LF Refining Furnace Intelligent Steelmaking System Design

Based on the research background of the 120tLF refining furnace control system of a steel mill,this thesis describes the research and design process of the electrode control system.In the refining production process,operators only rely on experience to adjust,which will lead to long adjustment time,untimely control,low accuracy and large energy consumption.The electrode system is a complex and changeable,strongly coupled,nonlinear,time-varying system,which accounts for the largest proportion of energy consumption in the refining process.Therefore,the production mode and electrode control system of LF refining furnace are studied,the problems affecting energy consumption are analyzed,the problems affecting large disturbance and slow adjustment speed in electrode control are analyzed and compared,the optimal electrode working point in the electrode control system is designed,and the LSTM predictive control method is proposed to improve the accuracy,work efficiency and reduce energy consumption of electrode control.The main work of this thesis is as follows:(1)For the problem of large power consumption caused by poor electrode working point,this thesis explains the characteristics of each working point of the electrode by analyzing the power characteristic curve of the LF refining furnace,the working method of the typical working point has a single nature,which can not meet the requirements of many aspects of actual production,and this thesis designs the optimal electrode working point suitable for this thesis in combination with the actual production needs.Due to the complex variability of the electrode system,this thesis analyzes and compares the electrode control method,adopts the constant impedance control method suitable for this thesis,and establishes the relationship equation between arc length and resistance value on this basis,and establishes a mathematical model according to each link of the electrode system.(2)Because the disturbance of the electrode system is very serious,especially in the stage of descending the electrode at the beginning of refining,the disturbance is particularly serious during the control process of the electrode arcing to stable work in the slag layer,and the control effect is not ideal by using the traditional PID control method.Aiming at the fact that the electrode system is a kind of nonlinear system,this thesis proposes a neural network predictive control method to solve the electrode control problem,and uses LSTM network,SVM and ELM network to carry out online direct multi-step prediction and recursive multi-step prediction,and verifies the effect of LSTM network recursive multi-step prediction by simulation.Comparing the LSTM predictive control method with the traditional PID control method,the simulation results show that the LSTM predictive control method has better control effect,and has good anti-interference and fast adjustment ability,which verifies the feasibility of the method.(3)According to the actual situation and process requirements,based on the Lab VIEW software development platform,the electrode control method proposed in this thesis is realized by writing a program.This thesis realizes the real-time recording,monitoring and intelligent control of electrode information data in the whole production process of LF refining furnace,and proves through tests that the control method proposed in this thesis has high precision,fast adjustment speed and low energy consumption,which solves the practical problems in steelmaking production.