The invention discloses a high-dimensional random matrix-based electronic type transformer error state evaluation method, and aims to perform error state evaluation by means of determining whether statistical distribution of the measurement error of the electronic type transformer has abnormal changes or not based on the high-dimensional random matrix only according to the output of the electronic type transformer, without depending on a standard tool or a physical model. The error state evaluation method specifically comprises the steps of establishing the random matrix through the output data of the electronic type transformer based on a sliding time window; expanding the random matrix based on a Kalman filter; solving a non-Hermitian matrix; performing evaluation on matrix solving; and performing evaluation index calculation on a sample central moment and a sample original moment to carry out the electronic type transformer error state evaluation. Effective evaluation on the error state of the electronic type transformer can be carried out under the premise of not depending on the standard tool or the physical model; and the evaluation result is accurate and visualized, and the evaluation method is universal, effective and can be realized easily.