Research On Capacitor Voltage Balancing And Failure Diagnosis Methods Of MMC Based On Fundamental Frequency Sorting Strategy With Nearest Level Modulation

Modular multilevel converter(MMC)effectively avoids the series-connection of power switching devices and it processes the advantage of low harmonic,redundancy and easy expansion.With the increase of the direct current(DC)bus voltage in the flexible DC transmission project,MMC gradually replaces the traditional two-level and three-level converters and becomes the most popular topology in the flexible DC transmission converters.However,compared with the two-level and three-level converters,MMC has more IGBTs and floating capacitors.The balance of sub-module capacitor voltages is a major challenge for MMC.At the same time,the failure rate of converters caused by the damage of capacitors and power switching devices is more than 60%.Thus,it is of great significance to study the capactor voltage balance method and the on-line monitoring strategy for the health state of submodules' capacitor and power switches.In this paper,the operating characteristics of MMC under nearest level modulation(NLM)is analyzed in detail,and the charging characteristics of the fundamental-frequency-switched NLM driving signals is discussed.Based on thes analysis,a balancing strategy with fundamental frequency sorting strategy,an on-line capacitance estimation method and an open-circuirt failure diagnosis strategy are proposed.The specific research contents are as follows.Firstly,a capacitor voltage balancing method based on the fundamental frequency sorting strategy is proposed,which effectively reduces the sorting frequency and avoids the extra switching actions.From the perspective of the net charge transfer of the NLM driving signals,the charging and discharing patterns of the sub-module capacitors in one fundamental frequency period by the driving signals are revealed.Then,the capacitor voltage balancing method based on the fundamental frequency sorting strategy is proposed,which sorts the submodule capacitor voltages only one time in one fundamental frequency cycle and assigns the driving signals to each submodule according to the charging ability and the sorting results.This method greatly reduces the frequency of sorting process and avoids the extra switching actions of power switching devices,so it not only reduces the computational resource consumption of controller,but also further reduces the switching loss of MMC.Secondly,an online capacitance estiomation method is proposed to monitoring the capacitor health state.A mathematical model of the switching angle of the driving signal,the variation of the capacitor voltage and the capacitance is established.The method using submodule capacitor voltage variations in one fundamental frequency cycle as the input.Furthermore,an online capacitance estimation method based on recursive least squares(RLS)is presented,which avoids the influence of MMC arm circulating current and the low frequency capacitor voltage osscilations on the estimation results.Moreover,neither the injection of low frequency arm current,nor the measurement of capacitor current is required,which effectively reduces the complexity of the algorithm.Thirdly,a capacitance consistency evaluation method for the sub-modules and a modified fundamental frequency sorting strategy based balancing method,which considers the consistency of the capacitance value,are proposed.Aiming at the problem of different aging speeds of the sub-module capacitors,the effect of the capacitance inconsistency on the convergence of the capacitor voltage is deeply analyzed.Based on the real-time estimation results of the capacitance of sub-modules,a method based on quantile analysis for the consistency evaluation and detection of abnormal capacitance is proposed.Then a modified fundamental frequency sorting balancing method,which takes into account the inconsistencies of the capacitance,is furtherly proposed.This method aims at minimizing the overall fluctuation of the capacitor voltages in one bridge arm.By selecting the best driving signal distribution method,it ensures that the MMC can maintain a good voltage balancing performance when the capacitance of some submodules are greatly different.Finally,an open-circuit failure diagnosis method based on the observation of capacitor voltage variation is proposed,which achieves fast open-circuit failure detection and location.Aiming at the open-circuit failure of power switching devices,this paper analyzes the behavior characteristics of capacitor voltage of MMC submodules in the case of power switching devices open-circuit failure.Combined with the technical advantages of the fundamental frequency sorting strategy based balancing method,and taking the charging capability of the driving signals as a reference,an open-circuit failure diagnosis method based on the observation of capacitor voltage variation is proposed.This method can detect and locate the open-circuit failure of submodules with no extra hardwares and no informantrion of the arm current.Moreover,this method can realize failure detection and location of submodule simultaneously,which is simple and efficient.Based on the detailed analysis of charging characteristics of the NLM driving signals,this paper takes MMC as the research object,and puts forward a group of control methods for the MMC.The proposd methods are based on the net charge transfer of the driving signals and the variation of sub-module capacitor voltage.The online estimation of the sub-module capacitance,the diagnosis of sub-module open-circuit failure,and the balance control of capacitor voltages with consistent or inconsistent capacitance are realized.The reliability of MMC is effectively improved.The effectiveness of the proposed control methods is verified by the Matlab/Simulink platform and a 9-level three-phase experimental prototype.