The Research On Power Plant Boiler Performance Diagnosis And Optimization System

As the electricity market competition and the growing national emphasis on energy conservation, it has become an urgent need for thermal power units to improve the boiler combustion system performance to achieve the purpose of energy saving and make the thermal power units have a lot of competition in an invincible position.The boiler combustion system has many complex characteristics,such as multi-variable, strong coupling, strong interference and large delay. Because of this, the paper presents a multi-objective optimization approach of using neural network to establish the boiler combustion model and using genetic algorithm to optimize the input parameters of neural network model and obtain optimal input parameters combination. The number of experiment samples is so large that there are a lot of redundant information in them. The paper uses the fuzzy clustering method to streamline the original sample set, discard useless information for the final classification of the sample conditions, to ensure the neural network prediction accuracy while reducing the training time. Before the optimization, we combine the actual operation of the boiler with a combination of fuzzy comprehensive evaluation method to establish the performance of the boiler combustion system diagnostic model. Finally,the paper finishes optimization of poor conditon by Genetic Algorithms.It constructs the fitness function based on the result of the fuzzy comprehensive evaluation model, which makes multi-objective optimization problem transformed into a variety of objective optimization problem solving to achieve the varioua kinds of optimal operation of the boiler combustion.The paper applied the theory above to the simulation and actual operation and develop the power plant boiler combustion optimization system. These results showed that the use of intelligent optimization algorithms to optimize the combustion system can achieve high efficiency, low milling unit consumption, low emissions and multi-objective optimization. It also can achieve the purpose of saving energy and reduce emissions.