CAJ | 학술논문

要对系统进行准确、及时的控制,就必须精心设计有效的控制方法.在控制领域,最常用的是PID控制方法.这种控制方法能够有效地提高被控系统的稳定性,加快系统对输入的响应,以及能减小静态误差.但是,P、I、D三个环节相互影响和制约,它们之间的关系很复杂,并不是简单的线性关系,再加上实际系统的非线性和时变不确定性,三个参数的人工整定比较困难.BP(back propa-gation)神经网络理论上可以逼近任何非线性函数,将它与传统的PID控制方法相结合可以达到良好的控制效果.通过采用BP神经网络,以闭环反馈系统的误差作为神经网络的学习误差,可以实现PID控制器参数的自适应整定.
요대계통진행준학、급시적공제,취필수정심설계유효적공제방법.재공제영역,최상용적시PID공제방법.저충공제방법능구유효지제고피공계통적은정성,가쾌계통대수입적향응,이급능감소정태오차.단시,P、I、D삼개배절상호영향화제약,타문지간적관계흔복잡,병불시간단적선성관계,재가상실제계통적비선성화시변불학정성,삼개삼수적인공정정비교곤난.BP(back propa-gation)신경망락이론상가이핍근임하비선성함수,장타여전통적PID공제방법상결합가이체도량호적공제효과.통과채용BP신경망락,이폐배반궤계통적오차작위신경망락적학습오차,가이실현PID공제기삼수적자괄응정정.
To correctly control a system in time you must design effective control method. In the control province, the most often-employed control method is PID-control method. It can efficienfly improve the stability; quicken the response to input and decrease the static error of the controlled system. But the three factors P, I and D, which are proportion, integral and differential, influence and restrict each other. Their connections are so complex that simple linear connections do not work. In addition to the non-linear and uncertainty of the practical system, it is hard to set the three parameters manually. BP-Neural-Net can tbeoreticaUy approach any non-linear function. It can gain good control effect to combine BP-Nerve-Net with conventional PID control method. Although the traditional study mode of BPNerve-Net is with-mentor-study we can realize without-mentor-study while we make use of the closed-loop and feedback structure.