Research On Power System Fault Diagnosis Method Based On Bayesian Networks

Always influenced by natural and human factors in the operation of the power system, power system inevitably leads to failure.At the same time, at the time of failure will have a large number of alarm information to dispatching center, to dispatch personnel to bring huge difficulties accident analysis and accident treatment. So if we can dispatch personnel to a automatic processing of alarm system, will make dispatch staff get twice the result with half the effort on the analysis of the accident and accident treatment. But the traditional artificial intelligence methods mainly focus on how to deal with the alarm information uncertainty in space, and how to build model, in the time on the study of the uncertainty exists serious deficiencies, at the same time not to build a model can simultaneously deal with the uncertainty of space-time. Therefore, choose a bayesian network for the problem, based on temporal causal bayesian network model is proposed.First,this article elaborated the grid fault diagnosis research significance, research status quo and existing problems.Secondly introduces the related theory of bayesian network, described the importance of independence in bayesian network and Leaky Noisy- Or how to reduce the complexity of modeling, node model and simple approximate reasoning algorithm of bayesian network is introduced.Then, in view of the fault diagnosis research to research the problem of insufficient time uncertain causal bayesian network model is proposed based on time, the introduction of fuzzy theory is a causal relationship between fault and alarm, set up two models used with "or"- "and" node, then gives the conditional probability of nodes and the calculating method for the probability of failure hypothesis, finally through the case analysis to confirm the accuracy of the model.Finally, based on SCADA and puts forward a new SOE information grid fault diagnosis technology based on the time causal bayesian network power grid fault diagnosis technology. Introducing new elements in the model- switch, search optimization method is put forward based on the breadth for correlation path to determine switch state, finally analyzed by an example, to verify the accuracy of the model. This model can be used very good solve the scheduling of grid fault diagnosis technology, has the very good robustness...