CAJ | 학술논문

对影响电力负荷因素之间的非线性，有效提高电力负荷的预测精度，本文提出了一种最小二乘支持向量机（LSSVM）和粒子群优化技术（PSO）相结合的电力负荷预测方法。以历史负荷数据气象因素等作为输入，建立预测模型，对未来时刻电力负荷进行预测。该模型利用结构风险最小化原则代替传统的经验风险最小化，以充分挖掘原始数据的信息，并采用粒子群优化算法来优化最小二乘支持向量机的参数，旨在提高预测模型的训练预测精度。实际算例表明，使用PSO-LSSVM方法进行电力负荷预测，具有良好的可行性和有效性，与BP神经网络和LSSVM方法的预测结果相比，所提出的PSO-LSSVM模型预测平均误差仅为0.85%，具有更高的精度，适用于电力负荷预测。
대영향전력부하인소지간적비선성，유효제고전력부하적예측정도，본문제출료일충최소이승지지향량궤（LSSVM）화입자군우화기술（PSO）상결합적전력부하예측방법。이역사부하수거기상인소등작위수입，건립예측모형，대미래시각전력부하진행예측。해모형이용결구풍험최소화원칙대체전통적경험풍험최소화，이충분알굴원시수거적신식，병채용입자군우화산법래우화최소이승지지향량궤적삼수，지재제고예측모형적훈련예측정도。실제산례표명，사용PSO-LSSVM방법진행전력부하예측，구유량호적가행성화유효성，여BP신경망락화LSSVM방법적예측결과상비，소제출적PSO-LSSVM모형예측평균오차부위0.85%，구유경고적정도，괄용우전력부하예측。
In response to the non-linear effect on the power load factors, to improve the prediction accuracy of power load, this paper puts forward the power load forecasting method between LSSVM and PSO optimization al-gorithm. By taking historical load data, and meteorological factors, etc. as input, we can build the predictive mod-els to forecast the future power load. We make the structural risk minimization principle take the place of the tradi-tional empirical risk minimization in the model, in order to fully exploit the raw data. And we use the particle swarm to optimize the parameters of the least squares support vector machine, so as to improve the forecasting model training speed and forecast accuracy. Practical examples show that the PSO-LSSVM power load forecasting method has a good feasibility and effectiveness. When the prediction results are compared to the BP neural net-work and LSSVM method, the proposed PSO-LSSVM model predicts an average error of only 0.85% and has more high accuracy for power load forecasting.