Control System Design, Based On Neural Networks Simple Guided Bombs

The task of the flight control system is to stabilize the bomb angle movement and flight along with the scheme ballistic trajectory until hitting the target. The control signal is formed by comparing the practical ballistic trajectory with scheme ballistic trajectory with control law, then the rudder is driven accordingly so that the bomb flight to the target. From the point view of the task and characteristics of the control system of guided bomb, the following topics are studied in the paper:Firstly, by studying the longitudinal and lateral dynamics of the guided bomb, the perturbation equations, and thus the transfer functions are derived.Secondly, the variation characteristic of coefficients of transfer functions with the variation of flight parameter in the entire course of the entire flight course is analyzed. The traditional PID controller parameters on the lattices of the flight parameter field are proposed, so that preparation for controller design based on neural network is fulfilled in order to overcome the shortage of fight control with single set of controller parameter in the entire flying course of the bomb.Thirdly, the principle and training algorithms of BP neural network are introduced. In view of the slow convergence rate and part convergence, several improved training algorithms of BP are imported, and several BP neural network algorithms are studied with simulation method.Fourthly, Utilizing the approximation of neural network to nonlinear functions, the controller adaptive to flight parameter variation in considerable large field is designed with BP neural network, the simulation is proposed to demonstrate its effectiveness.