Non-Parametric Model Predictive Control For DC Microgrid Clusters

Microgrid is a small-scale power supply and distribution system built by distributed energy,energy storage system,local load and monitoring and protection devices.It is a form of transition from the traditional power grid to the smart grid.The DC microgrid has a friendly interface with the distributed resource and energy storage system,and at the same time,the control is simple,so it is superior to the AC microgrid in terms of transmission efficiency,power quality and cost reduction,and is an important way to realize the sustainable development of green energy.Meanwhile,multiple DCMGs can be networked as clusters to further take advantage of DC distributed power systems.To manage complex DCMGs cluster systems,the hierarchical control paradigm that consists of three levels,i.e.,primary,secondary,and tertiary,was previously proposed with decoupled design of the controllers at different levels.The primary and secondary controllers are applied to achieve voltage regulation and current sharing within every single DCMG,whereas the tertiary controller focuses on power flow among multiple DCMGs in the cluster.Traditional parameterized tertiary controller design relies excessively on a priori models of the controlled system(e.g.,transfer functions,state space expressions,etc.).Therefore,it is a serious challenge to build dynamic models for complex high-order systems.To this end,this paper focuses on the non-parametric model predictive control strategy and accomplishes the following three major tasks.Firstly,the overall architecture of the DC microgrid cluster is analyzed,namely the communication layer and the physical layer,and the coordination of both is a prerequisite for the stable operation of the complex cluster system.Further,the hierarchical control topology adopted in this paper is decoupled at different layers,and an equivalent model is constructed with two DC/DC converters connected in parallel to form a network to establish a mathematical model for the proposed hierarchical control strategy,which can verify the effectiveness of hierarchical control.Secondly,to reduce the modeling difficulty and optimize the control system,while ensuring the reliable and stable operation of the DC microgrid cluster,a non-parametric model predictive tertiary control strategy(NPMPTC)based on the dynamic matrix control(DMC)is introduced in this paper.The proposed strategy only utilizes the step response data obtained at the start-up instant of the DC microgrid converter output current to construct a non-parametric model and use it to complete the design of the model predictive controller,which effectively reduces the complex modeling work of the tertiary controller.By constructing a simulation model and completing the testing of relevant cases,the proposed control strategy is demonstrated to be competent in handling various transient events.Further,the proposed control algorithm is improved by using the time-optimal strategy and verified by simulation results.Finally,a hardware-in-the-loop experimental platform and an experimental platform are built to verify the proposed NPMPTC strategy.The first one verified the effectiveness of the proposed strategy by adopting the form of ‘simulation of main power circuit,hardware transmission of control loop’,which laid the foundation for the experimental platform.The latter builds an experimental platform by using programmable power supply,interface converter,resistance load,DSP28335 digital controller and other related instruments,and completes the test of related cases,which provides a realistic basis for the application of the proposed control strategy in actual working conditions.The relevant experimental test cases show that the proposed NPMPTC strategy inherits the advantages of model predictive control to complete online optimization on the one hand,and can significantly reduce the modeling workload of complex systems such as DC microgrid clusters on the other hand.