【摘要】：隨著國家電網(wǎng)建設(shè)規(guī)模的日益增加,保障電力系統(tǒng)的安全穩(wěn)定運(yùn)行就顯得尤為重要。但是由于電力系統(tǒng)的特殊性,很多實驗環(huán)節(jié)只能借助計算機(jī)仿真實現(xiàn),此外當(dāng)電網(wǎng)出現(xiàn)短路、次同步振蕩等故障現(xiàn)象時,需要電力保護(hù)系統(tǒng)能夠快速的做出響應(yīng)并使電網(wǎng)在短時間內(nèi)恢復(fù)穩(wěn)定,這樣就要求系統(tǒng)有快速準(zhǔn)確的數(shù)據(jù)處理能力,因此傳統(tǒng)、單一的計算機(jī)串行仿真已經(jīng)難以滿足電力系統(tǒng)復(fù)雜性和數(shù)據(jù)實時處理的要求。鑒于DSP在并行運(yùn)算和大數(shù)據(jù)量處理上的優(yōu)勢,目前許多大型的電力系統(tǒng)并行運(yùn)算機(jī)群都是基于DSP實現(xiàn)的,但同時也帶來了研發(fā)成本高、需要龐大的硬件設(shè)施支持的不足,而市場專門應(yīng)用于中小型規(guī)模的電力系統(tǒng)環(huán)節(jié)實時仿真的產(chǎn)品并不多。此外本文通過對UDP和MAC協(xié)議方式的以太網(wǎng)通信在進(jìn)行短幀數(shù)據(jù)交互時的開銷進(jìn)行驗證比較后,得出基于MAC地址的網(wǎng)絡(luò)傳輸機(jī)制的數(shù)據(jù)交互,短幀數(shù)據(jù)的通信開銷約為34微秒,這個時間為實現(xiàn)實時仿真提供了重要保障。針對電力系統(tǒng)控制環(huán)節(jié)是有向有環(huán)圖的特點,本文在查閱了大量調(diào)度算法文獻(xiàn)的基礎(chǔ)上,以各任務(wù)之間通信關(guān)聯(lián)矩陣為基礎(chǔ),對系統(tǒng)模型進(jìn)行了任務(wù)分割并按照相關(guān)規(guī)則對通信矩陣進(jìn)行化簡整理消除關(guān)聯(lián)項,得出完成并行運(yùn)算需要的最優(yōu)處理器的數(shù)量,并以Blackfin548為硬件核心,計算出每個子任務(wù)的運(yùn)算時間,按照并行運(yùn)算動態(tài)負(fù)載平衡的策略對系統(tǒng)模型進(jìn)行了仿真研究,為基于嵌入式DSP并行運(yùn)算應(yīng)用在中小型電力系統(tǒng)實時仿真中的可行性和實用性提供了一定參考依據(jù)。 本文首先以同步發(fā)電機(jī)為研究對象,建立起同步發(fā)電機(jī)的數(shù)學(xué)模型并在MATLAB/SIMULINK里面進(jìn)行仿真,得出仿真曲線。然后將驗證過的同步發(fā)電機(jī)數(shù)學(xué)模型轉(zhuǎn)換成用傳遞函數(shù)表示的系統(tǒng)結(jié)構(gòu)圖,根據(jù)傳遞函數(shù)之間的邏輯關(guān)系進(jìn)行任務(wù)分解,最終化簡成7個任務(wù),并根據(jù)任務(wù)之間的通信關(guān)聯(lián)矩陣關(guān)系,進(jìn)行矩陣行列式化簡,將7個任務(wù)分配在3塊不同的DSP上,節(jié)點間的數(shù)據(jù)交互通過以太網(wǎng)接口以全雙工通信的方式實現(xiàn)。在任務(wù)分配中采用動態(tài)調(diào)度負(fù)載平衡算法,從而盡量保證了并行運(yùn)算時的負(fù)載平衡。文章最后以同步發(fā)電機(jī)三相短路故障為例,采用四階龍格庫塔法在單臺DSP上進(jìn)行了實時仿真并對仿真步長進(jìn)行了比較分析,得出了實時仿真較為理想的仿真步長。
[Abstract]:With the increasing scale of national power grid construction, it is particularly important to ensure the safe and stable operation of power system. However, because of the particularity of power system, many experimental links can only be realized by computer simulation. In addition, when the power network is short circuit, sub-synchronous oscillation and other fault phenomena, The power protection system needs to be able to respond quickly and restore stability to the power grid in a short period of time, which requires the system to have rapid and accurate data processing capability. The single computer serial simulation can not meet the requirements of power system complexity and real-time data processing. In view of the advantages of DSP in parallel operation and large amount of data processing, many large power system parallel computing clusters are based on DSP at present, but it also brings high R & D cost and needs huge hardware facilities to support. But the market specially applies to the medium and small scale power system link real-time simulation product is not many. In addition, after verifying and comparing the cost of short frame data interaction between UDP and MAC protocol, the data exchange of network transmission mechanism based on MAC address is obtained. The communication overhead of short frame data is about 34 microseconds. This time provides an important guarantee for the realization of real-time simulation. In view of the fact that the control link of power system is directed and cyclic graph, this paper, based on the reference of a large number of dispatching algorithms, is based on the communication correlation matrix among the tasks. The system model is divided into tasks and the communication matrix is simplified according to the relevant rules to eliminate the correlation items. The optimal number of processors needed to complete the parallel operation is obtained, and the Blackfin548 is the core of the hardware. The computation time of each sub-task is calculated, and the system model is simulated according to the strategy of parallel operation dynamic load balancing. It provides a certain reference for the feasibility and practicability of the application of parallel operation based on embedded DSP in the real-time simulation of small and medium power system. In this paper, the mathematical model of synchronous generator is established and simulated in MATLAB/SIMULINK, and the simulation curve is obtained. Then, the verified mathematical model of synchronous generator is converted into a system structure diagram represented by transfer function, and the task is decomposed according to the logical relation between transfer functions, and finally reduced to seven tasks. According to the communication relation between tasks, the matrix determinant is simplified, and seven tasks are assigned to three different blocks of DSP. The data interaction between nodes is realized by means of full duplex communication through Ethernet interface. In task allocation, dynamic scheduling load balancing algorithm is used to ensure load balance in parallel operation as far as possible. Finally, taking the three-phase short-circuit fault of synchronous generator as an example, the four-order Runge-Kutta method is used to carry out real-time simulation on a single DSP, and the simulation step size is compared and analyzed, and the ideal simulation step size is obtained.
【學(xué)位授予單位】：昆明理工大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2012
【分類號】：TM743;TP338.6

