收稿日期:2022-08-19
基金項(xiàng)目:陜西省教育廳一般專項(xiàng)科研計(jì)劃項(xiàng)目(21JK0770);國(guó)家自然科學(xué)基金(41701442;41977059)
通信作者:張淑花(1988—),女,博士、副教授,主要從事復(fù)雜地形區(qū)地表太陽(yáng)輻射模擬等方面的研究。shuhuazhang@xust.edu.cn
DOI:10.19912/j.0254-0096.tynxb.2022-1246 文章編號(hào):0254-0096(2023)12-0150-12
摘 要:根據(jù)太陽(yáng)輻射預(yù)測(cè)使用的數(shù)據(jù)及預(yù)測(cè)方法,將目前地表太陽(yáng)輻射短期預(yù)測(cè)方法歸納總結(jié)為4類:基于地面觀測(cè)數(shù)據(jù)預(yù)測(cè)方法、基于衛(wèi)星遙感觀測(cè)預(yù)測(cè)方法、基于地基云圖觀測(cè)預(yù)測(cè)方法以及基于數(shù)值天氣預(yù)報(bào)模式預(yù)測(cè)方法。分別闡述4類地表太陽(yáng)輻射短期預(yù)測(cè)方法的研究進(jìn)展,并對(duì)不同方法適用性及其優(yōu)缺點(diǎn)進(jìn)行評(píng)價(jià),最后對(duì)未來(lái)短期地表太陽(yáng)輻射預(yù)測(cè)方法進(jìn)行展望。
關(guān)鍵詞:太陽(yáng)輻射;遙感;預(yù)測(cè);統(tǒng)計(jì)模型;全天空成像儀;數(shù)值天氣預(yù)報(bào)模式
中圖分類號(hào):P40 文獻(xiàn)標(biāo)志碼:A
0 引 言
地表太陽(yáng)輻射(surface solar radiation,SSR)是由太陽(yáng)以電磁波的形式向外發(fā)射,經(jīng)過(guò)大氣吸收和散射后,地表所接收的短波輻射[1-2]。它是地球能量收支的重要組成部分,是地球系統(tǒng)的主要驅(qū)動(dòng)力,也是清潔能源之一[3-4]。同時(shí),SSR是農(nóng)業(yè)、生態(tài)、水文、氣象等領(lǐng)域模型的重要輸入?yún)?shù),也是太陽(yáng)能發(fā)電工程的重要指標(biāo)[5-6],對(duì)地表過(guò)程研究、太陽(yáng)能資源評(píng)估和光伏發(fā)電設(shè)計(jì)等具有重要意義。由于太陽(yáng)輻射受云、氣溶膠、臭氧、水汽等大氣環(huán)境因素的影響,尤其因云運(yùn)動(dòng)引起的遮擋,造成短期SSR波動(dòng)性大、隨機(jī)性強(qiáng)[7-10]。因而SSR短期預(yù)測(cè)對(duì)太陽(yáng)輻射波動(dòng)性模擬更為重要。
最初對(duì)于太陽(yáng)輻射預(yù)測(cè)的主要數(shù)據(jù)來(lái)源于地面站點(diǎn)觀測(cè)。因而本研究將第1類SSR短期預(yù)測(cè)方法歸納為基于地面觀測(cè)數(shù)據(jù)的預(yù)測(cè)方法。此類模型主要利用太陽(yáng)輻射歷史觀測(cè)數(shù)據(jù)本身的統(tǒng)計(jì)特征或與易獲取的影響太陽(yáng)輻射的因子觀測(cè)數(shù)據(jù),如氣象因子等,通過(guò)對(duì)影響因子和太陽(yáng)輻射擬合形成統(tǒng)計(jì)模型,從而用于短期SSR預(yù)測(cè)。其中根據(jù)太陽(yáng)輻射歷史觀測(cè)數(shù)據(jù)構(gòu)建預(yù)測(cè)模型,核心為提取歷史數(shù)據(jù)的時(shí)間序列特征,如自相關(guān)性、周期性等,從而構(gòu)建統(tǒng)計(jì)模型[11-14]。此類模型有自回歸模型(autoregressive,AR)、自回歸滑動(dòng)平均模型(autoregressive moving average,ARMA)以及小波變換等。而利用與其相關(guān)因子構(gòu)建模型的方法主要在于因子選擇,如最高和最低溫度、風(fēng)速、降雨量、露點(diǎn)溫度和大氣壓力等常作為回歸因子。擬合多元線性回歸、神經(jīng)網(wǎng)絡(luò)、深度學(xué)習(xí)、混合模型等多因子線性及非線性預(yù)測(cè)模型為基于地面觀測(cè)數(shù)據(jù)的常用預(yù)測(cè)方法[15-19]。
由于觀測(cè)站點(diǎn)分布稀疏,較難獲取大范圍代表性觀測(cè)數(shù)據(jù),因此利用地面觀測(cè)數(shù)據(jù)形成統(tǒng)計(jì)模型對(duì)于獲取SSR的空間分布較為困難。而衛(wèi)星遙感則能快速且大面積觀測(cè),可較好地獲取SSR時(shí)空變化,因此衛(wèi)星遙感觀測(cè)成為研究SSR的重要手段[20-22]。云是影響SSR短期精確預(yù)報(bào)的關(guān)鍵因子,目前較多衛(wèi)星能獲得云光學(xué)厚度、云量以及云反射率等信息,為基于衛(wèi)星觀測(cè)數(shù)據(jù)預(yù)測(cè)SSR提供了可靠數(shù)據(jù)源。因而本文將第2類方法歸納為基于衛(wèi)星觀測(cè)數(shù)據(jù)的預(yù)測(cè)方法,且該類方法最常用氣象衛(wèi)星,因?yàn)闅庀笮l(wèi)星產(chǎn)品具有較高的時(shí)間分辨率,可達(dá)到短期預(yù)測(cè)的目的。在基于衛(wèi)星的預(yù)測(cè)方法中,主要預(yù)測(cè)手段為: 從連續(xù)的衛(wèi)星圖像獲得云運(yùn)動(dòng)矢量(cloud motion vector,CMV),結(jié)合經(jīng)驗(yàn)或統(tǒng)計(jì)模型進(jìn)行短期預(yù)報(bào)。如楊麗薇等[23]通過(guò)從風(fēng)云-4衛(wèi)星圖像獲得連續(xù)的云反照率,采用粒子圖像測(cè)速法獲得云運(yùn)動(dòng)矢量,根據(jù)晴空模型以及Heliosat-2預(yù)測(cè)了30~180 min的總太陽(yáng)輻照度。
全天空成像儀(total sky imager,TSI)可實(shí)時(shí)提供半球天空云特征信息,因而它對(duì)特定點(diǎn)SSR的短期預(yù)測(cè)更精確,甚至可進(jìn)行超短期預(yù)測(cè)[24-25]。本研究將第3類方法歸納為基于地基云圖的預(yù)測(cè)方法。目前此方法主要基于地基云圖觀測(cè),結(jié)合簡(jiǎn)單統(tǒng)計(jì)模型、機(jī)器學(xué)習(xí)方法、混合模型預(yù)測(cè)短期甚至超短期太陽(yáng)輻射[26-30]。如蔣俊霞等[31]基于全天空成像儀獲取的天空?qǐng)D像所計(jì)算云分?jǐn)?shù),分別采用線性回歸和反向傳播神經(jīng)網(wǎng)絡(luò),預(yù)測(cè)了1~5 min的太陽(yáng)輻照度。由于全天空成像儀價(jià)格相對(duì)較高,因此利用數(shù)碼相機(jī)如高分辨率數(shù)碼單反相機(jī)(digital single-lens reflex,DSLR)等結(jié)合魚(yú)眼鏡頭獲取天空?qǐng)D像從而代替全天空成像儀觀測(cè),同樣可用于短期太陽(yáng)輻照度預(yù)測(cè)[32]。
以上基于不同數(shù)據(jù)源的預(yù)測(cè)方法主要依賴于數(shù)據(jù)變化規(guī)律,利用簡(jiǎn)單統(tǒng)計(jì)模型或智能化算法進(jìn)行預(yù)測(cè),缺少一定的物理機(jī)制。而數(shù)值天氣預(yù)報(bào)模式,則根據(jù)大氣動(dòng)力學(xué)和熱力學(xué)原理,給定大氣的初始狀態(tài)下可推算出預(yù)測(cè)時(shí)刻的大氣狀態(tài),從而預(yù)測(cè)太陽(yáng)輻射[33]。目前,數(shù)值天氣預(yù)報(bào)模式已成為太陽(yáng)輻射預(yù)報(bào)有利工具,因此本研究將第4類短期預(yù)測(cè)方法歸納為基于數(shù)值天氣預(yù)報(bào)模式預(yù)測(cè)方法。該類模型具有完備的物理機(jī)制,從動(dòng)力學(xué)角度預(yù)測(cè)未來(lái)地表太陽(yáng)輻射。同時(shí)遙感數(shù)據(jù)以及全天空成像儀觀測(cè)數(shù)據(jù)都可能成為數(shù)值天氣預(yù)報(bào)模式的輸入數(shù)據(jù),從而使其短期預(yù)測(cè)更加精確。如Arbizu-Barrena等[34]將從MSG衛(wèi)星圖像得到的云量指數(shù)與數(shù)值天氣預(yù)報(bào)模式相結(jié)合,預(yù)測(cè)了未來(lái)6小時(shí)總太陽(yáng)輻照度和直接輻照度。
