Abstract: Although the traditional probability integration method can invert the parameters more accurately， it has the disadvantages of high requirements for the type of working face and the layout of measuring station， large calculation workload and low efficiency.The intelligent algorithm provides a new method for accurately determining the optimal parameters of probabilistic integration model. In order to explore the performance of different intelligent optimization algorithms in the process of calculating the parameters of probabilistic integration method， operational procedures of the six models including modular vector method， genetic algorithm， culture-particle swarm algorithm， particle swarm algorithm， fruit fly algorithm and ant colony algorithm were written by using Matlab programming language. Through simulated tests，comparison and analysis of the above six algorithms are done from the aspects of algorithm inversion accuracy， stability， anti-error interference ability， global optimization ability and operating efficiency. The results show that when the initial value of the parameter is close to the true value， the inversion accuracy and efficiency of the modular vector method is the highest； when the initial value of the parameter differs greatly from the true value， the modular vector method will fall into a local optimal solution. The algorithm has the strongest inversion accuracy and stability. Considering the accuracy and efficiency of parameter inversion， when the parameter range is known， the optimal algorithm is the modular vector method； when the parameter range is unknown， the selection order of the optimal algorithm is culture-particle swarm algorithm， genetic algorithm， fruit fly algorithm， particle swarm algorithm， ant colony algorithm and modular vector method.

Key words: mining subsidence prediction, intelligent optimization algorithm, probability integral method, parameter inversion