Research On Condition Evaluation And Prediction Methods For The Metering Error Of Capacitor Voltage Transformer

As one of the key devices for signal acquisition in power system,capacitor voltage transformers are widely used in the grid with a voltage of 110 kV and above due to its superiority in insulating property and economy.The higher the voltage is,the greater the advantage is.Compared with traditional potential transformers,the structure of capacitor voltage transformers is more complex,and its metering error condition is more easily affected by many factors during its operation.As a measuring device,the long-term stability of metering error is regarded as one of the key parameters to evaluate the performance of a capacitor voltage transformer.The traditional method is to conduct an error test with a standard at a fixed interval when the power transmission is cut off,which is not easy to be achieved.As a result,a large number of capacitor voltage transformers have been serving for the power system without being verified at the established interval and have risks of exceeding the error limit,effecting the fairness of power trading.Thus,the traditional method cannot meet smart substations' requirement of on-line monitoring on the key equipment and how to evaluate and predict the error condition of capacitor voltage transformers during their operation is becoming a vital problem in terms of power system's safety,stability and economy,which will be studied in this paper:(1)The key issues of evaluating the metering error condition without the power off is to choose of a new standard quantity.A new idea has been proposed in this paper: excavating the parameters whose variation regulation can be described during the operation,and take it as a standard quantity of evaluation.According to the output of capacitor voltage transformer,the method of mathematical statistics is used to determine whether these parameters are invariable to the known rules.Thus,the evaluation of metering error for capacitor voltage transformer is realized.According to this idea,an evaluation model for metering error condition based on the cyber-physical correlation analysis is proposed.Based on the short-time invariant characteristics of the three-phase unbalance,the standard quantity is set up.And at that time,the secondary output of the three-phase capacitor voltage transformers is linearly correlated.Then the secondary output is analyzed by the method of principal component analysis to set up the Q statistics.The condition evaluation of metering error is mapped to the condition evaluation of Q statistics under the cyber-physical correlation constraint.An experiment is designed to simulate the different bulk of metering error.The experiment results show that the amplitude error can be evaluated by proposed model is better than 0.1%,and the phase error can be evaluated is better than 4.(2)In the view of time-varying characteristics of the power system,the three phase unbalance is not changed in a small fixed range in a long time.The metering error of the capacitor voltage transformer will be misjudged in the long term evaluation by the traditional principal component analysis.Therefore,an improved condition evaluation method based on variable N-step moving window principal component analysis is proposed.The condition evaluation model is updated and selected adaptively based on the real-time change of the secondary output of a capacitor voltage transformer.Collecting the running data of the substation for a long time and set it as the voltage source.The rate of misjudgment is obviously reduced by the evaluation experiment in normal condition.The frequency experiment and temperature experiment is designed to simulate the mutation error and gradient error of the capacitive voltage transformer.The experiment results show that the method has good detection ability for the mutation error and the gradient error.(3)In the view of condition prediction,the Q statistical and its statistical control threshold are selected as the prediction object on the basis of the condition evaluation.Then the condition prediction of capacitor voltage transformer is mapped to the condition prediction of the cyber-physical correlation of the power grid.Then the influence of the primary voltage changed on the prediction results is eliminated.According to the time series analysis method,the prediction model of metering error condition is set up as the Q-ARMA.According to the historical evaluation data,the error condition prediction is realized at the next sampling time or some time in the future.The experimental results show that the error between the prediction value and the actual value of the metering error is about 5% under the single-step prediction.(4)A high-precision device for signal acquisition under synchronous is developed for the analog-digital conversion of the secondary signals of a capacitive voltage transformer.After the calibration test at the national high voltage metering station,its accuracy can reach up to 0.05 class.Based on the development of the high-precision device for signal acquisition under synchronous,an evaluation and prediction system for metering error is established to monitoring the condition of 12 capacitor voltage transformers in a 220 kV substation.The field application shows that the method proposed can identify the abnormal condition of the metering error.This paper focuses on the method of evaluating and predicting the error condition of capacitor voltage transformers during their operation,and the proposed technology route and method can also be used for evaluating the condition of other sensors in power system,which is has great theoretical and engineering value.