發(fā)布時(shí)間：2021-12-28 23:14

　　最優(yōu)潮流計(jì)算（OPF）是電力系統(tǒng)經(jīng)濟(jì)運(yùn)行的基礎(chǔ)。其一般具有大規(guī)模非凸約束而難以求解,解決方法之一是利用松弛技術(shù)將該約束轉(zhuǎn)化成凸約束再用分布迭代方法求解。但松弛技術(shù)涉及到復(fù)雜計(jì)算,且收斂速度慢�，F(xiàn)有的其他方法通常是同步迭代形式,每次迭代均需收集齊所有子區(qū)域的迭代結(jié)果,因此優(yōu)化計(jì)算非常困難、效率較低。本文提出了一種新的電力系統(tǒng)分布式最優(yōu)潮流計(jì)算方法。文中具體采用了交替方向乘法器（ADMM）算法,通過對(duì)罰參數(shù)的比較選擇,確立了最佳罰參數(shù),可以完全分布式地求解非凸OPF問題,解決了 OPF收斂速度慢的問題。建立了基于節(jié)點(diǎn)電壓變量復(fù)制的分布式最優(yōu)潮流計(jì)算模型。主要對(duì)N個(gè)節(jié)點(diǎn)的網(wǎng)絡(luò)進(jìn)行劃分和分解,通過分離負(fù)責(zé)節(jié)點(diǎn)間耦合的變量,將計(jì)算向下分配到各工作節(jié)點(diǎn)。在工作節(jié)點(diǎn)上觀察到相鄰電壓的實(shí)部和虛部,并引入輔助變量,進(jìn)行交替迭代計(jì)算和更新局部電壓、功率、主變量和雙變量。本文通過不同IEEE算例系統(tǒng)的仿真驗(yàn)證了 ADMM的電力系統(tǒng)分布式最優(yōu)潮流計(jì)算方法的有效性。選擇了 IEEE30、57和118節(jié)點(diǎn)算例,本文分析了 OPF模型中的有功功率、無功功率等變量收斂過程,對(duì)殘差進(jìn)行了分析,并觀察了懲罰參數(shù)對(duì)原始?xì)埐?.. 

【文章來源】：華北電力大學(xué)(北京)北京市 211工程院校 教育部直屬院校

【文章頁數(shù)】：75 頁

【學(xué)位級(jí)別】：碩士

【文章目錄】：
摘要
ABSTRACT
Abbreviations
Notations
CHAPTER 1 INTRODUCTION
    1.1 IMPORTANCE OF OPTIMAL POWER FLOW
    1.2 OPTIMIZATION IN POWER SYSTEM
    1.3 RESEARCH STATUS
    1.4 MOTIVATION TOWARD TOPIC
    1.5 SOME NATURAL QUESTIONS
    1.6 THESIS FLOW CHART: METHODOLOGY
    1.7 CONTRIBUTION OF THE THESIS
CHAPTER 2 DISTRIBUTED OPTIMAL POWER FLOW PROBLEM
    2.1 INTRODUCTION TO OPTIMAL POWER FLOW PROBLEM
        2.1.1 Standard OPF Problem
        2.1.2 Objective Function
        2.1.3 Variables
        2.1.4 Constraints
    2.2 OPTIMAL POWER FLOW CHALLENGES
    2.3 OPF SOLUTION METHODOLOGIES
        2.3.1 Conventional Methods
        2.3.2 Intelligent Methods
    2.4 RISE OF DISTRIBUTED OPF
    2.5 METHODS TO SOLVE DISTRIBUTED OPF
    2.6 OPF FORMULATION FOR DISTRIBUTED ADMM
    2.7 NODE-BASED FORMULATION
    2.8 SUMMARY
CHAPTER 3 OPTIMIZATION THEORY AND ADMM
    3.1 DISTRIBUTION OPTIMIZATION
    3.2 DUAL ASCENT
    3.3 METHOD OF MULTIPLIERS
    3.4 ALTERNATING DIRECTION METHOD OF MULTIPLIERS
    3.5 CONVERGENCE RESULTS
    3.6 OPTIMALITY CONDITIONS AND STOPPING CRITERION
    3.7 ADMM AND QUADRATIC PROGRAMMING
    3.8 ADMM Implementation
        3.8.1 X-Update
        3.8.2 Y-Update
        3.8.3 Dual Update
    3.9 PSEUDOCODE FOR DISTRIBUTED ALGORITHM
    3.10 STOPPING CRITERIA
    3.11 DYNAMIC PENALTY PARAMETER
    3.12 SUMMARY
CHAPTER 4 NUMERICAL RESULTS
    4.1 TEST NETWORKS
    4.2 INITIALIZATION AND PARAMETER TUNING
    4.3 PERFORMANCE EVALUATION
    4.4 CONVERGENCE BEHAVIOR
    4.5 SUMMARY
CHAPTER 5 CONCLUSION
APPENDIX A
ACKNOWLEDGEMENT
REFERENCES



本文編號(hào)：3554921