A Simulation Research On Multi-variable Model Predictive Control Algorithm

Among varieties of control systems, there may be many approaches to sort them out. According to the features of model predictive control, controlled systems can be divided into fast varying ones and slowly varying ones. For the slowly and fast varying system, a typical chemical industry object-- resid fluid catalytic cracking unit fractionator, and a typical motor object—unduction motor are respectively selected to be the subjects in this thesis.China’s major crude oil is the heavy one with high content of hydrogen which contains much low-sulfur, low-metal and paraffin base. Being the equipment to lighten the heavy oil, resid fluid catalytic cracking unit fractionator has become the core equipment and the pillar of the economy of crude oil deep processing in petrochemical enterprises. The control of heavy oil catalytic cracking processes is very difficult for it complicated dynamic mechanism, multiple variables, long time-delay in every control loop and strong coupling. However, the heavy oil catalytic cracking fractionators cost so much energy. What is worse, the majority of the cost are not for the fractionation process, which causes huge pressure and potential of energy saving.As a typical representative of the AC drive systems, induction motors have been widely used with the prosperous development of power electronic technology. Because of the simple, solid and durable structure, low cost in production and maintenance, induction motors has been widely promoted. It is the bottleneck of further development that induction motors maybe low efficient at light load so that they cannot deal with a wide range of load changes. Induction motor efficiency optimization has attracted strong attention of experts and scholars around the world due to its great ownership and evident advantages.In this thesis, a novel multi-variable dynamic matrix control algorithm, with the idea of error adjustment, is proposed based on the state space description, and its stability and robustness are proved according to Lyapunov second stability theorem. This algorithm combines PID and DMC strategies, making the controller have such advantages as simple structure, clear physical meaning of PID and advantages as output prediction and rolling optimization of DMC. Heavy oil catalytic cracking fractionator simulations prove that multi-variable PIDDMC can obtain good control performance under both model matching and model mismatching circumstances, and have the advantages of fast response, small overshooting and steady-state errors and good decoupling performance.A novel multi-variable PEPFC controller, which is based on the 3rd generation model predictive control method-- Predictive Function Control, and combines PI, is given in this thesis for induction motors. The controller has structured inputs and the advantages of PI. In the maximum torque per ampere of induction motor control system, such good control performance has been obtained via multi-variable PIPFC controller as fast response of motor speed, quick tracking of stator current and good decoupling of flux and speed at light load.It can be seen from the simulation of typical slowly-varying and fast-varying system, multi-variable PIDDMC and multi-variable PIPFC both have good control performance, which demonstrate the feasibility and superiority of multi-variable model predictive control.