1#. • Basis Weight Control System
'JH Identification parameters of denoised SOPTD with Gaussian noise by running the four algorithms ten times 5BCMF '01%5SFTVMUTGPS 1 VTJOHGPVS140 - CBTFEBMHPź SJUINT
5BCMF 401%5SFTVMUTGPS 1 VTJOHGPVS140 - CBTFEBMHPź SJUINT Methods Min Max Mean Std
Best parameter K F =1.016, T F =2.989, L F =5.473 K F =1.016, T F =2.991, L F =5.473 K F =1.016, T F =2.987, L F =5.476 K F =1.016, T F =2.990, L F =5.473
Best parameter K S =1.008, T S 1 =1.954, T S 2 =2.069,
Min
Max
Mean
Std
Methods
173.5
1323.7
366.3
350.8
Classical PSO
315.1
1251.7
610.6
371.5
Classical PSO
L S =4.221 K S =1.008, T S 1 =2.059, T S 2 =1.962, L S =4.234 K S =1.009, T S 1 =1.919, T S 2 =2.185, L S =4.149 K S =1.008, T S 1 =2.011, T S 2 =2.014, L S =4.226
315.1
1251.7
496.9
319.7
PSO-TVAC
173.3
2093.7
630.7
627.8
PSO-TVAC
315.1
500.3
338.1
57.8
MPSO
175.8
1257.1
323.4
317.5
MPSO
315.1
470.3
316.3
51.7
SNPSO
173.1
958.4
290.5
290.4
SNPSO
algorithms, the tables also include a measure of performance: the integral of absolute error (IAE), where the error in each case is the difference between the "real" process and model output. The statistical results demonstrate that SNPSO can determine the best
approximation parameters, and the average convergence performance of SNPSO is better than other methods. Fig. 14 and Fig. 15 illustrate the best FOPDT and
7PM /P
Made with FlippingBook - professional solution for displaying marketing and sales documents online