電氣專業(yè)畢業(yè)設(shè)計(jì)外文翻譯--基于人工智能的長(zhǎng)期電力負(fù)荷預(yù)測(cè)

1、<p>　　附錄A: 外文文獻(xiàn)及譯文</p><p><b>　　第一部分：原文</b></p><p>　　Artificial intelligence in long term electric load forecasting</p><p>　　K. Metaxiotis, A. Kagiannas, D. Askounis

2、, J. Psarras</p><p>　　Abstract: Intelligent solutions, based on artificial intelligence (AI) technologies, to solve complicated practical problems in various sectors are becoming more and more widespread now

3、adays. AI-based systems are being developed and deployed worldwide in myriad applications, mainly because of their symbolic reasoning, flexibility and explanation capabilities. This paper provides an overview for the res

4、earcher of AI technologies, as well as their current use in the field of long term electric loa</p><p>　　Keywords: Artificial intelligence; Electric load forecasting; Energy</p><p>　　1. Introduc

5、tion</p><p>　　In the past two decades. AI has been defined as the study of how to make computers do things that, at the moment, people do better. AI provides powerful and flexible means for obtaining solutio

6、ns to a variety of problems that often cannot be solved by other, more traditional and orthodox methods. </p><p>　　This review bears witness to the application of AI technologies in the field of long term el

7、ectric load forecasting (LTELF). Certainly, this is not the first paper to review the application of AI based systems in energy related problems with varying success. In general, AI developments in the field of energy ha

8、ve been reviewed by several authors from various points of view. </p><p>　　Taylor and Lubkeman reviewed the applications of knowledge based programming to power engineering problems, describing prototype pro

9、jects developed at North Carolina State University, while the survey of Zhang et al. concerned the use of ES technology in electric power systems. Ypsilantis and Yee presented a review of ESs for SCADA based power applic

10、ations and Lubarskii et al. discussed the use of ESs for power networks. Since that time, several other survey papers have been written in various e</p><p>　　However, this paper has a different focus. Writin

11、g a fully comprehensive survey of AI applications in energy systems is objectively impracticable. For this reason, our paper aims to create a large knowledge base for the researcher, introducing him/her to the specific a

12、rea of AI applications in LTELF and indicating other fields fertile for research. </p><p>　　2. AI applications in long term electric load forecasting</p><p>　　2.1 Expert systems</p><p

13、>　　ESs are one of the most commercially successful branches of AI. Welbank defines an ES as follows: An expert system is a program, which has a wide base of knowledge in a restricted domain, and uses complex inferenti

14、al reasoning to perform tasks, which a human expert could do.</p><p>　　In other words, an ES is a computer system containing a well organised body of knowledge, which emulates expert problem solving skills i

15、n a bounded domain of expertise . The system is able to achieve expert levels of problem solving performance, which would normally be achieved by a skilled human, when confronted with significant problems in the domain.&

16、lt;/p><p>　　The first works in ES application in LTELF were implemented by Rahman and Bhatnagar and Jabbour et al. The objective of these approaches was to use the knowledge, experience and analytical thinking

17、of experimental system operators. Park et al. made a further step by using fuzzy logic in an ES for a LTELF problem. In 1990, Ho et al. presented the use of a knowledge based ES in long term load forecasting of a Taiwan

18、power system, while in 1993, Rahman and Hazim tried to generalize his first work. </p><p>　　In 1995, Kim et al. implemented a long term load forecaster by using ANNs and a fuzzy ES, while later, Mori and Kob

19、ayashi presented an optimal fuzzy inference approach for the LTELF problem. Ranaweera et al. proposed a fuzzy logic ES model for the LTELF problem, which used fuzzy rules to incorporate historical weather and load data.

20、These fuzzy rules were obtained from historical data using a learning type algorithm.</p><p>　　A back propagation neural network with the output provided by a rule based ES was designed by Chiu et al. for th

21、e LTELF problem. To demonstrate that the inclusion of the prediction from a rule based ES of a power system would improve the predictive capability of the network, load forecasting was performed on the Taiwan power syste

22、m. The evaluation of the system showed that the inclusion of the rule based ES prediction significantly improved the neural network’s prediction of power load.</p><p>　　2.2 Artificial neural networks</p&g

23、t;<p>　　ANNs are an information processing technique based on the way biological nervous systems, such as the brain, process information. The fundamental concept of ANNs is the structure of the information process

