Fuzzy Neural Network based Voltage Security Assessment with Structure and Weight Initialization

The main objective of power system planning and operation is to maintain the system security while fulfilling certain constraints and contingencies. With the global trend towards deregulation, the frequency and complexity of security checks are increasing in order to accommodate the market trends. If the system is found to be insecure, timely corrective measures need to be taken to prevent system collapse. The paper proposes a fuzzy neural network (FNN) based approach for voltage security assessment employing a severity index, based on bus voltage violations for accurate prediction of the system state. Conventional artificial neural network (ANN) presents an opaque structure to the user without giving any insight to the output generation process. A method based on fuzzy curves has been employed to determine significant inputs, to initialize the structure, initial weights and rules for the security assessment problem from the input-output data of the system. Once properly initialized, the FNN trains much faster as compared to an ANN. The effectiveness of the proposed method has been demonstrated on IEEE 30-bus system.