data that is continuously generated during normal operation, allowing the system to improve its predictive accuracy automatically without requiring additional complex infrastructure.
3Ease of operation
If static limit values are used for endpoint determination, then ease of operation is improved, but adaptability deteriorates
Solution Approach 1:
The patent replaces static limit values with dynamic threshold values that are continuously updated based on real-time process conditions and historical data. The regression models adapt to changing process characteristics by learning from ongoing measurements, allowing the endpoint determination criteria to evolve with the process while maintaining ease of automated operation.
Solution Approach 2:
The system implements feedback mechanisms where prediction results and actual process outcomes are continuously compared. This feedback loop allows the regression models to adjust their parameters and the endpoint thresholds to adapt to process variations, ensuring both ease of automated operation and adaptability to changing conditions.
4Manufacturing precision
If blowing process is extended to ensure target values, then manufacturing precision is improved, but productivity deteriorates
Solution Approach 1:
The patent uses regression models to predict the trajectory of carbon and phosphorus content reduction during the blowing process. By analyzing real-time data and projecting future states, the system can determine the optimal endpoint before the actual endpoint is reached, avoiding unnecessary extension of the blowing process while ensuring target values are achieved.
Solution Approach 2:
The patent replaces conservative mechanical process extension with intelligent prediction based on acoustic, optical, and chemical sensor data combined with regression analysis. This allows for precise determination of the optimal blowing endpoint, eliminating the need to over-blow to ensure target values are met, thereby maintaining productivity while achieving manufacturing precision.
Data Source
Figure 1
AI summary
The present invention relates to a method for predicting, controlling and/or regulating steelworks processes, comprising the steps of monitoring at least two input variables related to a target variable, determining the relationship between the at least two input variables and at least one target variable by means of regression analysis or classification methods, and using the determined target variable for predicting, controlling and/or regulating the steelworks process.