考慮氣象因素的PCA-BP神經(jīng)網(wǎng)絡(luò)短期負(fù)荷預(yù)測

 

王海峰，姜雲(yún)騰，李萍

（寧夏大學(xué) 物理與電子電氣工程學(xué)院，寧夏 銀川 750021）

 

    摘 要：為有效提高電力系統(tǒng)短期負(fù)荷預(yù)測精度及效率，提出一種基于主成分分析的BP神經(jīng)網(wǎng)絡(luò)短期負(fù)荷預(yù)測優(yōu)化算法。利用主成分分析法將多個原始變量降維成少數(shù)彼此獨立的變量作為輸入，并根據(jù)各主成分的貢獻(xiàn)率來確定網(wǎng)絡(luò)的結(jié)構(gòu)，有效解決BP網(wǎng)絡(luò)預(yù)測精度與效率不高問題。在考慮氣象因素的影響下通過對某地區(qū)歷史負(fù)荷數(shù)據(jù)進(jìn)行訓(xùn)練仿真，平均預(yù)測精度接近98%，預(yù)測程序運行效率提高兩倍以上，仿真結(jié)果表明，該模型在效率和預(yù)測精度方面優(yōu)于BP神經(jīng)網(wǎng)絡(luò)模型。

    關(guān)鍵詞：主成分分析；負(fù)荷預(yù)測；BP 神經(jīng)網(wǎng)絡(luò)

    中圖分類號：TM715    文獻(xiàn)標(biāo)識碼：A     文章編號：1007-3175(2018)07-0038-04

    Abstract: In order to effectively improve the accuracy and efficiency of short-term load forecasting, this paper proposed a back propagation(BP) neural network short-term load forecasting optimization algorithm based on the principal component analysis. The principal component analysis method was used to reduce a number of original variables into a few independent variables as input, and to determine the network structure according to the contribution rate of the main components, and effectively solve the problem of low prediction accuracy and efficiency of BP network. Taking the influence of meteorological factors into consideration, the results of training and simulation of historical load data in a certain area show that the average prediction accuracy is close to 98%, which is more than two times of the running efficiency of the forecast program. The simulation results show that the model is superior to the BP neural network model in efficiency and prediction accuracy.

    Key words: principal component analysis; load forecasting; back propagation neural network