Research On Inter-turn Short Circuit Fault Detection Method Of Dry-type Air-core Reactor Based On IFRA And PSO-SVM

As a pivotal equipment that realizes reactive power compensation and stabilizes power quality in power systems,dry-type air-core reactor(DAR)play a vital role in power systems.However,the internal winding fault of DAR,especially the inter-turn short circuit fault,is one of the main reasons that cause the DAR to catch fire.Therefore,timely diagnosis and evaluation of the internal winding status of the DAR to prevent the occurrence of burning incidents are key technical issues that need to be resolved urgently.Traditional inter-turn short circuit fault detection methods generally have the problem of low sensitivity.Therefore,this paper proposes a DAR inter-turn short circuit fault detection method based on impulse frequency response analysis(IFRA)and particle swarm optimization support vector machine(PSO-SVM).Firstly,the equivalent circuit models of DAR in normal operation and inter-turn short circuit were analyzed,and the voltage-current matrix equation was obtained.Considering the selection conditions of the key pulse parameters in IFRA from many aspects,the pulse signal amplitude is 1k V,the frequency is 1Hz,the pulse width is500 ns,and the front and back edges are kept at 10～50ns.Secondly,based on the proposed pulse parameter requirements,a pulse excitation device was developed.The development process of the device was explained from four aspects: theory analysis,simulation verification,physical design and device testing.Considering the variability of subsequent experimental applications,the output parameters have a certain margin.The output performance of the device is: the amplitude of the pulse excitation signal is 0～4k V,the pulse width is 100ns～500ns,the frequency is 1Hz～1k Hz,and the front and back edges are kept at about 20 ns.Thirdly,the DAR inter-turn short circuit experiment platform was built to simulate the internal winding faults of different states,different degrees and different positions of the inter-turn short circuit,and characteristics of the frequency response curve under different faults were studied.The results show that,the breakdown of the inter-turn insulation and the inter-turn short circuit will cause the resonance point in the frequency response curve to move to high frequency.As the severity of the inter-turn short circuit increases and the location of the inter-turn short circuit approaches the middle,the frequency response curve moves to high frequency.Additionally,compared with the inter-turn short circuit of the same strand,the effect of different strands is greater.Finally,correlation coefficient(CC),euclidean distance(ED),absolute sum of the logarithmic error(ASLE)and sum of squares of errors(SSE)were used to quantify the frequency response curve.Combined with the diagnosis and analysis of the measured frequency response curve by PSO-SVM,the diagnosis results are maintained above98%.After that,the uncertainty sensitivity analysis of PSO-SVM was carried out.The results show that as the number of feature parameters increases,the diagnostic accuracy rate will also increase.The lack of non-support vectors does not change the diagnosis results,while the lack of support vectors will greatly reduce the diagnostic accuracy.In summary,the new method proposed in this paper can effectively detect inter-turn short circuit fault,has great application potential and research value.