风电场内机组优化调度研究

发布时间：2018-05-09 00:20

本文选题：疲劳损伤 + 相空间重构　；参考：《华北电力大学》2014年博士论文

【摘要】：随着电力系统中风电并网比例的增加,风能的随机波动性对传统电力系统经济调度和安全运行带来挑战。研究在风电功率预测与电力系统的负荷约束条件下,风电场内机组优化调度问题,不仅能减少风力发电机组的冗余运行和磨损浪费,避免机组的频繁启停,还可以降低运行成本,提高风电场输出功率的电能质量,有效减轻风电波动性对电网的影响,从而在保证电力系统安全性的前提下,提高电力系统的消纳风电能力和经济效益。以风电场功率预测数据为基础,重点研究了以降低风力发电机组疲劳载荷损伤相对量和降低集电系统损耗为目标的风电场内机组优化调度的算法,完成了以下研究工作。 (1)提出了不同运行工况下风力发电机组关键部件相对疲劳损伤量的计算方法。根据华北某风电场的风资源数据,利用瑞利分布的风速累积分布函数,基于GH-Bladed模拟了1.5MW风力发电机组的疲劳载荷,利用雨流循环计数法,得到风力发电机组各个部件的疲劳载荷谱,然后根据仿真计算和Miner法则得到的风力发电机组关键部件的相对疲劳损伤量,可为风电场内机组优化运行提供评价准则。 (2)基于相空间重构的神经网络风电功率预测算法的应用。风电场内机组优化调度是以风力发电机组的短期和超短期功率预测值为研究基础,根据混沌-相空间重构的原理可知风力发电机组的风速和风电功率时间序列数据具有混沌的属性的基础上,将相空间重构与神经网络相结合,建立混沌-Elman、混沌-BP和混沌-Volterra级数的风电功率预测模型,经实例验证,分析比较得出混沌-Elman模型的预测效果相对较好,能够提高预测的精度和稳定性。 (3)建立以风电场集电系统网损最小为目标的机组优化调度模型。以风电场内集电系统网损最小为目标函数,电网调度要求、风力发电机组有功输出的功率上下限、风力发电机组无功输出的功率上下限、风力发电机组端电压上下限、变压器变比上下限等为约束条件,建立机组优化调度的数学模型,分别采用粒子群优化算法和粒子群-遗传优化算法进行寻优。结果表明,粒子群-遗传算法在优化效果和运算效率方面均优于单一粒子群算法。 (4)建立了以风电场内机械损伤量最小为目标的机组组合优化模型。基于前述的相对疲劳损伤量模型,建立机组组合模型,合理配置机组启停方案,以期在调度期内风电场整体机械损伤最小,延长机组运行效率和使用寿命。然后利用改进二进制粒子群优化算法(BPSO)、遗传优化算法(GA)、粒子群-遗传混合优化算法(BPSO-GA),进行优化求解。结果表明,BPSO-GA比单一GA和BPSO提高了优化性能,运行期间总疲劳损伤量最小；引入粒子群优化参数的BPSO-GA算法的计算时长相对BPSO算法略长,但比GA算法计算时长要短；三种模型的计算时长从大到小依次为：GA, BPSO-GA, BPSO。 (5)建立了基于机组优先级分类的风电场内功率分配模型。以风力发电机组发电功率、风速平均值和均方根差值作为特征值,分析机组发电性能,并分别采用SOFM神经网络算法与基于模拟退火遗传算法的模糊C均值聚类算法建立机组优先级分类模型。将发电性能较好的一类作为优先执行发电计划的机组,计及线路损耗后的发电计划,对风电场内其余机组进行两层优化,外层是以风力发电机组相对疲劳损伤量最小为目标的机组优化出力,内层是确定机组间负荷满足电网调度要求的最优功率分配。通过两层分配,得到既满足电网调度需求,又降低风电场运行损耗的功率分配结果。结果表明,基于遗传模拟退火算法的模糊聚类算法分类方法的疲劳损伤量比自组织特征映射神经网络分类方法的疲劳损伤量较小,SAGA-FCM分类方法下停机的机组台数较多。风力发电机组分类后优化调度,能够使风电场机组运行优化,提高风电场输出电能质量。
[Abstract]:With the increase of the proportion of apoplexy in the power system, the stochastic volatility of wind energy brings challenges to the economic dispatch and safe operation of the traditional power system. Under the condition of wind power forecasting and the load constraint of the power system, the optimal scheduling problem of the unit in the wind farm can not only reduce the redundant operation and the wear wave of the wind turbine. To avoid the frequent start and stop of the unit, it can also reduce the operating cost, improve the power quality of the output power of the wind farm, reduce the influence of the light wind wave on the power grid, and improve the power system's ability to eliminate wind and electricity and the economic benefit on the premise of ensuring the safety of the power system.
Based on the prediction data of wind power, this paper focuses on the optimization scheduling algorithm of wind farms in wind farms aiming at reducing the relative amount of fatigue load damage and reducing the loss of the collecting system. The following research work has been completed.
(1) the calculation method of the relative fatigue damage of the key components of the wind turbine under different operating conditions is proposed. According to the wind resource data of a wind farm in North China, the cumulative distribution function of the wind velocity distribution is used to simulate the fatigue load of the 1.5MW wind turbine based on GH-Bladed, and the wind power generation is obtained by the rain flow cycle counting method. The fatigue load spectrum of each component of the unit, and the relative fatigue damage of the key components of the wind turbine based on the simulation calculation and the Miner rule, can provide the evaluation criteria for the optimal operation of the unit in the wind farm.
(2) the application of the neural network wind power prediction algorithm based on phase space reconstruction. The optimization scheduling in the wind farm is based on the short-term and ultra short term power forecast of the wind turbine. According to the principle of the chaotic phase space reconstruction, the wind speed and the wind power time series data are chaotic. On the basis of the attribute, the phase space reconstruction and the neural network are combined to establish the wind power prediction model of chaotic -Elman, chaotic -BP and chaotic -Volterra series. The results of the analysis and comparison show that the prediction effect of the chaotic -Elman model is relatively good, and can improve the accuracy and stability of the prediction.
