Synchronous generator parameters identification on-line using small population-based particle swarm optimization
Xiuge Zhang,Ye Ren,Q. Hu
2014-01-01
Abstract:The parameters of synchronous generators are the basis of power system analysis and operating control. Parameters identification of synchronous generator plays a key role for the power system stability analysis. In this paper, a small population-based particle swarm optimization (SPPSO) approach is used to acquire synchronous generator on-line model quickly and accurately. In the proposed approaches, three operations are introduced to improve the performance of the algorithm, namely mutation operation, DE-acceleration operation and migration operation. Furthermore, the synchronous generator practical model and the PMU data are adopted. The simulation results of the model obtained by SPPSO have been compared with hybrid genetic algorithm and PSO. The SPPSO algorithm shows better performance on the convergence as well as computation time and effort. Key words—PMU data,synchronous generator, parameter identification, SPPSO _____________________________________________________________________________________________ INTRODUCTION The system parameter identification (System parameter Identification) is a branch of modern control theory. After processing input and output data of the control system, it is to estimate the mathematical model of system structure. Power system is a huge artificial complex system, and the transient stability calculation and real-time control of system are closely related to the generator, synchronous generator and load model .So, it has a very realistic significance to research the parameter identification of the power system. Synchronous generator (SG) is an important part of the grid stability analysis. A precise on-line synchronous generator model is a foundation of the power grid stability analysis. Previously, parameters identification of synchronous generator is offline. The signals used for the parameters identification of synchronous generator are the generator's dynamic response to the step change under the condition of no-load. In this case, the obtained models are usually not accurate. With the development of PMU technology, it is possible to obtain the dynamic data and it provided the conditions for real-time online parameter of power system model. Thus, on-line parameter identification for synchronous generator becomes possible. Synchronous generator modeling plays an important role in system planning, operation and post-disturbance analysis[1-3]. Some algorithms have been proposed such as least square method[4],extended Kalman filtering[5],volterra series[6],and colony optimization algorithm. The least square method has simple objective function and quick computation speed but sensitive to selection of initial value and of measurement noise. Extended Kalman filtering has poor searching efficiency. Identification of based on Volterra series needs to compute a large number of kernel function. Recently, The parameter identification of synchronous generator is considered as an optimization process with a fitness function minimizing the errors between the estimated values and measured values. The literatures [7-8] using genetic algorithm(GA) to make identification of synchronous generator parameters but GA is poor in computing efficiency and the search process needs more time, not suitable for online parameter identification. In [9], it combines particle swarm optimization algorithm with Quantum Operation (PSO– Xiuge Zhang et al J. Chem. Pharm. Res., 2014, 6(5):1689-1695 ______________________________________________________________________________ 1690 QO) to solve both offline and online parameters estimation problem for SG. The hybrid algorithm is proposed to increase the convergence speed and identification accuracy, but, PSO has premature convergence problem [10-12]. Motivated by above discussions, the paper presents a small population-based particle swarm optimization (SPPSO) method to identify synchronous generator parameters based on the PMU data. In this method, the parameters identification of synchronous generator is formulated as an optimization problem of input-output system. A small population-based particle swarm algorithm has less computation, fast convergence speed, the identification accuracy is high, it is suitable for real-time online parameter identification of power system. and the synchronous generator parameters identification becomes easy. II. SYNCHRONOUS GENERATOR MODEL A. Synchronous Generator Model Description For the parameters identification of the synchronous generator, the first step is to establish the suitable mathematical model which can express different working conditions of synchronous generator precisely. Various types of synchronous generator are supplied by different manufacturers. Corresponding mathematical models of these synchronous generator s have been established by the manufacturers. However these models are not practical in the on-line stability analysis of power system. Choosing an appropriate model is the key point for the parameters identification of synchronous generator. With the information provided by the power plant, the five-step voltage synchronous generator practical model is adopted in this paper. The following equations show the basic aspects of the model. D-Axis electric model: ( )