24、ing system. Composed of a large number of highly interconnected processing units (“neurons”) connected into networks, a neural network system uses the human-like technique of learning by example to resolve problems. Ever

25、y neuron applies an input, activation and an output function to its net inp</p><p>　　The first researchers who introduced the ANN application in LTELF were Lee et al., who proposed an innovative ANN methodol

26、ogy for the LTELF problem. Park et al. proposed the use of a multilayer network with three layers, i.e. one input, one hidden and one output. The training of the network was performed through a simple back-propagation al

27、gorithm. Using load and weather information, the system produced three different forecast variables. Lee et al. treated electric load demands as a non-station</p><p>　　In 1992, Peng et al. presented a search

28、 procedure for selecting the training cases for ANNs to recognize the relationship between weather changes and load shape, while Ho et al. implemented a multilayer neural network with an adaptive learning algorithm.</

29、p><p>　　Chen et al. proposed an ANN for weather sensitive long term load forecasting, while an alternative technique using a recurrent high order neural network was considered by Kariniotakis et al. Papalexopou

30、los et al. proposed the inclusion of additional input variables, such as a seasonal factor and a cooling/heating degree into a single neural network.</p><p>　　Czernichow et al. used a fully connected recurre

31、nt network for load forecasting in which the learning database consisted of 70,000 patterns with a high degree of diversity. Mandal et al. applied neural networks for LTELF in which the inputs consisted of the past load

32、data only, and no weather variables were used, while Sforna and Proverbio investigated the application of ANNs in LTELF, through a research project at ENEL, and confirmed their positive contribution. </p><p>

33、;　　In 1997, Kiartzis et al. presented the Bayesian combined predictor, a probabilistically motivated predictor for LTELF based on the combination of an ANN predictor and two linear regression predictors. The method was a

34、pplied to LTELF for the Greek Public Power Corporation dispatching center of the island of Crete. Ramanathan et al. made several comparisons of statistical, time series and ANN methods for the LTELF.</p><p>

35、　　In 1998, Sforna reported the implementation of a software tool, called NEUFOR, based on ANN technology and specifically designed to meet the operational needs of utility power system dispatchers regarding online operat

36、ion, while Papadakis et al. continued to improve their previous work. The same goes for Drezga and Rahman. The development of improved neural network based LTELF models for the power system of the Greek island of Crete,

37、as well as radial basis function networks and fuzzy neural typ</p><p>　　3. Conclusions</p><p>　　Electricity long term load forecasting is important for the power industry, especially in the dere

38、gulated electricity market. Proper demand forecasts help the market participants to maximize their profits and/or reduce their possible losses by preparing an appropriate bidding strategy. Traditional statistics based li

39、near regression methods need modification to capture the more and more non-linearities in demand signals under the market conditions.</p><p>　　What emerges from this discussion is that AI based systems are b

40、ecoming more and more common decision making tools in LTELF. AI methods for forecasting have shown an ability to give better performance in dealing with the non-linearities and other difficulties in modeling the time ser

41、ies. The ESs as well as the ANNs have been found to be the most popular for this field. The advantage of these technologies over statistical models lies in their ability to model a multivariate problem without making<

42、/p><p>　　Concluding, we can say that AI techniques, like all other approximation techniques, have relative advantages and disadvantages. There are no rules as to when a particular technique is more or less suit

43、able for LTELF. Based on the survey presented here, it is believed that AI offers an alternative “philosophy" which should not be underestimated at all.</p><p><b>　　第二部分：譯文</b></p>&l

44、t;p>　　基于人工智能的長(zhǎng)期電力負(fù)荷預(yù)測(cè)</p><p>　　K. Metaxiotis, A. Kagiannas, D. Askounis, J. Psarras</p><p>　　摘要：基于人工智能( AI )技術(shù)的智能解決方案，由于是為了解決復(fù)雜的實(shí)際問(wèn)題，所以在社會(huì)各界得到越來(lái)越廣泛的重視?；贏I系統(tǒng)的開發(fā)和發(fā)展在世界各地得到應(yīng)用，主要是因?yàn)樗鼈兊耐评?，靈活性和解釋能

