Parameter Identification and Calculation of Return Voltage Curve Based on FDS Data

Abstract

Frequency domain spectroscopy (FDS) method based on dielectric response measurement technique is being investigated as a potential tool for condition assessment of transformer insulation. It is a simple and nondestructive method to monitor insulation status of transformer. This paper establishes a mathematical model with the data of frequency domain spectroscopy for solving parameter of dielectric response equivalent circuit, which is a group of nonlinear equations. It is difficult to solve the problem using nonlinear least square optimization method. The problem solving for nonlinear equations is transformed to the constrained optimization problem using nonlinear least square method, and then, the bacterial foraging algorithm is used for solving the problem. According to identified circuit parameters, the FDS curve is obtained, which compares the calculated data with the measured. The relative error is controlled in 1%, which demonstrates the parameter identification method is successful. The return voltage curve can be calculated from the estimated circuit parameters for verifying insulation condition. Therefore, the transformer insulation condition can be diagnosed using multi-physical parameters with FDS and RVM data comprehensively, to improve the accuracy of diagnosis.