计及警报信息时间特性的故障诊断模型

电力系统发生故障时的警报信息具有时间特性，如果能够充分与合理地利用这一特性，可以 提高故障诊断结果的准确性和加快诊断速度。但至今，还没有一个电力系统故障诊断方法能 够系统地处理警报信息的时间特性。以外展推理(abductive inference)和简洁覆盖集理 论(parsimonious set covering theory)为基础，对计及警报信息时间特性的电力系统故 障诊断问题做了一些初步的研究工作。采用时间图表示元件故障与警报信息及其时间特性之 间的关系，时间图中的节点表示警报信息，节点间的有向支路表示这两个节点所代表的警报 信息出现的先后次序。在此基础上，首次建立了描述这一问题的优化模型(0-1整数规划模型 )，给出了用Tabu搜索技术求解该问题的算法。最后，用一个简单的例子说明了所建立的故 障诊断模型的正确性和Tabu搜索方法的可行性。

A NEW FAULT DIAGNOSIS MODEL CAPABLE OF DEALING WITH THE TEMPORAL INFORMATION OF ALARM MESSAGES

It has been recognized that the temporal information of alarm messages in power systems plays an important role in fault diagnosis.Regretfully,in the existing power system fault diagnosis methods the temporal knowledge has not been well explored.As an initial attempt,this paper presents a new fault diagnosis model capable of dealing with the temporal information of alarm messages based on the abductive inference model and parsimonious set covering theory. At first, a temporal graph is employed to represent the relationship among element fault, relevant alarms and their temporal information. In the temporal graph,each node denotes an alarm, and the directed arcs between nodes denote temporal precedence of these alarms. A 0-1 integer programming formulation is next established for describing this problem, and a solving algorithm based on the Tabu search (TS) technique is then presented.A simple example is used to demonstrate the correctness of the developed fault diagnosis model and the feasibility of the TS based algorithm.