In order to solve the problem that the speed and stability of maglev ball system with traditional controller are easy to be disturbed, a RBF neural network supervisory controller based on cloud adaptive particle swarm optimization (CAPSO) was established. After learning and tuning the output of PD controller through RBF neural network, the three parameters of RBF network were normalized and dynamically optimized by using cloud adaptive particle swarm optimization algorithm. The original RBF neural network gradient descent method, particle swarm optimization algorithm and cloud adaptive particle swarm optimization algorithm were used in the contrast experiment respectively, and then the contrast control simulations were carried out.The results show that the optimized hybrid control algorithm can realize the adaptive control of the magnetic levitation ball system, and the dynamic performance and steady-state performance of PSO-RBF algorithm are improved.