Classification And Evaluation Of Three Phase Voltage Sag Based On Phase Space Reconstruction And Comprehensive Weighting

Excellent power quality is the premise of maintaining the security and stability of power market.On the one hand,the rapid development of economy and the rapid change of science and technology make a large number of precision instruments become the core of the normal production and operation of the power system.This batch of instruments is very dependent on the high-quality power environment,especially the immunity to voltage sag disturbance is very low.On the other hand,the tense international energy situation,climate change and environmental pollution and so on all urge people to turn their attention to clean energy.Therefore,the proportion of renewable energy,mainly solar and wind energy,in the power supply increases.However,the voltage sag disturbance seriously hinders the process of clean energy grid connection.As the power quality problem with the highest frequency and the most serious harmfulness,the management of voltage sag has become the consensus of the industry.Since voltage sag classification and evaluation are the key to reasonable,effective,comprehensive and scientific management of voltage sag disturbance,this thesis focuses on related technologies of voltage sag classification and evaluation,and the main research contents are as follows:(1)An improved voltage sag classification method based on phase space reconstruction was proposed.Firstly,the delay factor was obtained by using Takens theorem,and the embedding dimension was obtained by comparing the G-P algorithm with the Cao method.Thus,the phase space reflecting the original attractor topological structure was constructed.Then the undisturbed three phase signals were transformed into concentric circles by phase space transformation and space rotation,therefore the transformation matrix was obtained.Hence the classification results of voltage sag were obtained according to three phase phasor diagram.On this basis,to solve the problem that the manual extraction process of voltage sag features was complicated and it was difficult to realize automatic classification of voltage sag,the obtained voltage sag phase space reconstructed image set was combined with Goog Le Net for classification purpose.Firstly,the Goog Le Net network was fine-tuned.Then the main network was selected for feature training and the Ada Boost-DT classifier was used to further improve the accuracy of network classification.The experimental result showed that this classification method was feasible.(2)In view of the traditional voltage sag assessment did not deeply analyze the severity of three phase voltage sag,and there were problems such as difficulty in qualitative and quantitative evaluation and unreasonable quantification in the evaluation process,an improved three phase voltage sag assessment method based on comprehensive weighting was proposed.Firstly,the index evaluation system was established and graded.Then,the subjective weight Pareto optimal solution was obtained by linear programming.Then,the improved entropy weight was combined with correlation coefficient to form CRITIC method,so as to obtain objective weight.Then,the principle of minimum identification information was utilized to construct comprehensive weighting objective function and in view of artificial approximation and calculation when comprehensively weighting the voltage sag was not easy and traditional particle swarm optimization method was difficult to jump out of local optimal problem,the Levy flight mode was introduced to get the improved parameters so that the objective function solution was obtained.Then,the final evaluation result was obtained by referring to the health index in the form of effective value.Finally,a voltage sag example of a substation was taken as the evaluation object to verify the rationality of the evaluation method.