Research On Reliability Evaluation Method For Relay Protections In Smart Substation

New techniques like Gigabit Ethernet, high speed communication, accurate synchronous sampling, non-conventional inductors are applied in smart substation, making the system structure become more complex, which has a negative impact on the system reliability. As the first defense line of power system, relay protection equipment is the key to ensure safe and stable operation of the power grid, and comprehensive modeling and evaluation for its reliability is of particular importance. This paper discussed reliability evaluation method for relay protections in smart substation from three aspects.Relay protection system in smart substation is of high reliability which often results in lack of failure data. This paper first discussed reliability evaluation based on historical and statistical data, and proposed a reliability evaluation method based on support vector machine under small sample condition. Then, via utilizing grid search and cross validation, optimize the parameters of support vector machine, thus improving the estimation accuracy. Simulation and analysis show that the algorithm has high precision under small sample condition, and can inhibit the effect of outliers to some extent.After acquiring basic reliability parameters of the various devices, reliability modeling and analysis were put forward based on the system structure of relay protection. Focusing on the complicated networking modes and diverse maintenance strategies, reliability analysis method was proposed considering maintenance of the protection system. The approach can effectively solve the state space explosion problem caused by the complicated system structure of smart substation through the combination of reliability block diagram and Monte-Carlo simulation.Online condition evaluation is the premise of condition based maintenance. This paper discussed online condition evaluation of protection devices in the end. Condition evaluation based on fuzzy support vector machine was proposed due to the advantage of smart substation in condition monitoring. Via the pretreatment of the historical operation records and remaining life in percentage, the feature vector is formed and samples of the same type of devices is utilized as supplement at the same time. Then fuzzy support vector machine algorithm is used to differentiate the sample weights and generate the evaluation model by training the samples, while condition evaluation factors from online monitoring are pretreated simultaneously. After that, the online monitoring data can be evaluated by using the condition evaluation model generated offline, realizing online condition evaluation of protection devices finally.The above methods use the historical statistical data, the static structure information, maintenance strategy and online monitoring data of relay protection in smart substation respectively. Both the similarity and personality of the devices are focused on in the reliability assessment, thus achieving comprehensive assessment for the reliability of relay protection.