Amirkabir University of Technology Amirkabir Journal of Mechanical Engineering 2008-6032 56 4 2024 06 21 Controller design based on the use of state observers for blood glucose regulation in patients with type 1 diabetes Controller design based on the use of state observers for blood glucose regulation in patients with type 1 diabetes 467 496 5512 10.22060/mej.2024.22825.7683 FA Mohamadreza Homayounzade Mechanical Engineering department- Fasa University-Fasa-Iran. Journal Article 2023 11 25 In this paper, an observer-based nonlinear controller for regulating blood glucose concentrations (BGC) in type 1 diabetes mellitus (T1DM) is proposed. The virtual patient model considered is the extended Bergmann minimal model, which is augmented by a meal disturbance and adapted to represent the insulin-glucose homeostasis of T1DM. The backstepping (BS) technique is used to design a closed-loop feedback controller. The proposed controller does not need to measure insulin, and plasma concentrations while improving control performance and robustness against uncertainty. Insulin concentration and plasma levels are estimated using state observers. These estimations are used as feedback to the controller. The asymptotic stability of the observer-based controller is proved using the Lyapunov theorem. Moreover, it is proved that the system is bounded input-bounded output (BIBO) stable in the presence of uncertainties generated by uncertain parameters and external disturbances. For realistic situations, we consider only the BGC to be available for measurement, and additionally, inter-and intra-patient variability of system parameters is considered. The results confirm that the proposed controller can asymptotically regulate BGC by appropriate injection of insulin under meal disturbance and ±%25 of variations in system parameters. In this paper, an observer-based nonlinear controller for regulating blood glucose concentrations (BGC) in type 1 diabetes mellitus (T1DM) is proposed. The virtual patient model considered is the extended Bergmann minimal model, which is augmented by a meal disturbance and adapted to represent the insulin-glucose homeostasis of T1DM. The backstepping (BS) technique is used to design a closed-loop feedback controller. The proposed controller does not need to measure insulin, and plasma concentrations while improving control performance and robustness against uncertainty. Insulin concentration and plasma levels are estimated using state observers. These estimations are used as feedback to the controller. The asymptotic stability of the observer-based controller is proved using the Lyapunov theorem. Moreover, it is proved that the system is bounded input-bounded output (BIBO) stable in the presence of uncertainties generated by uncertain parameters and external disturbances. For realistic situations, we consider only the BGC to be available for measurement, and additionally, inter-and intra-patient variability of system parameters is considered. The results confirm that the proposed controller can asymptotically regulate BGC by appropriate injection of insulin under meal disturbance and ±%25 of variations in system parameters.

https://mej.aut.ac.ir/article_5512_f6493a7c8ea273d0bf975a225c520be9.pdf

Amirkabir University of Technology Amirkabir Journal of Mechanical Engineering 2008-6032 56 4 2024 06 21 Parametric Study of Model-Based Dynamic Control Methods for Enhancing Locomotion in Underactuated Biped Robots, Case study: Hybrid Zero Dynamics and Proportional-Derivative Feedback Parametric Study of Model-Based Dynamic Control Methods for Enhancing Locomotion in Underactuated Biped Robots, Case study: Hybrid Zero Dynamics and Proportional-Derivative Feedback 497 516 5513 10.22060/mej.2024.22874.7688 FA Roozbeh Ghanadi-Azar Shahid Beheshti University Mohammad Reza Haghjoo Faculty of Mechanical and Energy Engineering, Shahid Beheshti University, Tehran, Iran. Mostafa Taghizadeh Faculty of Mechanical and Energy Engineering, Shahid Beheshti University, Tehran, Iran. Journal Article 2023 12 20 The parametric study of model-based dynamic control methods holds significant importance in biped robot motion control. This research delves into a detailed examination of the parameters of model-based dynamic control methods, specifically the Hybrid Zero Dynamics (HZD) and Proportional-Derivative (PD) feedback control methods, to enhance the locomotion of underactuated biped robots. A three-link underactuated biped robot without a knee joint with three degrees of freedom is used as a case study, and the dynamic equations for this model are extracted in continuous and impact phases. Robot simulations are executed in MATLAB software by comparing and analyzing the control parameters in the two mentioned methods, and the results are compared and discussed. Furthermore, the effect of variations in control parameters in the Proportional-Derivative feedback control method is evaluated and compared. The results indicate that the Hybrid Zero Dynamics method generates more symmetrical and uniformly paced movements than the Proportional-Derivative feedback control method, with lower control effort. Increasing the control parameters in the Proportional-derived feedback control method brings its results closer to those of the hybrid zero dynamics method, accompanied by a reduction in control effort. In addition to presenting results, this study meticulously examines and analyzes control parameters, which can enhance bipedal robot performance. The parametric study of model-based dynamic control methods holds significant importance in biped robot motion control. This research delves into a detailed examination of the parameters of model-based dynamic control methods, specifically the Hybrid Zero Dynamics (HZD) and Proportional-Derivative (PD) feedback control methods, to enhance the locomotion of underactuated biped robots. A three-link underactuated biped robot without a knee joint with three degrees of freedom is used as a case study, and the dynamic equations for this model are extracted in continuous and impact phases. Robot simulations are executed in MATLAB software by comparing and analyzing the control parameters in the two mentioned methods, and the results are compared and discussed. Furthermore, the effect of variations in control parameters in the Proportional-Derivative feedback control method is evaluated and compared. The results indicate that the Hybrid Zero Dynamics method generates more symmetrical and uniformly paced movements than the Proportional-Derivative feedback control method, with lower control effort. Increasing the control parameters in the Proportional-derived feedback control method brings its results closer to those of the hybrid zero dynamics method, accompanied by a reduction in control effort. In addition to presenting results, this study meticulously examines and analyzes control parameters, which can enhance bipedal robot performance.

