【作者】 于晗；

【导师】 张建华；

【作者基本信息】 华北电力大学（北京） ， 电力系统及其自动化， 2009， 博士

【摘要】 风电作为一种可再生能源,对于缓解世界范围内的能源危机具有重大意义,近年来发展迅速。风电本身所固有的随机性和问歇性,给电力系统的规划和运行增添了新的不确定因素。含有风电场的电力系统存在更大的潮流波动,因此传统的确定性输电网规划模型不适用于含有风电场的输电系统。在此背景下,为了获得稳健的电网规划方案,本文提出用概率的方法来解决含有风电场的电网的输电系统扩展规划问题。在传统的确定性输电系统扩展规划模型的基础上,本文提出了一种输电系统规划的快速计算方法。该方法用线路过负荷的百分比来代替传统模型中的切负荷大小,避免了启发式优化方法中对众多待选方案进行的线性规划计算,极大的节约了计算时间,而规划结果与传统模型完全相同。为了在输电系统规划中考虑负荷和风电场输出功率不确定性,本文通过演化经典确定性的输电系统规划模型,提出了一种考虑负荷和风电场有功出力的概率分布、基于概率潮流计算的输电系统机会约束规划模型。在求解过程中,采用蒙特卡罗和解析法相结合的方法,极大降低了计算量,使风电模型结合在概率潮流和机会约束规划中成为可能。为了进一步降低计算量,设计了一种两步的遗传算法。本文提出的基于机会约束规划方法的输电网扩展规划模型与传统的确定性规划模型相比,可以在规划中考虑更多的对规划方案影响很大的信息,如风速、风机参数、负荷分布等。本文所提出的规划模型可以有效考虑这些不确定信息并且能够在规划过程中提供更丰富的信息,如不确定性对规划结果的影响、投资和输电网过负荷风险的关系等。利用本文提出的模型可以得到更稳健、有效益的规划方案。概率潮流计算是考虑各种不确定因素评估电网稳态运行情况的有效工具,在输电网规划中得到了广泛应用。采用与随机采样相结合的蒙特卡罗模拟法的概率潮流计算具有应用灵活、计算精度高的优点,缺点是计算量过大。本文提出了用拉丁超立方采样与Cholesky分解法相结合的概率潮流计算方法。与简单随机采样和与随机组合相结合的拉丁超立方采样相比,本文的方法可以改善采样值对输入随机变量的分布空间的覆盖程度、提高蒙特卡罗模拟法的采样效率,降低采样规模,同时保留蒙特卡罗模拟法的优点,具有广泛的应用前景。更多还原

【Abstract】 In order to deal with the world-wide energy crisis,wind farms have been developed rapidly in recent years to generate electric power from renewable wind power.However,wind power is variable and intermittent and it therefore introduces an extra factor of uncertainties for power system operation and planning. The network with large proportion of wind power will have more power flow fluctuations.Therefore,the deterministic transmission network expansion methods, which only account for one operation scenario,are unsuitable for the planning with stochastic power output from wind farms.This paper proposes to tackle the uncertainties of wind farms in transmission network expansion planning with probabilistic methods to obtain robust transmission network expansion planning schemes.First,a fast deterministic transmission network expansion planning model is proposed based on the traditional model.The way to reflect the overload level of a candidate planning scheme in the new model is changed by replacing the loss of load item with total overload percent.The linear optimization using in determining the loss of load item avoided and hence the computational time is shorter.Although different formulations are used to penalize the overload schemes,the proposed and traditional models can obtain the same optimal solution with minimized cost if a planning scheme without load curtailment and overload can be achieved finally.In order to take uncertainties of both load variation and wind farm power output into consideration,a chance constrained transmission network expansion planning formulation is proposed by expanding the deterministic formulation into probabilistic area.Different from the conventional Monte Carlo simulation approach to obtain the final probability distribution,this paper has also proposed a solution approach which combines the Monte Carlo simulation with the analytical method.The combined method reduces the computational cost greatly and makes it possible to introduce the wind farm model and probabilistic power flow in the proposed chance constrained transmission network expansion planning.In order to accelerate the optimization process,a two step genetic algorithm has been used and found to be efficient.The study examples have shown the necessity in including the uncertainties of load and wind farm in the transmission network expansion planning and has illustrated that the proposed chance constrained method for transmission network expansion planning can consider these uncertainties and provide much more comprehensive information,including the effects of these uncertainties in planning schemes and the relationship between the investment cost and the risk of overload,which are essential in order to achieve a robust and cost-effective planning scheme.The probabilistic load flow evaluation is a powerful approach to investigate the steady-state power system operation characteristics under various possible uncertainties which has been widely used in transmission network expansion planning.Monte Carlo simulation combined with simple random sampling is one of the most popular mathematical methods adopted in probabilistic problems and also has been widely used in PLF and many other power system analyses.However,the high accuracy of solutions can only be achieved by a large number of repeated calculations.This paper proposes the use of an efficient sampling method,Latin hypercube sampling combined with Cholesky decomposition method,into Monte Carlo simulation for solving the PLF problems.The proposed method can achieve a better sampling efficiency than simple random sampling and Latin hypercube sampling combined with random permutation,and makes it possible for Monte Carlo simulation to get an accurate simulation result with a much smaller simulation size.The proposed method is found to be robust and flexible and has the potential to be applied in many power system probabilistic problems.更多还原