Research On Fuzzy Model Of Grinding Classification Control System

Mineral separation means removed from the original ore and gangue contains harmful elements, which is useful to enable accumulation of mineral or minerals to make a variety of symbiotic mutual separation. In these operations, we can get one or several products which are useful minerals. Mineral separation operations include crushing and screening, grinding and classification, sorting, concentrate dehydration aspects. The process of grinding is the continuity of crushing mill and the last process of the mineral selection. The purpose of grinding is to make a useful component of ore in all or most of the monomer to achieve separation, while at the same time try to avoid over grinding situation, and the purpose of selecting the size of other requirements for sorting ore recycling effective in a useful component of create the conditions.Mechanism of grinding process is complicated, which is more than one variable input and output systems. The production processes is slow and lagging a long time, at the same time which is nonlinear time-varying characteristics and interference characteristics of many factors. These features bring a lot of difficult to system control and optimization. The traditional industrial control is often necessary given precise mathematical model, but the grinding is a typical non-linear process, so the traditional control method of the results is unsatisfactory.With the development of the intelligent control theory and the extensive application of high-performance computers, the new control strategy was introduced to the industrial process control. Fuzzy control is a rule-based fuzzy inference system, which is based on fuzzy set theory and fuzzy rules and fuzzy inference calculate model, whose fuzzy logic is a description of the proposition. Artificial neural network has the ability of simulating the human brain structure and function, which is a large number of processing units connected by a wide range of network systems. With self-organizing and self-learning ability, artificial neural network is particularly suitable solution taking into account the necessary lot of factors and conditions, the imprecise and vague information processing problem. Expert system is an intelligent computer system of substantial expertise and experience, which is able to simulate the thought processes of experts in the field, so that the computer looks like a smart as human experts to solve practical problems.Clearly known characteristics and control objects of grinding process and based on the use of intelligent control technology, thesis design and implement grinding classification operations intelligent control system. This thesis pay close attention to the process of grinding and classification for the fuzzy inference system modeling process, and makes a new method to solve the right to adjust the value of fuzzy rules which is based on genetic algorithm.First of all, the paper analyzes the complexity of the grinding classification in details, and has known optimization of the key and difficult goal of grinding classification. With these analyzing, thesis proposed a series of coping strategies and designed control system for grinding and classification of the overall structure. Using fuzzy logic variables express the relationship between the coupling; fuzzy rule base used to set up historical data extraction based on cluster analysis, the expert experience of a complementary manner; designed based on fuzzy rule base of fuzzy control module as the core system control module; overall the use of expert system architecture on the way, the prediction and control of the extension of contracting to resolve the grinding of large time-delay system problem.This paper focused on modeling the process of fuzzy control. Fuzzy model is mainly built up by the knowledge base and fuzzification and fuzzy solution and fuzzy inference module. As a target, grinding process is analyzed on conduct data and feature extraction, which is established the number of input and output variables and membership function type and related parameters; target data through cluster analysis of samples set up fuzzy control rule base; set up for the final grinding process fuzzy control model. Using of Matlab Fuzzy Toolbox tectonic-style command-line fuzzy controller to implement logic operations functions, through the use of COM component technology with the Matlab implementation. NET platform mixing language programming, and introduce a cross-platform programming matters they should pay attention.In optimizing process of grinding system, there may appear to run in the same state, a number of rules to control its guidance, there may appear to run in the same state, a number of rules to control its guidance, these rules overlap between cross and even paradoxical situation. Solution to this problem is a fuzzy controller is initialized for one of the fuzzy weights, and adjust the weights of fuzzy rules for the purpose of regulation is to achieve the overall optimum, we need to adopt a global optimization method to avoid local solution. Genetic algorithms are based on natural selection and natural genetic principle of the search algorithm, its advantage lies in its robustness and easy GM, genetic algorithm global search characteristics to enable it to efficiently find the global optimal solution or near optimal solution.This paper presents an innovative method of adjusting the weights of fuzzy rules based on genetic algorithm, which is solving the fuzzy rule applies to the priority problem. Related research work focused on two aspects:First, setting up neural network in order to analog computing functions of fuzzy controller. M language of the rules of fuzzy controller weights are solidified in the "black box operation", that is, after the completion of the design of fuzzy controller should not randomly adjust the rules in the controller weights. To the language of M, adjusting the value of the fuzzy rule right, we must re-fuzzy module compile, this approach obviously can not be a means of adjusting fuzzy rules weights. Therefore, this article designed fuzzy neural network, fuzzy logic was putted into the neural network, so that the neural network has the function of logic. fuzzy rules as a neural network hidden node will be the right value of fuzzy rules as the rules After the pieces of the activation function embedded in neural networks, when the right to try another value of the control sequence, the updated neural network hidden layer's activation function can, Do not need to change the network structure, as flexible and realize the right to adjust the value of fuzzy rules without the need the overall change in the purpose of computing modules.Second, neural network as a computing module, the right of the value of the chromosome sequence as a genetic operation. Using real-coded on chromosome coding; according to the operator the number of groups of settings, initialization chromosome groups; to teachers approaching the target signal for the stability of the design of adaptive function; using roulette wheel selection genetic algorithm, in accordance with the initial set up set value of the crossover and mutation of chromosomes, the optimal sequence weights. In this paper, grinding and classification of man-machine interface control system is introduced. The using method of adjusting value of fuzzy rules is showing.Research of this paper is aimed at stability control of grinding and classification, off-line adjustment of the core of the fuzzy controller performance. The focus of future work will be allowing the system to self-regulating and self-study, be able to achieve intelligent control of grinding and classification Optimize adjust online.