Research On Case-Based Reasoning Optimization Control Of Pellets Roasting Process

With the incessant development of iron and steel industry, pellet has become an indispensable material for blast furnace and the grate-kiln production system has been widely used. In the process of production, to establish a rational grate-kiln pellet roasting process of thermal systems, and to achieve all of the grate temperature and the kiln temperature control is the key to guarantee the quality of pellet.The grate-kiln system is a multi-variable, strongly coupled non-linear system, human experience, often can not accurately adjust the wind valve opening and pulverized coal injection rate to control roasting temperature in the best value. Therefore, on the basis of In-depth analysis and study of the grate-kiln of pellet production technology process, this paper uses the basic idea of case-based reasoning to establish an optimal Control model of grate-kiln roasting process of pellets. It aims at increasing the grate-kiln roasting temperature control automation level, overcoming man-made temperature fluctuations, eliminating habitual violate compasses operation, realizing the grate-kiln roasting temperature balance stability control and eventually reducing the coal consumption, improving pellets quality and reducing manufacturing costs of pellets effectively.Firstly, this paper discusses methods and principles of the pellets roasting process optimization control case expression on the condition that the condition features is temperature difference value, temperature value in the each section and corresponding wind valve opening, and the decision features is the wind valve opening adjustment quantity and pulverized coal injection rate adjustment quantity, the case expression based frame is designed. The organizational method and the maintenance algorithms of the case base are researched, and then the case base of the roasting process optimization control is designed. Secondly, aiming at the case expression characteristics and organizational principles, this paper constructs a two class case base, case retrieval using the nearest neighbor strategy, the first representative of the case base to find the most similar representative case, then in the corresponding sub-case base further search. This can greatly reduce the search range of cases, improve the efficiency. After the analysis of the problems of the existing algorithms such as poor objectivity and high complexity, a method based on coverage for determining the case feature weights is proposed, which calculates the case feature weights by the affecting of each attribute to the case average coverage, and it improves the precise of the system retrieve. Thirdly, based on practical expertise, it designs the size of a case maximum similarity to select the method of case adaptation strategy to adapt to the grate-kiln pellets roasting process characteristics of the purpose, and proposes two kinds of self-learning mechanism. Lastly, simulation results show that optimal control of roasting process of case-based reasoning is feasible and the strategy can control the roasting temperature in the best condition.