The Research On Distributed Model Predictive Control Algorithms

With the rapid development of modern technology, modern industry processes are characterized by their large-scale, strong interactions and the presence of constraints. With its ability to systematically handle these issues, distributed model predictive control(DMPC) is a promising approach for the control of such systems. The advantages of applying DMPC to control large scale system include:(1) reducing the computational load of each subsystem; (2)improving the scalability; (3) strong fault-tolerant capability, ect. The aim of DMPC algorithms is to ensure as good performance as possible with limited communication load among subsystems. Moreover, convergence of the DMPC algorithms and the close loop stability should be ensured. In order to cope with the problmes related to the design of DMPC algorithms, the following issues are addressed in this thesis:1. The system decomposition before implementing DMPC algorithm is adressed. A genetic algorithm(GA) based system decomposition method is proposed. A two-stage approach, including the Input Clustering Decomposition (ICD) as well as the Input-Output Pairing decomposition (IOPD), to decompose a large MPC system is considered. A new decomposition index for ICD representing both the coupling of subsystems and the computational balance is defined. As for the performance index of IOPT, only coupling effect needs to be considered. GA is utilized to solve the specific decomposition problem for both ICD and IOPD.2. The issue of model predictive control design of distribution systems using a popular singular value decomposition (SVD) technique is addressed. Namely, projection to a set of conjugate structure is dealt with in this paper. The structure of the resulting predictive model is decomposed into small sets of subsystems. The optimal inputs can be separately designed at each subsystem in parallel without any interaction problems. In addition, the design of distribution model predictive control (DMPC) with constraints using the SVD framework is also presented. The unconstraint inputs are checked in parallel in the conjugate space. Without solving the QP problem of each subsystem, the suboptimal solution canbe quickly obtained by selecting the bigger singular values and discarding the small singular values inthe singular value space. The convergence condition of the proposed algorithm is also proved.3. A novel fast DMPC approach based on a distributed active set method and offline inversion of the Hessianmatrix to efficiently solve a constrained distributed quadratic program is proposed. A dual-mode optimization strategy based on the value of unconstrained optimal solution is developed to accelerate the computation of control action. The proposed method allows for the optimization to be terminated before convergence to cope with the fast sampling periods. Furthermore, a warm-start strategy based on the solution of the previous sampling instant is integrated with the approach to further improve convergence speed. The approach is highly parallelized as constraints can be checked in parallel.4. A non iterative distributed model predictive control design for cascade processes is proposed by fully exploiting the structure of cascade processes, i.e. the interation among subsystem is only the upstream units can influence the downstream units. The proposed algorithms is based on traditional iterative distributed algorithm, but the proposed algorithm can greatly save computational burden for cascade processes.5. A novel approach to progress improvement of the economic performance in model predictive control systems is developed. This approach can cope with the shortcoming of traditional LQG based economic performance assessment, i.e. without the ability to tuning control parameter. Its optimal performance is achieved by solving economic performance design (EPD) problem and optimizing the MPC performance iteratively. The ILC strategy is proposed to adjust the MPC tuning parameter based on the sensitivity analysis. The extension of the method to distributed MPC is also discussed.