基于CLPSO优化LSSVM的风数据缺失部分插补

Interpolation of wind partial missing data based on CLPSO optimized LSSVM

风电场的实测风数据普遍存在着数据缺失的问题。由于风数据的非线性特点,常用的插补方法难以跟踪风的变化趋势,导致风数据缺失部分插补值精度低的问题。针对此问题,采用最小二乘支持向量机(LSSVM)模型插补缺失的风数据,使用综合学习粒子群算法(CLPSO)优化影响LSSVM模型性能的参数,从而形成了CLPSO-LSSVM插补模型。为了进行对比,另外使用了风切变指数模型(WSC)、自回归滑动平均模型(ARMA)、自适应神经模糊推理系统模型(ANFIS),对测试数据和风数据缺失部分进行插补。仿真结果表明:CLPSO-LSSVM模型的测试数据插补值精度最高,对风数据缺失部分插补值的相关系数也较大,综合指标最优,验证了该插补模型的有效性。

There is a problem that wind data are existed missing in the field of wind power, because of the nonlinear characteristics of wind data, the common interpolation methods are difficult to track the trend of wind, resulting in low accuracy of interpolation in the missing part of the wind data. Aiming at these problems, use the least squares support vector machine（LSSVM） model to interpolate the wind missing data, use the comprehensive learning particle swarm optimization（CLPSO） algorithm to optimize the parameters of the LSSVM model, thus, the CLPSO-LSSVM interpolation model is formed. In order to compare the results of the test data and missing wind data, the wind shear coefficient（WSC）model, autoregressive moving average（ARMA）model and adaptive neuro fuzzy inference system（ANFIS） model were used,the simulation results show that the accuracy of the CLPSO-LSSVM model is the highest, and the correlation coefficient is also relatively larger, and the effectiveness of the interpolation model is verified.