Study Of Clustering Analysis For Power Load Series And Fault Type Recognition Of The Transmission Line Based On Ant Colony Optimization Algorithm

In this paper, the clustering analysis of power load series and the recognition ofthe fault type of transmission line based on Ant Colony Optimization Algorithm(ACOA) are first time studied and presented. AOCA has very successful applicationin many related fields. But its application in power system is just started. ACOA hasstrong advantages in large-scale complex combined optimization, and proved itsvalidity. For better mastering basic principle of ACOA and its algorithm, to achieve thetwo studying subjects of this paper, the clustering theory and its methods based datamining and the research trends of domestic and foreign are summarized necessarily.So does ACOA at the next step. Above jobs are the important foundations of theresearch in the paper. External random factors to power loads will lead to the reducing of short-term loadforecasting (STLF) accuracy. STLF has been done by combining clustering and patternrecognition before. But the performance of clustering does not correspond with theexternal weather factors perfectly. So the clustering results are not satisfied byforecasting precision. In this paper, the clustering analysis of power load series basedon ACOA is first time presented. Compared with Kohonen neural network used inpower load clustering usually, the load clustering performance of ACOA in actualload system has shown its superiority, which has more sensitivity and resolution toclimatic factors and high temperature area;to festival and holiday condition andwhich has more minute and even of the clustering feature on the similarity of loadcurve profile. The above clustering performance has a most important significance toimprove the accuracy of STLF. The recognition of the fault type of transmission line is very important to relayprotection and reclosure. Now the processing to transient signal based FourierTransform has limitation and does not ensure the accuracy of fault recognition. In thispaper the recognition of the fault type of transmission line is realized by using AOCAat first time. Three phases current data and zero-sequence current data after the faulttime are directly used. With transformed, these data composed the characteristicvectors. The clustering to the vectors indicate that AOCA can achieve the fault typerecognition of transmission line reliably and accurately, and the result is notinfluenced by fault type, operation of power system, transition resistance, initial phaseangle, or the location of fault.