Abstract:
The operating conditions of the valve of the high-pressure diaphragm pump are complex, and the vibration signal is in the characteristics of non-linear and non-stationary, which makes it difficult to extract the signal characteristic information and the fault state to be difficult to identify. In order to extract the non-linear dynamic characteristics of the operating state of the one-way valve and improve the recognition accuracy and generalization ability of the fault diagnosis model, a method based on Multi-scale Permutation Entropy (MPE) and Random Vector Functional Link network (RVFL) was proposed. Firstly, the vibration signal of the check valve which was collected under the working condition was decomposed by Variational Mode Decomposition (VMD) to obtain certain Intrinsic Mode Function (IMF) components. Then, the multi-scale permutation entropy of IMF component was calculated, and the eigenvalue vector representing the operation state of one-way valve was constructed. Finally, based on the eigenvalue vectors of the operating state, a regularized RVFL fault diagnosis model was established, which could be applied to the monitoring and identification of the operating state of one-way valves. The experimental analysis showed that the fault diagnosis model could accurately identify the fault type of one-way valve and the accuracy rate was 98.89%.