In order to solve the problems of dramatic changes for loads, high energy consumption and low efficiency when a boom-type roadheader cut coal and rock in random distribution, using genetic algorithm is used to control fuzzily the boom swing speed to achieve constant power control of the cutting motor, a simulation model of the boom horizontal swing system was established, and simulation of the model was done with Matlab. Results show that the speed fluctuation of the system is reduced by 25.9%, the power overshoot of cutting motor is decreased by 5.0%, adjustment time is decreased by 0.1s compared with the traditional one, which indicates that the system has faster response, higher accuracy, as well as good control characteristics, it can automatically adjust the swing speed of roadheader according to different coal and rock, reduce the power loss of the machine and improve the working efficiency.