论文部分内容阅读
针对激光制导炸弹滚转控制通道的时变特性，基于神经网络具有定性和定量多模态控制能力能够实现多个常规控制器的功能，融合炸弹投放过程中的多种工作状态参数信息，设计了非线性神经网络控制器，给出神经网络控制器与常规控制器的功能等价性分析。该控制器具有鲁棒性，能适应时变系统参数大范围的变化，而且方法简单，实现容易。利用该方法对某型激光制导航空炸弹进行仿真，并与炸弹的变结构控制器相比，从根本上解决了变结构控制器的抖振问题，结果表明，该神经网络控制器具有良好的控制性能。
Aiming at the time-varying characteristics of the laser-guided bombs’ rolling control channels, based on the neural network’s qualitative and quantitative multi-modal control ability, it can realize the functions of many conventional controllers and integrate the information of various working parameters in the process of bombs dropping. The nonlinear neural network controller gives the functional equivalence analysis of neural network controller and conventional controller. The controller is robust and can adapt to a wide range of time-varying system parameters, and the method is simple and easy to implement. Using this method to simulate a laser-guided air bomber, and compared with the bombshell variable structure controller, this paper fundamentally solved the chattering problem of the variable structure controller. The results show that the neural network controller has good control performance.