基于PSO-ICA-BP神經(jīng)網(wǎng)絡的短期風電功率預測
王帥哲1,王金梅1,2,王永奇1,馬文濤1
(1 寧夏大學 物理與電子電氣工程學院,寧夏 銀川 750021;
2 寧夏沙漠信息智能感知自治區(qū)重點實驗室,寧夏 銀川 750021)
摘 要:針對傳統(tǒng)的BP神經(jīng)網(wǎng)絡對短期風電功率預測精度不高的缺點,提出粒子群算法改進帝國競爭算法(PSO-ICA),通過PSO算法改進殖民地同化操作提高ICA 算法的全局尋優(yōu)能力,輸出全局最優(yōu)解作為BP神經(jīng)網(wǎng)絡初始權(quán)值閾值。同時用主成分分析法降維壓縮輸入數(shù)據(jù),提高網(wǎng)絡泛化能力。利用PSOICA-BP預測模型對某風電場實際風電功率數(shù)據(jù)進行預測,仿真結(jié)果表明該模型預測誤差更小,對短期風電功率預測更有效。
關(guān)鍵詞:帝國競爭算法;粒子群算法;BP神經(jīng)網(wǎng)絡;風電功率預測
中圖分類號:TM614;TM715 文獻標識碼:A 文章編號:1007-3175(2019)02-0007-05
Short-Term Wind Power Forecast Based on PSO-ICA-BP Neural Network
WANG Shuai-zhe1, WANG Jin-mei1,2, WANG Yong-qi1, MA Wen-tao1
(1 School of Physics and Electronic-Electrical Engineering, Ningxia University, Yinchuan 750021, China;
2 Key Laboratory of Ningxia Desert Information Intelligent Perception Autonomous Region, Yinchuan 750021, China)
Abstract: In view of the shortcomings of the traditional BP neural network for short-term wind power prediction, the particle swarm optimization algorithm (PSO) was proposed to improve the Empire competition algorithm (PSO-ICA), to improve the diversity of colonial assimilation, and to optimize the initial weight threshold of the BP neural network by the output of the global optimal solution. The principal component analysis method was used to reduce dimension and to compress input data and improved the network generalization ability. The PSO-ICA-BP prediction model was used to predict the actual wind power data of certain wind farm. The simulation results show that the prediction error of this PSO-ICA-BP model is smaller and more effective for the short-term wind power prediction.
Key words: imperial competition algorithm; particle swarm optimization; BP neural network; wind power forcast
參考文獻
[1] 王焱,汪震,黃民翔,等. 基于OS-ELM和Bootstrap方法的超短期風電功率預測[J]. 電力系統(tǒng)自動化,2014,38(6):14-19.
[2] 張彥恒,鄭玉玉. 基于RBF神經(jīng)網(wǎng)絡的風電場功率預測研究[J]. 南方農(nóng)機,2018,49(7):192.
[3] 田淑慧,于惠鈞,趙巧紅,等. 基于經(jīng)驗模態(tài)分解的PSO-SVM風電功率短期預測[J]. 湖南工業(yè)大學學報,2018,32(3):59-64.
[4] 周松林,茆美琴,蘇建徽. 基于主成分分析與人工神經(jīng)網(wǎng)絡的風電功率預測[J]. 電網(wǎng)技術(shù),2011,35(9):128-132.
[5] 王強,汪姚,胡紅颯,等. 基于BP神經(jīng)網(wǎng)絡算法的風電功率預測[J]. 科技和產(chǎn)業(yè),2014,14(4):143-146.
[6] 劉帥, 劉長良. 基于帝國競爭算法的主汽溫控制系統(tǒng)參數(shù)優(yōu)化研究[J]. 系統(tǒng)仿真學報,2017,29(2):368-373.
[7] 楊曉博, 李陽, 肖朝霞, 等. 改進粒子群算法的自動阻抗匹配技術(shù)[J]. 重慶大學學報,2016,39(6):41-48.
[8] 朱曉青,馬定寰,李圣清,等. 基于BP神經(jīng)網(wǎng)絡的微電網(wǎng)蓄電池荷電狀態(tài)估計[J]. 電子測量與儀器學報,2017,31(12):2042-2048.
[9] 馬廣慧,馬豆豆,邵秀麗. 基于遺傳BP神經(jīng)網(wǎng)絡的三七價格預測[J]. 天津師范大學學報( 自然科學版),2017,7(6):76-80.
[10] LI Dong jie, LI Yang yang, LI Jun xiang, et al. Gesture Recognition Based on BP Neural Network Improved by Chaotic Genetic Algorithm[J]. International Journal of Automation and Computing,2018,15(3):267-276.
[11] 張曉東,楊圣祥. 基于PCA與NARX的市政工程造價組合預測[J]. 控制工程,2017,24(12):2485-2490.
[12] 李亞,蔣偉,樊汝森,等. 基于BP神經(jīng)網(wǎng)絡的智能臺區(qū)識別方法研究[J]. 電測與儀表,2017,54(3):25-30.
[13] 張立影,孟令甲,王澤忠. 基于雙層BP神經(jīng)網(wǎng)絡的光伏電站輸出功率預測[J]. 電測與儀表,2015,52(11):31-35.