CAJ | 학술논문

针对传统的PID控制算法参数整定困难、控制效果不理想,一般BP神经网络算法收敛速度慢,容易使运行系统局部陷入极小及稳态时震荡等缺点,将神经网络算法与PID控制法相结合,引入动量因子自适应学习速率算法改进PID神经网络,并在Matlab中进行了实时仿真训练。仿真结果表明,此控制策略在被控对象未知或参数变化的情况下,具有很强的适应性和鲁棒性,相对于其他PID和神经网络调整参数方法,本算法有其实用性。
침대전통적PID공제산법삼수정정곤난、공제효과불이상,일반BP신경망락산법수렴속도만,용역사운행계통국부함입겁소급은태시진탕등결점,장신경망락산법여PID공제법상결합,인입동량인자자괄응학습속솔산법개진PID신경망락,병재Matlab중진행료실시방진훈련。방진결과표명,차공제책략재피공대상미지혹삼수변화적정황하,구유흔강적괄응성화로봉성,상대우기타PID화신경망락조정삼수방법,본산법유기실용성。
The traditional PID control algorithm has difficulty in tuning satisfactory controlling effect. While a general BP neural network has parameters, leading to an un- a low convergence rate and is prone to fall into a local minimum and unsteady state. To solve the aforementioned problems, this pa- per combines the BP neural network algorithm and PID control by introducing an adaptive learning method for the momentum factor to improve the PID neural network. As well, a real-time training is simulated with MATLAB. Simulated results show that the proposed BP-PID controller has stronger anti-interference ability and more robustness in the ease of unknown controlled targets and changing parameters. Compared with other PID controllers or parameters adjustment methods with neural net- work, the proposed model has better practicality.