(3) set up an optimal scheduling model for the minimum loss of the wind electric field collection system, with the minimum loss of the net loss in the wind farm as the objective function, the requirements of the power grid dispatching, the power upper and lower limits of the active output of the wind turbine, the power upper and lower limits of the reactive output of the wind turbine, the upper and lower limit of the end voltage of the wind generator set and the variable pressure The mathematical model of the optimal scheduling of the unit is established, and the particle swarm optimization algorithm and particle swarm optimization algorithm are used to optimize the optimal scheduling of the unit. The results show that the particle swarm optimization algorithm is superior to the single particle swarm optimization in the optimization effect and the operation efficiency.
(4) set up a unit combination optimization model which aims at the minimum amount of mechanical damage in the wind farm. Based on the relative fatigue damage quantity model mentioned above, set up the unit combination model and rationally configure the unit start and stop scheme, in order to minimize the overall mechanical damage of the wind farm in the scheduling period, and the operation efficiency and service life of the long unit, and then use the improvement to improve the operation efficiency and service life of the long unit. Binary particle swarm optimization (BPSO), genetic optimization algorithm (GA), particle swarm genetic hybrid optimization algorithm (BPSO-GA) are used to optimize the solution. The results show that BPSO-GA is better than single GA and BPSO to improve the performance, and the total fatigue damage is minimal during operation, and the phase to BPSO calculation when the BPSO-GA algorithm is introduced into the particle swarm optimization parameters The length of the three models is shorter than that of the GA algorithm. The computation time of the three models is from GA to BPSO-GA, BPSO..
(5) the power distribution model in wind farm based on unit priority classification is established. The power generation power of the wind turbine, the average wind speed and the mean square root difference are used as the eigenvalues to analyze the generating performance of the unit, and the SOFM neural network algorithm and the fuzzy C mean clustering algorithm based on simulated annealing genetic algorithm are used to establish the unit priority. A class of class classification model, which takes a generating unit with better power generation performance as a priority to execute the power generation plan, takes into account the power generation plan after the line loss, and optimizes the two layers of the rest of the wind farm. The outer layer is the optimal output of the unit which aims at minimizing the relative fatigue damage of the wind turbine, and the inner layer is to determine the inter unit load to satisfy the power grid. The optimal power allocation required by the scheduling is obtained by two layers of distribution to obtain the power allocation results that meet both the demand of the power grid scheduling and the loss of the wind farm operation. The results show that the fatigue damage amount of the fuzzy clustering algorithm based on the genetic simulated annealing algorithm is better than the fatigue damage amount of the self organizing feature mapping neural network classification method. There are a lot of units in the SAGA-FCM classification method. The optimal scheduling of wind turbines can optimize the operation of the wind farm unit and improve the output power quality of the wind farm.

【学位授予单位】：华北电力大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM614

【参考文献】