神经网络解耦控制在某无人机组件测试系统的应用

无人机组件测试系统，主要用于对某型号无人机组件进行相关性能测试和试验验证。系统的控制参数温度、压力、流量间存在强耦合关系。为了解决常规控制方法难以控制的难题，设计了一种基于PID神经元网络的前向多层解耦控制器，并利用遗传算法对神经网络的权值进行训练。将此算法在MATLAB下进行仿真，解耦控制效果较为理想，随后经航空测试系统试验验证，该控制方法可以达到设计需求，有力地支持了相关型号的研制工作。

Application of Neural Network Decoupling Control in a UAV component testing system

UAV component testing system, mainly for a large unmanned aerial vehicle components for performance testing and test validation. The control parameters of the system have strong coupling between temperature, pressure and flow. In order to compensate for the shortcomings of conventional control methods, a forward multi-layer decoupling controller based on PID neural network is designed, and the weight of neural network is trained by genetic algorithm. This algorithm is simulated under MATLAB, and the decoupling control effect is ideal. Then, it is verified by the Aviation product testing system. The control method can meet the design requirements and supported the development of the relevant model.