45、力。本文為在人工智能技術(shù)領(lǐng)域研究者提供了一個(gè)概述，以及他們?cè)陂L(zhǎng)期電力負(fù)荷預(yù)測(cè)(LTELF) 領(lǐng)域的當(dāng)前應(yīng)用。人工智能在LTELF中運(yùn)用的發(fā)展概括，通過(guò)討論了各種方法，以及目前的研究方向。文章最后通過(guò)交流想法和估計(jì)人工智能的前途。這項(xiàng)研究結(jié)果顯示，雖然人工智能技術(shù)仍然被視為一種新的方法，但在許多應(yīng)用中，從實(shí)用的角度看，它已顯示出是一種成熟的方法。</p><p>　　關(guān)鍵詞：人工智能；電力負(fù)荷預(yù)測(cè)；能源</p

46、><p><b>　　1.引言</b></p><p>　　在過(guò)去二十年中，人工智能已經(jīng)被確定為研究如何讓電腦做的事情比人做得更好。人工智能提供了功能強(qiáng)大和靈活的手段解決那些往往不能單靠更傳統(tǒng)和正統(tǒng)的方法解決的問(wèn)題。</p><p>　　這次研究見證了人工智能技術(shù)在長(zhǎng)期電力負(fù)荷預(yù)測(cè)(LTELF)領(lǐng)域中的應(yīng)用。當(dāng)然，這不是第一篇研究人工智能系統(tǒng)與能源

47、有關(guān)的問(wèn)題成功的運(yùn)用的論文?？傮w上講，人工智能在能源領(lǐng)域的發(fā)展已經(jīng)被別的作者從不同的觀點(diǎn)引入。</p><p>　　Taylor和Lubkeman的研究是以基于電力工程規(guī)劃問(wèn)題的學(xué)科應(yīng)用，當(dāng)Zhang等人在調(diào)查ES技術(shù)在電力系統(tǒng)中的相關(guān)應(yīng)用時(shí)，描述了北卡羅萊納州立大學(xué)的原型設(shè)計(jì)的發(fā)展。Ypsilantis和Yee發(fā)表基于電力應(yīng)用的SCADA的專家系統(tǒng)， Lubarskii討論了電網(wǎng)應(yīng)用的專家系統(tǒng)。自那時(shí)起，在不同

48、的能源相關(guān)領(lǐng)域，發(fā)表了幾篇其它調(diào)查論文。</p><p>　　但是，這編論文有著不同的重點(diǎn)。寫一個(gè)人工智能在能源系統(tǒng)中應(yīng)用的全面調(diào)查，，客觀上是行不通的。為此，我們本文的目的是為研究者創(chuàng)建一個(gè)大型知識(shí)庫(kù)，介紹人工智能在LTELF特定領(lǐng)域中的應(yīng)用，指明在別的領(lǐng)域的廣泛應(yīng)用。 </p><p>　　2．人工智能在長(zhǎng)期電力負(fù)荷預(yù)測(cè)中的應(yīng)用</p><p><b>

49、;　　2.1 專家系統(tǒng)</b></p><p>　　專家系統(tǒng)是人工智能分支中商業(yè)化最成功之一。Welbank定義專家系統(tǒng)如下：專家系統(tǒng)在某一特定領(lǐng)域是一個(gè)具有廣泛的知識(shí)基礎(chǔ)的程序，并使用復(fù)雜的推理來(lái)執(zhí)行人類專家可以做到的任務(wù)。</p><p>　　換句話說(shuō)，專家系統(tǒng)是一種計(jì)算機(jī)系統(tǒng)，它包含一個(gè)有組織的知識(shí)體，它是模擬專家解決問(wèn)題的技巧，在一個(gè)有界域的專長(zhǎng)。該系統(tǒng)能夠?qū)崿F(xiàn)專家解決

50、問(wèn)題的水平，這通常是通過(guò)熟練的人在解決某一領(lǐng)域中的重大問(wèn)題。</p><p>　　第一次將專家系統(tǒng)運(yùn)用于長(zhǎng)期電力負(fù)荷預(yù)測(cè)是Rahman、Bhatnagar和Jabbour等人。這些方法的目的是實(shí)驗(yàn)工作者利用知識(shí)、經(jīng)驗(yàn)和分析性思維作出進(jìn)一步加強(qiáng)，采用模糊邏輯的專家系統(tǒng)來(lái)解決LTELF問(wèn)題。1990年，Ho等發(fā)表一篇使用一種基于知識(shí)的專家系統(tǒng)在臺(tái)灣的電力系統(tǒng)的長(zhǎng)期負(fù)荷預(yù)測(cè)，而在1993年，Rahman和Hazim試圖

51、概括他的第一部作品。Markovic和Fraissler提出了一種專家系統(tǒng)的辦法(基于Prolog語(yǔ)言)滿足長(zhǎng)期負(fù)荷預(yù)測(cè)的真實(shí)性檢查公布的需求。</p><p>　　1995年, Kim等人實(shí)施了長(zhǎng)期負(fù)荷預(yù)測(cè)，通過(guò)利用神經(jīng)網(wǎng)絡(luò)和模糊專家系統(tǒng)，而后來(lái)，Mori和Kobayashi提出了最優(yōu)模糊推理方法來(lái)解決LTELF的問(wèn)題，Ranaweera等人提出一個(gè)應(yīng)用長(zhǎng)期電力負(fù)荷預(yù)測(cè)的模糊邏輯專家系統(tǒng)模型，用模糊規(guī)則去結(jié)合歷

52、史天氣和負(fù)荷數(shù)據(jù)。這些模糊規(guī)則被從歷史數(shù)據(jù)運(yùn)用學(xué)習(xí)型算法所包含。</p><p>　　基于專家系統(tǒng)的規(guī)則，Chiu等人為解決LTELF問(wèn)題設(shè)計(jì)出一種后臺(tái)傳播輸出的神經(jīng)網(wǎng)絡(luò)。一個(gè)包含基于電力系統(tǒng)的專家系統(tǒng)預(yù)測(cè)的演示證明了網(wǎng)絡(luò)的預(yù)測(cè)能力。負(fù)荷預(yù)測(cè)工作是針對(duì)臺(tái)灣的電力系統(tǒng)。系統(tǒng)的評(píng)價(jià)顯示，基于專家系統(tǒng)規(guī)則的預(yù)測(cè)大大提高了神經(jīng)網(wǎng)絡(luò)預(yù)測(cè)電力負(fù)荷的能力。</p><p>　　2.2 人工神經(jīng)網(wǎng)絡(luò)<

53、;/p><p>　　神經(jīng)網(wǎng)絡(luò)是一種基于生物神經(jīng)系統(tǒng)方式的信息處理技術(shù)，就像大腦對(duì)信息的處理。人工神經(jīng)網(wǎng)絡(luò)的基本概念是這樣的，結(jié)構(gòu)中的信息處理系統(tǒng)由一大批高度互連的處理單元(“神經(jīng)元”)連成網(wǎng)絡(luò)組成，一個(gè)神經(jīng)網(wǎng)絡(luò)系統(tǒng)采用像人一樣的來(lái)解決問(wèn)題。每一個(gè)神經(jīng)元，經(jīng)適量的輸入，激活和輸出，它的輸入計(jì)算出其輸出。神經(jīng)網(wǎng)絡(luò)是一個(gè)具體應(yīng)用的配置，如數(shù)據(jù)分類或模式識(shí)別，學(xué)習(xí)的過(guò)程稱為“訓(xùn)練” 。</p><p>

54、;　　第一個(gè)介紹了神經(jīng)網(wǎng)絡(luò)在LTELF中的應(yīng)用的研究者是Lee等人，他們提出一項(xiàng)創(chuàng)新神經(jīng)網(wǎng)絡(luò)方法論來(lái)解決LTELF的問(wèn)題。Park等人建議采用的一種新的多層網(wǎng)絡(luò)，有三個(gè)層次，即輸入，隱含層和輸出。訓(xùn)練中的網(wǎng)絡(luò)進(jìn)行了簡(jiǎn)單的BP算法。利用負(fù)荷和天氣信息，該系統(tǒng)制作了三套不同的預(yù)測(cè)變量。Lee等人處理的電力負(fù)荷需求，作為一個(gè)非平穩(wěn)時(shí)間序列，并參照它們的負(fù)荷分布的神經(jīng)網(wǎng)絡(luò)。</p><p>　　1992年，彭等人發(fā)表了一

55、個(gè)程序?qū)⑸窠?jīng)網(wǎng)絡(luò)進(jìn)行精確的訓(xùn)練，并且承認(rèn)天氣變化和負(fù)荷變化形狀的關(guān)系，與此同時(shí)，Ho等人實(shí)行多層神經(jīng)網(wǎng)絡(luò)的自適應(yīng)學(xué)習(xí)算法。</p><p>　　Peng等人提出了神經(jīng)網(wǎng)絡(luò)對(duì)天氣敏感的長(zhǎng)期負(fù)荷預(yù)測(cè)，而另一種技術(shù)，Kariniotakis和Papalexopoulos建議采用高階神經(jīng)網(wǎng)絡(luò)增加輸入變量，例如季節(jié)因素和冷卻/加熱程度形成的單一神經(jīng)網(wǎng)絡(luò)。</p><p>　　Czernichow等人

56、采用全連接鏈型網(wǎng)絡(luò)的負(fù)荷預(yù)測(cè)，其中學(xué)習(xí)數(shù)據(jù)庫(kù)包含70,000模式，具有高度的多樣性。Mandal等人應(yīng)用神經(jīng)網(wǎng)絡(luò)進(jìn)行LTELF，其中輸入只用過(guò)去的負(fù)荷數(shù)據(jù)，沒有任何天氣變量使用，而Sforna和Proverbio研究ANN在LTELF 中的應(yīng)用，通過(guò)在ENEL中的研究項(xiàng)目，其積極的貢獻(xiàn)得到確認(rèn)。</p><p>　　1997年，Kiartzis等人發(fā)表了貝葉斯組合預(yù)測(cè)，在一種概率方法預(yù)測(cè)LTELF的基礎(chǔ)上，結(jié)合人

57、工神經(jīng)網(wǎng)絡(luò)預(yù)測(cè)和兩個(gè)線性回歸預(yù)測(cè)。該方法已用于希臘公共電力公司調(diào)度中心的克里特島上。Ramanathan等人多次比較、統(tǒng)計(jì)，對(duì)時(shí)間序列分析和神經(jīng)網(wǎng)絡(luò)方法進(jìn)行了研究。</p><p>　　1998年，Sforna報(bào)道了一個(gè)執(zhí)行軟件工具，稱為NEUFOR,基于人工神經(jīng)網(wǎng)絡(luò)技術(shù)，專門為滿足業(yè)務(wù)需要的公用電力系統(tǒng)調(diào)度人員網(wǎng)上作業(yè)，雖然Papadakis等人繼續(xù)改善他們以前的工作。Drezga和Rahman也在做同樣的工作

58、。開發(fā)改進(jìn)型神經(jīng)網(wǎng)絡(luò)LTELF模型，對(duì)希臘克里特島上的電力系統(tǒng)，Kodogiannis和Anagnostakis在1999年就提出并討論徑向基函數(shù)神經(jīng)網(wǎng)絡(luò)和模糊神經(jīng)網(wǎng)絡(luò)類型。在2000年和2001年，一些研究者應(yīng)用ANN對(duì)LTELF的問(wèn)題進(jìn)行處理都取得了不同程度的成功。</p><p><b>　　3. 結(jié)論</b></p><p>　　電力長(zhǎng)期負(fù)荷預(yù)測(cè)對(duì)電力行業(yè)來(lái)說(shuō)

59、是重要的，尤其是在開放的電力市場(chǎng)。通過(guò)準(zhǔn)備一個(gè)合適價(jià)格，正當(dāng)需求的預(yù)測(cè)，可以幫助市場(chǎng)參與者獲得最大的利潤(rùn)或減少其可能的損失。傳統(tǒng)統(tǒng)計(jì)的線性回歸方法需要改進(jìn)，以便掌握更多的非線性需求信號(hào)下的市場(chǎng)條件。</p><p>　　提出這種討論是基于人工智能的系統(tǒng)正變得在LTELF中越來(lái)越成為普遍決策工具。人工智能預(yù)測(cè)方法已顯示有更好的能力來(lái)處理非線性和在建模的時(shí)間序列中遇到的其他困難，專家系統(tǒng)和神經(jīng)網(wǎng)絡(luò)一直被認(rèn)為是最熱門的

60、領(lǐng)域。利用這些技術(shù)，對(duì)統(tǒng)計(jì)模型的優(yōu)勢(shì)在于它們有能力模擬多元問(wèn)題而在輸入變量時(shí)不用依賴復(fù)雜的假設(shè)。此外，通過(guò)對(duì)訓(xùn)練資料的學(xué)習(xí)，神經(jīng)網(wǎng)絡(luò)從輸入變量中提取非線性關(guān)系。</p><p>　　最后，我們可以說(shuō)，人工智能技術(shù)，像所有其他逼近技術(shù)，有相對(duì)優(yōu)點(diǎn)和缺點(diǎn)。當(dāng)當(dāng)某一特定技術(shù)，或多或少都適合LTELF時(shí)，它們沒有規(guī)律。 基于這種情況，在這里, 我們相信人工智能提供了一種可供選擇的方法，其作用是不容低估的。</p&g