摘要
提出了1种基于混沌分析和支持向量回归机的短期空调负荷预测建模方法。通过研究实际空调负荷序列的混沌特性,确定其混沌特征参数并选取支持向量回归机进行预测。支持向量机建模过程使用粒子群算法进行参数寻优。仿真结果表明,空调负荷序列具有一定的混沌特性,使用混沌支持向量机方法的预测精度比单一支持向量机法预测结果 EEP指标降低了31.4%,预测精度有了明显提升。
An effective short-term air-conditioning load forecasting method based on Chaos Analysis and Support Vector Machine was proposed. Its chaos characteristic parameters were determined by study on the actual airconditioning load and selection of SVR for forecasting. Particle swarm optimization algorithm was used for parameter optimization in the building process of a support vector machine model. Simulation results revealed that the airconditioning load had some chaos characteristic and the expected error percentage index by chaos analysis and support vector regression was reduced by 31. 4% than by the simple SVR method,while the prediction accuracy was improved significantly. This method provided new ideas for air conditioning load forecasting and a theoretical basis for air-conditioning energy conservation.
出处
《建筑科学》
CSCD
北大核心
2016年第6期102-107,共6页
Building Science
基金
广东省科技厅支撑项目"中央空调运行能效在线检测与优化控制"(2011B061200043)
关键词
短期空调负荷预测
混沌时间序列
支持向量回归机
粒子群算法
short-term air-condition load forecasting
chaotic time series
support vector regression（SVR）
particle swarm optimization algorithm
