Adaptive inverse control adopts the method in signal processing to deal with the problem in control, it is a kind of new idea in obtaining satisfied system response and restrain disturbance. However, when control nonlinear object, adaptive inverse control need create object model and object inverse model. It is infeasible to use linear adaptive filter to create nonlinear model. In contrast, the neural network displayed superiority obviously in this aspect. The neural network has the capability to approximate arbitrarily non-linear mapping, after trained we obtain forward model and reverse model.Particle swarm optimization is a population-based, self adaptive search optimization technique. As a kind of intelligent algorithm, it can be used to solve various optimization problems and shows great potential in practice. Now, it has been widely applied in many other areas, such as artificial neural network and fizzy system control. Study indicated that particle swarm optimization algorithm is a potential algorithm in training neural network. It reserve of based on genus swarm, the collateral global hunting strategy, the model of velocity-displacement adopted is simple to operate, and easy to achieve. We build nonlinear double inverted pendulum system based on s filtering LMS of the adaptive inverse control in this paper. The inverted pendulum is a typical experiment platform to study the theory and methods of control system. This paper will study the capability of the particle swarm optimization algorithm through comparing some different particle swarm improvement algorithms. The paper will integrate the neural network and particle swarm algorithm. With using the best result of particle swarm algorithm as the starting value of the BP, the training time and training efficiency can be improved greatly.This experiment show that adaptive inverse control has the good capability to resolve the deeply non-linear problems, and adaptive inverse control system can restrain the perturbation. |