Differential Protection Scheme Based On Singular Entropy Of Sparse Morphological Gradient For Power Transformers

As a bridge between different voltage levels in the power system,a transformer is responsible for the important tasks of voltage conversion and energy transmission,so its operating state can directly affect the operation state of the entire power grid.As we known,transformers are expensive to construct,difficult and long-time-consuming to repair,and susceptible to get an irreversible damage when a fault occurs because of the lower insulation levels compared to transmission lines of the same voltage level.Since differential protection has the advantages of fast action and strong selectivity,it has been selected as the main protection of large transformers.However,the saturation characteristics of ferromagnetic materials will bring the problems of transformer inrush current and current transformer saturation,which will affect the reliability of differential protection.At present,under the conditions of inrush and saturation,how to ensure the accuracy of the fault recognition is the key to improve the correct rate of the differential protection of power transformer.With the in-depth research on the key issues of transformer differential protection at home and abroad,some achievements have been made.But in face of complex situations of power systems,there are still some shortcomings in the existing schemes.Based on the theory of greyscale mathematical morphology,this paper first develops the basic theoretical framework of sparse grey-scale mathematical morphology,then researches on the performance of the sparse grey-scale mathematical morphology.And then the proposed sparse bipolar morphological gradient operator is used to prove that sparse expansion of mathematical morphology is effective in terms of noise immunity and efficiency.After that,combining the multi-scale sparse bipolar morphological gradients with singular entropy,a differential protection scheme based on singular entropy of sparse morphological gradient was proposed for transformers,where a three-stage process including rapid identification,accurate identification and supplementary identification is designed.In order to verify the superiority of the scheme,a simulation system for transformers based on PSCAD / EMTDC is established in this paper.The recognition rate of the scheme was analysed by simulating and verifying under the disturbance and fault cases.The results show that the scheme not only has a high accuracy of recognition,but also can accurately identify most fault conditions fast.Compared with other algorithms,the transformer differential protection scheme has higher accuracy of recognition on fault in both identification and rapid identification.Especially,it can improve the recognition accuracy of the no-load closing on the turn-to-turn fault situation in the first cycle of the current wave,which is of high practical value.