计算机仿真
計算機倣真
계산궤방진
COMPUTER SIMULATION
2014年
5期
419-422,431
,共5页
比例积分微分控制算法%神经网络%网络控制系统%丢包
比例積分微分控製算法%神經網絡%網絡控製繫統%丟包
비례적분미분공제산법%신경망락%망락공제계통%주포
PID control algorithm%Neural network%Network control system%Data dropout
基于BP神经网络整定参数的PID控制算法有着广泛的应用，然而现有的研究主要针对控制器和被控对象在同一位置的点对点系统。实际中网络控制系统被广泛采用，由于通信网络的存在使得控制系统的数据经常发生丢失现象，对控制性能造成很大影响。针对上述问题，将数据丢包描述为一个随机的伯努利序列，在此基础上给出了存在数据丢包的神经网络PID控制算法。仿真结果表明，当控制系统存在一定的数据丢失时，神经网络PID控制算法仍然可以保证系统的稳定性，但输出性能随着数据丢失程度的增加变差。
基于BP神經網絡整定參數的PID控製算法有著廣汎的應用，然而現有的研究主要針對控製器和被控對象在同一位置的點對點繫統。實際中網絡控製繫統被廣汎採用，由于通信網絡的存在使得控製繫統的數據經常髮生丟失現象，對控製性能造成很大影響。針對上述問題，將數據丟包描述為一箇隨機的伯努利序列，在此基礎上給齣瞭存在數據丟包的神經網絡PID控製算法。倣真結果錶明，噹控製繫統存在一定的數據丟失時，神經網絡PID控製算法仍然可以保證繫統的穩定性，但輸齣性能隨著數據丟失程度的增加變差。
기우BP신경망락정정삼수적PID공제산법유착엄범적응용，연이현유적연구주요침대공제기화피공대상재동일위치적점대점계통。실제중망락공제계통피엄범채용，유우통신망락적존재사득공제계통적수거경상발생주실현상，대공제성능조성흔대영향。침대상술문제，장수거주포묘술위일개수궤적백노리서렬，재차기출상급출료존재수거주포적신경망락PID공제산법。방진결과표명，당공제계통존재일정적수거주실시，신경망락PID공제산법잉연가이보증계통적은정성，단수출성능수착수거주실정도적증가변차。
The BP neural network based PID control algorithms have been widely applied in many systems. How-ever, the existing applications only consider the point-to-point control systems, and the controllers and plants are as-sumed at same position. In practical systems, networked control systems are widely introduced. Due to the failure of communication channel, data dropouts often occur in networked control systems, which results in worse output per-formance. This paper described the dropout rate as Bernoulli random variables, and then the BP neural network based PID control was given in the framework of data dropouts. Simulation results show that the control algorithm can guar-antee the stability of systems even though some data are missing, and the output performance becomes worse as the in-creasing of data dropout rate.
