文档介绍：同步发电机近似动态规划励磁控制的研究摘要同步发电机稳定运行是电力系统安全的基本条件之一。传统的控制同步发电机的方法是对线性化后的系统进行有效的控制,但对于同步发电机这种非线性、动作快的系统,这种传统的控制方法难以对同步发电机进行精确的控制。为此,本文研究了一种以人工神经网络为实现工具,结合动态规划与强化学习的综合控制算法,即近似动态规划算法,对同步发电机的励磁系统进行控制。近似动态规划采用非线性函数拟合逼近动态规划的性能指标的方法以获得最优解或次优解,可以有效的避免“维数灾’’问题。本文首先阐述了近似动态规划和同步发电机及其励磁系统在国内外研究的现状。其次,简单介绍了同步发电机及其励磁系统和电力系统稳定器的结构及工作原理。本文的重点部分是将近似动态规划的两种算法,执行依赖启发式动态规划算法和二次启发式动态规划算法分别应用于同步发电机励磁系统的自动调压器和电力系统稳定器中,达到对同步发电机控制的目的。在Matlab/ Simulink环境下的仿真结果表明,基于近似动态规划算法的控制效果优于模糊控制的控制效果,优于传统控制的控制效果。而基于近似动态规划算法的电力系统稳定器同样优于传统电力系统稳定器。关键词:近似动态规划同步发电机励磁系统仿真 RESEARCH OF APPROXIMATE DYNAMIC PRoGRAMMING ABOUT SYNCHRONoUS GENERATOR EXCITATION CONTRoL ABSTRACT The steady operation ofsyncllIIonous generator is one ofthe assurances on power system securi吼As the generator wim characteristics of nonlineari够and quick mobili劬the仃aditional availably con仃ol baSed on linear equation can’tdeten】[1ineme precise mathematical t11isp印er,artificial neural ne铆ork aS t11e implementation tool, combined wim dyn锄ic progr姐lllling andreinfo:rcementlea玎血g,is proposed. Approximate Dyna戚cProgr锄ming (ADP) uses non—linear丘mction fittmg performance ofdyIl锄ic pro掣ammir培method to searCh suboptimal solution or omimal solution,wKch c锄e虢ctiVely avoid ‘‘curse of dimensionali哆”brought bydyn锄ic programming. Firstly,me status ofresearCh athome and abroad of ADP and synchronous generator and excitation system ,吐le蛐rIlcture a11dwo瓶ng principle ofthesynchronous generator,excitation system arei砷沁duced. Automatic voltage regulator and po、№r System stabilizer inexcitation system of synchonous generator are controlled respectiVely by ADP (Action—DependentHe面sticDyn锄ic Progra棚ming and Dual He嘶stic Dyn锄ic Programming), in order to achieve me control of synchronous generator,which is tllefocus aIld core of resultsshow excitation syStem of syncllronous generator controlled based on ADP issuperior t0 conVentional control and如zzy systems讪ilizerbased on ADP is superior toconventional con订01. KE