To accurately and rapidly identify the key power flow transfer section following a cleared line due to its fault or overload

a new fast search method is proposed based on the random fuzziness clustering algorithm and the shortest path algorithm. Firstly

the vectors of line power components obtained by power flow tracking are classified using the random fuzziness clustering method with the capability of adaptive determination of the optimal cluster number

leading to an initial transfer section formed by parallel line set of fault or overloaded line

to avoid the problem probably of multi-selection or missed-selection lines of the random fuzziness clustering algorithm due to the inappropriate classification number given in advance. Then for each branch in the initial transfer section

the shortest path is obtained by searching from one end of fault or overloaded line to another end and simultaneously passing the corresponding branch. All the branches in the shortest path

having the same power flow direction as that of the fault or overloaded line

are considered as the candidate lines of the key transfer section. The key power flow transfer section is finally checked out by the simple distribution factor. The method proposed has significantly improved the recognition accuracy by taking advantage of both the branch power component clustering key section selecting method and the shortest path selecting method. The results of a number of samples show its effect and validity.