aiming at the defect of single feature selection and low accuracy of fault identification in traditional OLTC fault diagnosis, a high dimensional collection of time domain, energy and composite multi-scale permutation entropy is constructed from the perspective of multi-feature fusion