Improving domain generalization by hybrid domain attention and localized maximum sensitivity.

Domain generalization has attracted much interest in recent years due to its practical application scenarios, in which the model is trained using data from various source domains but is tested using data from an unseen target domain. Existing domain generalization methods concern all visual features, including irrelevant ones with the same priority, which easily results in poor generalization performance of the trained model. In contrast, human beings have strong generalization capabilities to distinguish images from different domains by focusing on important features while suppressing irrelevant features with respect to labels. Motivated by this observation, we propose a channel-wise and spatial-wise hybrid domain attention mechanism to force the model to focus on more important features associated with labels in this work. In addition, models with higher robustness with respect to small perturbations of inputs are expected to have higher generalization capability, which is preferable in domain generalization. Therefore, we propose to reduce the localized maximum sensitivity of the small perturbations of inputs in order to improve the network's robustness and generalization capability. Extensive experiments on PACS, VLCS, and Office-Home datasets validate the effectiveness of the proposed method.

Tuning of PID/PIDD2 controllers for integrating processes with robustness specification.

Tuning of PID/PIDD2 controllers are proposed in the paper for integrating processes with time delay via the state space pole placement method. The tuning formulas give the controller parameters in terms of a given maximum sensitivity. An observer-based PID structure is proposed to implement the ideal PID or PIDD2 controllers. The structure utilizes a model-independent observer to estimate the various order of the derivatives of the plant output, thus the sensitivity of the derivatives on the measurement noise can be reduced. Simulation results show that the tuning formulas can achieve good compromise among robustness, disturbance rejection, and noise attenuation for the integrating processes.

IMC based robust PI/PID controllers for time-delayed inverse response processes.

In the last decade, several works have been reported for stable and integrating processes to achieve a specified maximum sensitivity. Also, internal model control (IMC) is a popular controller design strategy as it has only one tuning parameter. IMC based controllers are available in literature for time-delayed inverse response processes but none of the reported works provide guidelines for selecting the tuning parameter. In the present work, IMC-PI/PID controllers for time-delayed inverse response processes are reported. To achieve a specified maximum sensitivity in the range of 1.4 to 2.0, set of tuning rules is proposed for the tuning parameter. Normalized form of the transfer functions are used in the present method which simplifies the design procedure. Novelty of the proposed approach is that the user can not only tune the tuning parameter for desired maximum sensitivity, but can also switch from smooth (Ms = 1.4) to tight (Ms = 2.0) control. Simulation results illustrate the effectiveness of the proposed tuning rules.

Design of fractional order PID controller for load frequency control system with communication delay.

This work explores a frequency-domain approach to design a fractional order proportional-integral-derivative (FO-PID) controller cascaded with a first-order filter for the load frequency control (LFC) system with communication delay. The proposed method is based on suitable reference model development in the direct synthesis (DS) approach, followed by frequency response matching technique. The reference model is developed for robust control-loop performance using the stability-margin and time-domain specifications. The values of the fractional orders of the integral and derivative terms are obtained according to the dynamics of the nominal system. The proposed controllers have been designed for some LFC systems taken from the literature that have different dynamics with reheat, non-reheat and hydraulic turbines and performances with non-linearity like generation rate constraint (GRC), generation dead band (GDB) along with noise have been compared favorably with that of some controllers prevalent in the literature. The proposed controllers have been shown to work efficaciously for the decentralized multi-area IEEE 39-bus New England test system along with variable communication delay. To show the efficacy of the proposed controllers the load-disturbance responses along with the frequency and time domain performance indices have been evaluated for comparison.

A novel robust Virtual Reference Feedback Tuning approach for minimum and non-minimum phase systems.

In real-world applications, it is often desired that the design of a closed-loop system must attain not only high performance but also robustness. This paper presents a novel robustness-based formulation for control for discrete time minimum and non-minimum phase systems using the Virtual Reference Feedback Tuning (VRFT) framework. The proposed idea is to design robust lower and higher order Proportional Integral and Derivative (PID) controller using Maximum Sensitivity (Ms) based closed-loop reference model by minimization of the VRFT objective function (JVR) The efficacy of the approach is verified on various systems in simulation and also demonstrated on the Temperature Control Process and Level Control Process experimental setups. The proposed VRFT approach provides improved performance and robustness for set point tracking and disturbance rejection. Also, the proposed approach ensures stability and specific robustness of the closed-loop system. Further, the fragility of the controller is investigated for perturbations in the controller parameters.

Optimal controller synthesis for second order time delay systems with at least one RHP pole.

An optimal H2 minimization framework is proposed in this paper for devising a controller of PID in nature, based on a refined IMC filter configuration. The tuning strategy is for controlling time delay system with at least one pole which falls on the right half of the s-plane. An underdamped model based filter is used in place of the unity damping ratio (critically damped) filter available in the literature to improve the reset action. The method has a single adjustable closed loop tuning parameter. Guidelines have been provided for choosing the pertinent tuning parameter based on the sensitivity function. Simulation work has been executed on diverse unstable models to support the advantages of the proposed scheme. The proposed controller yields improved performances over other recently reported tuning techniques in the literature. Experimental implementation is carried out on an inverted pendulum for demonstrating the practical applicability of the present method. The efficacy of the intended controller design is quantitatively analyzed using the time integral performance index.

Enhanced IMC based PID controller design for non-minimum phase (NMP) integrating processes with time delays.

Internal model control (IMC) with optimal H2 minimization framework is proposed in this paper for design of proportional-integral-derivative (PID) controllers. The controller design is addressed for integrating and double integrating time delay processes with right half plane (RHP) zeros. Blaschke product is used to derive the optimal controller. There is a single adjustable closed loop tuning parameter for controller design. Systematic guidelines are provided for selection of this tuning parameter based on maximum sensitivity. Simulation studies have been carried out on various integrating time delay processes to show the advantages of the proposed method. The proposed controller provides enhanced closed loop performances when compared to recently reported methods in the literature. Quantitative comparative analysis has been carried out using the performance indices, Integral Absolute Error (IAE) and Total Variation (TV).

Two-degree-of-freedom fractional order-PID controllers design for fractional order processes with dead-time.

Recently, fractional order (FO) processes with dead-time have attracted more and more attention of many researchers in control field, but FO-PID controllers design techniques available for the FO processes with dead-time suffer from lack of direct systematic approaches. In this paper, a simple design and parameters tuning approach of two-degree-of-freedom (2-DOF) FO-PID controller based on internal model control (IMC) is proposed for FO processes with dead-time, conventional one-degree-of-freedom control exhibited the shortcoming of coupling of robustness and dynamic response performance. 2-DOF control can overcome the above weakness which means it realizes decoupling of robustness and dynamic performance from each other. The adjustable parameter η2 of FO-PID controller is directly related to the robustness of closed-loop system, and the analytical expression is given between the maximum sensitivity specification Ms and parameters η2. In addition, according to the dynamic performance requirement of the practical system, the parameters η1 can also be selected easily. By approximating the dead-time term of the process model with the first-order Padé or Taylor series, the expressions for 2-DOF FO-PID controller parameters are derived for three classes of FO processes with dead-time. Moreover, compared with other methods, the proposed method is simple and easy to implement. Finally, the simulation results are given to illustrate the effectiveness of this method.