两级切换网络化系统的模型预测控制及在微网中应用研究

For power grid fault attack and existence of the intermittent renewable energy power generation, the micro-grid suffer from dual stochastic switching problems with respect to operation mode and topology structure, which will cause the inaccuracy of micro-grid control model. The traditional controller design technologies will be with low control precision, instability, or even failure control for this kind of stochastic networked systems, which threatens the safety operation of micro-grid. Therefore, it is very challenging to study the optimal control problem for micro-grid systems with two-level switching factor. This project adopts Markov random process describe the two-level switching process within operation mode and topological structure. The maximum likelihood parameter identification technology will be used combining with adaptive principle to consider the identification of unknown switching parameters, and one two-level switching networked control system model is established. The optimal control problem considered for micro-grid is based on the model predictive controller design technology for two-level switching networked systems, which is capable for handling the inequality constraints within dynamics of micro-grid system as well. Meanwhile, the equality constraint of the system are treated by the spatial mapping method. Finally, one integrated experimental platform of micro-grid networked system is constructed, which can be applied in experimental verification and performance evaluation of the new controller design method. Then, a preliminary theoretical method and application basis are established for solving the optimization-based controller design problem of micro-grid.

在智能微网中，针对配电网故障突发与分布式多种可再生能源间歇性发电，导致微网运行方式与系统拓扑结构的双重随机切换问题，引发微网控制模型的不准确性，造成控制精度的降低或失稳，甚至失灵，危害电网的运行安全。因而对含有两级切换因子微网，如何实现在线网络化最优控制是极具挑战性研究课题。本项目采用Markov随机过程分别描述微网的运行方式与拓扑结构的两级切换过程，采用最大似然参数辨识技术融合自适应原理考虑未知切换参数在线估计问题，建立微网的两级切换网络化系统。同时采用模型预测控制技术设计两级切换网络化系统控制器，在空间映射法辅助下，可有效处理微网系统中的等式、不等式约束条件，实现微网在线优化控制。本项目将计划构建微网综合网络化系统实验平台，对控制器设计新方法进行实验验证与性能评估。为解决双重随机因子交叉影响下微网最优化控制器设计问题奠定初步的理论方法与应用基础。

本项目旨在针对电网在遭受随机时延、攻击时，网络拓扑结构随机改变，借助电网状态滤波器的设计，开展电网的控制与状态监测理论的研究。项目研究过程中，分析电网的演化机理与特征，建立随机因子影响下的电网模型。基于电网模型通常建立于频域内的特性，本项目从频域辨识方法研究出发，构建模型的极大似然频域参数辨识法；其次，因电网的状态估计器的设计关系到电网状态反馈估计器设计的有效性，针对电网多信道网络离散事件系统开展状态估计问题研究。此外针对现代智能电网中，网络攻击作为威胁电网安全的因素之一，其隐秘性、危害性更强，更能触发电网拓扑结构随机改变。研究过程中，针对电网遭受连续拒绝服务攻击时，提出了一种适用拒绝服务攻击的改进无迹卡尔曼滤波方法；最后，以风电系统为例，研究了部分负荷条件下变速风力发电机组的转子转速跟踪问题，建立了慢速和快速控制之和的局部状态反馈控制器。