Experimental investigation on cavitation and cavitation detection of axial piston pump based on MLP-Mixer

• Published in: Measurement : journal of the International Measurement Confederation 2022-08, Vol.200, p.111582, Article 111582
• Main Authors: Lan, Yuan, Li, Zhijie, Liu, Shengzheng, Huang, Jiahai, Niu, Linkai, Xiong, Xiaoyan, Niu, Chenguang, Wu, Bing, Zhou, Xu, Yan, Jinbao, An, Siyuan, Lv, Jishuang
• Format: Article
• Language: English
• Subjects: Axial piston pump; Cavitation detection; MLP-Mixer; Vibration signal processing
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2022.111582

Hydraulic pump constitutes the key component in hydraulic power system and its fault diagnosis is of great importance. Among all types of pumps, the axial piston pump is widely used, but due to its high speed and high pressure working condition, its cavitation characteristics are not so good as those of other hydraulic pumps. Therefore, in this article, a new cavitation detection framework that contains both experimental investigations and numerical signal processing is proposed to detect cavitation intensity of the axial piston pump. Cavitation intensity of the axial piston pump has been divided into three stages: normal state, cavitation development state and cavitation severe state. The cavitation detection model based on MLP-Mixer is introduced to recognize cavitation intensity of the axial piston pump with given working conditions. The testing accuracy of optimized cavitation detection model is quite high with above 99% for all given working conditions. •A novel cavitation detection framework for axial piston pumps•The framework contains both experimental investigations and numerical analysis•The cavitation intensity has been divided into three stages•The detection model is based on MLP-Mixer•The model successfully recognizes cavitation intensity with given working condition