Research on the Wavelet Neural Network Pattern Recognition Technology for Chemical Agents

• Main Authors: Minghu Zhang, Dehu Wang, Shijun Lv, Enzhong Quan, Shaojie Chen, Yingsong Li
• Format: Conference Proceeding
• Language: English
• Subjects: analysis and simulation; Artificial neural networks; Chemicals; Feature extraction; model design thought; Pattern recognition; Training; Wavelet analysis; wavelet neural network; Wavelet transforms
• DOI: 10.1109/ISME.2010.189

Aims at the characteristics of the chemical agents signals, the pattern recognition method based on the wavelet neural network is put forward. Firstly, the model design thought of the wavelet neural network pattern recognition technology for chemical agents are analyzed, which shows that the wavelet analysis has good time-frequency features due to its capability of localizing and differentiating the high and low frequencies of a signal and keeping the time domain features of the original signal, as a result, the wavelet transform can effectively extract the feature of the chemical agents signals; Secondly, the model and learning algorithm for the wavelet neural network are constructed, which implements that the pattern recognition method for the chemical agents combines with the advantages of the wavelet and neural network, the wavelet transform method as a fore processing medium is used to extract the feature which reflects the information of the chemical agents, and the features are fed into the neural network as the input patterns for training and classifying, to achieve the intelligence distinguishing; Lastly, the results of the examples and simulated test show that: this method is workable, and of the high identification accuracy, remarkable generalization capability, good stability, and high speed and high reliability. At the same time, this method has the general application value.