基于Elman神经网络系统辨识柴油机调速算法研究  

Study of Advanced Control Based on the Elman Neural Network Indentification Theory for Diesel Engine Speed Control

作　　者：姚崇[1] 龙云 马子焜 YAO Chong;LONG Yun;MA Zi-kun(College of Power and Energy Engineering,Harbin Engineering University,Harbin 150001,China;Shanghai Marine Diesel Engine Research Institute,Shanghai 201108,China)

机构地区：[1]哈尔滨工程大学动力与能源工程学院,黑龙江哈尔滨150001 [2]中国船舶集团有限公司七一一研究所,上海201108

出　　处：《控制工程》2021年第6期1122-1129,共8页Control Engineering of China

摘　　要：为了提高柴油机的瞬态调速性能,基于Elman神经网络辨识理论,提出了一种新的柴油机自适应调速算法,即Elman-PID控制算法。该算法运用神经网络提取柴油机转速的雅可比信息,通过转速的雅可比信息反映负载扭矩变化的大小,在线调整PID控制参数,使调速控制能更好地适应柴油机外部负载突变的情况,增加神经网络收敛速度,减小网络对初始权值的依赖。同时,在柴油机整机控制策略中,加入了网络在线学习判定规则,避免运行过程中过度学习、稳态控制性能恶化等问题。最后,建立柴油机平均值模型,验证该算法的实用性。In order to improve the transient control performance of diesel engine speed control,based on the Elman neural network identification theory,a novel adaptive speed control algorithm for diesel engine is proposed in this paper,namely Elman-PID control algorithm.The proposed algorithm extracts the Jacobian information of the diesel engine speed through the neural network,reflects the change of the load torque and adjusts the PID control parameters online by using the Jacobian information.The algorithm can better adapt to the sudden change of the diesel engine external load,which increases the convergence speed of the neural network and reduces the dependence of the network on the initial weight.At the same time,in the diesel engine control strategy,the network online learning decision rule is added to avoid the problems such as over-learning and deterioration of the steady-state control performance during the operation of diesel engine.Finally,the mean value models of diesel engine is established,and the practicability of the algorithm is verified.

关 键 词：柴油机 ELMAN神经网络 系统辨识 调速控制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]