【作者】 徐兵；

【导师】 梁军；

【作者基本信息】 山东大学 ， 电力系统及其自动化， 2013， 硕士

【摘要】 电力系统数字仿真是电力系统设计、规划、运行的主要工具,仿真结果的准确性对电力系统的安全、可靠、经济运行具有重要的影响,作为数字仿真的基础,负荷模型的准确与否对仿真结果影响很大。然而由于负荷自身的复杂性、分散性和随机性,使得负荷建模研究一直以来都是电力学界的一个难题。本文对负荷建模实用化进程中的若干问题进行了探讨。充分的数据来源是负荷建模的基础,本文比较了负荷建模所需数据的几种获取方式,包括负荷特性测量装置、数据采集与监控系统(SCADA)、广域测量系统(WAMS)、故障录波监测系统(FRMS)等。随着故障录波技术的发展,提出采用广泛普及的故障录波监测装置进行负荷建模研究,不仅投资小而且使大量的故障录波数据得到充分应用,能够有效解决负荷时变性和地域分散性等难题。传统的用于静态负荷建模的统计综合法、稳态试验法限于人力物力难以经常进行,从而无法解决负荷的时变性难题。本文提出采用故障录波器记录的稳态数据进行在线静态负荷建模,得到各个时刻的负荷静态特征系数。结合统计综合法的建模思路,按时间特性进行分类,形成相应各类的负荷模型参数库,然后根据相应的稳态数据采用递推最小二乘法在线修正各模型参数库,得到不同时间尺度下的负荷模型,方便用户根据需要进行选择,能够有效解决负荷时变性难题。采用日照地区夏季典型日的数据进行仿真验证,表明本文所提方法正确有效。动态负荷建模中,针对时变性,研究一直围绕分类与综合的方法展开,然而这种方法没有结合负荷特性从模型应用角度研究分类的效果,给实际应用造成了困难。本文提出采用故障录波器记录的扰动数据进行在线动态负荷建模,采用与在线静态负荷建模相同的建模思路,得到各个时间区间的负荷动态特征参数。在动态负荷模型参数修正时,提出一种基于渐进学习的电力负荷递推修正建模方法,该方法在由历史样本辨识得到的模型参数基础上,每采集一组扰动数据对原模型参数进行在线修正,可只保存修正后的模型参数而无需保存所有历史数据,提高了建模的运行效率。采用PSASP仿真数据进行建模分析,结果表明,该方法正确有效,同时与参数加权平均法和基于标准实测样本聚类中心法进行比较,结果表明该方法实现更为简便,精度更高。分布式发电的接入,给传统的负荷建模研究带来了困难,针对含有分布式电源的广义电力负荷建模,本文采用PSASP仿真软件在EPRI-9节点系统基础上搭建含分布式发电的动态仿真系统。比较当分布式电源容量比例不同时对负荷建模的影响,进而采用异步机并联静态负荷的综合负荷模型来描述含有分布式电源的区域负荷特性,同时在不同的负荷水平下对模型适应性进行检验。仿真结果表明,异步机并联静态负荷的综合负荷模型结构在描述含分布式电源的区域负荷时是正确有效的。更多还原

【Abstract】 The digital simulation of power system is the main tool for power system planning、 design and operation. The accuracy of simulation results is important for the safety, reliability and economic operation of the power system. As the basis of digital simulation, the accuracy of the load model is a great impact on the simulation results. However, due to the complexity, dispersion and randomness of the load itself, the research of the load modeling is always a difficult problem in power field. This paper discusses some problems in the process of load modeling.Sufficient data are the basis of the load modeling. This paper compares several access methods of the required data for load modeling, including the load measurement device, SCADA, WAMS, FRMS. With the development of fault recording technology, propose using the widespread fault recording device for load modeling, not only small investment and a lot of fault recorder data can be fully applied. This way can effectively solve the time-variation and geographic dispersion problem of the load.The traditional statistical synthesis method, steady state test method for static load modeling is limited to carrying out regularly, unable to solve the time-variation problem of the load. The on-line static load modeling is proposed on the basis of steady-state data of the fault recording device in the paper. Combining with the statistical synthesis method, classify according to the time characteristic, form the load model parameter library of the corresponding types, then according to the corresponding real-time data using the recursive least squares method correct the parameters of each model library on-line, obtain load models under different time scales, so it can be convenient for the user to choose models according to need and can effectively solve the time-variation problem of the load. Simulating with the summer typical daily data from Ri-Zhao, the results show that the proposed method is correct and effective.For dynamic load modeling, aiming at the time-varying, studies were focused on the classification and comprehensive method. However, this method did not study on the effect of classification from the view of application with the load characteristic, moreover this method cause difficulty in practical application. The on-line dynamic load modeling is proposed on the basis of disturbance data of the fault recording device in the paper, using the same modeling method with online static load modeling. For the parameters correction of the dynamic load model, the method of recursive correction load modeling based on incremental learning is proposed in the paper. On the basis of the model parameters obtained by the historical samples, correcting the original model parameters online with each new disturbance data. By this way we can only save the model parameters without saving all the historical data, so that we can improve the efficiency of load modeling. Load modeling with the PSASP simulation data, the results show that this method is correct and effective. Comparing this method with the parameters weighting-mean method and the clustering center method based on standard test data, the results show that this method is more convenient to be carried out and the precision is much higher.The access of distributed generation brings difficulty to the traditional research of the load modeling. For the generalized load modeling with distributed generation, dynamic simulation system with distributed generation is built based on EPRI-9node system by PSASP in this paper. Comparing the influence of the different distributed power capacity ratio on load modeling, then the synthesis load model of asynchronous machine in parallel with static load is adopted to describe the regional load characteristics with distributed generation, at the last the applicability of the model is verified on different load level. The simulation results show that synthesis load model of asynchronous machine in parallel with distributed generation is correct and effective in describing the regional load with distributed generation.更多还原