| Capacitor voltage transformer(CVT)has been the most widely used voltage transformer in high-voltage power system,the operation status of CVT directly relates to the accuracy of power trade and the stability of power system.However,the internal structure of CVT is much more complicated,and its operation status is easily affected by the environmental factors,therefore,exceeding measurement error and abnormal internal insulation are prone to occur during long-term operation.According to statistics,CVT faults related to abnormal measurement error accounted for more than 50%,and faults caused by abnormal internal insulation status accounted for more than 20%,measurement errir status and internal insulation status are two keys to ensure the long-term stability of CVT.In order to evaluate the operation status of CVT,offline calibration based on standard device is adopted by the power system as the evaluation method for measurement error status and internal insulation status.However,due to the difficulty of power outages in high-voltage substations and the heavy test work,lots of CVTs cannot be calibrated on time,the long-term operation stability cannot be guaranteed.In recent years,some data-driven online evaluation methods for CVT operation status had been proposed.However,the evaluation reliability of all these methods is not high enough to meet the requirement of actual operation.Aiming at these problems above,CVTs of same voltage level in a high-voltage substation is regarded as the evaluation group,and the electrical-physical correlations between CVT groups are digged deeply in this thesis,a measurement error online evaluation method which is suitable for actual working condition is proposed.Then the correlation between the internal insulation status of CVT and its measurement error status is studied,combining the physical modeling and data modeling,the internal insulation status evaluation method based on fuzzy analysis of measurement error status is proposed.Therefore,the real-time sensing of measurement error and internal insulation status of CVT is realized.The main research works of this thesis are listed as follows:(1)The differences between the actual measurement error and the offline calibration results are revealed,and the compensation method for offline calibration results is proposed.The condition difference between offline calibration and field operation is studied,and the influence mechanism and effect of each factor is proved;The finite element simulation models and equivalent circuit model of CVTs are established,and the differences between the actual measurement error and the offline calibration results under the multi-factor coupling effect is quantitatively analyzed,in addition,the above results are verified through field experiments;according to the quantitative analysis results,a CVT offline calibration results compensation method is proposed,so that the offline calibration results could represent the actual measurement error of CVT more accurately.(2)Various electrical-physical correlations in the CVT evaluation group are revealed,a measurement error online evaluation method using the outlier detection method of unsupervised learning to monitor the measurement error parameters is proposed,which realize the accurate evaluation of the individual CVT measurement error status under the condition of complex output voltage data distribution.This method reveals both the rigid correlation based on the electrical connection relationship and flexible correlation constrainted by the operation mode of power system,then the inter-phase correlation parameters and the inter-group correlation parameters are proposed,which could represent the measurement error status of CVTs.The unsupervised learning method based on local distribution density is studied to monitor the changes of these parameters,then the measurement error status of CVT is evaluated accurately under the actual working conditions with complex output voltage distribution characteristics.According to the experimental results based on the field output voltage data of CVTs,this method has good adaptability to the variable conditions in high-voltage substations,and the evaluation reliability is significantly higher than the existing data-driven evaluation method,this method could evaluate the measurement error status of CVTs with 0.2 accuracy class.(3)The degradation mechanism of the internal insulation status of CVT is studied,and the relations between CVT’s internal insulation and measurement error are revealed.The structure of CVT’s insulation is studied,then it is found that the deterioration of internal insulation is the main reason for the abnormal measurement error status of CVT,moreover,the influence mechanism of various factors on the internal insulation of CVTs is studied,the main forms of degradation of internal insulation is pointed out;the equivalent physical model considering the internal insulation parameters is established,and both the influence of voltage dividing capacitor breakdown and abnormal dielectric loss on the measurement error status is derived.A single-phase simulative CVT is designed,and the correlation between the internal insulation and measurement error status is verified.(4)Based on the relations between CVT’s internal insulation and measurement error,an internal insulation evaluation method based on measurement error fuzzy analysis is proposed,which realizes the identification of both the abnormal type and degree.This method calculated the additional measurement error caused by typical abnormal internal insulation status by the CVT equivalent physical model,then a typical status sample set containing various typical internal insulation is established.The fuzzy clustering method is studied to match the parameters of CVT’s measurement error status with the typical internal insulation status,then the long-term online evaluation of the internal insulation status of CVT is realized.In addition,a fuzzy clustering center initialization method based on local distribution density is proposed,the data points with highest local distribution density is used as the initial clustering center,which improves the stability of fuzzy clustering significantly when evaluating the internal insulation status of CVT.A simulative CVT internal insulation degradation experimental platform is developed,which verified that this method could evaluate the internal insulation status of CVT timely and accurately.The research results of this thesis realize the real-time sensing of CVT’s measurement error status and internal insulation status,systematically solve the long-term online evaluation of two key operation statuses that affect its operating stability.The research results could provide effective CVT online operation status evaluation methods for power equipment maintenance department,which could reduce the power trade settlement disputes,and avoid CVT operation safety accidents. |
