Traveling Wave Fault Location Algorithm For Cross-bonded Cable

With the development of cities,the cable rate of high-voltage lines is increasing year by year.The cross-bonded cable belongs to the multi-conductor structure and there is a crossinterconnection between the sheaths,which causes a lot of interference to the propagation of the traveling waves;the current offline fault location method has disadvantages such as complicated operation and long power outage time.Aiming at the problems of cross-bonded cable fault location,this paper proposes an online fault location method that combines unsupervised learning with double-ended fault location.The following research work was completed:(1)The structure of cross-bonded cables and the main fault types are introduced.In addition,the generation and propagation process of traveling waves when the main insulation fault occurred are analyzed.The simulation analysis is performed on the distortion and variable wave speed characteristics of the traveling waves.Finally,according to the cable data used in the simulation,the refraction and reflection coefficient of the traveling wave at the cross interconnection point is calculated,and the reflection of the traveling wave at the cross interconnection is summarized.(2)This paper selects the expanded Clarke transformation matrix,uses the method of multiconductor analysis to solve the electromagnetic coupling existing in the cross-bonded cable,and then analyzes the propagation characteristics of each modulus and the cross-bonded structure in the direct-buried laying method in the modulus domain.(3)Through the analysis of the various modulus and traveling wave propagation characteristics of the cross-connected cables,it is proposed to use the sum of the three-phase sheath current including modulus 1 and modulus 4 as the fault location signal.In order to reduce the influence of traveling wave dispersion on fault location,the method proposes to establish a high-dimensional data matrix by monitoring the three-phase sheath current of each cross-bonded section,and use principal component analysis and density clustering algorithm to find out The data characteristics of the sheath current matrix can be used to determine the fault section;finally,the fault location method is used to determine the arrival time of the traveling wave head by using the VMD algorithm.(4)The 110 k V cross-bonded cable model is established in software PSCAD/EMTDC,and the simulation data is analyzed through MATLAB and Python to verify the influence of fault occurrence time,fault resistance,fault point location and earth resistivity on the ranging method.The simulation results show that the cross-bonded cable fault location method based on unsupervised learning proposed has a high accuracy rate,and the absolute error of the distance measurement is less than 0.4%.In addition,the method is compared with three commonly used traveling wave head calibration and recognition algorithms,and the results show that the method has higher ranging accuracy.