A Pso-based Neural Network Ensemble with the Application of Flame Combustion Diagnosis

In thermal power station, pulverized coal furnace is widely used. It requires the chamber of furnace holding steady uniform flame, and ensureing that strong full combustion. This study has developed a neural network ensemble model to perform the judgement of combustion diagnosis based on the spectral distribution of the light intensity pulse signal of the flame. Compared with the single neural network, the two-stage integrated model of neural network ensemble, based on Bootstrap and electoral cooperative particle swarm optimization, can found the internal relations among inputs and outputs according to the learning of internal rules, and weaken the human factors in the weights determination. Based on the experiments on real scenario, the results show that the proposed model outperforms all the compared ones in perspective of the convergence speed of total error, and also obtains stable classification effect.