多机电力系统附加NNPSS的在线学习神经网络逆励磁控制器  

Online learning ANN-inversion excitation controller with NNPSS for multi-machine power system

作　　者：徐庆宏[1] 戴先中[1] 

机构地区：[1]东南大学自动化学院复杂工程系统测量与控制教育部重点实验室,江苏南京210096

出　　处：《电力自动化设备》2010年第1期25-31,共7页Electric Power Automation Equipment

基　　金：国家自然科学基金资助项目(60574097)~~

摘　　要：为提高发电机机端电压和转速的综合控制性能,设计了附加神经网络电力系统稳定器(NNPSS)的在线学习神经网络逆(OLANNI)励磁控制器。针对多机系统同步发电机组模型,根据逆系统方法得到发电机励磁系统的逆系统的表达形式,并通过离线训练得到发电机励磁系统的神经网络逆系统。借鉴传统的AVR/PSS控制方法,并考虑到其对电力系统不确定性的自适应能力的不足,在离线训练的基础上分别设计了自适应的OLANNI、NNPSS以取代传统的AVR、PSS,给出了基于在线梯度算法的OLANNI和NNPSS的在线学习算法,并根据Lyapunov稳定性理论证明了OLANNI和NNPSS在线学习的收敛性。将设计的控制器应用于一个典型的2区域4机系统,仿真研究结果表明:在系统遭受扰动时,所设计的控制器较AVR/PSS和OLANNI控制器具有更好的综合控制性能。To improve the comprehensive control performance of generator terminal voltage and rotor speed, an OLANNI（Online Learning ANN-Inversion） excitation controller supplemented with NNPSS（Neural Network Power System Stabilizer） is designed. For the synchronous generator set model of multi-machine power system,the inverse system expression of excitation system is deduced using inverse system method,and its ANN-inversion system is derived by offline training. Motivated by the conventional AVR/PSS control scheme and with the consideration of its bad adaptability to the uncertainties of power system, the adaptive OLANNI excitation controller and NNPSS are respectively designed based on offline training,to replace the conventional AVR and PSS. An online learning algorithm based on the online gradient descent method is proposed for the OLANNI and NNPSS,and its convergence is proved by Lyapunov stability theory. Case study is fulfilled for a typical two-area four-machine power system. Simulation results show that the proposed controller has better comprehensive control performance than both AVR/PSS and OLANN[ when the controlled system suffers disturbances.

关 键 词：神经网络 逆系统 在线学习 励磁控制 神经网络电力系统稳定器 

分 类 号：TM571[电气工程—电器] TM712