Circulating Fluidized Bed Coal-Saving Optimization Control Method - Intelligent Computing for Sustainable Energy and Environment - page 309

Fig. 4. The step responses of the CFB boiler

4.2 Constrained DMC

Consider online computation, we choose control horizon as M=1 optimization
horizon P = 100 ,and the weighting matrices Q = blkdiag ( Q c , Q c , Q c , Q c ), Q c =
diag (1 ,..., 1 , 100)

R 100 × 100 . Here, we impose the terminal weighting parame-
ter of the terminal output of the optimization horizon as 100 to improve the
system stability. In addition, weighting matrix R = blkdiag (0 . 1 R c , R c , 0 . 1 R c ),
R c =[1] 1 × 1 The physical constraints of the system include: the amount of coal
feeding should be within [10, 30](t/h); the primary air should be within [10000,
80000](Nm 3 /s); the secondary air should be within [10000, 40000](Nm 3 /s). The
control goal is to steer the boiler from the current operating point to a new one
as ω r = [800 , 4 , 470 , 3 . 7] T , where the data represents the bed temperature, steam
pressure, steam temperature, and flue gas oxygen content. The control results
of constrained DMC control are shown in Fig. 5. It can be seen from Fig. 5 that
from a given initial state to a new one, the amount of coal consumption, primary
air and secondary air eventually converge to a steady vector. Bed temperature,
steam temperature and steam pressure reached stable, and basically reached the
preset reference value. There is gap between flue gas oxygen content and the
preset reference value, which is within 3% to 5%. This gap is resulted by the
character of the three-input and four-output system.

∈

4.3 RTO/DMC Optimal Control

According to the settings and operation requirements, set δ =

1 . 5] T , δ =[20 , 0 . 01 , 5 , 0 . 8] T , λ 1 = diag (0 . 01 , 0 . 01 , 0 . 01 , 0 . 01).
For the same control goal as Section 4.2, we choose the same parameters as
the constrained DMC to achieve a fair comparison and let λ 2 = 100 to design
the RTO/DMC optimal controller. The control results are shown in Fig. 6.
As shown in Fig. 6, the dynamic performance of the RTO/DMC control is

[ − 20 , − 0 . 01 , − 5 , −