Application and research of the genetic neural network with wavelet transform in the boiler water temperature modeling

As the boiler combustion is a complex physical and chemical process, it has greater delay, multivariable coupling and other complex non-linear characteristics; therefore, the BP neural network algorithm is selected for modeling and analyzing the boiler combustion process in this thesis. This method regards the combustion system as a black box, rather than make a concrete analysis of the internal mechanism of the combustion process. In this context, establish a reasonable model structure. As the BP algorithm has some shortcomings such as slow convergence rate and easily to fall into the local extremum, we choose the genetic algorithm to optimize the network and make full use of its global searching ability to optimize the weights and thresholds of the BP neural network and establish a reasonable GA-BP network model. This algorithm uses the historical operating data to training and testing the network. From the experimental results we found that using of genetic algorithm to optimize BP neural network can improve the model accuracy and the convergence rate, it makes the model do a better reflection on the characteristics of the boiler combustion.