Abstract: A fault noise signal monitoring model of axial piston pump is established by using non-contact acoustic array, and an intelligent fault identification method is proposed based on CNN-SVM model. Firstly, the signal model of subarray translation is used to filter the signal, and the time-frequency graph sample is generated by continuous wavelet transform (CWT). RGB picture is synthesized by multiple subarrays as the fault acoustic signal sample. Secondly, the fault identification model of multi-subarray acoustic signal fusion based on CNN-SVM is established using SVM replaced Softmax classifier. Finally, four kinds of faults such as plunger fault, plate fault, swash plate fault and return plate fault are designed and verified by experiments. The results show that the classification accuracy of the proposed method reaches 97.5% in the running noise environment, which is 1.1% higher than that of the single channel time-frequency sample.