In order to monitor the deformation state of transformer winding

we proposed a method for state evaluation of transformer winding deformation based on information fusion.We proposed that the wavelet energy-matrix norm transformation method was employed to extract vibration feature

and the reactance identification method was employed to extract the reactance feature of transformer.Multiple information evaluation index system was built based on the matter-element theory

and weight coefficients of index were computed by adopting an information entropy method.Taking the type S–11–M–500/35 transformer as an object

we designed a multi-information collection system of transformer winding deformation state

experimentally studied the short circuit current impact

and designed three deformation fault windings.Moreover

we experimentally studied the state evaluation on loosing winding after short circuit current impact and deformation fault windings.Aiming at severity state winding

we proposed a comprehensive evaluation strategy of evaluation-diagnosis

and diagnosed fault types based on relevance vector machine.Results indicate that the evaluation method based on information fusion can evaluate the loosing and severe deformation states of winding effectively

the diagnostic accuracies on three deformation fault types of winding

including radial stretch and compression

axial insulation falling off

and end folding

are more than 92%.