高超音速飞行器的神经网络跟踪控制

针对最小相位高超音速飞行器非线性系统的调节器方程无法求出解析解的情况,提出利用三层神经网络来近似输出反馈控制器中的前馈函数的方法,得到基于神经网络的输出反馈控制器可以解决高超音速飞行器系统的跟踪控制问题的结论.

Neural network tracking control of hypersonic vehicle

The longitudinal dynamics model of a hypersonic vehicle is a complex nonlinear system. Since its regulation equation cannot be accurately solved, the tracking problem of this system is usually difficult. A recent research uses six neural networks to approximate the solution to the regulator equation of the hypersonic vehicle system. Notice that the feedforward function in the tracking controller relies only on the external reference signal, it can be approached directly instead of finding the approximation of the regulator equation, and hence, the computational complexity can be significantly reduced. By doing so, we first prove that the regulator equation of hypersonic vehicle has a local solution, and then design the output feedback controller in which the neural network is used to approximate the feedforward function. It is shown that the tracking control problem of hypersonic vehicle can be solved accordingly and the the calculation efficiency can be improved.