Research On Control Ctrategy And Fault Diagnosis Tchnology Of Modular Multilevel Converter

As one of the most promising topologies in the field of medium and high voltage,modular multi-level converter(MMC)is getting more and more attention.Compared with the traditional multilevel converter,MMC has many advantages,such as good scalability,simple sub module structure,rich application scenarios,high power density and so on.However,there are many control objectives of MMC,and different objectives are coupled with each other,which makes the steady-state control of MMC very difficult.At the same time,due to the large number of fragile switching devices in MMC,the fault problem of sub module needs to be further studied.This paper focuses on the steady-state control and fault diagnosis of MMC.The topology,working principle and mathematical model of MMC are studied.Aiming at the key problems of MMC,such as output current,circulating current and frequency reduction,a hierarchical model predictive control strategy based on increasing level is proposed.The strategy decouples output current control and circulating current suppression by introducing intermediate variables,reduces the computational complexity of increasing level algorithm,and realizes voltage stabilization by using sequencing method,After that,the switching frequency is reduced according to the sorting result of the previous switching state correction.In order to solve the problem of high probability of open circuit fault of switching devices in modular multilevel converter,a multi-point open circuit fault diagnosis strategy based on feature reconstruction is proposed by comparing the characteristics of normal sub module and different fault sub module.Through feature reconstruction,the detection and location of multi-point,multi leg and multi type open circuit faults are realized without changing the topology and increasing the amount of calculation.On the basis of feature reconstruction,an open circuit fault diagnosis strategy based on artificial neural network is proposed.Each sub module is scanned one by one,and the fault diagnosis of the corresponding sub module is completed by fast classification of reconstructed eigenvectors.This paper studies the operation principle of MMC and establishes the relevant mathematical model,optimizes the adding level hierarchical model predictive control strategy,and puts forward the open circuit fault diagnosis strategy based on feature reconstruction,which realizes the fast diagnosis of multi-point fault without additional hardware circuit,and realizes the efficient MMC fault diagnosis.An open circuit fault diagnosis strategy based on artificial neural network is proposed,which realizes the fast diagnosis of multi-point fault without artificial design parameters,and improves the accuracy of fault diagnosis.The simulation platform is built based on Matlab / Simulink for simulation analysis,and the experimental verification of related algorithms is completed based on the three-phase MMC physical platform.