基于Tube的模型预测控制及其在风力发电系统中的应用

模型预测控制是工业控制领域中应用最广泛的先进控制策略之一。由于实际工业过程普遍存在着参数不确定及扰动不确定，鲁棒模型预测控制近年来逐渐成为研究的热点。基于Tube的模型预测控制，通过设计一个参数化的Tube不变集序列，使得持续有界扰动下的系统状态始终保持在以标称系统状态为中心的Tube不变集序列内，从而能够有效地抑制有界的不确定性。结合概率论、数理统计和随机过程等数学概念和工具，通过构造适合随机系统分布特性的概率约束，构建基于Tube的稳定高效的非线性随机预测控制策略。通过基于深度学习的风电场风速特征提取、随机过程参数辨识及风力发电系统的精确建模，设计符合风场风速随机分布特性的概率约束，构造基于Tube的风力发电系统模型预测控制，使得实际风力发电系统状态保持在Tube不变集内，满足系统物理约束，提高风速扰动下风电机组并网能力，减小对电网的冲击，保证风电机组运行的稳定性和经济性。

Model predictive control (MPC) is one of the most widely used advanced control strategies in industrial process control. In recent years, robust model predictive control has become a research hotspot due to the existing parameter uncertainty and disturbance uncertainty in industrial processes. Under continuous bounded perturbation, the Tube MPC can always keep the system state within the parameterized tube invariant set sequence centered on the nominal system state, which can effectively suppress bounded uncertainty. Incorporating mathematical concepts and tools, i.e., probability theory, mathematical statistics and stochastic process, a stable nonlinear stochastic predictive control strategy based on tube is provided by constructing the probability constraint suitable for the distribution characteristics of stochastic system. With the deep-learning-based feature extraction, parameter identification, and accurate modeling of wind power generation system, the probability constraint is designed in accordance with stochastic distribution characteristics of wind speed. Meanwhile, the Tube MPC is constructed, which makes the actual state to be within the tube invariant set and meet the system physical constraints. Consequently, the Tube MPC overcomes the wind speed disturbance to guarantee the stability and economy of the wind turbine operation. Furthermore, it improves the grid-connection ability with decreasing the impact of the stochastic wind power on power grid.