Distributed Cooperative Predictive Control Of Smart Power Converters In Microgrids

Energy drives the economic prosperity of modern life because it powers the technologies that make both life and livelihood safer,smarter,and simpler.Paradoxically,the catalyst that oils the wheels of global economies also creates sustainability problems that threaten the harmonious existence of humans on Earth.Among other sources of energy,the production processes for global electrical energy account for 42%of total annual CO2 emissions,leading all other energy sources.This is because today,fossil fuels account for more than 60%of electricity generation sources.Although the well-known pathway to sustainable power generation is to replace the fossil fuels in the electricity life cycle with sustainable energy sources,progress appears to be slow.In the past 30 years.the amount of fossil fuels in global electricity generation decreased by only 2%.Whereas,in the same period,the per capita CO2 emissions from electricity and heat generation increased by more than 80%.Therefore,more concerted global political and technical efforts are necessary to accelerate the emergence of an electricity sector with net-zero emissions by 2050(in accord with the Paris Climate Agreement).Microgrids have been identified as key enablers of sustainable and reliable electricity generation and distribution.They facilitate the integration of distributed renewable energy sources(DER)to the electrical power grid.and can also work in standalone mode during periods of grid uncertainties.The potential of microgrids can be fully realized by the deployment of smart power electronic converters;power converters with self-awareness.adaptability.autonomy,cooperativeness.and plug and play features.Nonetheless.the multivariable and multi-timescale requirements of microgrids make the conventional linear control of power converters cumbersome.Also,the cascaded linear control loops result in slower dynamic performance at the higher control hierarchies.Hence,the application of multivariable,nonlinear optimal control techniques like model predictive control(MPC)has become necessary.MPC is an optimal control method that engages the system model to predict the states.and input control sequence over a pre-defined horizon of prediction.MPC’s superior support for multi-objective constrained optimization.and high performance dynamic control position it advantageously over linear control methods.Nonetheless,MPC-based microgrid control presents drawbacks in two major categories.The first is the sensitivity of MPC to parameter uncertainties and sensor measurement noise in the control of voltage source converters(VSC).The second is the poor coordinated control of frequency,voltage and reactive power sharing in multiple DERs which are interfaced by power electronic converters.Therefore,the overall objective of this thesis is to contribute novel solutions to these two key challenges of MPC-based microgrid control.The first major category of challenges.parameter uncertainties and sensor measurement noise,is resolved by four new contributions.First,data-based variance sensitivity analysis and neuro-fuzzy filter parameter identification are utilized to achieve robust predictive control of VSCs in microgrids.Second,a novel parallel extended state observer(PESO)and an improved cascade extended state observer(CESO)are designed using a proposed multi-frequency extended state observer(ESO)model.CESO,in particular,provides a model-free predictive control technique with superior high frequency measurement noise suppression for power converters.Third,in order to improve the disturbance rejection during noise suppression,the novel hybrid parallel-cascade ESO(PC-ESO)and cascade-parallel ESO(CP-ESO)are introduced.Extensive theoretical analyses and experimental validation demonstrate their superior robust predictive control performance over both the conventional ESO and CESO.Fourth,structurally-adaptive ESO(SAESO)control schemes are proposed to enhance the disturbance rejection capabilities of CESO and CP-ESO.These SAESOs produce better robust control predictive control with up to 70%lower measurement noise dc current ripples(and 20%lower noise total harmonic distortion(THD)in ac current)than model-based predictive control.The performances of all these control methods are validated by analytical formulations,PLECS and MATLAB/Simulink simulation,and laboratory experiments on a grid-connected power converter and a bidirectional dc-dc boost power converter.The second major category of contributions proposes solutions to overcome the poor coordinated control of power converter-interfaced DERs.The distributed secondary consensusbased control of voltage and frequency in an AC microgrid,with parallel-connected inertiaemulating VSCs,is realized.It is shown that the proposed control scheme effectively reduces the load change-induced rate of change of frequency(ROCOF)by up to 89%and also has very fast and accurate dynamic response.MPC supports robust and rapid recovery from external disturbances.The problem of disproportionate reactive power sharing in VSC-interfaced DERs was solved by a novel virtual capacitance regulation of accurate reactive power sharing,and restoration of DER output voltages.The solution utilizes a dynamic average consensus algorithm for low-bandwidth communication among the neighbor DERs.The effectiveness of the control scheme was validated through stability analysis,offline simulations and real-time hardware-in-the-loop tests on a microgrid system.