CAJ | 학술논문

为了提高风电功率的预测精度，研究了一种基于粒子滤波(PF)与径向基函数(RBF)神经网络相结合的风电功率预测方法。使用PF算法对历史风速数据进行滤波处理，将处理后的风速数据结合风向、温度的历史数据，归一化后构成风电功率预测模型的新的输入数据；利用处理后的新的输入数据和输出数据，建立PF-RBF神经网络预测模型，预测风电场的输出功率。仿真结果表明，使用该预测模型进行风电功率预测，预测精度有一定的提高，连续120 h功率预测的平均绝对百分误差达到8.04%，均方根误差达到10.67%。
위료제고풍전공솔적예측정도，연구료일충기우입자려파(PF)여경향기함수(RBF)신경망락상결합적풍전공솔예측방법。사용PF산법대역사풍속수거진행려파처리，장처리후적풍속수거결합풍향、온도적역사수거，귀일화후구성풍전공솔예측모형적신적수입수거；이용처리후적신적수입수거화수출수거，건립PF-RBF신경망락예측모형，예측풍전장적수출공솔。방진결과표명，사용해예측모형진행풍전공솔예측，예측정도유일정적제고，련속120 h공솔예측적평균절대백분오차체도8.04%，균방근오차체도10.67%。
To improve accuracy of wind power prediction,this paper proposes a short-term wind power prediction method combining a particle filter (PF)and a radial basis function (RBF) neural network.Historical wind speed data are first processed with a particle filter.The processed wind speed data combined with the historical data of wind direction and temperature are used as in-put to the model.A PF-RBF neural network of wind power output prediction model is established using the new input data.Simulation results show that the proposed model is accurate in wind power prediction.The mean absolute percentage prediction error in a period of 120 hours has been reduced to 8.04%,and the root mean square error is 10.67%.