近年來(lái),較多研究者對(duì)太陽(yáng)輻射的不同預(yù)測(cè)方法進(jìn)行了總結(jié),其中周勇等[35]對(duì)使用機(jī)器學(xué)習(xí)方法預(yù)測(cè)總太陽(yáng)輻照度進(jìn)行了全面系統(tǒng)總結(jié)。Guermoui等[36]對(duì)太陽(yáng)輻射預(yù)測(cè)的混合模型進(jìn)行了論述,并對(duì)不同的混合模型進(jìn)行比較研究。Aicardi等[37]對(duì)基于衛(wèi)星的每小時(shí)太陽(yáng)輻射預(yù)測(cè)方法進(jìn)行了綜述,總結(jié)了不同方法的最優(yōu)參數(shù)及其性能評(píng)估和比較。Lin等[38]系統(tǒng)綜述了基于全天空成像儀的小時(shí)內(nèi)太陽(yáng)輻射預(yù)報(bào)的最新進(jìn)展以及可用的公開(kāi)數(shù)據(jù)集。Chu等[39]對(duì)小時(shí)內(nèi)太陽(yáng)輻照度預(yù)測(cè)方法的進(jìn)展進(jìn)行了全面回顧和簡(jiǎn)明總結(jié)。以上綜述文章主要針對(duì)不同類型模型分別進(jìn)行論述總結(jié),缺少對(duì)現(xiàn)有方法的總結(jié)歸納及對(duì)比。本文根據(jù)太陽(yáng)輻射預(yù)測(cè)使用的數(shù)據(jù)以及采用的預(yù)測(cè)方法總結(jié)了太陽(yáng)輻射短期預(yù)測(cè)方法的研究進(jìn)展,并總結(jié)對(duì)比不同方法的優(yōu)缺點(diǎn)及其適用性,為未來(lái)太陽(yáng)輻射短期預(yù)測(cè)模型改進(jìn)及其發(fā)展提供參考。
由于不同應(yīng)用領(lǐng)域?qū)Χ唐陬A(yù)測(cè)的時(shí)效性定義不同,如氣象領(lǐng)域美國(guó)國(guó)家海洋和大氣管理局(National Oceanic and Atmospheric Administration,NOAA)定義小于6 h預(yù)報(bào)為短期預(yù)報(bào)[40],而中國(guó)在光伏應(yīng)用領(lǐng)域相關(guān)國(guó)家標(biāo)準(zhǔn)中將短期預(yù)測(cè)時(shí)效性定義為72 h以內(nèi)[41],因此本文所涉及短期預(yù)測(cè)的時(shí)效性綜合考慮以上兩個(gè)定義的標(biāo)準(zhǔn)。
1 SSR短期預(yù)測(cè)方法
1.1 基于地面觀測(cè)數(shù)據(jù)預(yù)測(cè)方法
基于地面觀測(cè)數(shù)據(jù)的預(yù)測(cè)方法主要為統(tǒng)計(jì)模型。使用統(tǒng)計(jì)模型進(jìn)行SSR短期預(yù)測(cè)的基本原理為:通過(guò)分析歷史觀測(cè)數(shù)據(jù)資料的相關(guān)性、周期性等特征,建立統(tǒng)計(jì)模型,以歷史太陽(yáng)輻射值或氣象因子觀測(cè)數(shù)據(jù)為輸入數(shù)據(jù),實(shí)現(xiàn)對(duì)未來(lái)時(shí)刻的太陽(yáng)輻射值的預(yù)測(cè)。傳統(tǒng)的統(tǒng)計(jì)方法多以線性預(yù)測(cè)方法為主。典型模型如[I(h,nj)]模型,[I(h,nj)=A+B×][e-μ(nj)x(h)×cos(2πh24)e-μ(nj)x(h-12)],其中[h]為一天中的某小時(shí),[nj]為一年中的某天,[x(h)]為第[h]小時(shí)的太陽(yáng)光在大氣層中傳播的距離,[μ(nj)]為第[nj]天的衰減系數(shù)。此模型是對(duì)研究點(diǎn)半天內(nèi)的觀測(cè)數(shù)據(jù)擬合獲得未知參數(shù)[A]和[B]進(jìn)而預(yù)測(cè)一年中任意一天任意小時(shí)的太陽(yáng)輻射[42]。另一類典型模型為自回歸差分移動(dòng)平均模型(autoregressive integrated moving average,ARIMA)、自回歸移動(dòng)平均模型(autoregressive moving average,ARMA),根據(jù)歷史數(shù)據(jù)自相關(guān)特征,建立回歸模型,從而預(yù)測(cè)未來(lái)時(shí)刻太陽(yáng)輻射。徐維[43]基于歷史地表太陽(yáng)輻射數(shù)據(jù)的自相關(guān)性,建立4種時(shí)間尺度的ARMA模型,用于預(yù)測(cè)未來(lái)不同時(shí)間尺度太陽(yáng)輻射。Reikard[44]、Das[45]同樣利用ARIMA模型實(shí)現(xiàn)了15和30 min太陽(yáng)輻射的預(yù)測(cè)。
然而在實(shí)際獲取的數(shù)據(jù)中,由于云、氣溶膠等大氣因素的影響,導(dǎo)致太陽(yáng)輻射觀測(cè)時(shí)間序列具有突變和強(qiáng)波動(dòng)性。針對(duì)此現(xiàn)象,更多學(xué)者提出利用隨機(jī)過(guò)程模型、神經(jīng)網(wǎng)絡(luò)模型、混合模型等對(duì)其突變和時(shí)間序列波動(dòng)性進(jìn)行模擬。如張雪松等[46]利用經(jīng)小波變換處理的太陽(yáng)輻射觀測(cè)數(shù)據(jù),建立太陽(yáng)輻射短期預(yù)測(cè)神經(jīng)網(wǎng)絡(luò)模型和支持向量機(jī)模型。隨著計(jì)算性能及人工智能算法的快速發(fā)展,構(gòu)建混合智能化算法模型預(yù)測(cè)短期太陽(yáng)輻射成為目前的主要發(fā)展趨勢(shì)。蔣鋒等[47]將奇異譜分析(singular spectrum analysis,SSA)技術(shù)、支持向量機(jī)(support vector machine,SVM)和K-均值聚類結(jié)合,形成太陽(yáng)輻射短期預(yù)測(cè)框架。Mukhtar等[48]提出混合神經(jīng)網(wǎng)絡(luò):卷積神經(jīng)網(wǎng)絡(luò)/人工神經(jīng)網(wǎng)絡(luò)(convolutional neural network/artificial neural network,CNN-ANN)和卷積神經(jīng)網(wǎng)絡(luò)/長(zhǎng)短時(shí)記憶網(wǎng)絡(luò)/人工神經(jīng)網(wǎng)絡(luò)(convolutional neural network/long short-term memory/artificial neural network,CNN-LSTM-ANN)對(duì)歷史太陽(yáng)輻射數(shù)據(jù)進(jìn)行網(wǎng)絡(luò)訓(xùn)練,進(jìn)而形成小時(shí)尺度太陽(yáng)輻射預(yù)測(cè)模型。
上述模型主要利用太陽(yáng)輻射觀測(cè)數(shù)據(jù)本身統(tǒng)計(jì)特征進(jìn)行預(yù)測(cè),基于地面觀測(cè)數(shù)據(jù)的另一種模型則為利用與太陽(yáng)輻射相關(guān)因子構(gòu)建統(tǒng)計(jì)模型并預(yù)測(cè)太陽(yáng)輻射。此類模型更多采用多元線性回歸、機(jī)器學(xué)習(xí)等多元線性及非線性擬合方法。周滿國(guó)等[49]、倪超等[50]將多個(gè)氣象參數(shù)作為輸入變量,分別用基于深度學(xué)習(xí)的混合預(yù)測(cè)方法和雙向長(zhǎng)短期記憶網(wǎng)絡(luò)預(yù)測(cè)模型對(duì)未來(lái)15分鐘太陽(yáng)輻照度進(jìn)行了預(yù)測(cè)。朱婷婷等[51]對(duì)影響太陽(yáng)輻射量的相關(guān)氣象數(shù)據(jù)進(jìn)行相關(guān)性分析,選出最優(yōu)指標(biāo)作為預(yù)測(cè)方法的輸入,采用集成模型預(yù)測(cè)未來(lái)15分鐘太陽(yáng)輻照度。
總結(jié)以上研究進(jìn)展可看出早期的太陽(yáng)輻射預(yù)測(cè)主要為簡(jiǎn)單的統(tǒng)計(jì)預(yù)測(cè)模型,但太陽(yáng)輻射受到多種氣象因素影響,會(huì)出現(xiàn)強(qiáng)烈的波動(dòng)性,導(dǎo)致該類模型的預(yù)測(cè)結(jié)果不理想。因此較多學(xué)者將非線性的模型用于太陽(yáng)輻射的預(yù)測(cè),其中神經(jīng)網(wǎng)絡(luò)、深度學(xué)習(xí)等智能化算法以及多算法混合模式成為主要的非線性統(tǒng)計(jì)預(yù)測(cè)模型。但以上無(wú)論是單一統(tǒng)計(jì)模型還是混合統(tǒng)計(jì)模型,均需大量歷史太陽(yáng)輻射數(shù)據(jù)及與太陽(yáng)輻射相關(guān)因子進(jìn)行擬合或訓(xùn)練模型參數(shù),對(duì)于歷史觀測(cè)數(shù)據(jù)質(zhì)量較高的區(qū)域(獲取數(shù)據(jù)時(shí)間分辨率高且觀測(cè)質(zhì)量高)是較好的選擇。而對(duì)于缺少觀測(cè)數(shù)據(jù)區(qū)域,如地形復(fù)雜的山區(qū)以及荒漠區(qū),仍存在一定局限性,同時(shí)構(gòu)建該類模型所選擇因子大都與太陽(yáng)輻射的關(guān)系是間接的,因此這類模型無(wú)法較好地考慮大氣環(huán)境以及地面環(huán)境對(duì)太陽(yáng)輻射的直接影響。且太陽(yáng)輻射預(yù)測(cè)主要與當(dāng)?shù)氐臍夂驐l件有關(guān),而模型中的參數(shù)通常為本地?cái)?shù)據(jù)擬合獲得[52],使得此類模型較難直接用于其他地區(qū),因而其普適性較差。
1.2 基于衛(wèi)星遙感觀測(cè)預(yù)測(cè)方法
