Optimized Voltage Stability for Maximum Loadability Using Neural Networks

This paper proposes a Neural Network-Based method for on-line maximum loadability estimation, for an optimized power system voltage stability profile. A simulated annealing optimization technique for optimal voltage stability profile through out the whole power network was used. The minimization of the voltage stability index at each individual load bus as well as the global voltage stability indicator is obtained through adjustment of real power and reactive resources control devices. Optimal load buses voltages and angles at the input layer and the maximum MVA loading level at the output layer accomplished the training of the Radial Basis Function Neural Network (RBFNN). The generalization capability of the designed Neural Networks under large number of operation conditions and contingencies has been tested for power systems. Fast performance, accurate evaluation and good prediction for maximum loadadbility level have been obtained. Results of tests conducted on the Six-bus Wale and Hale system are presented and discussed.