Fast transient stability calculation by pattern recognition method

Prevention of large-scale blackout is one of the most important design goals for power system engineers. The Tokyo Electric Power Co. and the Tokyo Shibaura Electric Co. have jointly developed an automatic security monitoring system. This system checks the security of power systems under several hundred prespecified fault conditions from viewpoints of transient stability, power flow, voltage and frequency. The conventional step-by-step method of calculating transient stability requires a long computing time and therefore it is almost impossible to apply the step-by-step method for on-line security monitoring. The pattern recognition method developed recently has many favorable features in a fast stability calculation. In the pattern recognition method, a small number of observed prefault values of state variables such as bus voltages, generator powers, etc., is substituted into the so-called classifier. If the calculated value of the classifier for an assumed fault condition exceeds a threshold value, the power system under consideration is judged to be unstable. To apply the pattern recognition method, it is necessary to calculate in advance the values of the classifier under various fault conditions. This requires a repetition of the transient stability calculation by the step-by-step method. Further, the probability that the stability of the powermore » system is misjudged is reported to be considerably high. The possibilities of improving the calculation accuracy and reducing misjudgment probability and off-line preparatory computing time were studied and are reported. Examples of applications of the improved method are given to show its effectiveness.« less