云是影響地表太陽(yáng)輻射衰減的主要因素之一,因此地表太陽(yáng)輻射短期預(yù)測(cè)的關(guān)鍵在于未來(lái)時(shí)刻云對(duì)太陽(yáng)遮擋的預(yù)測(cè)及遮擋程度的量化[53-54]。目前,較多研究利用遙感觀測(cè)獲取云屬性信息,并采用云運(yùn)動(dòng)矢量場(chǎng)估算云運(yùn)動(dòng)速度,獲取云動(dòng)態(tài)過(guò)程,實(shí)現(xiàn)云運(yùn)動(dòng)過(guò)程預(yù)測(cè),結(jié)合經(jīng)驗(yàn)或統(tǒng)計(jì)模型進(jìn)而預(yù)測(cè)未來(lái)太陽(yáng)輻射量。Hammer等[55]以及Lorenz等[56]分別基于地球靜止衛(wèi)星Meteosat及歐洲第二代靜止軌道氣象衛(wèi)星(meteosat second generation,MSG)提供的云屬性數(shù)據(jù),利用云運(yùn)動(dòng)矢量場(chǎng)、線性模型以及Heliosat方法,分別預(yù)測(cè)了未來(lái)30分鐘和4小時(shí)太陽(yáng)輻射。Cros等[57]、楊麗薇等[58]分別利用MSG衛(wèi)星搭載的自旋增強(qiáng)可見(jiàn)光和紅外成像儀(spinning enhanced visible and infrared imager,SEVIRI)和風(fēng)云-4A觀測(cè)所獲取的云數(shù)據(jù),將晴空模型與計(jì)算出的云運(yùn)動(dòng)矢量場(chǎng)結(jié)合,實(shí)現(xiàn)太陽(yáng)輻射短期預(yù)測(cè)。目前,計(jì)算云運(yùn)動(dòng)矢量場(chǎng)的方法主要有:Lucas-Kanade方法、Farneb?ck方法、Horn和Schunck方法、TVL1方法、粒子圖像測(cè)速(particle image velocimetry,PIV)法、塊匹配算法等[59-60]。
與地面觀測(cè)數(shù)據(jù)的預(yù)測(cè)方法相似,混合模型以及人工智能算法目前也較多用于基于衛(wèi)星觀測(cè)數(shù)據(jù)的太陽(yáng)輻射短期預(yù)測(cè)。該類方法利用衛(wèi)星觀測(cè)作為輸入,同時(shí)用地面觀測(cè)數(shù)據(jù)用于訓(xùn)練或驗(yàn)證模型,進(jìn)而形成基于衛(wèi)星觀測(cè)數(shù)據(jù)的智能化算法及混合模型算法。如Nielsen等[61]提出基于衛(wèi)星的神經(jīng)網(wǎng)絡(luò)模型,簡(jiǎn)稱IrradianceNet,對(duì)歐洲未來(lái)4小時(shí)的地表太陽(yáng)輻照度進(jìn)行了預(yù)測(cè)。Marquez等[62]基于衛(wèi)星觀測(cè)數(shù)據(jù)得到云分?jǐn)?shù)、云量指數(shù)和云運(yùn)動(dòng)矢量場(chǎng),作為人工神經(jīng)網(wǎng)絡(luò)模型的輸入數(shù)據(jù),預(yù)測(cè)了30、60、90和120 min的總太陽(yáng)輻照度。Dong等[63]基于遙感云量指數(shù),利用混合指數(shù)平滑狀態(tài)空間(exponential smoothing state space,ESSS)模型和人工神經(jīng)網(wǎng)絡(luò)(artificial neural networks,ANN)預(yù)測(cè)小時(shí)尺度的太陽(yáng)輻照度。
基于遙感的SSR短期預(yù)測(cè)模型的優(yōu)點(diǎn)為能夠?qū)SR的時(shí)空分布進(jìn)行較好的預(yù)測(cè)。但受衛(wèi)星自身傳感器、大氣以及地形的影響,導(dǎo)致其獲取的云信息存在一定的不確定性,且遙感影像空間分辨率有限,空間上代表一定區(qū)域,而不是精確的點(diǎn),對(duì)于特定站點(diǎn)的短期預(yù)測(cè)也存在一定的局限性[64]。而且衛(wèi)星捕捉云的快速演變及其運(yùn)動(dòng)的能力相對(duì)有限,因此對(duì)未來(lái)時(shí)刻云位置預(yù)測(cè)同樣存在一定的不確定性。
1.3 基于地基云圖觀測(cè)預(yù)測(cè)方法
使用全天空成像儀(total sky imager,TSI)進(jìn)行SSR短期預(yù)測(cè)的基本原理為: 全天空成像儀可提供可見(jiàn)天空的整體狀態(tài)以及瞬間的云特征信息,尤其是與云量、云層分類和組成、以及所觀察到的平均移動(dòng)方向有關(guān)的信息[65]。通過(guò)TSI獲得的圖像計(jì)算出云運(yùn)動(dòng)矢量以及云分?jǐn)?shù)等,并將其作為經(jīng)驗(yàn)或統(tǒng)計(jì)模型輸入,形成短期太陽(yáng)輻射預(yù)測(cè)模型;或利用深度學(xué)習(xí)等智能化統(tǒng)計(jì)方法直接建立TSI獲取的天空?qǐng)D像與輻照度之間的關(guān)系,進(jìn)行短期預(yù)測(cè)。Marquez等[66]通過(guò)對(duì)全天空成像儀獲得的云圖進(jìn)行處理,計(jì)算云運(yùn)動(dòng)矢量場(chǎng)和云分?jǐn)?shù),結(jié)合持久性模型預(yù)測(cè)了15分鐘內(nèi)直接輻照度。Feng等[67-68]基于全天空成像儀獲取的天空?qǐng)D像,分別用深度卷積神經(jīng)網(wǎng)絡(luò)模型和深度學(xué)習(xí)方法,預(yù)測(cè)未來(lái)1小時(shí)和10~60分鐘太陽(yáng)輻照度。
然而,全天空成像儀價(jià)格昂貴,無(wú)法在低成本的情況下進(jìn)行廣泛應(yīng)用[69],隨著價(jià)格低廉的數(shù)碼相機(jī)和更強(qiáng)大的圖形處理技術(shù)的出現(xiàn)。一些學(xué)者使用數(shù)碼相機(jī)結(jié)合魚(yú)眼鏡頭代替全天空成像儀獲取數(shù)據(jù),得以實(shí)現(xiàn)低成本獲取天空?qǐng)D像。Alonso-Montesinos等[70]通過(guò)RGB和HSV顏色空間將天空相機(jī)中的數(shù)字圖像轉(zhuǎn)換為輻照度,利用云運(yùn)動(dòng)矢量場(chǎng)和最大互相關(guān)法預(yù)測(cè)未來(lái)1~180分鐘的輻照度值。Ai等[71]基于低成本的魚(yú)眼相機(jī),通過(guò)稀疏光流法、自適應(yīng)閾值方法以及GHI和云分?jǐn)?shù)的經(jīng)驗(yàn)公式實(shí)現(xiàn)了極短期太陽(yáng)輻照度的預(yù)測(cè)。
使用地基云圖預(yù)測(cè)短期太陽(yáng)輻射的方法優(yōu)點(diǎn)是其時(shí)間間隔可設(shè)定,采樣頻率較高,因此可進(jìn)行超短期預(yù)測(cè),對(duì)于單點(diǎn)太陽(yáng)能預(yù)測(cè)是較好的選擇[72],但同樣缺點(diǎn)是空間范圍的適用性非常有限,且該方法是由云圖的采樣頻率和時(shí)間跨度決定了預(yù)測(cè)精度和算法的復(fù)雜度。云圖采樣頻率高,需處理的圖像較多,增加了算法的復(fù)雜度。若為了提高運(yùn)算效率,降低云圖的采樣頻率,則可能存在確定云運(yùn)動(dòng)速度和方向預(yù)測(cè)不夠準(zhǔn)確的問(wèn)題,如何平衡算法復(fù)雜度和預(yù)測(cè)精度是該方法的關(guān)鍵點(diǎn)和難點(diǎn)[73]。
1.4 基于數(shù)值天氣預(yù)報(bào)模式的預(yù)測(cè)方法
數(shù)值天氣預(yù)報(bào)(numerical weather prediction,NWP)模式是通過(guò)物理過(guò)程參數(shù)化來(lái)描述不同尺度的天氣過(guò)程[74],同時(shí)它也可通過(guò)復(fù)雜的物理模型實(shí)現(xiàn)對(duì)太陽(yáng)輻射的預(yù)測(cè)。數(shù)值預(yù)報(bào)模式可分為:全球NWP模式:可描述全球范圍的大氣過(guò)程,但分辨率較粗,無(wú)法顯示小尺度的天氣特征,如大氣環(huán)流模式(general circulation models,GCM)和歐洲中央氣象中心(European Centre for Medium-Range Weather Forecasts,ECMWF)模式;而中尺度NWP模式(或區(qū)域性模型): 其應(yīng)用可限于區(qū)域,分辨率相對(duì)較高,如第五代中尺度模型(the fifth generation mesoscale model,MM5)、天氣研究與預(yù)報(bào)(weather research and forecasting,WRF)模式等均被用于太陽(yáng)輻射預(yù)報(bào)中[75-76]。
目前WRF及其衍生的模型較常用于地表太陽(yáng)輻射短期預(yù)測(cè)。其中WRF-Solar、MAD-WRF、WRF-CLDDA等為近年來(lái)常用的太陽(yáng)輻射短期預(yù)測(cè)模式。WRF-Solar是基于WRF模式開(kāi)發(fā)的擴(kuò)展模型,專門(mén)為太陽(yáng)能資源評(píng)估和預(yù)報(bào)需求設(shè)計(jì)的數(shù)值天氣預(yù)報(bào)模式。它改進(jìn)了氣溶膠-輻射反饋、云-輻射反饋以及云-氣溶膠相互作用,在很大程度上減少了輻照度預(yù)測(cè)的誤差[77-80]。MAD-WRF是將MADCast模型的基本理念與WRF-Solar中的云-氣溶膠-輻射物理學(xué)相結(jié)合,具有云初始化系統(tǒng)和更好的物理性能,能提供云的三維分析以及云演變,從而得到更精確地云的位置,為更好地短期輻照度預(yù)測(cè)提供了可能[81-82]。如Jiménez等[83]基于GOES-16衛(wèi)星中的云微物理參數(shù),結(jié)合MAD-WRF模型,預(yù)測(cè)了0~6小時(shí)的總太陽(yáng)輻照度。而WRF-CLDDA是使用WRF架構(gòu)和直接云同化系統(tǒng),可以直接同化衛(wèi)星獲得的云信息,從而改進(jìn)云天條件下的預(yù)報(bào)精度 [84-85]。
