毕业论文：单神经元自适应PID控制.doc

摘 要 在现代工业控制领域，PID控制以其原理简单，鲁棒性好，可靠性高等优点被广泛应用。PID控制的各个参数是为特定的被控对象而整定的，在系统模型参数变化不大的情况下,其控制性能优良，操作简单，能满足工程控制的要求，然而在工业实际控制中，被控对象往往还具有高度的非线性，不确定性和参数时变等特点，在这种情况下单纯依靠PID控制是不能达到要求的。神经网络控制系统作为新兴发展的智能控制系统，能很好地解决上述问题，已经在很多领域得到应用，同时也显示了它的优越性。 单神经元作为构成神经网络的基本单元，具有自学习和自适应能力，且结构简单而易于计算。基于以上分析，在PID控制中引入单神经元自适应算法，同时采用有监督的hebb学习规则来设计单神经元自适应PID控制器，这样既可以解决传统PID控制器的不足，同时又能充分利用PID控制技术成熟的优势。通过仿真对比发现单神经元自适应PID控制器比传统的PID控制器控制效果好。 关键词: 单神经元自适应控制； PID控制；神经网络；有监督的hebb学习规则 Abstract As the advantages of simple algorithm,good robustness ans stability,the PID controller is widely adopted in modern industrial fields.Becsuse of the parameters of the PID controller are set for the fixed objiect,when the system model changes very little,the performance of PID controller is very good,while the conventional PID controller can not performance very well if the system model is highly nonlinearity ,uncertainty and the parameters change usually,immediately those will lead to the conventional PID controller can not achieve good results.Neural network as one of developing intelligent controllers can solve the problem that conventional PID controller can not overcome,now the nerral network has been used in industrial fields and the control effects are satisfied. The single neuron as the basic element of neural network has ability of self-study and self-adaptive,At the same time ,the structure and algorithm of single neuron is also very simple ,immediately the calculation is not very complicated.Base on the analysis above,now I make a new controller combine the conventional PID controller algorithm and single neuron adaptive algorithm named single neuron adaptive PID controller ,I choose monitor hebb learning as the study rule .In this way,the single neuron adaptive PID controller has advantages both conventional PID and neural network.Simulation and experiment results show that the control effects of single neuron adaptive PID controller are much better than conventional PID controller. Keywords: single nerron