Application of Bayesian Network in Power Grid Fault Diagnosis

The rule base can be visually mapped into an initial network of starting leaning with noisy-or and noisy-and node models to effectively make use of the existing learning resources. After the rule base is transformed to the Bayesian network consisted of this node model, the parameters and structure can be learned and modified thus to realize the continuous perfection of diagnosis knowledge. The Bayesian network consisted of noisy-or and noisy-and nodes is applied into the power grid fault diagnosis and the general Bayesian network model of fault diagnosis facing the components is built, which has fewer parameter numbers than the common Bayesian network and can realize on-line rapid diagnosis of power grid fault.