盡管NWP模型考慮了主要物理過(guò)程,但不確定的初始條件、復(fù)雜的局部地形、數(shù)值誤差等影響仍會(huì)導(dǎo)致結(jié)果不準(zhǔn)確,且這些錯(cuò)誤可能存在一定程度的系統(tǒng)性誤差,使得NWP輸出無(wú)法達(dá)到預(yù)期精度[86-87]。因此,許多學(xué)者將NWP與后處理方法相結(jié)合,旨在從過(guò)去的NWP預(yù)測(cè)誤差中學(xué)習(xí),糾正未來(lái)偏差。目前,用于改善NWP輻照度預(yù)測(cè)的后處理方法主要有線性逐步多元回歸、支持向量回歸、局部回歸、神經(jīng)網(wǎng)絡(luò)等[88]。如Lauret等[89]通過(guò)使用人工神經(jīng)網(wǎng)絡(luò)對(duì)WRF模型預(yù)測(cè)的前一天(時(shí)間分辨率為1 h)總太陽(yáng)輻照度進(jìn)行偏差校正。
雖然數(shù)值天氣預(yù)報(bào)的預(yù)測(cè)方法考慮了大氣物理過(guò)程參數(shù)化方案,可實(shí)現(xiàn)不同尺度的天氣預(yù)報(bào),也有較多后處理方法進(jìn)行誤差校正,但數(shù)值模擬方法中的氣象和環(huán)境因素比較復(fù)雜,準(zhǔn)確模擬未來(lái)大氣狀況較為困難,所以預(yù)報(bào)的誤差不僅存在,而且對(duì)于短期大氣變化狀況較大的情況,太陽(yáng)輻照度預(yù)測(cè)準(zhǔn)確度會(huì)大大降低。同時(shí)限制NWP模式預(yù)測(cè)精度另一個(gè)因素是模式的空間分辨率,由于模式復(fù)雜,計(jì)算過(guò)程效率相對(duì)較低,因此其輸出結(jié)果空間分辨率相對(duì)較低[90]。
2 方法對(duì)比
通過(guò)總結(jié)基于地面觀測(cè)數(shù)據(jù)的統(tǒng)計(jì)預(yù)測(cè)模型可發(fā)現(xiàn),為了實(shí)現(xiàn)對(duì)SSR進(jìn)行幾分鐘到幾小時(shí)尺度內(nèi)的精確預(yù)測(cè),預(yù)測(cè)模型不斷發(fā)生變化,由線性方法到機(jī)器學(xué)習(xí)方法和多模型混合方法。最常用的模型包括線性回歸模型、自回歸滑動(dòng)平均模型、神經(jīng)網(wǎng)絡(luò)模型、支持向量機(jī)、深度學(xué)習(xí)等。為了更進(jìn)一步提高預(yù)測(cè)的精度,通過(guò)綜合幾個(gè)模型的優(yōu)勢(shì),組合形成混合模型,如Chaabene等[91]將卡爾曼濾波和自回歸滑動(dòng)平均模型結(jié)合起來(lái)進(jìn)行短期預(yù)測(cè)。Pedro等[92]基于k-最近鄰法和人工神經(jīng)網(wǎng)絡(luò)算法預(yù)測(cè)了15分鐘到2小時(shí)的總太陽(yáng)輻照度。
本文搜集了多種預(yù)測(cè)模型輸入數(shù)據(jù)、預(yù)測(cè)精度及所用模型,具體信息如表1所示。表中預(yù)測(cè)模型有單一模型也有多種混合模型,同時(shí)模擬時(shí)間尺度有1和6 h。對(duì)比兩個(gè)時(shí)間尺度模擬精度可發(fā)現(xiàn),6 h尺度的模擬精度明顯高于其他1 h預(yù)測(cè)精度。另外通過(guò)表1可發(fā)現(xiàn)基于簡(jiǎn)單統(tǒng)計(jì)模型、混合模型或神經(jīng)網(wǎng)絡(luò)等模型模擬結(jié)果都相對(duì)較好,相關(guān)系數(shù)可達(dá)0.9以上。雖然所列模型預(yù)測(cè)時(shí)間尺度不同且驗(yàn)證站點(diǎn)分布于不同區(qū)域,但總體可發(fā)現(xiàn)神經(jīng)網(wǎng)絡(luò)模型以及多個(gè)模型相結(jié)合的方法在近幾年更受關(guān)注。
注:[R2]表示決定系數(shù);[rRMSE]表示相對(duì)均方根誤差;[MPE]表示平均百分比誤差;[RMSE]表示均方根誤差;[MAE]表示平均絕對(duì)誤差;[nRMSE]表示歸一化均方根誤差;[R]表示相關(guān)系數(shù)。
本研究還搜集了基于多個(gè)氣象衛(wèi)星的短期太陽(yáng)輻射預(yù)測(cè)模型及其精度(表2)以及基于不同圖像分辨率和時(shí)間分辨率的全天空成像儀或天空相機(jī)的短期或超短期太陽(yáng)輻射預(yù)測(cè)方法及其誤差,如表3所示。從表2中可看出,目前基于衛(wèi)星圖像的短期太陽(yáng)預(yù)測(cè)方法主要是結(jié)合神經(jīng)網(wǎng)絡(luò)等智能算法,且使用神經(jīng)網(wǎng)絡(luò)等智能算法得到的預(yù)測(cè)結(jié)果精度更高。同時(shí)從表3也可發(fā)現(xiàn),基于全天空成像儀的方法中,使用神經(jīng)網(wǎng)絡(luò)等智能算法得到的預(yù)測(cè)結(jié)果精度也更高。另外可發(fā)現(xiàn)近年來(lái),隨著計(jì)算機(jī)性能的不斷提高以及人工智能算法的快速發(fā)展,其在短期太陽(yáng)輻射預(yù)測(cè)領(lǐng)域的應(yīng)用取得了較好的效果。
注:[PE]表示百分比誤差;[nMBE]表示相對(duì)平均誤差;[RMSD]表示均方根誤差。
3 總結(jié)與展望
本文對(duì)目前SSR短期預(yù)測(cè)方法歸納總結(jié)為4類進(jìn)行闡述,分別為基于地面觀測(cè)數(shù)據(jù)的預(yù)測(cè)方法、基于衛(wèi)星遙感的預(yù)測(cè)方法、基于地基云圖的預(yù)測(cè)方法以及基于數(shù)值天氣預(yù)報(bào)模式預(yù)測(cè)方法。綜合分析現(xiàn)有的基于地面觀測(cè)數(shù)據(jù)的統(tǒng)計(jì)模型,可發(fā)現(xiàn)無(wú)論從單一模型還是混合模型,大多數(shù)是基于歷史太陽(yáng)輻射數(shù)據(jù)或是將其與相關(guān)的氣象因子結(jié)合建立模型,此類模型選擇的因子與太陽(yáng)輻射的關(guān)系是間接的,較難模擬大氣中如氣溶膠、云等對(duì)太陽(yáng)輻射的直接影響,且此類模型與所使用的數(shù)據(jù)及當(dāng)?shù)貧夂蛱卣飨嚓P(guān),因此模型的可移植性相對(duì)較差。而衛(wèi)星遙感雖能夠快速、大面積觀測(cè),也能提供氣溶膠、云微物理屬性等大氣環(huán)境屬性信息。但該類模型仍受其時(shí)空間分辨以及反演參數(shù)不確定性的限制,因而存在一定的局限性。與衛(wèi)星圖像不同,全天空成像儀不僅能為特定點(diǎn)提供高空分辨率的天空?qǐng)D像,而且可獲取高時(shí)間分辨率(時(shí)間可根據(jù)研究設(shè)置)的實(shí)時(shí)云圖信息。因而在SSR的短期甚至超短期預(yù)測(cè)中的應(yīng)用越來(lái)越廣泛。此類方法的缺點(diǎn)是全天空成像儀為單點(diǎn)觀測(cè),因此空間覆蓋范圍小,大面積推廣較難,同時(shí)由于其采樣頻率較高,如何平衡算法的復(fù)雜度和精度成為一大難點(diǎn)。與以上3種方法不同,數(shù)值天氣預(yù)報(bào)的預(yù)測(cè)方法則是考慮了太陽(yáng)輻射的物理變化過(guò)程,可實(shí)現(xiàn)不同尺度的預(yù)測(cè),但受到輸入數(shù)據(jù)和模型的不確定性以及模式輸出結(jié)果分辨率的限制,導(dǎo)致NWP模型的預(yù)測(cè)結(jié)果往往存在偏差。雖然有較多的后處理方法被用于NWP模型預(yù)測(cè)結(jié)果的誤差校正,但數(shù)值模擬方法中的氣象和環(huán)境因素較復(fù)雜,較難準(zhǔn)確模擬。
總結(jié)以上方法,當(dāng)研究區(qū)域地面觀測(cè)數(shù)據(jù)質(zhì)量較高時(shí),可選擇利用基于觀測(cè)數(shù)據(jù)的統(tǒng)計(jì)模型進(jìn)行短期預(yù)測(cè)。單一統(tǒng)計(jì)模型精度有限,可綜合多個(gè)模型的優(yōu)點(diǎn),建立適合研究區(qū)域的混合模型來(lái)提高預(yù)測(cè)精度。而如果需進(jìn)行大面積SSR短期預(yù)測(cè)或?yàn)榱双@取其時(shí)空特征時(shí),基于衛(wèi)星數(shù)據(jù)的預(yù)測(cè)方法,則是較好的選擇。由于全天空成像儀既可調(diào)節(jié)時(shí)間分辨率也可獲取質(zhì)量較高的云圖,用于預(yù)測(cè)時(shí)更易得到理想的結(jié)果,因此如需獲取相對(duì)準(zhǔn)確且時(shí)效性要求較高的太陽(yáng)輻射預(yù)測(cè)結(jié)果,架設(shè)全天空成像儀是較好的選擇。而數(shù)值天氣預(yù)報(bào)模式主要用于預(yù)測(cè)時(shí)間范圍較長(zhǎng)的太陽(yáng)輻照度,目前只有較少的研究利用WRF及其衍生的模型進(jìn)行太陽(yáng)輻照度短期預(yù)測(cè)。
