Fault diagnosis of mine main ventilator based on multi-eigenvalue selection and data fusion

Abstract

As a strong nonlinear complex system, the mine main ventilator is difficult to select eigenvalues of the ventilator fault. This study proposed a multiple eigenvalue selection method based on ensemble empirical modal decomposition (EEMD). Firstly, an experimental platform of axial flow fan was built to simulate the bearing fault and fan blade fault. Secondly, the vibration and wind pressure data of the above three faults and the normal state were collected, and the vibration signals were decomposed by EEMD. The normalized Root Mean Square Error was used to evaluate the signal separation degree of different faults under the same eigenvalue and to select the eigenvalue. Then, combined with wind pressure information, multi-data information fusion was used to construct the failure indicator vector table. Finally, the BP neural network was trained and tested with a vector table, and the fault diagnosis of the mine main ventilator was achieved. The experiment results showed that the diagnostic accuracy of test results was 98.87% after eigenvalue selection and data fusion. The method has high accuracy and adaptability, and is very suitable for mechanical equipment such as mine ventilator.