基于神经网络的风洞马赫数预测控制仿真研究

针对2.4 m跨声速风洞很难用精确的机理模型表示系统的动态特性的问题,提出了基于神经网络模型的风洞马赫数预测控制策略.综合了模型预测控制和神经网络建模的优点,对于控制参数未知、非线性和时变系统具有很好的处理效果.利用基于径向基函数的神经网络模型预测系统的动态响应、非线性神经网络模型可以在训练过程中捕获系统的动态特性等措施,实现了将神经网络模型应用到MPC结构中.仿真结果表明,该控制策略具有很好的跟踪性能和控制效果.

Predictive Control Simulation Research of Mach Number in Wind Tunnel

It is hard to use precise mechanism to describe system dynamic feature of 2.4 m transonic wind tunnel. Put forwards wind tunnel Mach number predictive control strategy based on neural network. Combine the advanteges of model predictive control and nueral network modeling, it is good at processing control parameter unkown, unlinear system and time varing system. Use dynamic response of nueral network based on radial basis function and nonlinear neural network to capture system dynamic feature, apply nerual nwork model in MPC structure. The simulation results show that the control strtegy has a good control effect and trace performance.