摘要
针对于在非线性、时变和强耦合特性的特性下,PID控制效果不佳,以三自由度直升机模型系统作为研究对象,提出了粒子群优化神经网络控制器来实现对系统的控制。神经网络可以通过训练和学习自适应的调整参数,算法采用梯度学习法,初始权值随机产生。产生的权值在学习过程中能陷入局部最优值,因而粒子群算法优化得到的最优初始权值带入神经网络PID控制器能取得满意的效果,控制不仅迅速逼近控制目标,而且响应时间较短。仿真实验对比证明,粒子群优化神经网络PID算法具有优于PID算法的适应能力、鲁棒性和响应速度。
Taking three dof helicopter model system as the research object,or the characteristics of nonlinear, time-varying and strong coupling, PID control cannot work well, we proposed the particle swarm optimization neural network controller to realize control of the system. Neural network can be trained to learn, the parameters of the adaptive learning algorithm using gradient method, the initial weights randomly generated. The weights in the learning process can be trapped in local optimal value, the algorithm optimizes the initial weights into the neural network PID controller can obtain satisfactory result, the algorithm not only fast approaching control target, but the response time is shorter. Comparison of the simulation experiment proved, particle swarm optimization neural network PID algorithm is better than PID algorithm at the adaptability and robustness and response speed.
出处
《哈尔滨理工大学学报》
CAS
北大核心
2017年第4期13-17,共5页
Journal of Harbin University of Science and Technology