外界扰动对大停电幂律影响与电网故障长程相关性研究

发布时间：2018-04-12 20:20

本文选题：周期性 + 恶劣天气　；参考：《长沙理工大学》2014年硕士论文

【摘要】：在电力系统的各种事故中,大停电是危害最为严重的一种。新世纪以来,研究人员先后发现世界各国大停电规模服从幂律分布而并非经典可靠性理论假设的指数分布。研究表明,电力系统作为一个复杂系统具有自组织临界性。大停电作为电力系统的动态行为,其规模服从幂律分布,并具有长程自相关性。在外界扰动和负荷增长的推动下,电力系统会不断逼近自组织临界态。临界状态下,外界扰动导致线路跳闸后,在潮流转移作用下,可能引发连锁过负荷和继电保护系统的隐形故障,扩大事故范围,造成大停电事故的幂律分布特性。电网大停电时,常伴随有大量故障。研究表明,电网故障和大停电一样具有幂律分布特性和长程自相关性。但电网故障对大停电的幂律分布特性有何种作用,未见报道；电网故障长程自相关性的形成机理,也缺乏相关研究。论文针对此类问题展开研究。为探明外界扰动所致电网故障对大停电幂律特性的影响,文章从北美大停电记录数据和我国电网故障数据中找寻外界扰动独立导致大停电幂律特性的证据。首先,抓住幂律分布和指数分布的差异,分析了造成北美大停电幂律特性厚尾部分的29条极值大停电记录,发现其中有10起发生在恶劣天气条件下且仅限于配网故障。因配电网为环网设计、开环运行,发生连锁过负荷的概率远小于输电网,该部分极值大停电可能为外界扰动独立引起而并非连锁故障作用；其次,还分析了国内9处具有幂律分布特征的输、配电网故障数据。分析江西数据表明,该省冰灾期间负荷仅为正常情况下的一半,不会发生连锁过负荷。此外,该省输电网冰灾期间线路跳闸146次,而确定为继电保护误动的仅有1次,同样没有明显的连锁故障。因江西电网冰灾大停电系该省最大规模的停电事故,这从另一个角度证明了外界扰动可以在没有连锁故障的情况下引起大停电的幂律特性。为验证电网故障长程自相关性由恶劣天气引起的故障高峰所致,文章还分析了长沙和南昌电网故障数据。将故障数据按日极值故障数从高到低依次剔除最大日故障数据,并利用趋势波动法计算出各数据序列的Hurst指数,发现Hurst值随日故障最大值的减小呈下降趋势,当日故障最大值降至10或9时,Hurst指数值甚至小于0.5。这意味着电网故障的长程自相关性主要是由恶劣天气等外界扰动引起的故障高峰所致。
[Abstract]:Among all kinds of power system accidents, blackout is the most serious one.Since the new century, researchers have found that the power law distribution is not the exponential distribution of the classical reliability theory hypothesis.It is shown that the power system as a complex system has self-organizing criticality.As the dynamic behavior of power system, power outage is distributed according to power law and has long range autocorrelation.Driven by external disturbances and load growth, the power system will continue to approach the self-organized critical state.Under the critical condition, after the line tripping caused by the external disturbance, the cascading overload and the hidden fault of the relay protection system may be caused under the action of the power flow transfer, and the power law distribution characteristic of the blackout accident will be caused by enlarging the range of the accident.A large number of faults are often accompanied by power outages.It is shown that power law distribution and long range autocorrelation are similar to power outages.However, there is no report on the effect of power law distribution on power law distribution of power outages, and the formation mechanism of long range autocorrelation of power grid faults is also lack of relevant research.This paper focuses on this kind of problems.In order to find out the influence of power law characteristics caused by external disturbances on power law characteristics of power outages, this paper looks for evidence of power law characteristics caused by external disturbances from North American blackout record data and China power grid fault data.First of all, the difference between power law distribution and exponential distribution is grasped, and 29 extreme blackouts in the thick-tailed part of power law characteristic of power outages in North America are analyzed. It is found that 10 of them occur in severe weather conditions and are limited to distribution network failures.Because the distribution network is designed for loop network, the probability of occurrence of cascading overload is much smaller than that of transmission network. This part of extreme blackout may be caused by external disturbance but not cascading fault. Secondly,The fault data of power law distribution network in 9 places in China are also analyzed.The analysis of Jiangxi data shows that the load during ice disaster in this province is only half of that under normal conditions and there will be no interlocking overload.In addition, 146 trip trips were made during the ice disaster period of the provincial transmission network, but only one time was identified as relay protection maloperation, and there was also no obvious cascading fault.Since the ice disaster blackout in Jiangxi power grid is the largest power outage in the province, it is proved from another angle that the power law characteristic of the outage can be caused by external disturbance without cascading faults.In order to verify the fault peak caused by bad weather, the fault data of Changsha and Nanchang power networks are analyzed.The maximum daily fault data are eliminated according to the daily maximum number of faults from high to low, and the Hurst exponent of each data sequence is calculated by using trend fluctuation method. It is found that the Hurst value decreases with the decrease of daily maximum fault number.The maximum value of Hurst exponent is even less than 0.5.This means that the long range autocorrelation of power grid faults is mainly caused by the peak of faults caused by external disturbances such as bad weather.
【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM732

【共引文献】