Electric Field Feature Extraction Of Air Gaps From Interelectrode Paths And Equipotential Rings And Its Application To Insulation Prediction

Air gap is the main form of external insulation of power transmission and transformation system,and its insulation strength is an important basis for the selection of air clearance.Due to the high cost of testing and the inability to exhaust all gap types,the discharge voltage quantitative calculation method represented by empirical formula,physical model and data model was proposed to replace part of the real test.However,the above methods all have problems such as poor extrapolation,unclear physical parameters,and inconsistent input features.The discharge form of the long air gap is dominated by external excitation and atmospheric parameters.Under the action of a specific voltage waveform and standard atmospheric conditions,its insulation level is mainly determined by the gap structure.In this regard,based on the principle of mutual coupling between electrostatic field distribution and electrode arrangement,this paper proposes an interelectrode electric field feature set that characterize the air gap structure,and uses random forest(random forest,RF)to establish an insulation prediction model that maps it to the discharge voltage.The main research content and achievements of this paper are as follows:(1)An interelectrode electric field feature set is proposed to characterize the arrangement of air gap,involving sphere-plane long gap and transmission line-tower gap;Extract the 83 electric field features defined in the main path,auxiliary paths and equipotential rings as input,and use RF to establish their correlation model with the gap insulation state(withstand voltage or breakdown);Using Pearson correlation coefficient(PCC)and tuna swarm optimization algorithm for feature reduction,the basic principle and implementation process of insulation strength prediction is explained in detail.(2)For the sphere-plane long gap,the training and test sets are divided according to the electrode size.According to the PCC calculation results,some redundant features were eliminated,and the positive standard operation impulse discharge voltage prediction of the large-size sphere-plane long gap was successfully realized.Compared with K-nearest neighbor and support vector machine,the prediction performance of the RF model is better,and the test object can be extrapolated to the cap-ground gap and ring-ground gap.The research results verify the validity of the model when it is applied to the prediction of large-scale electrode gap insulation.(3)Aiming at the gap between transmission line-tower gaps,samples are divided according to the transmission voltage level.Training on the gap between ultra-high voltage tower gaps and successfully realized the discharge voltage prediction of UHV tower gaps.Combining PCC and TSO for two-stage feature selection,and analyzing the impact of different feature reduction methods and dimensions on the prediction results.Compared with the calculation results of the gap factor,the predicted values of the model in this paper are more consistent with the experimental values.The research results can provide guidance for the external insulation design of the gaps in power transmission and transformation projects.