The invention provides a method which comprises the steps of (1) collecting the voltage effective value, current effective value, active power, reactive power and other data in a grid electric energy quality monitoring device after connecting a large scale electrified railway load, (2) carrying out preprocessing on the data which reflects electric energy quality, and constructing an original sample set, (3) using an improved Gauss hybrid model based on a Figueiredo-Jain algorithm with a combined operator to train the original sample set, and establishing an electrified railway traction load electric energy quality distribution characteristic Gauss mixed model, and (4) based on the above model, identifying and analyzing the real-time collected data which reflects an electrified railway traction load electric energy quality distribution characteristic, and obtaining a corresponding load category. According to the method, through carrying out wavelet denoising on the grid data after connecting the large scale electrified railway traction load, the data characteristics in different loads are extracted from massive data by using the probability statistics theory of maximum likelihood estimation, and the electrified railway traction load electric energy quality distribution characteristic research is realized.