论文部分内容阅读
针对传统四旋翼PID控制器参数整定困难和控制效果较难达到最优的问题,综合了传统PID控制器工程意义明确、参数整定简单以及神经网络的非线性映射和自学习的优点,构造了四旋翼飞行器神经网络PID(PIDNN)控制器。利用神经网络的非线性映射特点和自学习能力优化了传统PID控制器的控制效果,借助PID控制器的结构,解决了神经网络层数、节点数和连接权重初值选取困难的问题。同时利用自适应调整比例神经元加权系数,增加了系统的响应速度。最后,通过非线性全数值仿真验证了算法的合理性和有效性。
Aiming at the difficulty of parameter tuning and control effect of traditional quadrotor PID controller, the traditional PID controller has the advantages of clear engineering meaning, simple tuning of parameters and nonlinear mapping and self-learning of neural network. Rotorcraft neural network PID (PIDNN) controller The neural network’s nonlinear mapping characteristics and self-learning ability are used to optimize the control effect of traditional PID controller. With the help of the structure of PID controller, the problems of difficult selection of neural network layers, nodes and connection weights are solved. At the same time, adaptive proportional proportional neuron weighting coefficient is used to increase the response speed of the system. Finally, the rationality and effectiveness of the algorithm are verified by nonlinear numerical simulation.