1#. • Basis Weight Control System
the output ŷ ( t ) to verify the robustness of the algorithm. Fig. 12 illustrates the SOPTD system parameters K S , T S 1 , T S 2 , and L S obtained from ten experiments using four algorithms. The parameter value of the estimated system obtained by SNPSO algorithm is closer to the parameter value of the real process. Fig. 13 illustrates that the wavelet denoising method improves the stability
and robustness of SNPSO. 6.3 Ȟ Transfer function P 3
When the real process is assumed to be the transfer function (Eq. (13)) the estimation models are as follows: G F ( s ) = K F T F s + 1 e ϖ L F s (17) and G S ( s ) = K S ( T S 1 s + 1) ( T S 2 s + 1) e ϖ L S s (18) Table 4 and Table 5 present the FOPDT and SOPDT model parameters for processes P 3 , respectively. To compare the identification capabilities of the four
the variations of objective function J for four algorithms during the optimization. The convergence speed and convergence precision of four PSO-based algorithms is clearly illustrated in this figure. SNPSO method converges faster and more precisely than other PSO-based algorithms. The noise signal n ( t ) with variance of 0.16 is added to 'JH Ȟ Variations of objective function J for SOPTD using four algorithms
'JH Identification parameters of SOPTD with Gaussian noise by running the four algorithms ten times
7PM /P
Made with FlippingBook - professional solution for displaying marketing and sales documents online