【參考文獻(xiàn)】

相關(guān)期刊論文 前9條

1 王占領(lǐng);鄭三立;;電力系統(tǒng)實時仿真技術(shù)分析[J];電力設(shè)備;2006年02期

2 王偉,劉為群,曾繼倫;用于AVR閉環(huán)測試的同步發(fā)電機(jī)組實時仿真研究[J];電力系統(tǒng)自動化;1999年23期

3 俞乾;鄭三立;;數(shù)字實時仿真在控制和保護(hù)系統(tǒng)測試中的應(yīng)用[J];電力系統(tǒng)及其自動化學(xué)報;2008年06期

4 王俊;張玉璽;李偉;;高性能DSP互聯(lián)方法[J];電子器件;2008年06期

5 郝曉平;;電力系統(tǒng)實時數(shù)字仿真技術(shù)及應(yīng)用進(jìn)展[J];湖北電力;2009年04期

6 許慶強(qiáng);徐賢;袁宇波;;江蘇電網(wǎng)實時數(shù)字仿真系統(tǒng)介紹[J];江蘇電機(jī)工程;2008年03期

7 鄭三立,黃梅,張海紅;電力系統(tǒng)數(shù)�；旌蠈崟r仿真技術(shù)的現(xiàn)狀與發(fā)展[J];現(xiàn)代電力;2004年06期

8 高朝暉,張曉斌,吳小華;基于MATLAB的飛機(jī)高壓直流電源系統(tǒng)數(shù)字仿真[J];系統(tǒng)仿真學(xué)報;2003年04期

9 易鵬;周紅;;系統(tǒng)仿真驗證車輛結(jié)構(gòu)設(shè)計[J];裝備機(jī)械;2009年01期

相關(guān)博士學(xué)位論文 前1條

1 馬丹;任務(wù)間相互依賴的并行作業(yè)調(diào)度算法研究[D];華中科技大學(xué);2007年

相關(guān)碩士學(xué)位論文 前10條

1 杜玉霞;基于網(wǎng)格的任務(wù)調(diào)度算法研究[D];山東師范大學(xué);2011年

2 袁亮;嵌入式快速邏輯控制系統(tǒng)的設(shè)計與開發(fā)[D];浙江大學(xué);2005年

3 馬保宏;空氣中目標(biāo)聲場建模與軟件實現(xiàn)[D];西北工業(yè)大學(xué);2006年

4 黃騫;雙DSP并行處理系統(tǒng)的設(shè)計與應(yīng)用研究[D];武漢理工大學(xué);2006年

5 王超;復(fù)合勵磁同步發(fā)電機(jī)技術(shù)研究[D];西北工業(yè)大學(xué);2007年

6 丁逸;基于層次策略的動態(tài)負(fù)載均衡算法研究[D];東南大學(xué);2005年

7 黃涌;網(wǎng)絡(luò)視頻電話終端系統(tǒng)雙核處理技術(shù)研究[D];華中科技大學(xué);2006年

8 杜杰;網(wǎng)格環(huán)境中基于DAG的并行任務(wù)調(diào)度算法研究[D];上海交通大學(xué);2009年

9 高梅;使用新型判據(jù)的水輪發(fā)電機(jī)組運(yùn)行穩(wěn)定性研究[D];合肥工業(yè)大學(xué);2009年

10 劉生;用于電力系統(tǒng)在線暫態(tài)穩(wěn)定數(shù)字仿真的調(diào)速器模型研究[D];天津大學(xué);2009年

，

本文編號：2372605