A probabilistic approach to alarm processing in power systems using a refined genetic algorithm

In this paper, a probabilistic approach Is proposed for alarming processing in power systems based on a genetic algorithm (GA). Although alarm processing and fault diagnosis have different explanations in the power system community, from the viewpoint of the artificial intelligence community, the alarm processing problem is a typical multiple fault diagnosis (MFD) problem. Thus, at first we attribute the alarm processing problem to an MFD problem and discuss the MFD problem in general with special emphasis on the evaluation criteria. Then, a probabilistic criterion is introduced to describe the alarm processing problem, which is deemed more reasonable than the currently used criteria. Thirdly, a novel method is developed to solve the alarm processing problem using a refined genetic algorithm. Finally, a sample example is used to demonstrate the feasibility and efficiency of the developed method. The key features of this proposed method are that it has solid mathematical foundation and can find multiple global optimal solutions directly and efficiently in a single run. This is very suitable for complex alarm processing problems especially for the situations with missing or false alarms, because different combinations of events can produce the same set of alarms under these circumstances. The test results, although preliminary, suggest that the developed GA-based probabilistic method to the alarm processing problem is promising.