Research On Voltage Level Automatic Identification Method Of High Voltage Transmission Line Based On SVM

High voltage overhead transmission line is the main way of long distance power transmission,and its live work has become an important part of live working.Although live working has strict operation specifications,the accident caused by negligence of the operator has occurred.The problem of early warning against live accidents in high voltage transmission lines is always a problem for the power sector.In order to realize the safety early warning of live working of high voltage transmission line,more and more high voltage security early-warning equipment is used.But these devices only achieve security early warning through fixed electric field threshold.It does not adaptively adjust the threshold according to different electric field strength at different safety levels.The electric field intensity on the safety distance of different voltage levels is different.Therefore,the equipment in the construction of different voltage levels,either early warning distance is too far away from normal construction,or early warning distance is not early warning effect.In order to make early warning system more accurate and safe early warning in high voltage transmission line work,it is necessary to recognize the voltage level before the operation.This paper studied the voltage identification method in the process of climbing tower of high-voltage transmission line of live working.This paper designed the power frequency electric field sensor and collect the electric field intensity and the height information of the process of climbing the tower with live work.30 sets of data are collected for each voltage level for three kinds of voltage levels of 10 kV,35kV and 110 kV.The parameters of Kalman filter are studied and determined,and the collected data are filtered by Kalman filter.The feature extraction method based on PAA feature extraction is used to extract the filtered data,and the eigenvalues of each group of data are obtained.The establishment and solving process of the SVM classifier model are analyzed,and 25 groups of eigenvalues are randomly selected as the training data of the SVM classification model.Three classification models of voltage level are obtained by the algorithm of LIB-SVM based SVM classification model.The remaining eigenvalues are used to test the classification model.The results show that the simple Kalman filter data processing method can not achieve the desired recognition effect.In order to improve the accuracy of voltage level recognition,the data processing method of information fusion is studied.The data are processed by the global optimal distributed fusion Kalman filtering algorithm.The process of feature extraction and recognition is repeated before modeling,solving and testing.In the end,the distributed fusion Kalman filtering algorithm is used to carry out the distributed fusion processing of the data,and then the SVM classification algorithm is used to realize the recognition of the voltage level.We carried out a major study in order to apply the proposed voltage level recognition algorithm in the actual security early-warning system.The proposed voltage level recognition algorithm is transplanted into the embedded system based on STM32F103RCT6.In the embedded system,the acquisition of electric field signal,data fusion,feature extraction,model reading and classification recognition are realized.It provides an effective solution for intelligent warning of different voltage levels for high-voltage electric shock warning equipment.