針對(duì)以上不同方法的優(yōu)缺點(diǎn),本文總結(jié)以下幾點(diǎn)展望:
1)發(fā)展基于多源數(shù)據(jù)的智能化算法融合模型。以上基于地面觀測(cè)數(shù)據(jù)、遙感數(shù)據(jù)以及地基云圖觀測(cè)數(shù)據(jù)預(yù)測(cè)模型,因數(shù)據(jù)采集頻率及空間范圍不同使得不同數(shù)據(jù)的適用性有限。其中地面站點(diǎn)觀測(cè)太陽(yáng)輻射以及地基云圖觀測(cè)云圖空間代表性有限,但其時(shí)間分辨率相對(duì)較高,而遙感觀測(cè)則可提供大范圍觀測(cè)但其時(shí)空分辨率相對(duì)較低,較難表達(dá)時(shí)空變化細(xì)節(jié)。因此融合3種數(shù)據(jù)源是提高短期太陽(yáng)輻射預(yù)測(cè)精度的有效方式,同時(shí)也可提高預(yù)測(cè)結(jié)果的時(shí)空表達(dá)。時(shí)間尺度上,由于地基云圖可進(jìn)行更高時(shí)間分辨率的云圖觀測(cè),可將其用于輔助遙感觀測(cè)時(shí)間分辨率的不足,進(jìn)而可獲得時(shí)間加密的云觀測(cè)??臻g尺度上,遙感獲取面數(shù)據(jù),地基云圖及輻射觀測(cè)數(shù)據(jù)可增加空間變化細(xì)節(jié)。同時(shí)通過(guò)不同模型的研究進(jìn)展可發(fā)現(xiàn),智能化算法,如神經(jīng)網(wǎng)絡(luò)以及深度學(xué)習(xí)等方法由于其具有對(duì)復(fù)雜數(shù)據(jù)強(qiáng)大的處理和特征解析能力等優(yōu)點(diǎn),近年來(lái)在太陽(yáng)輻射短期預(yù)測(cè)研究中廣泛應(yīng)用[108-110]。因此有效融合多種數(shù)據(jù)源并結(jié)合深度學(xué)習(xí)等智能化算法是未來(lái)短期太陽(yáng)輻射預(yù)測(cè)的發(fā)展方向之一。目前,已有部分研究通過(guò)結(jié)合不同數(shù)據(jù)進(jìn)行太陽(yáng)輻射短期預(yù)報(bào),包括通過(guò)將地面觀測(cè)數(shù)據(jù)與天空?qǐng)D像或者衛(wèi)星圖像數(shù)據(jù)進(jìn)行融合,同時(shí)采用神經(jīng)網(wǎng)絡(luò)、深度學(xué)習(xí)、集成學(xué)習(xí)框架等進(jìn)行短期太陽(yáng)輻照度的預(yù)測(cè)[111-114],有效提高了太陽(yáng)輻射短期預(yù)報(bào)精度。
2)增強(qiáng)NWP模型預(yù)測(cè)結(jié)果的適用性。隨著輸入數(shù)據(jù)分辨率以及計(jì)算機(jī)性能不斷提升,利用NWP進(jìn)行更高分辨率或更短時(shí)間范圍的預(yù)測(cè)成為可能[115-116]。將多源及不同時(shí)間分辨率的數(shù)據(jù)作為NWP預(yù)測(cè)模型輸入是提高預(yù)測(cè)結(jié)果的方法之一[117]。目前已有研究將遙感觀測(cè)數(shù)據(jù)或地基云圖數(shù)據(jù),作為NWP輸入,進(jìn)行太陽(yáng)輻射的預(yù)測(cè),用于提高其預(yù)測(cè)精度。同時(shí)對(duì)NWP預(yù)測(cè)結(jié)果進(jìn)行后處理使其形成高時(shí)空分辨率預(yù)測(cè)結(jié)果也是提高該類方法適用性的有效方法之一,如通過(guò)對(duì)NWP預(yù)測(cè)結(jié)果進(jìn)行合理時(shí)空降尺度[118],以及利用地面觀測(cè)數(shù)據(jù)及深度學(xué)習(xí)等智能化算法校正模型輸出誤差等,同樣可有助于形成高精度、高分辨率的SSR短期預(yù)測(cè)結(jié)果。
[參考文獻(xiàn)]
[1]"""" 呂寧. 近年來(lái)中國(guó)地表太陽(yáng)輻射時(shí)空變化及影響因素分析[D]. 北京: 中國(guó)科學(xué)院地理科學(xué)與資源研究所, 2009.
LYU N. Analysis of spatio-temporal variation of surface downward shortwave radiation and associated effecting factors over China in recent years[D]. Beijing: Institute of Geographic Sciences and Natural Resources Research, 2009.
[2]"""" 馬金玉. 中國(guó)地面太陽(yáng)輻射長(zhǎng)期變化特征及短期預(yù)報(bào)方法研究[D]. 南京: 南京信息工程大學(xué), 2011.
MA J Y. The long-term variation and short-term prediction method of surface" solar" radiation" in" China[D]. Nanjing: Nanjing University of Information Science and Technology,2011.
[3]"""" 黎微微, 胡斯勒?qǐng)D, 陳洪濱, 等. 利用MODIS資料計(jì)算不同云天條件下的地表太陽(yáng)輻射[J]. 遙感技術(shù)與應(yīng)用,2017, 32(4): 643-650.
LI W W, HUSI L T, CHEN H B, et al. Estimation of surface solar radiation using MODIS satellite data and RSTAR"" model[J]."" Remote""" sensing""" technology""" and application, 2017, 32(4): 643-650.
[4]"""" 王小潔, 陳珍莉, 王旭, 等. 基于相似日和Elman神經(jīng)網(wǎng)絡(luò)的逐時(shí)太陽(yáng)總輻射預(yù)測(cè)[J]. 海南大學(xué)學(xué)報(bào)(自然科學(xué)版), 2020, 38(4): 347-355.
WANG X J, CHEN Z L, WANG X, et al. Hourly total solar radiation prediction based on similar day and Elman neural" network[J]." Natural" science" journal" of" Hainan University, 2020, 38(4): 347-355.
[5]"""" 崔洋, 孫銀川, 常倬林. 短期太陽(yáng)能光伏發(fā)電預(yù)測(cè)方法研究進(jìn)展[J]. 資源科學(xué), 2013, 35(7): 1474-1481.
CUI Y, SUN Y C, CHANG Z L. A review of short-term solar photovoltaic power generation prediction methods[J]. Resources science, 2013, 35(7): 1474-1481.
[6]"""" 孫朋杰, 陳正洪, 成馳, 等. 太陽(yáng)能光伏電站發(fā)電量變化特征及其與氣象要素的關(guān)系[J]. 水電能源科學(xué), 2013, 31(11): 249-252.
SUN P J, CHEN Z H, CHENG C, et al. Relationship between meteorological factors and variation characteristics of electricity generation in solar photovoltaic" power" plant[J]. Water resources and power, 2013, 31(11): 249-252.
[7]"""" 郁云, 許昌, 曹瀟, 等. 云圖紋理分析結(jié)合SVM的地表太陽(yáng)輻射預(yù)測(cè)[J]. 西南師范大學(xué)學(xué)報(bào)(自然科學(xué)版), 2017, 42(12): 75-81.
YU Y, XU C, CAO X, et al. On ground solar radiation forecasting combined with textural feature extraction and SVM computer engineering and applications[J]. Journal of Southwest China Normal University (natural science edition), 2017, 42(12): 75-81.
[8]"""" DONG J, OLAMA M M, KURUGANTI T, et al. Novel stochastic methods to predict short-term solar radiation and photovoltaic"" power[J]." Renewable" energy," 2020," 145(C): 333-346.
