REAL TIME FAST PREDICT TRANSIENT STABILITY BASED CORRECTED POTENTIAL SURFACE CRITERION

A new prediction control method is proposed for improving transient stability of power systems in this paper. This method uses the combination of the practical measurement with prediction analysis to replace the traditional deductivecalculation. This paper presented a hybrid algorithm combining the temporal differences methods with BP algorithm basedsynchronization multi-parameter measurement. It helps to solve the computing problem incrementally in traditional BPalgorithm on multi-step predicting and has the ability ofstructural learning. The proposed scheme of high real timeprediction for out-of-step consists of control center basedcomputer and synchronization multi-parameter measuremenbased microprocessors. A number of numerical simulationsunder different swing conditions show that the scheme issatisfied with both correctness and fastness. The fault critical clearing time predicted by the scheme is almost equal to that calculated by the numerical integration program under the same fault conditions.