Study Of The Methods Amending Outliers In A DSP Based Electric Power Parameter

Along with the development of high and new technique, particularly the information technique, numerous modernization industrial and public control or management equipments based on the computer, microprocessor, and electric power electronics are more sensitive to electric power quality. The loss of economy and society, which is influenced by electric power quality, is increasingly serious. As a result, it is taken seriously that the electric power quality must be improved. In order to do so, we have to analyze the electric power parameter correctly; however, the outliers that are leaded in the electric power parameter reduce the analytical credibility. A great deal of domestic and international experiment expresses: even a little amount of outliers may cause analysis result deviate true appearance clearly. So, to remove outliers can make follow-up analysis correct, and it is also one of the important topics in electric power parameter analysis.According to the algorithms that are often used in amending outliers at present and the correlation of signals that we deal with in our project, this paper presents analgorithm-the method mending outliers based on the correlation of signals,whose arguments are based on the statistics. For measuring the good and bad of algorithms, we found an assessment system to do it. According to this assessment system, three algorithms that we discussed in the paper are evaluated professionally. Simulation calculation and engineering application in practice demonstrate that the algorithm is characterized by high reliability, fast and simpleness. It can eliminate the bad effect and improve accuracy in later analyses.Although the algorithm is mainly applied in the electric power parameter in this paper, it can be also applied in multi-frequencies and time-varying correlation of signal systems. So the method we have discussed in this paper may be of help and inspire to those who do similar work.