[9]"""" MORA-LóPEZ L, SIDRACH-DE-CARDONA M. Multiplicative ARMA models to generate hourly series of global irradiation[J]. solar energy, 1998, 63(5): 283-291.
[10]""" BEHERA M K, NAYAK N. A comparative study on short-term PV power forecasting using decomposition based optimized extreme learning machine algorithm[J]. Engineering science and technology, an international journal, 2020, 23(1): 156-167.
[11]""" 周海, 朱想, 金山紅, 等. 超短期太陽(yáng)輻照度多模型預(yù)測(cè)[J]. 中國(guó)科技論文, 2017, 12(23): 2695-2700.
ZHOU H, ZHU X, JIN S H, et al. Multi-model prediction for super short term solar irradiance[J]. China sciencepaper, 2017, 12(23): 2695-2700.
[12]""" MADUGU I S, OLUFEAGBA B J, ADEDIRAN Y A, et al. ARMA model for prediction of solar radiation in Kano-Nigeria[J]."" Zaria""" journal""" of""" electrical""" engineering technology, 2019.
[13]""" BACHER P, MADSEN H,NIELSEN H A. Online short-term solar power forecasting[J]. Solar energy, 2009,83(10): 1772-1783.
[14]""" WANG L H, SHI J R. A comprehensive application of machine learning techniques for short-term solar radiation prediction[J]. Applied sciences, 2021, 11(13): 5808.
[15]""" SEIDEL F C, KOKHANOVSKY A A, SCHAEPMAN M E. Fast and simple model for atmospheric radiative transfer[J]. Atmospheric measurement techniques discussions, 2010, 3(4): 1129-1141.
[16]""" 馬燕峰, 蔣云濤, 郝毅, 等. 基于非線性自回歸神經(jīng)網(wǎng)絡(luò)的GHI預(yù)測(cè)[J]. 太陽(yáng)能學(xué)報(bào), 2019, 40(3): 733-740.
MA Y F, JIANG Y T, HAO Y, et al. GHI forecast based on nonlinear autoregressive neural network[J]. Acta energiae solaris sinica, 2019, 40(3): 733-740.
[17]""" GUTIERREZ-COREA F V, MANSO-CALLEJO M A, MORENO-REGIDOR M P, et al. Forecasting short-term solar irradiance based on artificial neural networks and data" from" neighboring" meteorological" stations[J]. Solar energy, 2016, 134: 119-131.
[18]""" SORKUN M C, PAOLI C, INCEL ? D. Time series forecasting on solar irradiation using deep learning[C]// The 10th International Conference on Electrical and Electronics Engineering(ELECO). Bursa, Turkey, 2017: 151-155.
[19]""" ALI M, PRASAD R, XIANG Y, et al. Variational mode decomposition based random forest model for solar radiation forecasting: new emerging machine learning technology[J]. Energy reports, 2021, 7: 6700-6717.
[20]""" AYET A, TANDEO P. Nowcasting solar irradiance using an analog method and geostationary satellite images[J]. Solar energy, 2018, 164: 301-315.
[21]""" KALLIO-MYERS V, RIIHEL? A, LAHTINEN P, et al. Global horizontal irradiance forecast for Finland based on geostationary"" weather"" satellite"" data[J]." Solar"" energy, 2020, 198: 68-80.
[22]""" PINKER R T, FROUIN R, LI Z. A review of satellite methods to derive surface shortwave irradiance[J]. Remote sensing of environment, 1995, 51(1): 108-124.
[23]""" YANG L W, GAO X Q, HUA J J, et al. Intra-day global horizontal irradiance forecast using FY-4A clear sky index[J]. Sustainable energy technologies and assessments, 2022, 50: 101816.
[24]""" RODRíGUEZ-BENíTEZ F J, LóPEZ-CUESTA M, ARBIZU-BARRENA C, et al. Assessment of new solar radiation nowcasting methods based on sky-camera and satellite" imagery[J]." Applied" energy," 2021," 292(6): 116838.
[25]""" DU J A, MIN Q L, ZHANG P L, et al. Short-term solar irradiance forecasts using sky images and radiative transfer model[J]. Energies, 2018, 11(5): 1107.
[26]""" BERNECKER D, RIESS C, ANGELOPOULOU E, et al. Continuous short-term irradiance forecasts using sky images[J]. Solar energy, 2014, 110: 303-315.
[27]""" 蔣俊霞, 高曉清, 呂清泉, 等. 基于地基云圖的云跟蹤與太陽(yáng)輻照度超短期預(yù)報(bào)方法研究[J]. 太陽(yáng)能學(xué)報(bào), 2020, 41(5): 351-358.
JIANG J X, GAO X Q, LYU Q Q, et al. Study on cloud tracking and solar irradiance ultra-short-term forecasting based"" on"" TSI" images[J]." Acta" energiae" solaris sinica, 2020, 41(5): 351-358.
[28]""" EL ALANI O, ABRAIM M, GHENNIOUI H, et al. Short term solar irradiance forecasting using sky images based on a hybrid CNN-MLP model[J]. Energy reports, 2021, 7: 888-900.
[29]""" KAMADINATA J O, KEN T L, SUWA T. Sky image-based solar irradiance prediction methodologies using artificial neural networks[J]. Renewable energy, 2019, 134: 837-845.
[30]""" RYU A, ITO M, ISHII H, et al. Preliminary analysis of short-term solar irradiance forecasting by using total-sky imager and convolutional neural network[C]//2019 IEEE PES GTD Grand International Conference and Exposition Asia (GTD Asia), Bangkok, Thailand, 2019: 627-631.
[31]""" JIANG J X, LYU Q Q, GAO X Q. The ultra-short-term forecasting of global horizonal irradiance based on total sky images[J]. Remote sensing, 2020, 12(21): 3671.
[32]""" DEV S, SAVOY F M, LEE Y H, et al. Estimating solar irradiance using sky imagers[J]. Atmospheric measurement techniques, 2019, 12(10): 5417-5429.
[33]""" BAUER P, THORPE A, BRUNET G. The quiet revolution of numerical weather prediction[J]. Nature, 2015, 525(7567): 47-55.
[34]""" ARBIZU-BARRENA C, RUIZ-ARIAS J A, RODRíGUEZ-BENíTEZ F J, et al. Short-term solar radiation forecasting by" advecting" and" diffusing" MSG" cloud" index[J]." Solar energy, 2017, 155: 1092-1103.
[35]""" ZHOU Y, LIU Y F, WANG D J, et al. A review on global solar radiation prediction with machine learning models in a comprehensive perspective[J]. Energy conversion and management, 2021, 235: 113960.
[36]""" GUERMOUI M, MELGANI F, GAIRAA K, et al. A comprehensive review of hybrid models for solar radiation forecasting[J]. Journal of cleaner production, 2020, 258: 120357.
[37]""" AICARDI D, MUSé P, ALONSO-SUáREZ R. A comparison of satellite cloud motion vectors techniques to forecast intra-day hourly solar global horizontal irradiation[J]. Solar energy, 2022, 233: 46-60.
[38]""" LIN F, ZHANG Y, WANG J X. Recent advances in intra-hour solar forecasting: a review of ground-based sky image methods[J]. International journal of forecasting, 2023, 39(1): 244-265.
[39]""" CHU Y H, LI M Y, COIMBRA C F M, et al. Intra-hour irradiance forecasting techniques for solar power integration:"" a"" review[J]."" Iscience,"" 2021,"" 24(10): 103136.
[40]""" BARBIERI F, RAJAKARUNA S, GHOSH A. Very short-term photovoltaic power forecasting with cloud modeling: a review[J]. Renewable and sustainable energy reviews, 2017, 75: 242-263.
[41]""" GB/T 19964—2012, 光伏發(fā)電站接入電力系統(tǒng)技術(shù)規(guī)定[S].
GB/T 19964—2012, Technical requirements for connecting photovoltaic power station to power system[S].
[42]""" 陳文. 基于晴朗指數(shù)的超短期太陽(yáng)輻射預(yù)測(cè)算法的研究[D]. 武漢: 華中科技大學(xué), 2018.
CHEN" W." Research" on" ultra-short" term" solar" radiation forecast algorithm based" on" clearness" index[D]. Wuhan: Huazhong University of Science and Technology, 2018.
[43]""" 徐維. 基于時(shí)間序列分析的短時(shí)間尺度太陽(yáng)輻射模型研究[J]. 常州工學(xué)院學(xué)報(bào), 2009, 22(S1): 41-44.
XU W. The designing of establishing modeling simulation and forecasting methods for short-term solar irradiance based on time scale[J]. Journal of Changzhou Institute of Technology, 2009, 22(S1): 41-44.
[44]""" REIKARD G. Predicting solar radiation at high resolutions:" a" comparison" of" time"" series" forecasts[J]. Solar energy, 2009, 83(3): 342-349.
[45]""" DAS S. Short term forecasting of solar radiation and power output of 89.6 kWp solar PV power" plant[J]. Materials today: proceedings, 2021, 39: 1959-1969.
[46]""" 張雪松, 朱想, 趙波, 等. 基于 WT-PLSP 的太陽(yáng)輻射短期預(yù)測(cè)[J]. 干旱區(qū)資源與環(huán)境, 2017, 31(6): 142-146.
ZHANG X S, ZHU X, ZHAO B, et al. Short - term forecast of solar radiation based on WT-PLSP[J]. Journal of arid land resources and environment, 2017, 31(6): 142-146.
[47]""" 蔣鋒, 楊嘉偉. 基于多目標(biāo)優(yōu)化集成學(xué)習(xí)的短期太陽(yáng)輻射預(yù)測(cè)[J]. 云南大學(xué)學(xué)報(bào)(自然科學(xué)版), 2021, 43(3): 451-461.
