Voltage stability assessment using artificial neural network

Many techniques have been suggested to investigate the voltage instability phenomena and many trails have been carried out to find an accurate, fast, and reliable method to detect how the system is far from the instability boundary. In this paper artificial neural network has been suggested which can do the job in very fast and accurate way. A proper approach based on exact mathematical bases is used to generate the training set. A mathematical technique which relates the behavior of the voltage variation with the load conditions is used to generate the training set. In such way, the accuracy of the generated training set is guaranteed, also there will be no need for a big size data for the training set as it is deduced through exact mathematical relations. An index, which indicates the degree of the stability of the power system, has been used. This index has been deuced from the bifurcation diagram. The index used to detect the voltage stability varies between a large numerical value at no load condition to zero value at the stability boundary. The suggested technique has been applied to a sample power system and the results were found very satisfactory from the point of view of speed for obtaining the answer required and accuracy of the solution.