The Theory And Applications Of BP Neural Networks

Artificial neural networks (ANN) has been widely used in pattern recognition, signal and information processing, adaptive controlment, and many other fields because of its self-organize ability, self-adapt ability, robustness and fault tolerance, and which are coming from its parallel processing structure and distribute information storage. But at the same time, they also bring some problems such as the difficulty of getting optimized topology structure and the great training time-consuming, which restrict its reality use.This dissertation mainly stresses on the theory and applications of feed-forward neural networks (BP networks)1. Pre-knowledge applied to neural network weight initializing, training set making and using are being discussed in this dissertation. The results of the experiments have shown the importance of pre-knowledge applied to neural network.2. According to the characteristics of the license plate, we trained two kinds of modularized neural networks separately, integrate them under the fuzzy theory. The results of the experiments have shown that this method can recognize the characters of license plates quickly and accurately.3. According to the analysis of the characteristics of the aerial remote sensing hyper-spectral images' wave shape of the ocean red tide, we present a red tide analysis and recognition method based on the EMD and ANN. The comparative experiments proved that the method is effective. At the same time, the transitional water area can be detected by the information of the algae's thickness, which enables the forecasting of the red tide.