Online Evaluation Method Of Measurement Error Of Electronic Voltage Transformers Based On Large-dimensional Random Matrix

Electronic voltage transformer(EVT)is an electrical parameter measurement device used in smart substations.Its measurement accuracy directly affects the fairness and justice of electric energy trade settlement and the normal operation of measurement,control and protection devices.Evaluating the measurement error of EVTs,timely operation and maintenance of out-of-tolerance EVTs are of great significance to ensure the safety,stability and economic operation of the power system.In response to the difficulty of planned power outages in the periodic verification method,the live verification method has potential safety hazards and cannot cover all EVTs in the entire network.The data-driven evaluation method based on large-dimensional random matrix has attracted wide attention due to its simple calculation,wide modeling conditions,no decoupling,and no specific parameters.It is suitable for online and real-time evaluation of measurement error of EVTs in the entire network.However,the application of this evaluation method to the online evaluation of measurement error of EVTs has not yet solved the following key issues: random matrix convergence,evaluation index reliability,input parameter stability and self-adaptation to longterm changes.In this regard,this article expands from these four aspects,the main works of this thesis are listed as follows:(1)Aiming at the convergence of the random matrix and the reliability of the evaluation index,this paper combines the existing random matrix expansion methods and proposes a matrix expansion method suitable for a small number of EVTs to meet the requirements of random matrix convergence.Then,based on the KPCA reconstruction algorithm and the kernel density estimation method,the quantitative evaluation index describing single-ring graph and the evaluation threshold are extracted to realize the the transformation from manual evaluation to automatic algorithm evaluation.Finally,based on simulation experiment,it is verified that the evaluation index proposed in this paper has higher sensitivity and can be reliably applied to the measurement error evaluation of EVTs.(2)In order to obtain stable input parameters,this paper combines the knowledge-guiding& data-driven theory to study the general characteristics of EVTs: the internal correlation of front and near moments and same-phase correlation of output voltage.The characteristic parameters with stable distribution characteristics are extracted,and the evaluation method based on the above-mentioned characteristic parameters is discussed,thus,a widely applicable evaluation system of measurement error of EVTs is established,which can accurately identify abnormal EVT.(3)In order to realize the self-adaptation of the evaluation model to the long-term changes of the state of EVTs,this paper combines the sliding time window and proposes a real-time evaluation method of measurement error of EVTs,which can adapt to the long-term changes of the state of EVTs.Based on simulation experiments,the effectiveness of the evaluation model update mechanism is verified.(4)Designed and developed a big data platform for the electronic transformers,and carried out the pilot application.In the engineering application of the pilot substation,an abnormal EVT was found.Compared with the verification results of the traditional transformer,the effectiveness of the evaluation method is verified.The online evaluation method of measurement error of EVTs based on the largedimensional random matrix proposed in this paper can reliably find the EVT with abnormal measurement error for a long time without power failure or standard devices.It has important engineering application in ensuring the accuracy of electric energy measurement,control and protection.