Research On Sliding Mode Guidance Law Based On RBF Neural Network

With the development of aerospace technology,the speed of intercepted targets is getting higher,the maneuvering is getting stronger and the disturbance is getting more.Traditional guidance laws and modern guidance laws are difficult to meet the requirements of antimissile interception in warfare.In addition,some specific operational missions have certain requirements for the attack angle at the interception end.Therefore,new guided weapons adapted to the complex combat environment of the modern battlefields are an important means for countries to gain war dominance.Sliding mode variable structure control has invariance for system parameter uncertainties and external disturbances,so it is used to design the guidance law,but the chattering caused by the variable structure switch term will reduce the dynamic quality of the control system.Radial Basis Function(RBF)neural network could learn the non-linear relationship between the data autonomously,and has the ability to optimize.It is gradually used in the design of control systems.The control system designed based on the RBF neural network and variable structure control not only retains the advantages of strong robustness of the variable structure,but also weakens the chattering of the variable structure system.At first,this paper establishes a guidance control simulation system based on dynamics and kinematics models of missile and target,then analyzes the characteristics of traditional guidance laws,and finally gives the guidelines for selecting guidance laws.Secondly,a variable structure guidance law with fixed gain of the switch term is proposed.The guidance law treats target maneuvering as disturbance,which can effectively intercept high-speed and large maneuvering targets,but during the interception,there is a problem that the line-of-sight angular rate chattering is not conducive to the normal operation of the on-board mechanism.The RBF neural network's powerful nonlinear problem processing ability is used to adjust the gain of the switching term.It is expected that the line-of-sight angular rate chattering will be weakened while ensuring the accuracy of interception.Thirdly,a variable structure guidance law based on RBF neural network boundary layer regulation is proposed,in which a saturation function is used instead of a symbolic function.On one hand,the smaller the thickness of the boundary layer,the better the control effect,at the same time,it will increase the control gain and increase chattering;on the other hand,the larger the thickness of the boundary layer,the smaller the chattering,in the meantime,it will reduce the control gain and worsen control effect.In order to obtain the best control effect,the RBF neural network is used to adjust the thickness of the boundary layer.Compared with the proportional navigation guidance and the fixed gain variable structure guidance law,the two proposed guidance laws based on RBF neural network adjustment can effectively reduce the system chattering and improve the interception accuracy of the missile.Finally,considering the constraint of the attack angle when the missile hits the target,a sliding mode guidance law based on RBF neural network with attack angle constraint is proposed,which effectively weakened the chattering of the line-of-sight angular rate.Compared with fixed gain variable structure guidance law with attack angle constraint,the proposed guidance law can not only intercepts the target accurately,but also weakens the system chattering.