Power system security assessment using a hybrid expert system/neural network architecture

Presents a system in which artificial intelligence techniques and conventional analysis methods are combined to provide an analysis tool suitable for online power system security assessment. This system uses a hierarchical architecture which coupled an expert system (ES) with a set of neural networks (NNs) for assessing power system contingencies. The rule-based ES is used to examine various approximate measures to perform a preliminary coarse screening and direct the potentially harmful contingencies for further analysis by the NNs in a two-level structure. The first-level NN, called the severity evaluator, performs a finer screening by classifying the contingencies according to the severity of limit violations. The second-level NN, called the postcontingency predictor, estimates the postcontingency voltage drops and line flows. The proposed system is particularly useful for power systems which must rely on offline load flow results for voltage limit violations. A 17-bus power system model example is used to demonstrate the applicability of the system