南京理工大学学报（自然科学版）
南京理工大學學報（自然科學版）
남경리공대학학보（자연과학판）
JOURNAL OF NANJING UNIVERSITY OF SCIENCE AND TECHNOLOGY
2014年
3期
337-341
,共5页
改进粒子群优化算法%模糊径向基函数神经网络%退火炉%温度控制%径向基函数%权值%阀值%超调量%响应时间%稳态误差
改進粒子群優化算法%模糊徑嚮基函數神經網絡%退火爐%溫度控製%徑嚮基函數%權值%閥值%超調量%響應時間%穩態誤差
개진입자군우화산법%모호경향기함수신경망락%퇴화로%온도공제%경향기함수%권치%벌치%초조량%향응시간%은태오차
improved particle swarm optimization algorithm%fuzzy radial basis function neural network%annealing furnaces%temperature control%radial basis function%weights%thresholds%overshoot%response time%steady state errors
为提高对具有大滞后，强耦合的退火炉温度控制系统的控制精度，采用模糊径向基函数(RBF)神经网络控制炉温，并采用改进粒子群优化(PSO)算法进行优化。利用模糊推理过程与RBF 神经网络所具有的函数等价性，统一系统函数。在利用改进 PSO 算法对模糊 RBF 神经网络进行训练时，先利用改进 PSO 算法得到模糊 RBF 神经网络的初始权值和阀值，然后对其进行二次优化得到最终的权值和阀值。仿真结果表明：该文方法降低了超调量，缩短了响应时间，稳态误差很小，能够拟合参考模型的输出，控制效果明显优于常规 PID 控制。
為提高對具有大滯後，彊耦閤的退火爐溫度控製繫統的控製精度，採用模糊徑嚮基函數(RBF)神經網絡控製爐溫，併採用改進粒子群優化(PSO)算法進行優化。利用模糊推理過程與RBF 神經網絡所具有的函數等價性，統一繫統函數。在利用改進 PSO 算法對模糊 RBF 神經網絡進行訓練時，先利用改進 PSO 算法得到模糊 RBF 神經網絡的初始權值和閥值，然後對其進行二次優化得到最終的權值和閥值。倣真結果錶明：該文方法降低瞭超調量，縮短瞭響應時間，穩態誤差很小，能夠擬閤參攷模型的輸齣，控製效果明顯優于常規 PID 控製。
위제고대구유대체후，강우합적퇴화로온도공제계통적공제정도，채용모호경향기함수(RBF)신경망락공제로온，병채용개진입자군우화(PSO)산법진행우화。이용모호추리과정여RBF 신경망락소구유적함수등개성，통일계통함수。재이용개진 PSO 산법대모호 RBF 신경망락진행훈련시，선이용개진 PSO 산법득도모호 RBF 신경망락적초시권치화벌치，연후대기진행이차우화득도최종적권치화벌치。방진결과표명：해문방법강저료초조량，축단료향응시간，은태오차흔소，능구의합삼고모형적수출，공제효과명현우우상규 PID 공제。
In order to improve the control accuracy of temperature control systems of annealing furnaces with large time delay and strong coupling, the temperature of annealing furnaces is controlled by a fuzzy radial basis function ( RBF) neural network and optimized by an improved particle swarm optimization(PSO) algorithm. The system functions are unified using the function e-quivalency of the fuzzy inference process and RBF neural network. The initial weights and thresholds of the fuzzy RBF neural network are obtained by the PSO algorithm, and the final weights and thresholds are obtained by quadratic optimization when the fuzzy RBF neural network is trained by the improved PSO algorithm. The simulation results show that the method proposed here decreases the overshoot,shortens the response time,and the steady state error is small,which can fit the outputs of the reference model and is better than common PID control in control effects.
