Study And Project Implementation Of Metallurgical Crafts Expert System Based On BP Neural Network

In recent years, with the rapid development brings overcapacity in the steelindustry, and the rise in international and domestic ore prices of raw materials, compressthe metallurgical enterprise profit margin, the competition between enterprisesbecoming intensive. Therefore, in order to promote the competitive power, metallurgicalenterprises are increasingly concerned about the rate and comprehensive economicbenefits of using its production process resource.The research of metallurgical crafts expert system in this article is in favor of themetallurgical enterprises to optimize the actual production process, improve the rate andthe comprehensive economic benefits during the entire process of production. Theexpert system according to the characteristics of different aspects of metallurgyproduction, is divided into four functional modules, management of raw materials,ingredients optimization, granulating and sintering evaluation; the raw materialmanagement module stores basic information, various physicochemical information andpicture information of raw materials; ingredients optimization module using the linearprogramming method to fit the constraint condition of the chemical mixturecompositions and the particle size distribution, so we use the comprehensive cost asobjective function to optimize the proportion of mixture ingredients; Granulatingeffect prediction model is based on granulating evaluation module, BP neural networkand the concept of moisture capacity, so it is the best moisture content prediction model;sintering evaluation module using the balance between material and heat to calculate thebest carbon ratio during the sintering process, at the same time through the BP neuralnetwork comes the sinter performance prediction model which can be used to predictvarious physical and chemical indicators of sintered finished ore. The relevantparameters which calculate and predict by the metallurgical crafts expert system aremuch more accurate and flexible than the empirical formula or the parametersestablished by the metallurgical enterprise or the experts.This article is divided into three main parts:First of all, the background and significance of doing this research based on theanalysis of related domestic and foreign research, as well as the study of neural networkand linear programming.Secondly, analysis and research based on BP neural networks and linear programming theory, the application of BP neural networks and linear programming inmetallurgical technology blending, granulation and sintering processes in order tooptimize the entire process. Moreover, the corresponding algorithm and theestablishment of a metallurgical processing expert system model, giving some sampledata and algorithm of test results.Finally, designing and realizing the metallurgical processing expert system on thebasis of these studies.Currently, the expert system has come into service in several domesticmetallurgical enterprises. In the practical application, the expert system has optimizedthe parameters of ingredients’ proportion, granulating and sintering. And, from amacroscopic point of view, the system comes out better result from both the technicalindicators of entire process and the evaluation of economic benefit.