Improved disturbance rejection with modified Smith predictor for integrating FOPTD processes

Abstract

Improved disturbance rejection behaviour with modified Smith predictor is reported here for controlling integrating first-order plus time delay processes. Due to location of a pole at origin, process is said to be integrating in nature. In addition, due to presence of considerable dead time, it is very difficult to obtain the desired output from such processes using conventional control technique. In practice, a good number of chemical processes (e.g. distillation, evaporation, combustion, drying etc.) are integrating as well as delay dominating in nature. To ascertain desirable close-loop response for processes with large dead time, Smith predictor is a renowned methodology due to its simplicity and efficacy. But, this technique fails to perform satisfactorily for integrating processes with time delay. A good alternative can be considered as modified Smith predictor. This technique involves more than one controller for achieving desirable servo as well as regulatory responses. To avoid the tuning complexity of controllers, our proposed scheme involves comparatively less number of controllers with relatively simple tuning guide line. Distinct feature of the proposed tuning scheme is that process overshoot can be restricted within acceptable limit as well as improved load recovery can also be achieved. Efficacy of the proposed scheme is substantiated through performance assessment as well as stability study in comparison with well-known modified Smith predictor based tuning relations is also reported.

Modified Smith predictor‐based all‐proportional‐derivative control for second‐order delay‐dominated integrating processes

“…Furthermore, closed-loop response is also verified in the presence of measurement noise (noise power = 0.001), 22 and the related control actions for nominal (Equation 45) model are depicted in Figure 11. Robust stability is analysed by small gain theorem 28 by incorporating positive perturbations with process gain (ΔK), time delay (Δθ m ), and negative perturbation in…”

“…In addition, TV in control action is also computed to verify the smoothness in control action for each setting. To evaluate the noise sensitivity of the proposed scheme, noise signal with power 0.001 22 is introduced in the process output.…”

“…Robustness of proposed controller is verified with others' reported control schemes 15,19,20 by introducing +10% perturbation in process gain (K) and time delay (θ m ) as given by Equation 44. Stability analysis is conducted with small gain theorem 28 by incorporating positive perturbation in terms of process gain (ΔK), time delay (Δθ m ), and negative perturbation in time constant (Δτ u ) as depicted in Table II Moreover, closed-loop response in the presence of noise signal (noise power = 0.001) 22 of Model I associated with control actions for nominal (Equation 43) model is shown in Figure 7. Overall performance enhancement during servo and closed-loop responses is listed in Tables III and IV, respectively, for nominal and perturbed models.…”

“…Closed-loop performance is also tested in the presence of noise signal (noise power = 0.001) 22 for nominal (Equation 47) model as shown in Figure 15. Improved stability of the proposed scheme is established with reduced complementary sensitivity C(jω) in the presence of perturbed process gain (ΔK) and time delay (Δθ m ) as shown in Table V, and the corresponding magnitude versus frequency plot is provided in Figure 16.…”

“…In addition, robustness of the proposed scheme is also verified by incorporating perturbation in process parameters along with measurement noise. 22 Quantitative estimation of controller efficacy is also estimated through computing integral absolute error (IAE) and integral square error (ISE), integral time multiplied absolute error (ITAE) and integral time multiplied square error (ITSE) indices. 23 In addition, smoothness in control action is also assessed with total variation (TV) in control action 23 for the proposed scheme along with other reported dead-time compensating techniques 11,12,14,15,[18][19][20][21] for IFOPTD and DIPTD processes.…”