Ultrasonic echo waveshape features extraction based on QPSO-matching pursuit for online wear debris discrimination

Abstract The ultrasonic echoes reflected from debris in lubricant contain a lot of useful information, which can represent the size, material and geometric characteristics of the debris. Our preliminary simulation investigations and physical model analysis results show that the waveshape features are feasible and essential to discriminate debris in lubricant. An accurate waveshape features extraction method of debris echoes is presented based on the matching pursuit (MP). The dictionary of Gabor functions, which is suitable for ultrasonic signal processing, is adopted for MP. To seek faster and more accurate calculation of MP, quantum-behaved particle swarm optimization (QPSO) is introduced to optimize the MP algorithm. The simulation and experimental results reveal that the proposed method can effectively extract the waveshape features of debris echoes and air bubble echoes. Utilizing the extracted waveshape features, the debris with different shapes and air bubble can be distinguished.