逆变器约束优化及其高速在线算法的研究与应用

For the traditional inverter controllers, there have been some problems, i.e., the output current can not be optimized, system efficiency can not be controlled and constraints for system variables are not effectively coped. In this project, constrained model predictive control is used in inverters. It's aimed to improve the system performance by making use of the advantages of the controller such as rolling optimization, being able to deal with constraints and agility in the cost function design. Firstly the relationship between current tracking/system efficiency and switching in the inverter is analyzed. Then corresponding controller is proposed so that the concerned indexes of inverters can be expressed and system stability is studied consequently. Also robust stability and the robust controller are investigated for the closed inverter system with uncertainty. To solve the conflict between the huge computation burden of constrained optimization and limited hardware capability of inverters, high-speed online algorithm are proposed based on the viewpoint of fast gradient methods and so on. By the method of decoupling, a self-adaptive algorithm is given to tune the parameters of the controller so that system performance can be further improved. Also the efficiency of the online algorithm proposed above is improved based on the layering technology. Through simulations and experiments, feasibility and effectiveness of the research can be verified. The project achievement is of great practical value. It also can promote application and the on-line optimization algorithms development for constrained model predictive control.

现有逆变器存在电流和效率无法得到优化控制，以及不能有效处理变量约束的问题。本项目将约束预测控制算法引入到逆变控制中，利用其滚动优化、约束处理、目标函数设计灵活等特点，来提升系统性能。首先，对电流跟踪/效率等指标与逆变开关之间的关系进行分析，设计相应的约束预测控制器，将逆变器所关心的指标在控制器中得到表达，同时分析控制器的稳定性，以及系统存在不确定情况下的鲁棒稳定性，并提出逆变器的鲁棒控制器；为了解决约束优化问题在线求解的巨大计算量和逆变硬件有限的处理能力之间的矛盾，从高速梯度等角度，提出控制器的高速在线求解计算。进一步，分别从解耦或分层的角度，研究基于参数自适应调整的预测控制器的改进办法，以及在线算法的效率提升。通过仿真和实验，验证本项目成果的可行性和有效性。 本项目研究成果既具有重要的应用价值，对于约束预测控制在线优化计算方法的研究也具有促进意义。

本项目是将逆变对关键性能的期望通过约束预测控制器进行表达和控制，针对逆变平台微秒级的采样和控制周期，设计相应的高速约束预测控制在线算法，并进行相关分析。对控制量为若干离散向量的LCL型三相逆变，本项目研究了其开关切换同电流跟踪以及系统损耗之间的关系，设计了相应的约束预测控制器，并针对该复杂非线性优化问题，从分值定界的思路设计了高效的在线算法并进行分析。针对单相并网型逆变，我们从变系数的角度设计了相应的预测控制器，并给出相应的变系数设计办法。针对微电网群系统，集中式预测控制器在线计算量巨大的问题，我们从分布式的角度进行了研究，提出了相应的算法。我们所提出的各类控制器及其在线算法，都通过仿真以及物理实验平台进行了验证。通过本项目的研究，我们建立了比较完整的逆变器高速预测控制的设计思路和框架，并为理论成果的实际应用提供了算法基础。