Abstract:

In order to solve the problem that the identification accuracy of DC arc faults is not high under different working conditions,a gray wolf optimization least squares support vector machine (GWO-LSSVM) is proposed to diagnose DC arc under multi-load conditions.Firstly,the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) is applied to perform the intrinsic mode function (IMF) decomposition on the DC arc current signals obtained from the mixed load of the reference high-speed railway station under the different operating conditions.Secondly,the relevant components are screened and combined with multi-scale permutation entropy (MPE) to construct the feature vectors.Finally,in response to the slow convergence speed of the diagnostic model and the tendency of the model to fall into the local optima,the LSSVM model optimized by GWO is applied for the fault state recognition.The experimental results show that the accuracy reaches 98.33%.By comparing with other algorithms,the efficiency of the proposed method has been confirmed.

Key words: DC fault arc, multi-scale permutation entropy, grey wolf optimization (GWO) algorithm, fault diagnosis