Research On Key Technologies Of Intelligent Control Of The Whole Process Of Electrolytic Aluminum Production

China is the world's largest producer of electrolytic aluminum,and it is expected that the country's electrolytic aluminum output will reach 40 million tons in 2020.The full process control of aluminum production is the key process of electrolytic aluminum production,which mainly includes the two major links of aluminum production and aluminum distribution.At present,there are problems of extensive management of aluminum production and aluminum distribution control,which rely too much on manual experience.For this problem,this thesis has completed the following three aspects of work:First of all,a detailed systematic analysis of the production process of electrolytic aluminum is made,the process flow is described,and the information flow and logistics flow in production are clarified.On this basis,a comprehensive analysis of the aluminum and aluminum distribution of electrolytic aluminum is provided,which provides process explanations and scene descriptions for follow-up research and introduces the importance of aluminum and aluminum distribution.At present,the aluminum production plan for electrolytic production is mainly based on the technical personnel using the daily basic data to determine the aluminum production plan for each tank based on experience.Each manufacturer has different experience calculation methods.There is no unified calculation model for aluminum output,and the changes of operator will affect the production stability.In view of this,this thesis analyzes historical empirical data,systematically analyzes the data correlation of production process parameters that may affect the development of aluminum production tasks,and screens out six parameters including the most influential aluminum amount,cutting times,aluminum level,electrolyte level and bath temperature.The prediction model of aluminum output with 6 input nodes and 1 output node are established and then trained with MLP neural network algorithm to learn the knowledge contained in the excellent artificial experience and realize the aluminum output Computer intelligence decision.Aiming at the subsequent aluminum production process of electrolytic aluminum,an aluminum distribution optimization algorithm is proposed,with the shortest aluminum extraction path of the crown crane as the objective function,the limitation of the immune clone algorithm in solving the aluminum distribution problem,and the convergence of the basic immune clone algorithm is analyzed.Aiming at the problems of the slow convergence speed and easy localization of the local immune cloning algorithm,the idea of using simulated annealing algorithm to optimize it is proposed.At the same time,a direct gradient split strategy is adopted for high-impurity electrolytic cells to improve the performance of aluminum distribution algorithm.Aiming at the above work,the actual data of the electrolytic aluminum factory was used to complete the calculation test.The results have proved the effectiveness of the method in this thesis.The method in this thesis provides a reference scheme for the intelligent control of the whole process of aluminum electrolysis.