The invention discloses a knowledge graph-based defect analysis method for transformer equipment. The method comprises the steps of receiving defect phenomena of a transformer and forming a component characterization matrix; obtaining all characteristic states related to components and constructing and forming an association matrix; extracting the defect phenomena related to the characteristic states and constructing and forming a refined association matrix; extracting all magnitude characteristics corresponding to the defect phenomena and constructing a magnitude characteristic matrix; calculating weight factors by adopting a softmax algorithm and initial intermediate variables and forming a weight matrix by adopting all weight factors; calculating a fault cause vector according to the weight matrix and the magnitude characteristic matrix and adding one to the number of iterations; when the number of iterations is smaller than an iteration threshold, updating the initial intermediate variables and then returning to a weight factor calculation step; and when the number of iterations is greater than or equal to a set threshold, calculating probability vectors, corresponding to the magnitude characteristic matrix, of fault causes by adopting the last calculated fault cause vector.