| On the base of the former model, the article specially mends the shortcoming of the influence factor and influence degree of each factor to the combustion stability, structure and arithmetic of neural network. The mends are based on large academic research and uniting the experience of spot expert at the same time, and also using statistics and analytic method, multiplayer feed-forward neural network and improved BP arithmetic. By data processing and design of training set, the article chooses learning sample of neural network simultaneity. Comparing with the former system, the improved system can better reflect the practical changing tread of the single influencing factor and the combusting steady; the on-line evaluation of the combusting steady comes true; the modeling results are of agreement with the data from the power plant. Therefore, the system can instruct the operator purposefully and designedly, guarantee the boiler combusting steady and realize the optimal operating fashion. |
PDF Full Text Request