一种改进的指数平滑负荷预测方法研究

Optimized exponential smoothing for load forecast

提出了一种基于模糊理论优化的指数平滑法,该方法综合考虑环境温度、天气情况以及日期类型3个因素,基于一次指数平滑法的思想,建立了改进平滑法的预测模型。在充分理解一次指数平滑法基本原理的基础上,将天气情况和日期类型进行模糊化后,详细推导了改进平滑法的实现过程,并借助最小二乘法优化的模拟退火法得到了最优参数,最后将其运用到实际算例中。结果表明该方法预测精度高、数据平稳,有效减少了选择平滑系数所需的工作量,可以满足系统调度人员的工作需求,能够得到较好的实际运用。

A method for optimized exponential smoothing based on fuzzy theory is proposed, which considers the ambient temperature, weather conditions and type of dates three factors, and the optimized ex- ponential smoothing is estallished based on thought of single expenen- tial smoothing. Weather conditions and type of dates are, fuzzed based on undeltanding the basic principles of single exponential smoothing, and the implementation process of improved smoothing method is derived in detail. Simulated annealing based on least squares is adopted to obtain optimal parameters, finally which is used in actual engineering for ana- lyzing. Results show that the method has high precision forecast, and can reduce the required effort for selecting the smoothing parmneter, which meet the needs of the system operators and can get a better practical ap- plication.