Importance Sampling and the Impact of the Protection System on Power System Reliability

9634823索普对北美可靠性委员会五年报告的研究表明，保护系统经常在导致电力系统扰动的一系列事件中发挥作用。在未来放松管制的系统中，可靠性很可能会被视为一种商品。例如，独立系统运营商(ISO)由于一条关键输电线路的错误跳闸而未能促进生产者和负荷之间的电力谈判销售，可能导致支付罚款或以其他方式影响输电服务的价格。保护系统的故障对整个系统可靠性的影响虽然很重要，但还没有得到很好的研究。现有的保护系统具有多个保护区域和冗余系统，偏向于可靠性，即故障总是通过一些继电保护来清除。当今的中继系统被设计成以安全为代价而具有可靠性。虽然人们普遍认为，数字继电器的自我监控和自我检查功能将降低继电器未被发现或出现“隐性故障”的概率，但安装的微处理器继电器的数量仍然很少。此外，还没有对安装大量数字继电器对系统可靠性的影响进行定性评估。这项研究的目的是开发技术来评估保护系统对系统可靠性的影响，以及评估安装具有内置自监测和自检或保护系统其他修改的新的数字继电器的价值。电力系统运行的新范式似乎有可能为更现代化的保护系统提供动力。也就是说，传输系统可靠性的经济性将要求继电保护系统现代化。该方法的两个关键内容是使用“重要性抽样”和生成电力系统运行状态的大型数据库，如用于训练各种人工方案和评估新控制方案的性能的数据库。“重要性抽样”技术是一种模拟技术，用于研究重大电力系统扰动等“罕见”事件。为了训练各种人工智能方案，电力系统运行状态数据库的生成已成为最近研究的课题。它们已被用于暂态稳定的预防性控制，从实时相量测量预测不稳定，或使用实时相量测量控制直流线路。这项研究的实质是将这两种技术结合起来，即建立一个偏向于不太可能事件的数据库，因为概率是基于重要性抽样的。数据库生成被认为是一种统计方法，它可以通过随机抽样描述系统运行的参数来生成数据库。这里提出的工作将使用数据库作为潜在概率空间的代理。就像在重要抽样中改变概率一样，数据库将偏向不太可能发生的事件，以便通过模拟来研究减少“隐藏故障”的影响。拟议研究中涉及的具体任务是：*研究隐藏故障与系统条件的相关性的概率模型。*修改重要性抽样，使其更适合电力系统应用*检查适用于重要性抽样方法的性能指标*使用康奈尔生产超级计算机生成数据库*

9634823 Thorp The study of North American Reliability Council reports over a five year period indicates that protection systems frequently play a role in the sequence of events that lead to power system disturbances. It is likely that reliability will be treated as a commodity in the deregulated systems of the future. For example the failure of the Independent Systems Operator (ISO) to facilitate a negotiated sale of power between a producer and a load because of a false trip of a key transmission line could result in the payment of a penalty or otherwise effect the price of transmission services. In spite of its importance, the impact of protection system malfunctions on overall system reliability has not been well studied. The existing protection system with its multiple zones of protection and redundant systems is biased toward dependability, that is a fault is always cleared by some relay. Present day relaying systems are designed to be dependable at the cost of security. While it is generally believed that the self-monitoring and self-checking feature of digital relays will reduce the probability of undetected or "hidden failures" in relays, the number of installed microprocessor relays is still small. Further, there has been no qualitative evaluation of the effect on system reliability of the installation of large number of digital relays. The proposed research is to develop techniques to evaluate the impact on system reliability of the protection system and to appraise the value of the installation of new digital relays with built in self-monitoring and self-checking or other modifications of the protection system. It seems possible that the new paradigm for power system operation will provide the impetus for a more modern protection system. That is, the economics of transmission system reliability will require that the relaying system be modernized. The two key elements of the proposed approach are the use of "importance sampling" and th e generation of large data bases of power system operating conditions such as those used for training of various artificial schemes and evaluation of the performance of new control schemes. The "importance sampling" technique is a simulation technique to study "rare" events such as major power system disturbances. The generation of data bases of power system operating conditions for training of various artificial intelligence schemes has become a subject of recent study. They have been used for preventive control for transient stability, for prediction of instability from real-time phasor measurements, or the control of DC lines using real-time phasor measurements. The essence of the proposed research is to combine the two techniques i.e., to create a database biased toward the unlikely events as the probabilities are based in importance sampling. Data base generation has been recognized as a statistical approach where the data base could be generated by randomly sampling the parameters that describe system operation. The work proposed here would use the data base as a surrogate for the underlying probability space. The data base would be biased toward the unlikely events just as the probabilities are altered in importance sampling so that the impact of reducing "hidden failures" can be studied through simulation. The specific tasks involved in the proposed research are: * Investigation of probability models for the dependence of hidden failures on system conditions.. * Modifications of importance sampling to make it more appropriate for power system applications * Examination of performance indices appropriate to the importance sampling approach * Generation of data bases using the Cornell Production Super Computer ***