Research On Algorithm Of Compound Control System For Intercept Missile

Intercept maneuverable tactical ballistic missiles (TBM) is an importantorientation of the world military field. At a high altitude environment, it is far frombeing enough to get an accurate engagement if the interceptor missile is controlled bytraditional aerodynamic control. Instead, the compound control with lateral thrust andaerodynamic force for advanced interceptor missile is a key technique of guidance andcontrol in the terminal phase of interception. It is also the most effective way toimprove maneuverability and respond more rapidly to meet the new character ofmodern war. Therefore, this article analysed the problem of compound control systembased on the character of interceptor missile, and the following are the main researchs:The six-degree-of-freedom (6DOF) trajectory models of terminal interception areestablished. By analyzing the characteristics of normal pneumatic layout, the modelsof attitude control and the trajectory control jets engine, the configuration of pulse jetengine in attitude control is determine. According to analyzes the trajectory models,reasonable assumption and simplification, the nonlinear model of the lateral thrust andaerodynamics combined control system of the interceptor missile are constructed.Then the compound control law with dynamic inversion method is designed. In viewof the aerodynamic parameters of strongly nonlinear and the coupling between thecontrol channels, dynamic inversion method with the state feedback based on time-scale separation is designed. Because dynamic inverse control requires accurate model,the method used a neural network with an adaptive element to account for the inverseerror. But because of the effects of actuator saturation, the NN will wrong adapt to thecharacteristics. Pseudo-control hedging (PCH) was introduced to reduce the level andduration of actuator saturation.The improved particle swarm algorithm is used to optimize the back-propagation(BP) neural network (NN). Because of the disadvantages of BP-NN in its easily beingtrapped into local maxima, sensitive to the initial value and slow convergence speed, particle swarm optimization (PSO) was introduced to optimize the parameters of theBP-NN. Meanwhile, the inertial weight was regulated dynamically by the decayedexponential function to enhance the performance of the PSO. According to the abovemethods and theories, the system simulation model of pitch channel is constructed andthe performances of blended control law are verified by numerical simulations in theMATLAB/SIMULINK environment. The simulation results prove that thecompensation of the inverse error is effective and the new method avoids the neuralnetwork of local optimization and improved the learning efficiency.