含风电电力系统多时间尺度优化调度方法研究

发布时间：2018-09-09 10:01

【摘要】：能源革命已悄然发生二十年。风力发电作为目前技术最为成熟且基本实现商业化的新兴可再生能源技术,已在全球范围内迅速发展。风电大规模开发利用一方面有效缓解了能源可持续发展的困局,带来了巨大经济利益,但另一方面风电输出功率的随机性、不可控性及难以预测性给电力系统调度运行带来很大的困难。传统基于人工日前调度计划和自动发电控制2个时间尺度相结合的调度方式过于粗放,已无法适应大规模风电并网后的电网调度需求。鉴于此,本文针对含风电电力系统的多时间尺度优化调度问题展开研究,为大规模风电合理有效利用及并网调度提供理论支持。本文主要的研究内容和成果如下：首先研究了不同时间尺度下风电功率不确定性描述方法及其预测模型,在日前阶段采用场景处理技术；在超短期预测阶段采用一种非参数、离散的马尔可夫链(MarkovChain)预测技术,可以短时间内直接估计风电的概率分布,不依赖于数值天气预报系统和风机周围的复杂物理信息,可以适应实时调度的需求。然后分别阐述了日前调度、滚动调度、实时调度多个时间尺度调度的协调优化机制,提出日前调度的“鲁棒性”原则。基于“鲁棒性”思想,分别提出风电多场景的鲁棒备用调度模式和鲁棒经济调度模式。鲁棒备用调度模式关键描述了机组的“鲁棒运行轨迹”(robust operation trajectory);鲁棒经济调度模式引入鲁棒优化方法,采用1min-max函数建立了水、火机组出力与备用的协调优化模型；最后采用两阶段决策方式,统一了两种鲁棒调度模式并建立起整体的鲁棒调度模型。对所建模型基于Benders分解思想构建求解策略,并调用CPLEX线性规划包求解。算例结果表明,所定义鲁棒运行轨迹下的机组出力可以适应所有的风电波动场景,并能保证系统在不同场景下运行的安全性；所建立的鲁棒经济调度模式使系统最小化了在最坏情况下的发电成本,并保证实际发电成本均不大于该成本；通过机组出力与备用协调优化的方式,引导了系统备用容量在空间上和时间上的合理分布,有效避免了无效备用容量和冗余备用容量,实现了备用的优化配置。针对风电强随机性使日前计划精度急剧下降的问题,提出一种含风电电力系统日前发电计划的滚动修正策略,采用水、火发电机费用和电网弃风量的双重修正函数,并提出一种改进的非劣邻域免疫算法(NNIA, Nondominated Neighbor Immune Algorithm)求解。算例结果表明,所提滚动调度模型在日内过程中进一步了优化系统的运行成本,增强了系统消纳风电的能力；所提改进NNIA算法基于等微增率法和机组最大、最小技术出力、调节偏差约束定义搜索空间的抗体偏好关系模型,该模型对抗体进化方向的引导作用明显,使算法收敛至最终解的速度明显加快,局部搜索能力更强。提出一种风电时序动态修正的实时调度方法。该方法将风电时序的动态修正过程纳入实时调度模型,以5-15min为周期滚动计算风电功率的期望误差状态向量,并同步建立针对下一时段修正的目标函数。由于目标函数的凸性,将模型转化为凸二次规划及其拉格朗日对偶问题求解,并定义了各实时机组的负荷分配因子以给定算法的初始内点。算例结果表明,所建立实时调度模型增强了实时机组跟踪风电功率随机波动的能力,对于减小风电波动对电网的影响、积极消纳风电以及更精细化的优化电力系统运行成本十分必要；所构建基于原对偶内点法的求解算法收敛迅速、鲁棒性强,可以满足实时调度的计算需求。
[Abstract]:Wind power, as the most mature and commercialized renewable energy technology, has developed rapidly all over the world. On the one hand, the large-scale development and utilization of wind power has effectively alleviated the predicament of sustainable energy development and brought huge economic benefits, but on the other hand, wind power has brought about great economic benefits. The randomness, uncontrollability and unpredictability of output power bring great difficulties to power system dispatching and operation. The traditional dispatching method based on the combination of manual day-ahead dispatching plan and automatic generation control is too extensive to meet the grid dispatching requirements after large-scale wind power integration. The multi-time scale optimal dispatch problem of wind power system is studied to provide theoretical support for the rational and effective utilization of large-scale wind power and grid-connected dispatch.
Firstly, the uncertainty description method and prediction model of wind power under different time scales are studied. Scene processing technique is used in the day-ahead stage, and a non-parametric discrete Markov Chain prediction technique is used in the ultra-short-term prediction stage, which can directly estimate the probability distribution of wind power in a short time without depending on numerical value. The complex physical information around the weather forecast system and the fan can meet the needs of real-time scheduling. Then the coordination and optimization mechanisms of day-ahead scheduling, rolling scheduling and real-time scheduling with multiple time scales are described respectively, and the "robustness" principle of day-ahead scheduling is proposed.
Based on the idea of robustness, the robust standby dispatching model and the robust economic dispatching model for wind power in multi-scenarios are proposed respectively. The coordinated optimization model of unit output and reserve is established. Finally, two robust scheduling modes are unified and a global robust scheduling model is established by using the two-stage decision-making method. Unit output can adapt to all wind power fluctuation scenarios and ensure the safety of the system in different scenarios; the robust economic dispatching model minimizes the generation cost in the worst case and ensures that the actual generation cost is not greater than this cost; through the coordinated optimization of unit output and reserve The system spare capacity can be reasonably distributed spatially and temporally, and the invalid spare capacity and redundant spare capacity can be effectively avoided, and the optimal allocation of spare capacity can be realized.
Aiming at the problem that the randomness of wind power makes the precision of day-ahead generation plan decrease sharply, a rolling correction strategy for day-ahead generation plan of wind power system is proposed. The dual correction functions of water, thermal power cost and wind abandonment are adopted, and an improved non-inferior neighborhood immune algorithm (NNIA, Nondominated Neighbor Immune Algorithms) is proposed. The results show that the proposed rolling scheduling model can further optimize the operation cost of the system and enhance the ability of the system to absorb wind power during the day; the improved NNIA algorithm defines the antibody preference relation model of the search space based on the equal increment method and the maximum and minimum technical output of the unit, and the adjustment bias constraint. T-type has obvious guiding effect on the evolution direction of the antibody, which makes the algorithm converge to the final solution faster and has stronger local search ability.
This paper presents a real-time scheduling method for dynamic correction of wind power time series. The method incorporates the dynamic correction process of wind power time series into the real-time scheduling model, calculates the expected error state vector of wind power with 5-15 minutes rolling cycle, and establishes the revised objective function for the next period simultaneously. In order to solve convex quadratic programming and its Lagrange dual problem, the load distribution factors of each real-time unit are defined to the initial interior point of the given algorithm. The results show that the real-time dispatching model enhances the ability of real-time units to track the random fluctuation of wind power, and actively absorbs the wind to reduce the impact of wind power fluctuation on the power grid. It is necessary to optimize the operation cost of power system and more precisely, and the algorithm based on the primal dual interior point method has the advantages of fast convergence and strong robustness, which can satisfy the calculation requirements of real-time dispatching.
【学位授予单位】：武汉大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM614

【相似文献】