基于支持向量机的风电场风速预测方法研究
发布时间：2018-11-25 21:31
【摘要】：能源枯竭、资源匮乏已经成为一个全球性问题,从而可再生能源的开发与持续利用日益受到关注。其中,风能作为洁净、无污染和易于利用的可再生能源之一,更是在全世界范围内得到了广泛的应用。然而,由于风电发电自身的波动性和不稳定性,给电力系统的安全稳定和产能质量造成了不良的影响。解决这些问题的关键在于,对风电场风速和风电功率进行预测。通过对风电场风速的准确预测,可以降低风电功率的随机性,可有效缓解风速变化对电力系统造成的不利影响。风速预测方法近年来发展迅速,但是在预测方法的多元化和预测精度方面还有比较大的上升空间。本文通过建立和评估多种短期风速预测模型,发现单一的风速预测方法的预测精度相对不足,于是提出组合预测模型,并通过对对支持向量机方法的改进和研究,综合利用各风速预测算法的优点,提出了一种基于最小二乘支持向量机的组合预测模型。该预测方法首先利用模糊层次分析法,在若干单项预测模型中筛选出灰色预测算法,人工神经网络预测算法和时间序列—卡尔曼滤波混合算法;然后以这三种单项预测模型作为输入,并以实际风速值作为输出,进行训练最小二乘支持向量机;最终得出预测函数。本文还分别建立等权平均组合预测模型和最优加权组合预测模型,且以这两种组合预测模型为参照,来分析基于最小二乘支持向量机的组合预测模型的预测性能。本研究中,针对各模型的预测性能,采用预测平均绝对误差,平均绝对百分比误差以及误差平方和,这三个误差指标来比较分析。通过以内蒙古某风电场计算出的小时风速数据作为研究样本,运用MATLAB进行仿真,采用各模型对风速进行短期预测,验证了基于最小二乘支持向量机的风速组合预测模型的有效性。仿真试验表明,本文提出的基于支持向量机组合预测方法模型可进一步提升风速预测精度,而且相较于传统的两种组合预测模型,也具有比较大的精度优势。
[Abstract]:Energy depletion and resource scarcity have become a global problem, so the development and sustainable utilization of renewable energy have been paid more and more attention. Among them, wind energy, as one of the clean, pollution-free and easy to use renewable energy, has been widely used in the world. However, due to the volatility and instability of wind power generation itself, the safety and stability of power system and the quality of production capacity are adversely affected. The key to solve these problems is to predict wind speed and power of wind farm. Through the accurate prediction of wind speed in wind farm, the randomness of wind power can be reduced, and the adverse effect of wind speed variation on power system can be effectively alleviated. Wind speed forecasting methods have been developed rapidly in recent years, but there is still much room for improvement in the diversification of forecasting methods and prediction accuracy. Through the establishment and evaluation of several short-term wind speed forecasting models, it is found that the prediction accuracy of a single wind speed forecasting method is relatively insufficient, so a combined forecasting model is proposed, and the support vector machine method is improved and studied. A combined prediction model based on least squares support vector machine (LS-SVM) is proposed by synthesizing the advantages of wind speed prediction algorithms. In this method, the grey prediction algorithm, the artificial neural network prediction algorithm and the hybrid time series Kalman filter algorithm are selected from several single prediction models by using the fuzzy analytic hierarchy process (FAHP). Then the three single prediction models are taken as input and the actual wind speed is taken as the output to train the least squares support vector machine (LS-SVM) and finally the prediction function is obtained. This paper also establishes the equal weight average combination prediction model and the optimal weighted combination forecast model, and takes these two combination forecast models as the reference to analyze the prediction performance of the combination forecast model based on the least square support vector machine (LS-SVM). In this study, the average absolute error, average absolute percentage error and sum of error square are used to compare and analyze the prediction performance of each model. By taking the hourly wind speed data calculated from a wind farm in Inner Mongolia as the research sample, MATLAB is used to simulate the wind speed, and each model is used to predict the wind speed in the short term. The validity of the wind speed combination prediction model based on least squares support vector machine is verified. The simulation results show that the proposed combined forecasting model based on support vector machine can further improve the accuracy of wind speed prediction, and it also has a large accuracy advantage compared with the traditional two combined forecasting models.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TM614
[Abstract]:Energy depletion and resource scarcity have become a global problem, so the development and sustainable utilization of renewable energy have been paid more and more attention. Among them, wind energy, as one of the clean, pollution-free and easy to use renewable energy, has been widely used in the world. However, due to the volatility and instability of wind power generation itself, the safety and stability of power system and the quality of production capacity are adversely affected. The key to solve these problems is to predict wind speed and power of wind farm. Through the accurate prediction of wind speed in wind farm, the randomness of wind power can be reduced, and the adverse effect of wind speed variation on power system can be effectively alleviated. Wind speed forecasting methods have been developed rapidly in recent years, but there is still much room for improvement in the diversification of forecasting methods and prediction accuracy. Through the establishment and evaluation of several short-term wind speed forecasting models, it is found that the prediction accuracy of a single wind speed forecasting method is relatively insufficient, so a combined forecasting model is proposed, and the support vector machine method is improved and studied. A combined prediction model based on least squares support vector machine (LS-SVM) is proposed by synthesizing the advantages of wind speed prediction algorithms. In this method, the grey prediction algorithm, the artificial neural network prediction algorithm and the hybrid time series Kalman filter algorithm are selected from several single prediction models by using the fuzzy analytic hierarchy process (FAHP). Then the three single prediction models are taken as input and the actual wind speed is taken as the output to train the least squares support vector machine (LS-SVM) and finally the prediction function is obtained. This paper also establishes the equal weight average combination prediction model and the optimal weighted combination forecast model, and takes these two combination forecast models as the reference to analyze the prediction performance of the combination forecast model based on the least square support vector machine (LS-SVM). In this study, the average absolute error, average absolute percentage error and sum of error square are used to compare and analyze the prediction performance of each model. By taking the hourly wind speed data calculated from a wind farm in Inner Mongolia as the research sample, MATLAB is used to simulate the wind speed, and each model is used to predict the wind speed in the short term. The validity of the wind speed combination prediction model based on least squares support vector machine is verified. The simulation results show that the proposed combined forecasting model based on support vector machine can further improve the accuracy of wind speed prediction, and it also has a large accuracy advantage compared with the traditional two combined forecasting models.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TM614
【参考文献】
相关期刊论文 前10条
1 张涛;张明辉;李清伟;张sソ，
本文编号：2357432
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2357432.html
