Research On Robust Distributed Model Predictive Control Algorithms

The robustness of model predictive control strategies to model errors is a key factor for the successful application of these strstegies,as the controlled system may be affected by model uncertainty and external disturbance.And considering the geographical distribution of large-scale complex systems and the information exchange between subsystems in the actual industrial process,the distributed control method is more suitable for practical application.Therefore,two robust distributed model predictive control algorithms are proposed in this context.The first is a robust distributed predictive control algorithm with guaranteed H?performance,which is proposed for linear constrained systems suffering disturbance.This proposed algorithm can on-line manage the trade-off between satisfying constraints and achieving high performance.In case of unforeseen large disturbance,the performance is relaxed to meet hard constraints,and a better performance is recovered when the disturbance generally mild.The second is an improved robust distributed predictive control algorithm which is proposed for uncertain constrained systems.This algorithm expands the feasible domain of on-line optimization problem and reduces the conservation of the algorithm by using a full block S-procedure lemma,so as to obtain better control performance.Both the algorithms require decomposing the entire system into several subsystems,and design controllers for each subsystem according to the algorithm rules.A minimization optimization problem based on linear matrix inequalities is solved at each sampling time,and the control inputs are obtained and applied to subsystems.The simulation results of the two algorithms applied to the load frequency control of interconnected power system are given in this paper,and the simulation results are analyzed to demonstrate the effectiveness of the algorithms.Finally,a summary of the whole paper and the prospect of future research are given.