Abstract: In order to make up for the precocious convergence of the expected parameters of the standard genetic algorithm(SGA) for the probability integration method, it is easy to fall into the disadvantages of local optimal solution in the later stage. A variety of group genetic algorithms(MPGAs) were proposed to invert the parameters of probability integration, and the accuracy and reliability of the algorithm were studied. The simulation experiment shows that the inversion model of probability integral method based on MPGA can not only accurately obtain the expected parameters, but also have strong anti-jamming ability for random error, coarse difference and monitoring point loss in the station data. The probability integral method test shows that: in the case that MPGA only iterates for 57 times before convergence, while SGA iterates for 100 times before convergence, the fitting standard deviation of the subsidence value and horizontal movement value obtained by MPGA is 31 mm, and the fitting standard deviation of the subsidence value and horizontal movement value obtained by SGA is 32 mm.