Feature Extraction Of Magnetic Field Characteristics For Condition Assessment Of High Voltage AC Cable

Power cables with cross-linked polyethylene insulation as the main body have excellent electrical,mechanical and thermal properties,light weight,low price,and convenient construction.The application in domestic power transmission and distribution power lines is increasing rapidly,but the cables inevitably have defects such as air gaps and breakages during the manufacturing and use process.They will age due to thermal,mechanical,electrical and other factors.In severe cases,they will affect the reliable transmission of power.Therefore,it is necessary to carry out large-scale rapid testing and evaluation of cables to improve the management level of cable differential detection.This paper proposes a fast detection and condition assessment method for high-voltage AC cables based on magnetic field feature extraction.The main work is as follows:(1)The magnetic field characteristics of high-voltage AC cable were studied.Based on the finite element simulation model of 110 k V XLPE power cable body and accessories,the magnetic field characteristics of cable without defects and different types of deterioration were compared and analyzed,and the theoretical basis of cable magnetic field monitoring method was studied.The results show that when the water tree aging occurs in the cable insulation,the magnetic field around the water tree increases,and the magnetic field near the water tree area outside the cable increases.Compared with the water tree position,the amplitude of the external magnetic field of the cable is more sensitive to the size parameters of the water tree;When thermal aging occurs in the insulation of the cable body,the insulation magnetic field decreases,but the external magnetic field of the cable increases,and the magnetic field amplitude is approximately proportional to the degree of thermal aging;When the cable body is mechanically damaged,the magnetic leakage leads to the enhancement of the magnetic field near the damaged area outside the cable.When the main insulation of the cable accessory is scratched,the magnetic field in a small part of the area near the scratch outside the accessory will be slightly enhanced.When the sheath of the accessory is damaged,the magnetic field around the damaged area will surge.With the deepening of scratch and damage,the amplitude of external magnetic field of cable increases,but the influence of damage on the amplitude is more significant than scratch.(2)The on-line monitoring platform of cable condition based on magnetic field detection was built,and the influencing factors of cable magnetic field distribution characteristics and detection sensitivity under different deterioration types were analyzed.The results show that different deterioration types will distort the magnetic field distribution curve outside the cable to varying degrees;The core current is approximately linearly and positively correlated with the amplitude of the detection magnetic field.Increasing the core current can improve the detection sensitivity;The lift off value of the sensor is inversely proportional to the magnetic field amplitude,and the lift off value cannot be too large in actual measurement;The distortion degree of cable magnetic field distribution intensifies with the aggravation of aging.When the deterioration is more serious,the sensitivity of detecting cable deterioration is higher.(3)The cable state information database was established,the characterization parameters of cable state were extracted,and the cable state intelligent evaluation model based on genetic algorithm optimized support vector machine(GA-SVM)completed the accurate evaluation of cable state.47 cable magnetic field characteristic parameters in time domain,frequency domain,energy and fractal dimensions were extracted from the cable state information database,and the best feature set that can be used to evaluate the cable state,including 10 parameters:waveform index,minimum value,kurtosis index,mean value,maximum value,skewness index,DCC3 energy,frequency mean value,fractal dimension and frequency variance.Based on the principal component extraction algorithm,the fusion feature vector set was obtained from the optimal feature set as the index to evaluate the cable state.Taking the fused feature vector set of cable in different states as the input sample of the evaluation model,the evaluation model was trained to optimize the core parameters of the model.The accuracy and effectiveness of the evaluation model were verified through the test set,and the effective evaluation of cable state was realized.