Research Of Boiler Drum Level Basde On Model-Free Adaptive Control

Boiler is the equipment widely used in industrial production and daily life. As the boiler often runs in the high temperature and pressure production environment,it is very important to keep the drum water level steady in normal range.The process flow of boiler is described in this paper, and the control requirements of drum water level are determined. The model-free adaptive control based on partial format linearization is introduced, while the algorithm need to the I/O data of the controlled system only, doesn’t need to mathematical model of the control system. In addition, the improved algorithm is deduced. As to prove the availability of algorithm,and the large time-delay of the drum water level in actual industrial production, the linear objects with large time-delay is selected to be simulated in this paper, and the effects are analyzed among PID algorithm, basic model-free adaptive control algorithm and improved model-free adaptive control algorithm.In order to further close to the practical industrial application, in this paper, SMPT-100 (Advanced Multifunction Process Control Training System) is elected as an experimental platform, and the process of drum water level was simulated based on SIMATIC PCS7. Communications between field device and field control station by hardware configuration and software configuration are achieved in this paper, and control algorithm is configured by CFC/SCL programming language. In addition, traditional three impulses drum level control scheme is improved in this paper, and a MFAC-PID three impulses control scheme is formed, PID controller is used in inner loop as a main controller, and MFAC is adopted in outer loop as a second controller.The simulation results on MATLAB show that the improved algorithm has a relatively faster response and smaller overshoot relatively.For example,the overshoot of improved algorithm is less than by about 17% that of PID algorithm;and the rise time of improved algorithm is shorter than basic algorithm by around 10s. The simulation results on SMPT-1000 show that the algorithm and control scheme elected in this paper is feasible.