发电技术 ›› 2025, Vol. 46 ›› Issue (1): 103-112.DOI: 10.12096/j.2096-4528.pgt.23125

• 新能源 • 上一篇    

基于改进混合高斯模型的风速分布拟合与风机年发电量估算

王玲芝, 张新波   

  1. 西安邮电大学自动化学院,陕西省 西安市 710121
  • 收稿日期:2024-03-21 修回日期:2024-06-10 出版日期:2025-02-28 发布日期:2025-02-27
  • 作者简介:王玲芝(1981),女,博士,教授,研究方向为风电不确定性、风能资源评估、新能源发电功率预测技术等,wlzmary@126.com
    张新波(1998),男,硕士研究生,研究方向为风电不确定性,xb21042@163.com

Wind Speed Distribution Fitting and Annual Electricity Generation Estimation of Wind Turbine Based on Improved Mixture Gaussian Model

Lingzhi WANG, Xinbo ZHANG   

  1. School of Automation, Xi’an University of Posts and Telecommunications, Xi’an 710121, Shaanxi Province, China
  • Received:2024-03-21 Revised:2024-06-10 Published:2025-02-28 Online:2025-02-27

摘要:

目的 为解决混合高斯模型在低风速段、高风速段以及复杂峰值、波谷部分存在较大误差的问题,提出了一种改进的混合高斯模型。 方法 改进模型的所有子分量取相同的形状参数,用风速样本值代替位置参数。同时,采用非线性最小二乘法优化调整形状参数和子分量的权重,使得模型可以精确地逼近包括风速样本局部点在内的概率密度分布。基于国内外4组风速分布数据,将该模型与混合核密度模型、混合高斯模型进行拟合效果比较,并使用2种误差指标和卡方检验系数评估3种模型的拟合优度。 结果 改进的混合高斯模型对复杂风速分布的拟合效果得到了极大提升,而且能够准确地拟合低风速段、高风速段及峰值、波谷部分的风速分布概率。此外,通过比较基于3种模型的风机年发电量估算,进一步验证了改进模型的有效性和优越性。 结论 提出的更高精度的风速分布概率模型有助于准确评估风电场的发电潜力和经济效益,对风电场的规划设计具有重要的指导意义。

关键词: 风力发电, 风速概率分布, 混合高斯模型, 非线性最小二乘法, 拟合性能, 风机, 年发电量

Abstract:

Objectives To address the significant errors in low wind speed section, high wind speed section, and peak and trough sections of the mixture Gaussian model, this paper proposes an improved mixture Gaussian model. Methods In the improved model, all subcomponents have the same shape parameter, and the wind speed sample values are used to replace the position parameters. Meanwhile, the nonlinear least squares method is employed to optimize and adjust the shape parameters and the weights of the subcomponents, allowing the model to accurately approximate the probability density distribution, including local points of wind speed samples. Based on four sets of wind speed distribution data from both domestic and international sources, the fitting performance of this model is compared with that of the mixed kernel density model and the Gaussian mixture model. The goodness of fit of the three models is evaluated using two error metrics and the Chi-square test coefficient. Results The improved Gaussian mixture model significantly enhances the fitting performance for complex wind speed distributions, and it can accurately fit the wind speed distribution probabilities in low wind speed section, high wind speed section, and peak and trough sections. Additionally, by comparing the annual electricity generation estimation of wind turbines based on the three models, the effectiveness and advantages of the improved model are further verified. Conclusions The proposed wind speed distribution probability model of higher precision helps accurately evaluate the power generation potential and economic benefits of wind farms, providing crucial guidance for wind farm planning and design.

Key words: wind power generation, wind speed probability distribution, mixture Gaussian model, nonlinear least squares method, fitting performance, wind turbine, annual electricity generation

中图分类号: