【摘要】：隨著電力系統(tǒng)規(guī)模不斷擴(kuò)大,以及各區(qū)域電網(wǎng)的互聯(lián)工程的推進(jìn),造成傳統(tǒng)的集中式全網(wǎng)最優(yōu)潮流越來(lái)越難以滿足在線分析與實(shí)時(shí)控制的要求;同時(shí)集中式串行計(jì)算方式存在數(shù)據(jù)外泄風(fēng)險(xiǎn)高,數(shù)據(jù)傳輸量大,計(jì)算“維數(shù)災(zāi)”等問(wèn)題�？焖�、高效、穩(wěn)定的分區(qū)解耦分布式算法已成為解決大規(guī)模OPF的關(guān)鍵。為實(shí)現(xiàn)分區(qū)解耦計(jì)算,首先需要利用分區(qū)方法將一個(gè)大的電力系統(tǒng)分為多個(gè)區(qū)域系統(tǒng),接著利用分解協(xié)調(diào)算法對(duì)多區(qū)域系統(tǒng)進(jìn)行解耦分布式計(jì)算。因此,本文研究主要涉及兩個(gè)主要內(nèi)容,一是分區(qū)方法,二是分解協(xié)調(diào)算法。在電網(wǎng)分區(qū)方法上,本文提出了與分層分區(qū)結(jié)構(gòu)相適應(yīng)的電壓分區(qū)方法。500kV/220kV電磁環(huán)網(wǎng)會(huì)給電力系統(tǒng)運(yùn)行帶來(lái)諸多的不利影響,打開(kāi)電磁環(huán)網(wǎng)形成分層分區(qū)網(wǎng)架結(jié)構(gòu)將是電網(wǎng)的發(fā)展方向�；诜謱臃謪^(qū)結(jié)構(gòu)的特點(diǎn),提出了以220kV電壓等級(jí)為依據(jù)的分區(qū)方法,將大電網(wǎng)劃分為500kV及以上電壓等級(jí)的輸電網(wǎng)和多個(gè)220kV及以下電壓等級(jí)的高中壓配電網(wǎng)。這樣的分區(qū)方式能得到幾個(gè)規(guī)模適中、區(qū)域間連接耦合較弱的子網(wǎng),也與現(xiàn)實(shí)基于電壓等級(jí)的電力調(diào)度相適應(yīng)。在將全網(wǎng)數(shù)據(jù)按區(qū)域進(jìn)行歸類的實(shí)現(xiàn)上,提出了節(jié)點(diǎn)注入電流法。在求解多區(qū)域系統(tǒng)的最優(yōu)潮流問(wèn)題中,采用了兩種高效的分解協(xié)調(diào)算法——分解協(xié)調(diào)內(nèi)點(diǎn)法、改進(jìn)近似牛頓方向法。分解協(xié)調(diào)內(nèi)點(diǎn)法通過(guò)對(duì)邊界節(jié)點(diǎn)“一分為二”及增加耦合約束條件,實(shí)現(xiàn)了多區(qū)域模型的等價(jià)轉(zhuǎn)換,并利用魯棒性好的現(xiàn)代內(nèi)點(diǎn)法進(jìn)行求解,使各區(qū)域的內(nèi)部變量保持相互獨(dú)立,只需要邊界變量信息的傳遞協(xié)調(diào),從而實(shí)現(xiàn)了分區(qū)解耦后的最優(yōu)潮流分布式計(jì)算;改進(jìn)近似牛頓方向法是在近似牛頓方向法的基礎(chǔ)上,保留了簡(jiǎn)約修正方程中潮流方程的梯度部分,使獲得的近似牛頓方向更好地逼近于純牛頓方向,使其對(duì)較強(qiáng)耦合系統(tǒng)也能實(shí)現(xiàn)完全解耦。所提的兩種分解協(xié)調(diào)算法都具有一個(gè)顯著的特征,其修正方程都具有對(duì)角加邊的結(jié)構(gòu)。利用對(duì)角加邊結(jié)構(gòu)可實(shí)現(xiàn)修正方程的降階解耦,大大減少了修正方程的維數(shù),提高了計(jì)算速度和減少了計(jì)算機(jī)內(nèi)存占用,同時(shí)也實(shí)現(xiàn)了分區(qū)解耦計(jì)算。通過(guò)搭建多個(gè)具有分層分區(qū)結(jié)構(gòu)的電力系統(tǒng),首先利用所提出的節(jié)點(diǎn)注入電流法實(shí)現(xiàn)電力系統(tǒng)的快速分區(qū),進(jìn)而對(duì)分區(qū)后形成的多區(qū)域系統(tǒng)運(yùn)用兩種分解協(xié)調(diào)算法進(jìn)行最優(yōu)潮流的計(jì)算驗(yàn)證,所得計(jì)算結(jié)果表明這兩種解耦算法都能很好的提高計(jì)算效率,實(shí)現(xiàn)大網(wǎng)架的分區(qū)解耦計(jì)算。這種解耦計(jì)算方式對(duì)于其他大規(guī)模優(yōu)化問(wèn)題也具有廣泛的應(yīng)用前景。
[Abstract]:With the expansion of power system scale and the promotion of regional power network interconnection project, the traditional centralized optimal power flow becomes more and more difficult to meet the requirements of on-line analysis and real-time control. At the same time, there are some problems such as high risk of data leakage, large amount of data transmission, calculation of "dimension disaster" and so on. Fast, efficient and stable partition decoupling distributed algorithm has become the key to solve large scale OPF. In order to realize the divisional decoupling computation, a large power system should be divided into several regional systems by using the partition method, and then the distributed decoupling computing of the multi-area system is carried out by using the decomposition and coordination algorithm. Therefore, this paper mainly involves two main contents, one is partition method, the other is decomposition and coordination algorithm. In the area of power grid partition, this paper proposes a voltage partition method. 500kV / 220kV electromagnetic ring network, which is suitable for the hierarchical partition structure, will bring a lot of adverse effects to the power system operation. It will be the development direction of the power network to open the electromagnetic ring network to form a layered and partitioned grid structure. Based on the characteristics of hierarchical partition structure, a partition method based on 220kV voltage grade is proposed. The large power grid is divided into transmission networks with 500kV and above voltage levels and high voltage distribution networks with multiple 220kV and voltage levels below. In this way, several subnets with moderate scale and weak coupling between regions can be obtained, which are also suitable for power dispatching based on voltage level in reality. In the realization of classifying the whole network data by region, the nodal injection current method is proposed. In order to solve the optimal power flow problem for multi-region systems, two efficient decomposition and coordination algorithms, the decomposition coordinated interior point method, are used to improve the approximate Newton direction method. By dividing the boundary nodes into two and adding the coupling constraints, the decomposition and coordination interior point method realizes the equivalent transformation of the multi-region model, and uses the robust modern interior point method to solve the problem. The internal variables of each region are kept independent of each other, only the information of boundary variables is transferred and coordinated, thus the distributed optimal power flow calculation after partition decoupling is realized, and the improved approximate Newton direction method is based on the approximate Newton direction method. The gradient part of the power flow equation is preserved in the reduced correction equation, which makes the obtained approximate Newton direction approach to the pure Newton direction better, and it can also achieve complete decoupling for the strongly coupled system. Both of the proposed decomposition and coordination algorithms have a remarkable characteristic, and the modified equations have diagonal edge structure. The diagonal triangulation structure can be used to reduce the order decoupling of the modified equation, which greatly reduces the dimension of the modified equation, improves the computing speed and reduces the computer memory consumption, and also realizes the partition decoupling calculation. Firstly, the proposed node injection current method is used to realize the fast partitioning of power system. Then, two decomposition and coordination algorithms are used to verify the optimal power flow of the multi-area system formed after partitioning. The results show that the two decoupling algorithms can improve the efficiency of calculation and realize the partition decoupling calculation of large grid. This decoupling method also has a wide application prospect for other large-scale optimization problems.
【學(xué)位授予單位】：廣西大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TM744

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 汪超群;韋化;吳思緣;;計(jì)及風(fēng)電不確定性的隨機(jī)安全約束機(jī)組組合[J];電網(wǎng)技術(shù);2017年05期

2 呂仁周;白曉清;李佩杰;代景龍;林頌晨;;基于交替方向乘子法的電動(dòng)汽車分散式充電控制[J];電力系統(tǒng)自動(dòng)化;2016年16期

3 汪超群;韋化;吳思緣;楊健;;七種最優(yōu)潮流分解協(xié)調(diào)算法的性能比較[J];電力系統(tǒng)自動(dòng)化;2016年06期

4 陽(yáng)育德;龔利武;韋化;;大規(guī)模電網(wǎng)分層分區(qū)無(wú)功優(yōu)化[J];電網(wǎng)技術(shù);2015年06期

5 李佩杰;陸鏞;白曉清;韋化;;基于交替方向乘子法的動(dòng)態(tài)經(jīng)濟(jì)調(diào)度分散式優(yōu)化[J];中國(guó)電機(jī)工程學(xué)報(bào);2015年10期

6 許立雄;劉俊勇;劉洋;茍競(jìng);閆占新;張里;吳楊;;基于負(fù)荷電流場(chǎng)的電網(wǎng)分區(qū)方法[J];電網(wǎng)技術(shù);2015年04期

7 陽(yáng)育德;韋化;李濱;;求解暫態(tài)穩(wěn)定約束最優(yōu)潮流模型的遞推降階解耦算法[J];中國(guó)電機(jī)工程學(xué)報(bào);2015年02期

8 戶秀瓊;顏偉;趙理;余娟;趙霞;;互聯(lián)電網(wǎng)聯(lián)網(wǎng)最優(yōu)潮流模型及其算法[J];電力系統(tǒng)自動(dòng)化;2013年03期

9 黎蕾;;增廣拉格朗日函數(shù)的兩種分解方法比較[J];重慶科技學(xué)院學(xué)報(bào)(自然科學(xué)版);2012年06期

10 李智;楊洪耕;;基于鄰近中心算法的無(wú)功優(yōu)化分解協(xié)調(diào)計(jì)算[J];電力自動(dòng)化設(shè)備;2012年12期

相關(guān)碩士學(xué)位論文 前10條

1 高冬那;地區(qū)電網(wǎng)電壓無(wú)功分層分區(qū)綜合協(xié)調(diào)控制的研究[D];華北電力大學(xué);2013年

2 李泉;河北南網(wǎng)分層分區(qū)及其穩(wěn)定性研究[D];華北電力大學(xué);2011年

3 李磊;無(wú)功優(yōu)化的控制策略研究[D];西南交通大學(xué);2010年

4 阮解瓊;基于綜合靈敏度的無(wú)功規(guī)劃優(yōu)化研究[D];昆明理工大學(xué);2009年

5 佟永吉;沈陽(yáng)地區(qū)500kV/220kV電網(wǎng)電磁解環(huán)方法和技術(shù)的研究[D];東北大學(xué);2009年

6 岳曉華;基于全網(wǎng)分區(qū)的多目標(biāo)分布式無(wú)功優(yōu)化[D];山東大學(xué);2008年

7 溫力力;電力系統(tǒng)多區(qū)域無(wú)功優(yōu)化分布式算法研究[D];重慶大學(xué);2007年

8 李民;鎮(zhèn)江電網(wǎng)分層分區(qū)運(yùn)行的研究[D];東南大學(xué);2006年

9 蔡廣林;基于PC網(wǎng)格的水火電力系統(tǒng)最優(yōu)潮流分布式計(jì)算研究[D];廣西大學(xué);2005年

10 李曉華;基于電網(wǎng)分區(qū)和遺傳算法的無(wú)功規(guī)劃優(yōu)化[D];山東大學(xué);2005年

，

本文編號(hào)：2211289