The detection of traditional instrument transformer adopts the method of artificially reading error

which has low information level and low detection efficiency and cannot meet the requirements of automatic sampling inspection.A kind of detection method of instrument transformer

proposed in this paper on the basis of big data analysis

can read mass ratio and angle difference data of capacitor voltage transformer（CVT）. The feature data can be extracted by adopting the moving window method and the error limit can be updated adaptively. It can achieve automatically the accuracy assessment of the instrument transformer to be tested without the aid of the calibrator of instrument transformer. Finally

the CVT repetition test for measurement data of standard PT is performed. The test shows that this method can be used for long-term mass data analysis of CVT which

compared with traditional principal component analysis method

has higher accuracy and is free from the fluctuation of power grid. The method in this paper can provide some reference for calibration and inspection of instrument transformer.