Hybrid artificial neural networks for voltage instability monitoring in electric power systems

The authors present a method for voltage instability monitoring in electric power systems with hybrid artificial neural networks. A three-layered perceptron is used to estimate voltage instability indices while the Kohonen net is utilized to understand the trajectory of power system conditions in terms of power system security. Two indices are used for monitoring voltage stability. The two indices are based on the margin to critical conditions with a pair of multiple load flow solutions. They are expressed by the angle between the operational conditions and the nearest singular point. One is expressed in state space which corresponds to the voltage solution while the other is in parameter space which implies the specified value of the load flow calculation. The proposed method is demonstrated in a sample system