Aluminum Electrolytic Enterprises Production Benchmarking Optimization Algorithm Research

Aluminum electrolytic industry as the pillar industries for economic growth and social stability have a significant impact. However, aluminum electrolysis industry in the implementation of benchmarking of the main problems is the main benchmark based on past experience in artificial settings, such as the best, the best the world’s best companies, and other industries. Because these benchmarks required to produce a certain environment, uncertainty and chance, so we need to study in depth data mining and modern computer technology for production data analysis and forecasting, made for their own benchmark data.In this paper, in-depth benchmarking study optimization-related knowledge, including fuzzy clustering analysis, BP neural network and particle swarm optimization (PSO) and other related theories. Based on the theory of particle swarm optimization algorithm is proposed to improve optimization algorithm based on crossover and mutation APSO (MAPSO). Design and implementation of a set of standard management system, and the introduction of benchmarking MAPSO algorithm to optimize the selection. The main content and work is as follows:1. Learning data mining, artificial intelligence and other related theories. The use of fuzzy clustering analysis data preprocessing; by BP neural network training sample study, forecast production levels; combined PSO algorithm to optimize BP neural network structure, improve accuracy.2, improved PSO algorithm and put forward MAPSO algorithm. Introduced in PSO algorithm genetic algorithm selection, crossover and mutation mechanism, and defines the inertia weight adaptive function, improve PSO early fast convergence and slow convergence caused by the late local extreme properties greatly improved algorithm optimization capability.3, designed to achieve the standard management system, use MAPSO algorithm in the system analysis of enterprise historical production data to predict production benchmark. Select the system in accordance with the business needs of the standard elements of the establishment of the standard programs, including the standard units, benchmarking element and date. Analysis of historical data of selected benchmark standard feature, and to carry out benchmarking tables, graphs, and other ways to show results on the standard for enterprise management decision to provide a scientific basis.In this paper, the optimization algorithm based on particle swarm to be improved on, and combined with BP neural network algorithm proposed MAPSO. MAPSO algorithm analysis using production data, the establishment of benchmark optimization model, and applied to enterprise standard management process.