基于WT和黏菌算法的LSSVM短期风功率预测

Short-Term Wind Power Prediction Using LSSVM Based on WT and Slime Mould Algorithm

针对风电出力存在随机性、波动性等问题,建立基于小波变换(WT)与黏菌算法(Slime mould algorithm, SMA)优化最小二乘支持向量机(LSSVM)关键参数的风功率预测模型。首先利用小波变换将风功率信号以及风速信号分解为多个不同频率的平稳的子序列,并提出采用一种黏菌优化算法优化LSSVM的参数,同时引入气象因素,包括风速、风向、温度、气压、湿度作为输入,分别建立模型来预测风电功率。通过将各个模型预测结果加和得到完整的风功率预测值。使用某风电厂数据进行仿真验证,实验结果表明,所提出的WT-SMA-LSSVM预测模型在短期风功率预测中具有更高的预测精度。

The wind power prediction model based on the wavelet transform(WT)and the Slime mould algorithm(SMA)is developed to optimise the key parame ters of the least squares support vector mach ine(LSSVM).A wavelet transform was used to decompose the wind power signal and the wind speed signal into several smooth subseries with different frequencies,and a slime mould algo rithm was proposed to optimise the parameter s of the LSSVM,while meteorological factors,including wind speed,wind direction,temperature,air pressure and hu midity,were introduced as inputs to build separate models to predict wind power.The complete wind power prediction was obtained by summing the individual model predictions.The simula tions were carried out using data from a wind power plant and the experimental results show that the WT-SMA-LSSVM prediction model proposed in this paper has higher prediction accuracy in short-term wind power prediction.