Extraction Of Transient Dielectric Characteristics Parameters Of Nonlinear Insulation Based On Time Domain Spectra Of Excitation Voltage And Response Current

Nonlinear insulating dielectric is a dielectric whose dielectric property parameters such as conductivity and/or polarizability vary with the externally applied electric field and temperature,and its function of improving the uniformity of spatial electric field distribution in the insulating structure makes it widely used in various insulating systems.In this thesis,we propose an improved particle swarm intelligence algorithm based on the excitation voltage and response current time domain spectral information in Matlab software to complete the extraction of transient dielectric property parameters of nonlinear insulating media in the context of the transient dielectric property characterization.First,an equivalent three-branch circuit model for a nonlinear medium under transient excitation is developed,in which the nonlinearity of the voltage to time derivative is taken into account in the form of the parameter function of the relaxation branch.The simulation software Matlab is used to obtain the excitation voltage and response current time domain spectrum data covering the parameter information.The objective functions of the parameters,excitation voltage and response current are established,and the parameter extraction problem is transformed into a minimum value optimization problem.The feasibility of the particle swarm algorithm for the extraction of equivalent dielectric parameters of nonlinear media is verified by simulation,and the shortcomings of the standard particle swarm algorithm are found.Secondly,to address the shortcomings of the standard particle swarm algorithm,two improvement strategies,particle swarm own improvement and hybrid intelligent algorithm,are proposed.Firstly,the inertia weights of the particle swarm algorithm itself are focused on the improvement,and the effects of constant inertia weights,linear decaying inertia weights and nonlinear decaying inertia weights on the extraction parameters of the particle swarm algorithm are studied,and the inertia weight strategy with the best effect is determined from them.Then,based on the analysis of the advantages and disadvantages of the gray wolf algorithm and the genetic algorithm,the hybrid gray wolf-particle swarm algorithm,the hybrid genetic-particle swarm algorithm and the hybrid gray wolf genetic-particle swarm algorithm are designed.After comparative analysis,it is concluded that the hybrid particle swarm algorithm is better than the standard particle swarm algorithm,so as to determine the most suitable hybrid particle swarm algorithm for nonlinear medium parameter extraction.Finally,the effects of different excitation voltage waveforms on the extraction of nonlinear insulation dielectric parameters are studied,and the idea of adjusting the excitation voltage waveform with the principle of best matching the excitation voltage waveform with the nonlinear insulation properties is proposed.Specifically,according to different branch current to total current ratios,the excitation voltage amplitude and frequency are adjusted to meet the accuracy requirements of specific branch parameter extraction to achieve the goal of improving parameter extraction.The idea is applied to the parameter extraction of simulated and measured data to verify the feasibility of the idea.For the parameter extraction of DC conduction branch,a combined excitation waveform with a constant voltage section is designed to optimize the extraction of DC conduction parameters.