摘要
针对风机叶根载荷影响因素复杂、计算量大、非线性和强耦合,采用传统数理分析方法难以建模的问题。文章首先分析了叶根载荷的主要影响因素,并结合多元回归模型建立载荷预测模型;然后采用Bladed对2MW风机实验所得仿真数据划分训练数据集和测试数据集,并利用所得数据对Sine混沌映射改进麻雀算法优化的BP神经网络(Sine-SSA-BP)预测模型进行训练,使用训练后的模型进行叶根载荷预测;最后将预测结果与测试数据、BP神经网络预测模型和极限学习机(ELM)预测模型的预测结果进行对比分析。结果表明,Sine-SSA-BP预测模型性能更佳,预测精度更高,验证了所提方法的可行性和有效性。
To address the problems of complex,computationally intensive,nonlinear and strongly coupled influencing factors of wind turbine leaf root load,which are difficult to model using traditional mathematical analysis methods.Firstly,the main influencing factors of leaf root load are analyzed and the load prediction model is established by combining multiple regression model,then the simulation data obtained from the experiments of 2 MW wind turbine are divided into training data set and test data set by using Bladed,and the obtained data are used to train the BP neural network(Sine-SSABP)prediction model optimized by Sine chaotic mapping improved sparrow algorithm,using the trained the model is used to perform leaf root load prediction,and finally the prediction results are compared and analyzed with those of the test data,BP neural network prediction model and extreme learning machine(ELM)prediction model.The results show that the Sine-SSA-BP prediction model has better performance and higher prediction accuracy,which verifies the feasibility and effectiveness of the proposed method.
作者
张良
何山
艾纯玉
Zhang Liang;He Shan;Ai Chunyu(Xinjiang University,Urumqi 830047,China;Engineering Research Center of Ministry of Education for Renewable Energy Generation and Grid Control,Urumqi 830047,China)
出处
《可再生能源》
CAS
CSCD
北大核心
2023年第10期1322-1328,共7页
Renewable Energy Resources
基金
新疆维吾尔自治区高校科研计划项目(XJEDU2021I010)
国家重点研发计划项目(2021YFB1506902)
新疆大学课程思政标杆课项目(XJU2022BGK27)
国家自然科学基金项目(51767024)。
关键词
载荷预测
极限学习机
BP神经网络
麻雀算法
混沌映射
load prediction
extreme learning machine
BP neural network
sparrow algorithm
chaoticmapping
