含风电区域电网的可靠性评估模型与方法

发布时间：2018-07-20 18:22

【摘要】：风电的大规模接入势必对电网的安全性和可靠性造成影响,特别是对网架结构薄弱的区域电网影响较大,需要从理论上量化分析风电的可靠性及对电网的影响。根据某区域电网的特点,特别是受外界环境影响较大而且含大量风电的地区,分析天气、环境、自然灾害和运行工况等对风电机组、发电机、变压器和输电线路停运的影响,提出考虑天气、环境和电网运行工况的元件故障率计算方法,研究风电机组和主要电力系统一次设备的时变停运模型。针对风电机组的运行特点,基于马尔可夫链的解析方法,考虑运行、停运和降额状态,建立了风电机组的三状态故障模型,在此基础上,考虑风速的随机性、风电场尾流效应的影响建立了风电场的可靠性模型；基于此模型,应用蒙特卡罗方法对风电场有功出力的概率特性进行评估,给出了评估的方法和流程。考虑非序贯蒙特卡罗方法的不足,在风电机组三状态故障模型的基础上,基于状态持续时间抽样方法给出了风机状态的时间序列模型；基于风速的ARMA模型,考虑风电场复杂尾流效应的影响建立了用于序贯蒙特卡罗仿真的风电场可靠性模型；针对风电机组的三状态模型,提出双重抽样序贯蒙特卡罗方法,并基于此方法对风电场有功出力的可靠性进行评估。根据上述研究成果,应用分散抽样蒙特卡罗方法对某区域发电系统进行可靠性评估。首先考虑风电出力的随机性、常规发电机组的随机停运和负荷预测的随机性建立发电系统的可靠性评估模型；其次针对概率抽样中常规蒙特卡罗算法样本容量大、效率低等不足,提出应用分散抽样蒙特卡罗算法解决含风电发电系统可靠性评估问题,此算法将[0,1]区间分成若干子区间,在抽样后分别对每个子区间进行系统状态判断和指标计算,从而增加故障状态的抽样频率,提高抽样效率,在满足精度要求下,有效的减少了抽样次数。鉴于发电系统未涉及变压器、输电线路等电网元件,应用蒙特卡罗方法对含风电的发输电系统进行可靠性评估。首先考虑风速的随机性、多个风电场中风速的相关性和风机的停运及降额状态建立风电场的可靠性模型；其次考虑常规蒙特卡罗方法应用于大规模风电接入,特别是以单个小容量的机组接入的情况下会存在样本容量大、效率低等不足,本文提出在蒙特卡罗仿真的概率抽样环节中使用拉丁超立方采样和Cholesky分解相结合的方法,此方法使用拉丁超立方采样来改善样本值对输入随机变量的分布空间的覆盖程度、使用Cholesky分解来降低输入变量之间的相关性系数,从而提高了采样效率、增加收敛速度和提高评估精度。在MATLAB中建立了相关评估程序,并对150MW风电场算例、含2个风电场的10机发电系统算例和改进IEEE-RTS79发输电系统算例进行了仿真,通过对仿真结果的分析和研究验证所建模型和所提方法的有效性。
[Abstract]:Large-scale wind power access is bound to have an impact on the security and reliability of the power grid, especially on the regional grid with weak grid structure. Therefore, it is necessary to quantitatively analyze the reliability of wind power and its impact on the power grid theoretically. According to the characteristics of a certain regional power network, especially in areas which are greatly affected by the external environment and contain a large amount of wind power, the effects of weather, environment, natural disasters and operating conditions on the outage of wind turbines, generators, transformers and transmission lines are analyzed. This paper presents a method for calculating the failure rate of components considering weather, environment and power network operating conditions, and studies the time-varying outage model of wind turbine and primary equipment of main power system. According to the operating characteristics of wind turbine, based on the analytic method of Markov chain, the three-state fault model of wind turbine is established by considering the running, outage and reducing state. On this basis, the randomness of wind speed is considered. The reliability model of wind farm is established by the influence of wind farm wake effect. Based on this model, the probabilistic characteristics of active power output of wind farm are evaluated by Monte Carlo method, and the evaluation method and process are given. Considering the shortcomings of non-sequential Monte Carlo method, based on the three-state fault model of wind turbine, the time series model of wind turbine state and the ARMA model based on wind speed are given based on state duration sampling method. Considering the influence of complex wake effect of wind farm, the reliability model of wind farm for sequential Monte Carlo simulation is established, and a two-sampling sequential Monte Carlo method is proposed for the three-state model of wind turbine. Based on this method, the reliability of active power output of wind farm is evaluated. Based on the above research results, a decentralized sampling Monte Carlo method is used to evaluate the reliability of a regional power generation system. Firstly, considering the randomness of wind power output, the random outage of conventional generator sets and the randomness of load forecasting, the reliability evaluation model of generation system is established, and the sample capacity of conventional Monte Carlo algorithm in probability sampling is large. In this paper, a decentralized sampling Monte Carlo algorithm is proposed to solve the reliability evaluation problem of wind power generation system. The algorithm divides the interval of [0 ~ 1] into several sub-regions. After sampling, the system state judgment and index calculation are carried out for each sub-interval, thus increasing the sampling frequency of the fault state, improving the sampling efficiency, and effectively reducing the sampling times under the requirement of precision. In view of the fact that the generation system does not involve transformer, transmission line and other power network elements, Monte Carlo method is used to evaluate the reliability of wind power generation and transmission system. Firstly, considering the randomness of wind speed, the correlation of wind speed in many wind farms, the outage and reduction of wind turbines, the reliability model of wind farms is established, and the conventional Monte Carlo method is applied to large-scale wind power access. In particular, when a single unit with small capacity is connected, the sample size is large and the efficiency is low. In this paper, a method combining Latin hypercube sampling and Cholesky decomposition in probability sampling in Monte Carlo simulation is proposed. In this method, Latin hypercube sampling is used to improve the coverage of sample values to the distribution space of input random variables, Cholesky decomposition is used to reduce the correlation coefficient between input variables, and the sampling efficiency is improved. Increase convergence speed and improve evaluation accuracy. The relative evaluation program is established in MATLAB, and the examples of 150 MW wind farm, 10 generator power generation system with 2 wind farms and the improved IEEE-RTS79 generation and transmission system are simulated. The effectiveness of the proposed model and the proposed method is verified by the analysis and research of the simulation results.
【学位授予单位】：华北电力大学(北京)
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM614;TM732

【相似文献】