JIANG F, YANG J W. Short-term solar radiation forecast based on ensemble learning of multi-objective optimization[J]. Journal of Yunnan University (natural sciences edition), 2021, 43(3): 451-461.
[48]""" MUKHTAR M, OLUWASANMI A, YIMEN N, et al. Development and comparison of two novel hybrid neural network models for hourly solar radiation prediction[J]. Applied sciences, 2022, 12(3): 1435.
[49]""" 周滿國(guó), 黃艷國(guó), 段錦鋒. 基于GRU-RF模型的太陽(yáng)輻照度短時(shí)預(yù)測(cè)[J]. 太陽(yáng)能學(xué)報(bào), 2022, 43(7): 166-173.
ZHOU M G, HUANG Y G, DUAN J F. Short term prediction of soral irradiance based on GRU-RF model[J]. Acta energiae solaris sinica, 2022, 43(7): 166-173.
[50] 倪超, 王聰, 朱婷婷, 等. 基于CNN-Bi-LSTM的太陽(yáng)輻照度超短期預(yù)測(cè)[J]. 太陽(yáng)能學(xué)報(bào), 2022, 43(3): 197-202.
NI C, WANG C, ZHU T T, et al. Super-short-term forecast of solar irradiance based on CNN-Bi-LSTM[J]. Acta energiae solaris sinica, 2022, 43(3): 197-202.
[51]""" ZHU T T, GUO Y R, WANG C, et al. Inter-hour forecast of solar radiation based on the structural equation model and ensemble model[J]. Energies, 2020, 13(17): 4534.
[52]""" QUAN H, YANG D Z. Probabilistic solar irradiance transposition""" models[J]."" Renewable"" and""" sustainable energy reviews, 2020, 125: 109814.
[53]""" 艾逸陽(yáng). 基于地基云圖的超短期光伏電站輻照度預(yù)測(cè)研究[D]. 杭州: 浙江大學(xué), 2018.
AI Y Y. Very short-term solar irradiance forecast based on sky images[D]. Hangzhou: Zhejiang University, 2018.
[54]""" 陳志寶,丁杰,周海,等. 地基云圖結(jié)合徑向基函數(shù)人工神經(jīng)網(wǎng)絡(luò)的光伏功率超短期預(yù)測(cè)模型[J]. 中國(guó)電機(jī)工程學(xué)報(bào), 2015, 35(3): 561-567.
CHEN Z B, DING J, ZHOU H, et al. A model of very short-term photovoltaic power forecasting based on ground-based""" cloud""" images""" and""" RBF"" neural"" network[J]. Proceedings of the CSEE, 2015, 35(3): 561-567.
[55]""" HAMMER A, HEINEMANN D, LORENZ E, et al. Short-term forecasting of solar radiation: a statistical approach using satellite data[J]. Solar energy, 1999, 67(1/2/3): 139-150.
[56]""" LORENZ E, HAMMER A, HEINEMANN D. Short term forecasting of solar radiation based on satellite data[C]//EUROSUN2004(ISES Europe Solar Congress). Freiburg, Germany, 2004: 1.
[57]""" CROS S, BADOSA J, SZANTAI A, et al. Reliability predictors for solar irradiance satellite-based forecast[J]. Energies, 2020, 13(21): 5566.
[58]""" YANG L W, GAO X Q, HUA J J, et al. Very short-term surface solar irradiance forecasting based on FengYun-4 geostationary satellite[J]. Sensors, 2020, 20(9): 2606.
[59]""" ROY A, HAMMER A, HEINEMANN D, et al. Impact of tropical convective conditions on solar irradiance forecasting""""" based"""" on"""" cloud"""" motion""" vectors[J]. Environmental research letters, 2022, 17(10): 104048.
[60]""" OH M, KIM C K, KIM B, et al. Spatiotemporal optimization for short-term solar forecasting based on satellite imagery[J]. Energies, 2021, 14(8): 2216.
[61]""" NIELSEN A H, IOSIFIDIS A, KARSTOFT H. IrradianceNet: spatiotemporal deep learning model for satellite-derived solar irradiance short-term forecasting[J]. Solar energy, 2021, 228: 659-669.
[62]""" MARQUEZ R, PEDRO H T C, COIMBRA C F M. Hybrid solar forecasting method uses satellite imaging and ground telemetry as inputs to ANNs[J]. Solar energy, 2013, 92: 176-188.
[63]""" DONG Z B, YANG D Z, REINDL T, et al. Satellite image analysis and a hybrid ESSS/ANN model to forecast solar irradiance"" in"" the"" tropics[J]."" Energy"" conversion"" and management, 2014, 79(3): 66-73.
[64]""" PALMER D, KOUBLI E, COLE I, et al. Satellite or ground-based measurements for production of site specific hourly irradiance data: which is most accurate and where?[J]. Solar energy, 2018, 165: 240-255.
[65]""" SHIELDS J E, KARR M E, JOHNSON R W, et al. Day/night"" whole" sky"" imagers" for"" 24-h"" cloud"" and"" sky assessment: history and overview[J]. Applied optics, 2013, 52(8): 1605-1616.
[66]""" MARQUEZ" R," COIMBRA" C" F" M." Intra-hour" DNI forecasting based on cloud tracking image analysis[J]. Solar energy, 2013, 91: 327-336.
[67]""" FENG C, ZHANG J. SolarNet: a sky image-based deep convolutional neural network for intra-hour solar forecasting[J]. Solar energy, 2020, 204: 71-78.
[68]""" FENG C, ZHANG J, ZHANG W Q, et al. Convolutional neural networks for intra-hour solar forecasting based on sky image sequences[J]. Applied energy, 2022, 310: 118438.
[69]""" ZHANG C, DU Y, CHEN X Y, et al. Cloud motion tracking system using low-cost sky imager for PV power ramp-rate"""""" control[C]//2018""""" IEEE""""" International Conference on Industrial Electronics for Sustainable Energy Systems(IESES). Hamilton, New Zealand, 2018: 493-498.
[70]""" ALONSO-MONTESINOS J, BATLLES F J, PORTILLO C. Solar" irradiance" forecasting" at" one-minute" intervals for different" sky" conditions" using" sky" camera" images[J]. Energy conversion and management, 2015, 105: 1166-1177.
[71]""" AI Y Y, PENG Y G, WEI W. A model of very short-term solar irradiance forecasting based on low-cost sky images[C]//AIP Conference Proceedings. Hangzhou, China, 2017: 020022.
[72]""" 王聰. 基于數(shù)值和地基云圖的太陽(yáng)輻射超短期預(yù)測(cè)研究[D]. 南京: 南京林業(yè)大學(xué), 2021.
WANG C. Super-short term forecast of solar irradiance based on numerical and ground-based cloud images[D]. Nanjing: Nanjing Forestry University, 2021.
[73]""" 朱婷婷. 基于地基云圖的太陽(yáng)直接輻照度超短期預(yù)測(cè)研究[D]. 南京: 東南大學(xué), 2019.
ZHU T T. Research on inter-hour forecast of direct normal irradiance based on ground-based cloud images[D]. Nanjing: Southeast University, 2019.
[74]""" 馬雷鳴, 鮑旭煒. 數(shù)值天氣預(yù)報(bào)模式物理過(guò)程參數(shù)化方案的研究進(jìn)展[J]. 地球科學(xué)進(jìn)展, 2017, 32(7): 679-687.
MA L M, BAO X W. Research progress on physical parameterization schemes in numerical weather prediction models[J]. Advances in earth science, 2017, 32(7): 679-687.
[75]""" BLAGA R, SABADUS A, STEFU N, et al. A current perspective on the accuracy of incoming solar energy forecasting[J]. Progress in energy and combustion science, 2019, 70: 119-144.
[76]""" PEREZ R, KIVALOV S, SCHLEMMER J, et al. Validation of short and medium term operational solar radiation forecasts in the US[J]. Solar energy, 2010, 84(12): 2161-2172.
[77]""" JIMéNEZ P A, ALESSANDRINI S, HAUPT S E, et al. The role of unresolved clouds on short-range global horizontal" irradiance" predictability[J]." Monthly" weather review, 2016, 144(9): 3099-3107.
[78]""" RUIZ-ARIAS J A, DUDHIA J, GUEYMARD C A. A simple parameterization of the short-wave aerosol optical properties for surface direct and diffuse irradiances assessment in a numerical weather model[J]. Geoscientific model development, 2014, 7(3): 1159-1174.
[79]""" JIMéNEZ P A, HACKER J P, DUDHIA J, et al. WRF-SOLAR: description and clear-sky assessment of an augmented" NWP" model" for" solar"" power"" prediction[J]. Bulletin of the American Meteorological Society, 2016, 97(7): 1249-1264.
[80]nbsp;"" 董旭光, 陳艷春, 孟祥新. WRF-Solar 模式對(duì)山東太陽(yáng)總輻射的模擬效果檢驗(yàn)[J]. 海洋氣象學(xué)報(bào), 2021, 41(4): 36-51.
DONG X G, CHEN Y C, MENG X X. Performance evaluation of solar irradiance simulation over Shandong using WRF-Solar numerical model[J]. Journal of marine meteorology, 2021, 41(4): 36-51.
[81]""" MCCANDLESS T, JIMéNEZ P A. Examining the potential of a random forest derived cloud mask from GOES-R satellites"" to"" improve"" solar"" irradiance"" forecasting[J]. Energies, 2020, 13(7): 1671.
[82]""" HAUPT S E. A public-private-academic partnership to advance""" solar""" forecasting[C]//Proceedings""" of""" 42nd American Solar Energy Society Annual Conference. Baltimore, Maryland, 2013.
