A fault diagnosis method for rotating machinery based on improved variational mode decomposition and a hybrid artificial sheep algorithm

Due to the non-stationary and nonlinear characteristics of rotating machinery fault signals, it is difficult to identify different fault conditions using only traditional time-frequency domain analysis approaches. In this paper, a combined method based on improved variational mode decomposition (IVMD) and a hybrid artificial sheep algorithm (HASA) is proposed for rotating machinery fault diagnosis. In the proposed method, the IVMD is used to decompose the signal into several modes, which can determine the mode number k and update parameter adaptively. Then, the HASA is utilized for feature selection and parameter optimization, where the binary-valued artificial sheep algorithm (BASA) is employed to select the optimal features and the real-valued artificial sheep algorithm (RASA) is used to optimize the parameters of a support vector machine (SVM). Moreover, the HASA effectively extends the optimization range of the decision variables by the compound use of the BASA and RASA. Two experimental cases, including rolling bearing faults and rotor system faults, have been implemented to test the performance of the proposed diagnosis method. The experimental results demonstrate that the proposed method can achieve better classification accuracies in practical applications.