A Novel Method of Newton Iteration in Complex Field and Lattice Search for Locating Partial Discharges in Transformers

在变压器局部放电特高频定位的多传感器时间差测量中不可避免地存在误差,这往往导致时间差方程组在实数域内无解。为了得到时间差方程组的最优近似解,提出了复数域牛顿迭代网格搜索方法。该方法在复数域内进行迭代;当迭代结果为实数时就将其作为定位结果;当迭代结果为复数时,以其实部为中心点坐标,在该点周围局部区域内采用网格搜索法计算最优近似解。通过在变压器上开展试验检验了该方法的可行性和准确性。研究表明,在一定时间误差下,在实数域采用牛顿迭代法计算时不收敛,而在复数域可以收敛,平均定位误差为0.20 m。复数域牛顿迭代网格搜索方法能够在一定时间误差下求解时间差方程组,给出最优近似解,实现局部放电定位。

In the location of partial discharges （PDs） in transformers using the ultra-high-frequency （UI-/F） method, the errors of time-differences at different sensors are unavoidable, so there is no solution to the time-difference equations in real field. Thus, location of PDs cannot be obtained by newton＇s iteration method in real field which has always been used in PD location up till now. To work out the optimum approximate resolution of the time-difference equations with some time-difference errors, the novel method of Newton＇s iteration in complex field and lattice search is suggested.