Research on Fault Diagnosis of a Certain Launch Vehicle Based On RBF Neural Network

Abstract

In view of the multi-fault frequently occurring and cannot be efficiently diagnosed in dynamic system of a certain launching vehicle chassis, this paper analyzes the produce mechanism and sets up the experiment combining with the practical use. In this experiment, vibration sensors are installed in the appropriate location to collect vibration signals in different state and wavelet packet energy spectrum feature information is extracted. The diagnosis of multi-fault of dynamic system is attained by using the ability for RBF neural network to learn, train and test the feature information. Diagnosis results show that the wavelet packet energy spectrum can effectively describe the multi-fault information of the chassis of the launching vehicle and RBF network diagnosis technology is effective and high-accuracy.