New Method Of Modeling And Optimal Control For The Boiler Combustion System In Large Power Plants

With the rapid development of society and economy, the demand of automatic control system to the power industry is increasing greatly. The demands of the generation enterprises can't be satisfied only with the securely and steady production. Energy conservation and the efficiency have become the targets of power enterprises reformation in our country. However, at present the working conditions of most domestic thermal power plant boilers are complicated and variable. The practical working status differs greatly from what designed originally. It is incapable for modern power plants to meet the need of economical management only by traditional manipulating regulations and control methods.According to the requirement of modern power industry, this paper proposes a new optimal control method for the boiler combustion system, based on the study of neural network and optimization theory. And this method is applied to the 1150t/h boiler of 1 unit in Fuzhou power plant of Huaneng Inc by simulating. The off-line simulation indicates that this method can follow on the characteristic of boiler combustion to find out the optimal point-wise of working.The main contents of the research in this paper are as follows:1. Production process and control demands of large units are introduced, and the process of boiler combustion and its control system are introduced in detail as well.2. Based on the study of the basic theory of neutral network, the feasibility of the application of RBF neural network in nonlinear system modeling is analyzed.3. The optimal theory used in this paper and its engineering significance are briefly introduced and the SQP algorithm for optimal calculation and the process of realization with MATLAB are introduced in details.4. Based on uniform design and the cluster theory, a optimum selecting method of sample is proposed to solve the problem of sample selection.5. Based on the present control experience, let the boiler operate on a sub-optimal status first, and then seek out the better. In this way, the problem of local minimum can be avoided effectively while optimizing.