【作者】 张威；

【导师】 王雨虹； 包伟川；

【作者基本信息】 辽宁工程技术大学 ， 工程硕士（专业学位）， 2021， 硕士

【摘要】 随着全国电网的互联,电力系统规模日渐庞大。区域间电网长距离输电线路和快速励磁装置的大规模投入导致系统的总阻尼下降,因此易发生低频振荡,给电力系统造成极大的安全隐患。通过在励磁系统装设电力系统稳定器(Power System Sabilizer,PSS)可有效提升系统阻尼,抑制低频振荡。针对电力系统低频振荡辨识和稳定器设计问题,本文提出采用改进Prony法辨识低频振荡和使用天牛群算法优化PSS参数的方法。首先,以低频振荡的机理、分析辨识、抑制为思路开展研究。以负阻尼机理为研究基础,利用Prony分析法辨识系统振荡模式。针对Prony分析法对噪声敏感的问题,通过小波阈值去噪方法降噪,用测试算例验证了改进Prony法的可行性。结合负阻尼机理分析了PSS抑制低频振荡的原理,以实际工程中常用的PSS为研究对象进行参数优化。其次,PSS通过附加阻尼转矩提高系统振荡模式阻尼,机组参数选择不合适会使某些振荡模式阻尼增强,但也可能会使其他振荡模式的阻尼下降,系统总阻尼会受到影响,因此需合理整定PSS参数,考虑各参数之间的协调优化。本文提出利用改进天牛须搜索算法的PSS参数整定方法,针对天牛须搜索算法(BAS)全局搜索能力不强的缺陷,将BAS算法与粒子群算法(PSO)融合得到天牛群算法(BSO),采用三种典型测试函数验证算法性能,并将BSO算法与PSO算法和分层多子群的均匀分布混沌粒子群算法(HUCPSO)进行比较,结果表明BSO算法对高维函数的处理效果更好,可实现全局搜索,迭代次数更少,收敛所需时间更短。最后,在Matlab/Simulink中分别搭建单机无穷大系统模型和四机两区域系统模型,目标函数为系统特征值阻尼比的二次性能指标,以BSO算法为优化工具,将实际问题转化为PSS的参数优化问题,用去噪后改进的Prony法辨识采样数据,在相同条件下与PSO算法和HUCPSO算法进行对比。在单机无穷大系统中进行大小干扰分析,初步验证了该算法性能;在四机两区域系统中不同运行方式下进行大小干扰分析,进一步验证了该算法具有更好的阻尼特性和鲁棒性,可更加快速抑制低频振荡。该论文有图43幅,表11个,参考文献66篇。更多还原

【Abstract】 With the interconnection of the national power grids,the scale of the power system is becoming larger and larger.The large-scale investment of long-distance transmission lines and fast excitation devices in the inter-regional power grid has led to a decrease in the total damping of the system,so low-frequency oscillations is prone to occur,which causes great security risks to the power system.By installing power system stabilizer(PSS)in the excitation system,the system damping can be effectively improved and the low-frequency oscillation can be suppressed.Aiming at the problem of power system low frequency oscillation identification and stabilizer design,this thesis proposes an Improved Prony method to identify low frequency oscillation and the beetle swarm algorithm to optimize PSS parameters.Firstly,the mechanism,analysis,identification and suppression of low frequency oscillation are studied.Based on the research of negative damping mechanism,Prony analysis method is used to identify the system oscillation mode.Aiming at the problem that the Prony analysis method is sensitive to noise,the wavelet threshold denoising method is used to reduce the noise,and the feasibility of the improved Prony method was verified by a test example.Combined with the mechanism of negative damping,the principle of PSS’s suppression of low-frequency oscillation was analyzed,and PSS,which is commonly used in practical engineering is used as the research object to optimize the parameters.Secondly,PSS improves the damping of system oscillation mode by adding damping torque.Improper selection of unit parameters will enhance the damping of some oscillation modes,but it may also reduce the damping of other oscillation modes,and the total damping of the system will be affected.Therefore,it is necessary to reasonably adjust PSS parameters and consider the coordination and optimization of various parameters.This thesis proposes a PSS parameter tuning method using an improved beetle search algorithm.Aiming at the shortcomings of the beetle search algorithm(BAS)that the global search ability is not strong,the BAS algorithm and the particle swarm optimization(PSO)are combined to obtain the beetle swarm algorithm(BSO),used three typical test functions to verify the performance of the algorithm,and compared the BSO algorithm with the PSO algorithm and the layered fertility group of chaotic particle swarm optimization algorithm(HUCPSO),the results show that BSO algorithm is better for highdimensional functions,can achieve global search,has fewer iterations and takes less time to converge.Finally,build a single-machine infinity system model and a four machine two area system model in Matlab/Simulink.The objective function is the secondary performance index of the eigenvalue damping ratio of the system,and the BSO algorithm was used as an optimization tool,and the actual problem was transformed into a parameter optimization problem of PSS.The sampled data was identified by the improved Prony method after denoising,and compared with the PSO algorithm and the HUCPSO algorithm under the same conditions.The large and small interference analysis was carried out in the single-machine infinity system,and the performance of the algorithm was initially verified;the large and small interference analysis was carried out in the four-machine two-area system under different operation modes,which further verified that the algorithm has better damping characteristics and robustness,and can suppress low frequency oscillation more quickly.There are 43 figures,11 tables and 66 references in this thesis.更多还原