The electrical parameters identification is an important means to realize the winding condition monitoring of power transformers. Therefore

was proposed a method of on-line monitoring based on short-circuit reactance identification for three-winding transformer. In order to improve the accuracy of parameter identification

we investigated the relationship between different load distributions of the transformer and the condition number of the coefficient matrix.Using the condition number as a criterion

we adjusted the number of solution parameters according to different load states of the transformer. Furthermore

relying on the big data platform

we established an on-line monitoring method for transformer windings by reasonable screening data and identification scheme adaptive matching. Finally

numerical simulation was used to verify the influence of condition number on accuracy of parameter identification

while the engineering data were used to demonstrate the performance of on-line monitoring. The results show that the condition number of the coefficient matrix has an obvious influence on the accuracy of parameter identification

where the condition number is positively correlated with the error of the identification result. At the same time

the analysis of the actual engineering data by this method is consistent with the real state of the transformer. The results of this study can provide references for parameter identification and on-line status monitoring of three-winding transformer.