The invention discloses an evidence theory-based electromechanical product simulation model precision improvement method. The method comprises the following steps: 1, carrying out a small amount of test measurement on an electromechanical product, carrying out statistical analysis on measured response data, and analyzing the measured response data by utilizing an evidence theory; 2, establishing amechanical and electrical product simulation model, obtaining a calculation response, and establishing an objective function for identifying an electromechanical product modeling parameter interval range; 3, further dividing the identified modeling parameter interval into a plurality of sub-intervals, and preliminarily allocating a certain modeling parameter probability to each sub-interval; and4, establishing an objective function for identifying a modeling parameter quantum interval probability value by utilizing the electromechanical product simulation model and the allocated sub-intervalprobability, and identifying accurate probability allocation on the modeling parameter quantum interval through an optimization process. According to the method, the value interval and probability distribution of the modeling parameters can be accurately obtained under the small amount of measurement response, and the precision level of the electromechanical product simulation model is effectively improved.