[83]""" JIMéNEZ P A, DUDHIA J, THOMPSON G, et al. Improving" the" cloud" initialization" in" WRF-solar" with enhanced short-range forecasting functionality: the MAD-WRF model[J]. Solar energy, 2022, 239: 221-233.
[84]""" MATHIESEN P, COLLIER C, KLEISSL J. A high-resolution,""""" cloud-assimilating""""" numerical""""" weather prediction model for solar irradiance forecasting[J]. Solar energy, 2013, 92: 47-61.
[85]""" MATHIESEN P J, COLLIER C, KLEISSL J P. Development and validation of an operational, cloud-assimilating numerical weather prediction model for solar irradiance forecasting[C]//Proceedings of ASME 2012 6th international Conference on Energy Sustainability Collocated with the ASME 2012 10th International Conference on Fuel Cell Science, Engineering And Technology. San Diego, California, USA, 2013: 955-964.
[86]""" PIERRO M, BUCCI F, CORNARO C, et al. Model output statistics cascade to improve day ahead solar irradiance forecast[J]. Solar energy, 2015, 117: 99-113.
[87]""" GLAHN H R, LOWRY D A. The use of model output statistics" (MOS)" in"" objective"" weather"" forecasting[J]. Journal of applied meteorology and climatology, 1972, 11(8): 1203-1211.
[88]""" VERZIJLBERGH R A, HEIJNEN P W, DE ROODE S R, et al. Improved model output statistics of numerical weather prediction based irradiance forecasts for solar power" applications[J]." Solar" energy," 2015," 118:" 634-645.
[89]""" LAURET P, DIAGNE H M, DAVID M. A neural network post-processing"" approach""" to"" "improving""" NWP""" solar radiation forecasts[J]. Energy procedia, 2014, 57: 1044-1052.
[90]""" 劉唯佳. 基于云參量的太陽(yáng)輻射短時(shí)臨近預(yù)報(bào)模型的構(gòu)建及預(yù)報(bào)評(píng)估[D]. 南京: 南京信息工程大學(xué), 2021.
LIU W J. Construction and evaluation of cloud-property based" solar" radiation" short"" term"" forecast"" systems[D]. Nanjing: Nanjing University of Information Science and Technology, 2021.
[91]""" CHAABENE M, BEN A M. Neuro-fuzzy dynamic model with Kalman filter to forecast irradiance and temperature for solar energy systems[J]. Renewable energy, 2008, 33(7): 1435-1443.
[92]""" PEDRO H T C, COIMBRA C F M. Short-term irradiance forecastability" for" various" solar" micro-climates[J]." Solar energy, 2015, 122: 587-602.
[93]""" WANG Z, WANG F, SU S. Solar irradiance short-term prediction model based on BP neural network[J]. Energy procedia, 2011, 12: 488-494.
[94]""" MONJOLY S, ANDRé M, CALIF R, et al. Hourly forecasting of global solar radiation based on multiscale decomposition" methods: a" hybrid"" approach[J]. Energy, 2017, 119: 288-298.
[95]""" REIKARD G, HANSEN C. Forecasting solar irradiance at short" horizons:" frequency" and" time"" domain" models[J]. Renewable energy, 2019, 135: 1270-1290.
[96]""" 高碧軒. 基于機(jī)器學(xué)習(xí)算法的中短期太陽(yáng)輻照度預(yù)測(cè)研究[D]. 昆明: 云南師范大學(xué), 2020.
GAO B X. The medium and short term solar irradiance forecasting research based on machine learning algorithms[D]. Kunming: Yunnan Normal University, 2020.
[97]""" 黃小喬. 基于深度學(xué)習(xí)的超短期太陽(yáng)輻照度預(yù)測(cè)模型研究[D]. 昆明: 云南師范大學(xué), 2021.
HUANG X Q. Research on intra-hour solar irradiance prediction model based on deep learning[D]. Kunming: Yunnan Normal University, 2021.
[98]""" ALI-OU-SALAH H, OUKARFI B, MOUHAYDINE T. Short-term solar radiation forecasting using a new seasonal clustering" technique"" and"" artificial"" neural"" network[J]. International journal of green energy, 2022, 19(4): 424-434.
[99]""" NKOUNGA W M, NDIAYE M F, CISSE O, et al. Short-term multi horizons forecasting of solar irradiation based on artificial neural network with meteorological data: application"" in"" the"" north-west""" of"" Senegal[C]//2021 Sixteenth International Conference on Ecological Vehicles and Renewable Energies (EVER). Monte-Carlo, Monaco, 2021: 1-8.
[100]" MILLER S D, ROGERS M A, HAYNES J M, et al. Short-term solar irradiance forecasting via satellite/model coupling[J]. Solar energy, 2018, 168: 102-117.
[101]" SI Z Y, YANG M, YU Y X, et al. Photovoltaic power forecast based on satellite images considering effects of solar position[J]. Applied energy, 2021, 302: 117514.
[102]" QIN J, JIANG H, LU N, et al. Enhancing solar PV output forecast by integrating ground and satellite observations with deep learning[J]. Renewable and sustainable energy reviews, 2022, 167: 112680.
[103]" YANG H D, KURTZ B, NGUYEN D, et al. Solar irradiance forecasting using a ground-based sky imager developed at UC San Diego[J]. Solar energy, 2014, 103: 502-524.
[104]" QUESADA-RUIZ S, CHU Y, TOVAR-PESCADOR J, et al."" Cloud-tracking"" methodology"" for"" intra-hour"" DNI forecasting[J]. Solar energy, 2014, 102: 267-275.
[105]" NOU J, CHAUVIN R, EYNARD J, et al. Towards the short-term forecasting of direct normal irradiance using a sky imager[J]. IFAC-Papersonline, 2017, 50(1): 14137-14142.
[106]" CALDAS M, ALONSO-SUáREZ R. Very short-term solar irradiance forecast using all-sky imaging and real-time irradiance measurements[J]. Renewable energy, 2019, 143: 1643-1658.
[107]" TEERAKAWANICH N, LEELARUJI T, PICHETJAMR-OEN A. Short term prediction of sun coverage using optical flow with GoogLeNet[J]. Energy reports, 2020, 6: 526-531.
[108]" BUSTER G, BANNISTER M, HABTE A, et al. Physics-guided machine learning for improved accuracy of the National"" Solar"" Radiation"" Database[J]."" Solar"" energy, 2022, 232: 483-492.
[109]" CHU Y H, PEDRO H T C, COIMBRA C F M. Hybrid intra-hour"" DNI"" forecasts"" with"" sky" image" processing enhanced by stochastic learning[J]. Solar energy, 2013, 98: 592-603.
[110]" GALA Y, FERNáNDEZ á, DíAZ J, et al. Hybrid machine learning forecasting of solar radiation values[J]. Neurocomputing, 2016, 176: 48-59.
[111]" AJITH M, MARTíNEZ-RAMóN M. Deep learning based solar radiation micro forecast by fusion of infrared cloud images" and"" radiation" data[J]." Applied" energy," 2021, 294: 117014.
[112]" SHAN S, LI C X, DING Z T, et al. Ensemble learning based multi-modal intra-hour irradiance forecasting[J]. Energy conversion and management, 2022, 270: 116206.
[113]" 林艷彬. 基于異構(gòu)數(shù)據(jù)融合的超短期太陽(yáng)能預(yù)測(cè)[D]. 廈門(mén): 廈門(mén)大學(xué), 2019.
LINnbsp; Y" B."" Very-short-term"" solar" prediction"" based"" on heterogeneous data fusion[D]. Xiamen: Xiamen University, 2019.
[114]" 楊春熙. 基于多源數(shù)據(jù)融合的短期太陽(yáng)輻照度預(yù)測(cè)研究[D]. 南京: 南京信息工程大學(xué), 2020.
YANG C X. Short-term forecasting for solar irradiance based on multi-source data fusion[D]. Nanjing: Nanjing University of Information Science and Technology, 2020.
[115]" YANG D Z, QUAN H, DISFANI V R, et al. Reconciling solar"" forecasts:" temporal"" hierarchy[J]." Solar"" energy, 2017, 158: 332-346.
[116]" YANG D Z. Validation of the 5-min irradiance from the National Solar Radiation Database(NSRDB)[J]. Journal of renewable and sustainable energy, 2021, 13(1): 016101.
[117]" ZHANG G, YANG D Z, GALANIS G, et al. Solar forecasting with hourly updated numerical weather prediction[J]. Renewable and sustainable energy reviews, 2022, 154: 111768.
[118]" YANG D Z, WANG W T, GUEYMARD C A, et al. A review of solar forecasting, its dependence on atmospheric sciences and implications for grid integration: towards carbon neutrality[J]. Renewable and sustainable energy reviews, 2022, 161: 112348.
RESEARCH PROGRESS ON SHORT-TERM PREDICTION
METHODS OF SURFACE SOLAR RADIATION
Jin Cunyin1,Zhang Shuhua1,Li Xingong2,Tian Qianqian1,Wang Qianru1,Wang Mohan3
(1. College of Geomatics, Xi’an University of Science and Technology, Xi’an 710054, China;
2. Department of Geography and Atmospheric Science, University of Kansas, Lawrence 66045, USA;
3. Hebei Institute of Geological Survey, Shijiazhuang 050081, China)
Abstract:In this paper, according to the data sources and prediction methods used in short-term surface solar radiation prediction, the current short-term prediction methods of surface solar radiation are summarized into four categories: prediction methods based on ground observation data, based on remote sensing data, based on total sky images, and numerical weather prediction models. This paper presents the research progress of four short-term surface solar radiation prediction methods, and evaluates their applicability, advantages and disadvantages. Finally, the future development of the short-term surface solar radiation prediction methods is prospected.
Keywords: solar radiation; remote sensing; prediction; statistical models; total sky imager; numerical weather prediction model