改进蚁群化学聚类算法在短期负荷预测中的应用

摘要

In order to improve the precision of short-term load forecasting and overcome the disadvantages of large amount of data and easy to be locally optimized, an improved AntClust algorithm is presented, which uses the kernel function to optimize factors of load forecasting and to decrease the data amount in the training process based on AntClust algorithm. The method can increase the forecasting speed and improve the precision. Through the short-term load forecasting in a city of Hunan, the result shows that the algorithm has a better forecasting accuracy compared with the methods of SVM and neural network. Theoretical analysis and experimental evidences show the efficiency and feasibility of the algorithm.%针对现有电力系统短期负荷预测精度低、数据处理量大、易陷入局部寻优等缺点,提出了一种改进蚁群化学聚类方法.该方法通过在蚁群化学聚类算法的基础上引入核函数来优化负荷预测因素,减少在训练过程中的数据量,提高了预测速度和精度.经过对湖南某市的短期负荷预测,并与SVM及神经网络对比,其结果表明该预测模型精度高于SVM与神经网络模型精度.理论分析和实验数据验证了该算法具有一定的实用性和可行性.