Research On SVM Breakout Prediction System Based On Differential Evolution-grey Wolf Optimization Hybrid Algorithm Optimization

With the continuous casting technology keeps improving and developing,the high drawing speed as the core of high efficiency continuous casting technology has become the main research direction in the field of continuous casting technology.The improvement of billet pulling speed in the increase of production efficiency at the same time,triggered by the leakage of steel accidents occur more frequently,and the form of its occurrence presents a more complex trend.The continuous casting leakage is one of the most serious accidents in the process of steel production,once it happens,it will bring huge losses to the enterprise.Therefore,in order to curb the occurrence of steel leakage accidents,this paper develops a set of steel leakage forecasting system with fast response speed,good forecasting effect and high generalisation ability.The main work of this paper is as follows:First of all,after briefly analysing the reasons for the formation of various types of steel leakage accidents,it is determined that bonded steel leakage as bonded steel leakage as the main object of research,and targeted analysis of its causes and formation process.After summarising and comparing several main means commonly used in detecting bonded steel leakage,it is confirmed that the thermocouple temperature measurement method,which has the highest reliability and the widest application,is the main detection means of this research,and the continuous casting steel leakage forecast method based on the thermocouple temperature measurement method is obtained to provide a theoretical basis for the subsequent development of the steel leakage forecast system.Secondly,due to the actual production,the leakage data samples are limited,which is not enough to adequately train the BP neural network model,the support vector machine model leakage prediction model is established,and the Grey Wolf optimization algorithm has the advantages of high efficiency,fast convergence,and high accuracy,which improves the optimality searching speed of the continuous casting leakage prediction model.In order to solve the problem that the grey wolf optimization algorithm falls into local optimal solution due to the decrease of population variability during the iteration process,the differential evolution algorithm is introduced,which influences the initial population generation of grey wolves and maintains the variability of the population during the iteration period,forming a hybrid algorithm of differential evolution and grey wolf optimization,which solves the problem of the algorithm falling into local optimal solution.Combined with the actual data of the steel plant,the accuracy of the model was verified to be as high as 99.5%.Then,after analysing the structural characteristics of the crystallizer of a domestic steel plant’s continuous casting machine,a simplified assumption is established,and a three-dimensional model of the temperature field of the simplified crystallizer copper plate is built in the software.After calculating the heat transfer from each surface,the corresponding boundary conditions are imposed on the temperature field model,and the temperature field 3D model is solved.The temperature field model is verified to be accurate and usable after solving.Finally,a visualised continuous casting leakage forecasting system is established by programming the software and the secondary development function of the software.Determine the functions and structure required by the system,the forecast model and visualisation interface in the first two chapters of the secondary development,complete the system interface design,and get the structure and function of the complete visualisation of the continuous casting leakage forecasting system.And the offline test of the visualised continuous casting leakage forecasting system was carried out to verify the accuracy of the system.