Equivalence Analysis of Off-Line Test and On-Line Detection Technology Based on Data Accuracy

Abstract

To reduce the consumption of manpower and material resources on off-line test, it is necessary to research on how to extend the cycle of off-line tests to reduce the frequency of off-line tests. In this paper, the equivalence between off-line test and on-line detection technology is analyzed from the point view of data accuracy, which provides the quantized basis for evaluating the feasibility of replacing off-line test with on-line detection technology. As the cycle of off-line test is different from the cycle of on-line detection technology, neither quantity nor time series anastomoses. So after the original data of on-line detection technology are fitted by the Cubic Spline Interpolation, the estimated data should be taken value from the fitting curves according to the dates of off-line test. Then, the estimated data of on-line detection technology and the original data of off-line test can be compared. The Euclidean Distance and the Cosine Similarity of two datasets are computed to characterize the absolute error and relative error. Therefore, the specific and reasonable weights of absolute error and relative error are determined rationally according to regulations on the state evaluation of electrical equipment. And then the data accuracy compared on-line detection technology to off-line test, which is equivalence, is calculated. Finally, the improvement scheme of off-line test cycle is proposed according to the standard based on the value of equivalence index. The model in this paper is validated by the state variables data of a power transformer.