基于QGA-LSSVM的能源需求预测

能源需求预测是能源规划和政策制定的前提和基础,能源需求预测受到众多因素的影响。为了快速、有效的预测我国对能源的需求,采用量子遗传算法(QGA)对最小二乘支持向量机(LSSVM)的参数进行优化,建立最优的能源预测模型。收集1997—2011年我国能源需求的相关数据作为训练样本和测试样本,对影响能源需求的指标数据,利用因子分析,对关联程度较高的指标数据进行公共因子的提取,减少判别指标间信息交互,通过预测模型的检验,并对比其他预测模型,验证了该模型在能源需求预测中具有极低的误差率。

Energy Demand Forecast based on the QGA-LSSVM model

Energy demand forecast is the premise and foundation of energy planning and policy making, energy demand forecast influenced by many factors. In order to quickly and effectively forecast the demand for energy in our country, the quantum genetic algo- rithm（QGA） is adopted to the least squares support vector machine （LSSVM） parameter optimization, established the optimal energy forecast model. Collect relevant data of China＇s energy demand from 1997 to 2011 as the training sample and test sample, affecting the energy demand of index data, using the factor analysis, the high degree of correlation of common factor of index data are extracted, re- duce information interaction between discriminant index, through forecasting model test, meanwhile compared with other prediction models, this model was verified in energy demand forecast has very low error rate.