Research On Key Issues Of Model Predictive Direct Power Control In Three Level Grid-connected Inverters

Recently,the excessive growth of renewable energy has led to overcapacity,and distributed generation systems is regard as a good solution.As a bridge connecting renewable energy generation equipment and the grid,the performance optimization of grid-connected inverter plays a key role in improving the static performance and dynamic response of distributed generation system.Due to its easy implementation,fast dynamic response and multi-objective optimization,model predictive control(MPC)has attracted extensive attention among large numbers of control algorithms for grid-connected inverters.However,there are some problems in MPC,such as huge computational amount,variable switching frequency and so on.The huge computational amount limits its application in multi-level inverter.The variable switching frequency spreads the harmonics spectrum of the output voltage in the grid-connected inverter,which adds the difficulty in designing the AC filter.In this paper,several control algorithms,which belong to model predictive direct power control(MPDPC),are proposed to address the problem of huge computational amount and variable switching frequency in conventional MPC.T-type three-level grid-connected inverter is taken as an example to illustrate the superiority of these proposed algorithms.Besides,a model predicts flexible power control is developed to enable grid-connected inverters run stably in the case of grid failure.Specifically,the two step model predictive direct power control(TSMPDPC)is proposed to improve the optimization process of the conventional MPDPC.In this algorithm,the number of cycle computation is reduced from 25 to 12,and a new delay compensating approach is proposed to improve the static-state performance of grid-connected inverter.Although the TSMPDPC can reduce the computational amount,its switching frequency is still variable.Based on the TSMPDPC,the model precdictive power direct power control with dual vector(MPDPCDV)is provided to make the harominc frequency of output voltage in grid-connected inverter close to the sampling frquency.The switching sequence of this algorithm,including two vectors,is utilized to regulate the active and reactive power and balance the neutral point(NP)voltage.Comparing with the MPDPC,the harominc frequency of output voltage of grid-connected converter with adoption of MPDPCDV is close to the switching frequency.The ripples in the active and reactive power and grid current THD are decreased dramatically.However,comparing with the conventional MPDPC,the MPDPCDV can not decrease the computational amount.Absorbing the merits of the TSPMDPC and MPDPCDV,the four step model predictive direct power control(FSMPDPC)is proposed.In this algorithm,the optimal switching sequence including three vectors is used to replace the switching state and fix the switching frequency,meanwhile,a new cost function and four step predictive calalculation are provided to decrease the computational amount.Comparing with the conventional MPDPC,the FSMPDPC not only reduces the computational amount,but also addresses the problem of variable switching frequency.However,the switching frequency of FSMPDPC is higher than that of the conventional MPDPC,thus the adverse effect of dead-time on static performance(such as grid current THD)in grid-connected inverter is increased.An integrated approach with compensation process and elimination process is thus proposed to eliminate the adverse effect of dead-time.The elimination process with improving gate signals is applied when the phase current is large enough.In contrast,the compensation process is utilized when the phase current is relatively small and the elimination process is troubled by the misjudgement of the current polarity.When running in normal condtions,the inverter with above algirhtms has outstanding state-state performce and fast dynamic reponse.However,when the grid voltage is dropped or unbalanced,the grid-current's harmonics increases rapidly and its amplitude bumps up.Thus,a flexible power regulation scheme(FPRS)is proposed to reduce the grid-current hamonics and provide overcurrent protection.When the reactive power reference is equal to zero,the active power reference is divided into two parts,which are controlled to achieve the active power oscillation cancellation(APOC)and the reactive power oscillation cancellation(RPOC)respectively.Meanwhile,the relationship between the grid current peak value and the power ratio ? is derived.And the power ratio ? is regulated to prevent overcurrent damage to the power switches,reduce the grid current THD,and minimize the oscillation of active or reactive power.Finally,all the algorithms proposed in this paper are validated by experiments,and the results meet the expected goals.These algorithms have positive significance for the application and promotion of model prediction direct power control.