Research On The Data-driven Short-circuit Current Calculation Method For Distributed Systems With IIDGs

With increasingly serious problems of energy and environment,more and more distributed generations(DGs)are used in distributed systems,and most of them belong to inverter interfaced distributed generations(IIDGs).In spite of positive effects with the high penetration of IIDGs,the safe and stable operation of the distributed system faces new challenges.Due to the strong nonlinear characteristics of IIDGs,the shortcircuit current of IIDGs is significantly different from traditional generators.Thus,the traditional short-circuit current calculation methods is not applicable,which increases the difficulty of the settings of the relay protection.At present,the short-circuit current in a distributed system with IIDGs is mainly calculated by using physical modeling,which has drawbacks in the case of high IIDG penetration.In this dissertation,a datadriven short-circuit current calculation method for distributed systems with IIDGs and high penetration of IIDGs is proposed.The primary contents and results of this research include:(1)A data-driven short-circuit current calculation model for distributed systems with IIDGs is proposed.First,the overall framework and process of the model are given.The features of the distributed system related to short-circuit current are analyzed,and the short-circuit current when IIDGs are not connected to the distribution network is proposed as a key feature innovatively from the perspective of fault analysis theory.A sample generation method based on MATLAB / Simulink is proposed,with which data sets can be automatically accumulated.The proposed framework improves the realtime application capability of short-circuit current calculation methods with the procedure of offline training and online application.At the same time,modeling with data can eliminate errors caused by IIDG mechanism modeling with different simplified assumptions.(2)The above model is applied to actual calculation scenarios.First,a new datadriven short-circuit current calculation method is proposed for the calculation of the short-circuit current flowing through a specified measurement point after a fault occurs in a certain position in the distributed system.And two types of feature combination are given,each with corresponding applicable scenario.Second,according to the calculation of the short-circuit current flowing at a specified measurement point when a fault occurs at any position in active distribution networks,a new short-circuit current calculation method for 10 fault types based on XGBoost method is proposed.The case study shows that compared with short-circuit current calculation methods based on mechanism modeling,the above two methods have the advantages of fast online calculation,strong applicability,high accuracy,and easy programming,which facilitate the applications of adaptive setting and coordination of protection relays,fault management,etc.,which are based on short-circuit calculation.(3)For the further calculation scenario of the short-circuit current flowing through any node after any type of fault occurs at any position in the active distribution network,which can be considered as a multi-output regression problem,a short-circuit current calculation method based on multi-target regressor stacking algorithm and XGBoost algorithm is proposed.The correlation between different short-circuit currents is considered in model training.The case study shows that the proposed method is accurate and stable in the prediction of each short-circuit current.