| Most typical thermal processes have several manipulated variables and controlled variables, which are strongly coupled, dead-time and non-linear. Adopting traditional system control scheme, decentralized control structure composed by multiple SISO feedback control loops and based on PID linear control algorithm, is still widely used at present. However, the control structure cannot ensure long-time normal operation, so that the system operation parameters will deviate far from the economic indicator, and it will lead a succession of problems of causing vast energy loss, increasing equipment failure rate and seriously affecting economic benefits and safe and reliable operation for the system. Therefore it is of very important theoretical significance and practical value to study Smith predictor advanced control methods suitable for multivariable thermal processes, to effectively compensate time delay for multivariable processes and take it into automatic control and optimal operation. This paper studies the application research of the multivariable Smith predictor advanced control methods for the thermal processes based on the review of application research situation of the advanced control methods for thermal processes.Firstly, different design methods for multivariable dead-time compensator are studied based on the introduction of the Smith predictor algorithm principle and dead-time multivariable systems, and the design process for multivariable Smith predictor control system is summarized.Secondly, decoupling control methods for multivariable system are studed, aiming at the storage pulverized-coal system of steel ball coal mill, one of typical thermal processes. Based on the chief introduction of production process of for the ball mill coal pulverizing system, according to the typical model with strongly coupled and dead-time characteristices and the dynamic characteristics of each loop, this chapter proposes a decoupling control scheme for the ball mill coal pulverizing system after combining feedforward compensation decoupling with partial decoupling, and then carries out the system design and the simulation. From the simulation results, we can see that the partial feedforward compensation decoupling control can approximately divide the 3×3 process of the ball mill coal pulverizing system into a SISO process and a TITO process.Thirdly, a feasible Smith predictor control scheme based on the multivariable process of the ball mill coal pulverizing system is proposed after partial feedforward compensation decoupling. This chapter adopts two compensation methods to design the Smith predictor control system for the decoupled process by choosing appropriate turning methods for the PID controllers and a two degree of freedom PID controller scheme for Channer 2 and Channel 3 still with coupling effects, and carries out system simulation. The results show that the system output well follows the set point, and the disturbance response gets well restrained, and so the system can get a satisfied control effect.Finally, a modified design method of multivariable Smith predictor is proposed for TITO System with multiple time delays. A dead-time model for the TITO system is obtained first by robust decentralized model parameters identification method based on closed-loop step response. The computation or auto-tuning of the decoupler parameters, the decoupling equivalent object model parameters and the primary controller parameters are next accomplished automatically by integrating fuzzy self-tuning with auto-tuning technique of controller parameters. The simulation test is carried out on another typical thermal system, the distillation column system, and the results illustrate the effectiveness of the method, and significant performance improvement over the conventional multivariable Smith predictor control scheme has been achieved with the proposed approach.The last chapter summarizes all the works which have been done in the paper, and looks forward further works and research questions. |
PDF Full Text Request