根据小区用电数据特征，提出了一种基于用电影响因素回归的小区用电预测模型。居民用电量影响因素分为可预知因素和不可预知因素，可预知因素利用pearson积矩相关系数计算所对应用电量的偏相关系数。并用0.618优选算法选取最优可预知因素。不可预知因素通过可预知因素偏相关系数分配权重，最后利用最小二乘支持向量机得到预测模型回归方程。以某小区日用电量进行实际预测为例，通过与其他模型的比较，验证该小区用电预测模型在小区用电预测的有效性。%Based on the characteristics of residential electricity consumption data,it proposes a residential elec-tricity consumption forecasting model based on Pearson correlation electricity consumption factors. Residential consumption factors are divided into predictable and unpredictable factors,predictable factors use Pearson cor-relation coefficient to calculate corresponding electricity consumption partial correlation coefficient. Use 0. 618 preferred algorithm to select optimal predictable factors. Unpredictable factors assign weights by partial correla-tion coefficient of the predictable factors. Finally,it uses least squares support vector machine to get the regres-sion equation. Some experiments are carried out using the historical daily electric power consumption of a com-munity to compare the improved model with some other forecasting models, and it proves that the improved model is more effective than other models in daily electric power consumption forecasting.
展开▼
