Study On Recognition Of Sub-health State Based On RBF Neural Network

Indentification of sub-health state uses the examination and analysis of pulse signal information to judge whether a person is in healthy condition.No matter from physical or from the heart, the sub-health will seriously affect the quality of life; therefore, how to effectively identify the state of health has become a active field. The paper uses a method which based on Radial-basis Function to achieve the identification of sub-health as the main line. Aiming the problem of pulse signal easy to introduce interference, the low accuracy and identify the effect of the feature extraction. At present, on the base of the common methods, the paper uses a systematic research method of wavelet transform, time domain and spectral analysis, to realize the identification of sub-health state. Then a new method is proposed, it can improve the recognition accuracy and precision.The following is done in this thesis:(1)For the pulse signal of low signal-to-noise ratio and easy to introduce the interference noise problems as well as the current feature extraction method of signal feature extraction poor. In this paper, the method based on wavelet transform of pulse signal preprocessing, reduce or eliminate the interference noise, and time domain analysis and spectrum analysis to achieve the pulse signal feature extraction. The experimental results show that this method can effectively realize the pulse signal processing and improve the accuracy of feature extraction.(2)It is proposed which based on Radial-basis Function to achieve the identification of sub-health state. In the specific identification process, when output layer weights selection, respectively by means of pseudo-inverse method, LMS method and based on tansig function method, and compared to the quality of these three methods, so as to get a more effective identification method. The experimental results show that this method has a high recognition rate for the identification of the pulse signal, and thereby achieve the identification of sub-health state which has a certain application value.