There is influence of edge overlap among disturbed typical features on complex disturbance identification due to the limitations of feature extraction methods. Thus a multi-kernel support vector machine identification algorithm based on multi-domain feature optimization is proposed. First, a variety of feature extraction methods are used to obtain the complex perturbation multi-domain typical features. Secondly, in order to consider the correlation between high-dimensional features and target categories and the normalization of the measurement scale, an improved maximum correlation minimum redundancy criterion is used to select the key feature subset for identification, and then the multi-kernel SVM with radius information is used to identify the complex disturbance waveform. The simulation results show that the proposed algorithm can overcome the influence of spatial ambiguity of complex disturbance on identification accuracy, is less affected by noise and has good stability. This work is supported by the Science and Technology Project of the Headquarters of the State Grid Corporation of China (No. 52130421000S).