短期负荷预测"双周期加混沌"法中的多步法与气象因子的使用

短期负荷预测的"双周期加混沌"法是基于负荷记录数学性质的预测方法.为了进一步提高其预测精度而提出的三项改进为:(1)通过数值测试,优化了混沌子序列线性回归预测中三个参数,即嵌入空间维数、延时时间,和选择邻近矢量时采用的距离;(2)为了降低误差积累,采用了直接从邻近矢量回归出连续多个预测点的方法,并且采用离参考矢量最近的若干邻近矢量进行回归;(3)在选择直接多步法中的相空间邻近矢量时,考虑了与各负荷值相对应的当天气象因子.将改进后的双周期加混沌法用于负荷预测实例,结果表明预测精度有显著提高.

APPLICATION OF MULTI-STEP REGRESSION CONSIDERING CLIMATE FACTORS IN METHOD SYNTHESIZING DOUBLE PERIODS AND CHAOTIC COMPONENTS TO SHORT TERM LOAD FORECASTING

The method of synthesizing double periods and chaotic components for short term load forecasting is based on the mathematical characteristics of load records. To further improve its forecasting accuracy three modifications are put forward. The first is that through the numerical tests three parameters in the linear regression forecasting of chaotic subsequence, i.e., the dimensions of embedded space, the time delay and the distance to be used to choose adjacent vectors, are optimized. The second is that to reduce the accumulated error several continuous forecasting points are directly regressed from adjacent vectors and several vectors which are the nearest to the referential vectors are used for regression. The third is that during the selection of adjacent vectors in the phase space in direct multi-step regression method the climate factors corresponding to the value of load data in the very day are considered. Applying the method synthesizing double periods and chaotic components with above-mentioned modifications to practical load forecasting example, the result shows that the forecasted load data are more accurate.