Abstract: The accurate evaluation of the error status of the protection measurement circuit can help to understand the operation status of the protection devices, and to timely and effectively find the hidden dangers of the protection devices. This paper proposes an error status evaluation method for substation protection measurement circuit based on improved ED and CM-BPNN algorithms. Firstly, the normalization principle of different index data is formulated for the measurement data. Secondly, the improved Euclidean distance is introduced as the starting criterion for error evaluation. According to the degree of influence on the performance of the protection action, the error status is graded, and the cloud model is introduced to calculate the error status membership degree of the secondary index. Finally, the positioning mapping results corresponding to each error status are determined, and the BP neural network model is constructed, which is then used to evaluate and locate the unknown error status of the current/voltage of the protection devices. Based on PSCAD software, a 220kV substation model was constructed to verify the effectiveness of the proposed method.