Desgin On Stable Loop Of Missile Based On Neural Network Feedback Linearization Method

Bank-To-Turn (BTT) and Skid-To-Turn (STT) control techniques are widely used in current missile control systems. Bank-To-Turn control technology is a new missile's control technology and it has great meaning in improving missile's range, speed, accuracy and mobility. The characters of Bank-To-Turn control technology determine that the BTT missile's mathematical model contain a variety of coupling term. So the traditional three channel independent design autopilot's method can't meet the need of BTT missile's control. Searching an effective method for BTT missile's autopilot became an important topic to the scholars both at home and abroad.With the development of the nonlinear control theory and the neural network technology, feedback linearization became the main method to deal with the nonlinear system problem and meanwhile the neural network is good at dealing with the uncertain nonlinear system. This paper used the radiate basis function (RBF) and the dynamic inverse method which is one of the feedback linearization to deal with the nonlinear system problem design the BTT missile's autopilot. The main works of this paper are as fallows:First, mathematical model of the BTT missile is established. Because the model of BTT missile have the nonlinear, fast time-varying and strong coupling characteristics, when establishing the BTT missile's model, some simplifies and assumptions to be needed in that time. Considering the state variable is measurable, the three-channel mathematical model of the missile is established base on the information that been known.Second, dynamic inversion method is used in designing the BTT missile's autopilot. There are two methods to deal with the problem that the dynamic inversion method can't use in non-minimum phase system:first, input and output redefined method; second, time-scale separation. In order to use the dynamic inversion method on the BTT missile model, this paper applied the input and output redefined method to processing it firstly. At the end of this chapter proved the feasibility of this method.Third, for that the dynamic inverse method request accurate mathematical model but the mathematical model of BTT missile is uncertain, therefore the robustness of missile's stability loop which is designed by the method of dynamic inversion can't meet the requirement. This paper used neural network which has the characteristic of self-learning and adaptive design the dynamic inversion controller, then replaced the original controller by the new one which can effectively improve the robustness of the system. The results of simulation indicate that the missile's stability loop which is designed by neural network and dynamic inverse method have perfect performance in dynamic characteristics and tracking.