Partial Discharge Diagnosis And Condition Assessment Method Of DC Cable

As one of the carriers of DC transmission,DC Cross Linked Polyethylene(XLPE)cables play a pivotal role in large-capacity and long-distance DC transmission projects.In recent years,with the vigorous implementation of DC cable transmission projects,technologies such as fault diagnosis and condition assessment of DC cables have become urgent research topics.Partial discharge(PD)can be used as an important indicator to characterize the insulation status of power equipment,but the research on PD of DC cables is still in the early stage.Condition assessment is a comprehensive assessment of the overall operating conditions and insulation status of power equipment,which has important guiding significance for operation and maintenance.However,there are relatively few studies on the condition assessment of DC cables,which cannot make full use of operation and maintenance data and are difficult to reflect the status.In response to the above problems,the following researches are mainly carried out in this paper:Based on the equivalent circuit model of PD under DC voltage,the occurrence and recurrence of PD are analyzed,and the influencing factors and related relations of the characteristic parameters such as discharge amplitude and discharge repetition rate are studied.The common fault types of DC cables are summarized and four types of typical insulation defect models are designed: corona defects,air gap defects,scratch defects and creeping defects.Based on the electrical and thermal coupling model of the DC cable,the distribution of the electric field and temperature field in the designed defect cable models is studied by simulation.The above research provides theoretical support for the subsequent experiments and algorithm detection results.A test and detection platform for PD of DC cables is built,and PD data of typical defect DC cables under different aging levels are collected.By plotting the characteristic map of the discharge amplitude,the time interval between adjacent discharge pulses,and the discharge repetition rate,the correlation between PD and defect type along with aging degree is analyzed.And the changes in characteristics and statistical parameters of PD with increasing aging level of typical defect DC cables is revealed.Aiming at the PD signals of typical defect DC cables under different aging degrees,a classification study of defect pattern recognition and aging severity pattern recognition based on deep learning algorithms is carried out.Convolutional neural network(CNN)algorithm with adaptive learning capability is used to avoid errors and redundancy that may be introduced by artificially extracting statistical features.Using the H(q,?t)feature map of PD as input,the defect pattern is identified in a mixed scene of multiple aging levels.Moreover,pattern recognition research on the aging severity pattern under typical defect models is studied.In order to make full use of operation and maintenance data,an index parameter system for condition assessment of DC cables is established.In order to make full use of expert experience and follow the inherent characteristics of the data,a combination weighting method that fuses the fuzzy analytic hierarchy process and anti-entropy method is used to determine the weight of each index.In order to take into account the ambiguity of the state level and the randomness of the data,the membership function based on the cloud model is used,combined with PD pattern recognition results for aging severity patterns,to calculate the evaluation matrix of each index for the state level.Finally,an evaluation result which reflects the reasonable state of the DC cables is obtained.Moreover,a comprehensive evaluation software for DC cables including PD diagnosis and comprehensive condition assessment functions is developed,which provides a reasonable reference for intelligent fault diagnosis and comprehensive condition assessment of DC cables.