论文部分内容阅读
当前对于输变电工程不确定因素的研究集中于不确定因素识别,对其预测研究较少。不确定因素变化呈现出明显的非周期性及非平稳性,常规模型的预测精度较低。鉴于此,本文建立了EEMD-BP模型,将历史数据序列分解为随机分量和呈现明显波动周期的若干趋势分量。对各趋势分量采用BP算法预测,将各分量预测值叠加后得到趋势分量预测结果。对随机分量通过分析其相对于趋势分量的离散度得到其波动范围。综合考虑趋势分量预测值和波动区间得到最终的预测结果。选取PPI的历史数据进行算例分析,验证了方法的有效性,并对未来两个时点进行预测。采用同样方法对包括铜材价格、导线价格在内的其他不确定因素进行了预测。
The current research on uncertainties in power transmission and transformation projects focuses on the identification of uncertainties, and there are few studies on its prediction. Uncertainty changes show a significant non-periodic and non-stationary nature, and conventional models have lower prediction accuracy. In view of this, this paper establishes the EEMD-BP model, which decomposes the historical data sequence into random components and presents some trend components with obvious fluctuation cycles. The BP algorithm is used to predict each trend component, and the trend component prediction results are obtained after the component prediction values are superimposed. For the random component, its fluctuation range is obtained by analyzing its dispersion with respect to the trend component. The final forecast result is obtained by considering the trend component forecast value and the fluctuation range comprehensively. The historical data of PPI was selected to analyze the examples. The validity of the method was verified and the prediction was made at two future time points. In the same way, other uncertainties including copper price and wire price were predicted.