Chaotic Signal Pattern Recognition Using Orthogonal Wavelet Packet Method 2002-01-1400
In evaluating mobile hydraulic system health conditions, vibration signals are commonly collected for supporting system diagnosis and/or prognosis. However, the collected vibration signals are often presented in chaotic forms, and result in difficulties in performing sensitive and accurate diagnosis/prognosis. This paper proposes an orthogonal wavelet packet method to extract featured signals in support of fault diagnosis. This method can effectively distinguish the periodic signal, the chaotic signal, and the random noise based on energy distribution in the signal spectrum. Evaluation results indicate that the orthogonal wavelet packet method is capable of supporting system fault diagnosis based on a spectrum of chaotic signals.