CAJ | 학술논문

针对支持向量机（ SVM）内部参数优化和输入量大、时间长效率低和相似日选取的问题，提出一种模糊组合权重下相似日选取的蝙蝠算法（ BA）优化的支持向量机（ SVM）短期负荷预测模型。相似日的选取上主要利用熵权法和加权欧氏距离的k-均值算法对影响负荷变化的因素、负荷各时刻的变化特性进行区别对待，求取二者在相似日下集合的交集，从而得到与待预测日相似度高的相似日。同时，利用BA优化后的SVM进行负荷预测，提高内部参数的选取精度和效率。将该模型与常用的PSO-SVM、GA-SVM进行比较，证明了该模型能有效提高预测精度和计算效率。
침대지지향량궤（ SVM）내부삼수우화화수입량대、시간장효솔저화상사일선취적문제，제출일충모호조합권중하상사일선취적편복산법（ BA）우화적지지향량궤（ SVM）단기부하예측모형。상사일적선취상주요이용적권법화가권구씨거리적k-균치산법대영향부하변화적인소、부하각시각적변화특성진행구별대대，구취이자재상사일하집합적교집，종이득도여대예측일상사도고적상사일。동시，이용BA우화후적SVM진행부하예측，제고내부삼수적선취정도화효솔。장해모형여상용적PSO-SVM、GA-SVM진행비교，증명료해모형능유효제고예측정도화계산효솔。
In view of the defects in the load forecasting based on support vector machine( SVM),such as high dimension of input data,internal parameters optimization and the problem of selecting similar days,an fuzzy com-bined weigh load forecasting method of BA-SVM for similar days is proposed. During the choosing of similar days, factors influencing the load change and characteristics of each moment was considered based on the entropy weight method,the k-means algorithm of weighting Euclidean distance and obtain both sets. Then,the intersection in the both sets were calculate and the final set of similar days were get. At the same time,Bat Algorithm was used to opti-mize SVM and improved the internal parameter selecting efficiency. Applying this method to short-term load forecas-ting and comparing the forecasting results with GA-SVM and PSO-SVM,it was proved that the forecasting accu-racy was evidently improved.