| The high voltage direct current(VSC-HVDC)transmission subject based on voltage source converter is getting into a phase of rapid burgeon in the power area,Today,the grid facility has been promoted in status of voltage grades,transmittal capacity,and the space between transmittal lines across area.Power cable lines have the advantages of good safety,no space on the ground,and suitable for submarine power transmission.In fact,because of the causation of weather situation,and other causes,high-voltage transmittal line system faults arise constantly,which greatly affects the safety of electric energy transmission.Therefore,breakdown location of HVDC transmittal line system has developed into the discussed point of various proficients and scholars at this domain.Precise and available ranging technique can rise the level of power conveyance and decrease the wastage,which has tremendous realistic sense for China’s energy security.With the unceasing advance of AI technology,proficient ts and scholars often utilize the technique of intelligent algorithm and neural network to handle conveyance line fault location.The ranging exactness of such technique is tremendous restricted by the analysis accuracy of the fault transient signal by the intelligent algorithm and the designed validity effect of the neural network.At the moment,the most common way is the wavelet decomposition way to extract the characteristic quantity needed in the fault transient signal of HVDC system.When the conveyance line is the high-resistance grounding of the cable,the extraction exactness of this technique is apparent descend.In view of the traditional wavelet solution precision is not enough,and now available neural network location technique in training error of fault specimens not fully take into account the actual working condition.In this paper,account of S transform and PSO rise the function of generalized neural network,line fault location technique,and applied it in DC cable grounding breakdown location.The essential substances and verdicts are as below:In order to advantage analyze the fault issue of HVDC,According to the different grounding faults of the HVDC,the contact between wave spectrum energy and fault place of the HVDC line is resolved,and the corresponding mathematical contact between the grounding fault space of the HVDC and the natural frequency of wave is discover.Due to the interference of various bending and reflection in wave disseminate to the natural frequency,the S transform is proposed to gain the energy from the breakdown semaphore,which with advantage in prevent the issue of insufficient exactness when extracting the main frequency directly.In order to get the fault traveling wave energy spectrum and further distance,the relation between the GRNN on the characterization of each frequency band energy training specimens,to prevent enter local optimal place,take advantage of PSO to perfect the smooth factor of GRNN,in order to rise the network convergence speed and exactness.And carry on the fault location precisely.Finally,a ± 200 kV dual-terminal flexible HVDC transmission system model is construct.In the experiment process,the exactness of this method is usefully confirm by altering the breakdown location,transition resistance,interference noise and other different conditions.When the actual measurement error interference is introduced,the maximum error is still less than 1.5%.Contrast with the wavelet decomposition way and RBF network,the experiment outcomes certify that the exactness and robustness of the way are promoted,and the way is apposite for breakdown location with operating circumstance errors.The paper has 36 figures,15 tables,and 68 references. |