https://mej.aut.ac.ir/article_5513_3826ec89cd9f6ddcb7f35366bf7e3d73.pdf

Amirkabir University of Technology Amirkabir Journal of Mechanical Engineering 2008-6032 56 4 2024 06 21 Hydrodynamic investigation of industrial gas-phase polyethylene reactors in two different technologies Hydrodynamic investigation of industrial gas-phase polyethylene reactors in two different technologies 517 542 5518 10.22060/mej.2024.22927.7693 FA Peyman Karimzadeh-Soureshjani Center of Engineering and Multiscale Modeling of Fluid Flow (CEMF), Faculty of Chemical Engineering, Amirkabir University of Technology, Tehran, Iran Hamidreza Norouzi Center of Engineering and Multiscale Modeling of Fluid Flow (CEMF), Faculty of Chemical Engineering, Amirkabir University of Technology, Tehran, Iran Journal Article 2024 01 15 Gas-solid fluidized bed reactors are among the common methods to produce linear low-density polyethylene. The contact quality between the two phases and the mixing of the solid particles in these reactors have significant impacts on the polymerization reaction. In this research, the hydrodynamic behaviour of two reactors licensed by Basell and Mitsui companies was investigated using computational fluid dynamics. Two-fluid model with the kinetic theory of granular flow was used. The model was first validated using experimental data, and then the analyses of the Mitsui and Basell industrial reactors were carried out. The results showed that the uniformity of the gas phase volume fraction, which indicates the quality of the gas-solid contact, increases with the distance from the bed bottom in both reactors so that at a normalized height of 0.96, the phase homogeneity reaches its maximum. At this height, the coefficient of variations of volume fraction in the Basell and Mitsui reactors are 0.4% and 1.3%, respectively, and the phase homogeneity in the Mitsui reactor is always higher than that in the Basell reactor at different heights, indicating a better contact between particles and gas. The time-averaged axial velocity of the solid particles at different heights showed that the intensity of solid particle movement is higher in the Basell reactor, so that at a normalized height of 0.72, the axial velocity of the solid particles in the Basell reactor is approximately 2 m/s, while it is approximately 1 m/s in the Mitsui reactor. Considering the axial velocity parameter, it can be concluded that the quality of solid particle mixing in the Basell reactor is higher than that in the Mitsui reactor. Gas-solid fluidized bed reactors are among the common methods to produce linear low-density polyethylene. The contact quality between the two phases and the mixing of the solid particles in these reactors have significant impacts on the polymerization reaction. In this research, the hydrodynamic behaviour of two reactors licensed by Basell and Mitsui companies was investigated using computational fluid dynamics. Two-fluid model with the kinetic theory of granular flow was used. The model was first validated using experimental data, and then the analyses of the Mitsui and Basell industrial reactors were carried out. The results showed that the uniformity of the gas phase volume fraction, which indicates the quality of the gas-solid contact, increases with the distance from the bed bottom in both reactors so that at a normalized height of 0.96, the phase homogeneity reaches its maximum. At this height, the coefficient of variations of volume fraction in the Basell and Mitsui reactors are 0.4% and 1.3%, respectively, and the phase homogeneity in the Mitsui reactor is always higher than that in the Basell reactor at different heights, indicating a better contact between particles and gas. The time-averaged axial velocity of the solid particles at different heights showed that the intensity of solid particle movement is higher in the Basell reactor, so that at a normalized height of 0.72, the axial velocity of the solid particles in the Basell reactor is approximately 2 m/s, while it is approximately 1 m/s in the Mitsui reactor. Considering the axial velocity parameter, it can be concluded that the quality of solid particle mixing in the Basell reactor is higher than that in the Mitsui reactor.

https://mej.aut.ac.ir/article_5518_c07e572d7188a37a1c1a5313aa979f8b.pdf

Amirkabir University of Technology Amirkabir Journal of Mechanical Engineering 2008-6032 56 4 2024 06 21 Elastostatic Modeling and Optimal Design of Rhombic Compliant Mechanism Elastostatic Modeling and Optimal Design of Rhombic Compliant Mechanism 543 566 5517 10.22060/mej.2024.22905.7691 FA Mohammad Saeed Erami Department of Mechanical Engineering, Amirkabir University of Technology, Tehran, Iran Hamed Ghafarirad Department of Mechanical Engineering, Amirkabir University of Technology, Tehran, Iran Afshin Taghvaeipour Department of Mechanical Engineering, Amirkabir University of Technology, Tehran, Iran 0000-0002-0787-0979 Pouya Firuzy Rad Department of Mechanical Engineering, Amirkabir University of Technology, Tehran, Iran Journal Article 2024 01 02 Compliant mechanisms are designed and used for precise positioning and amplification of piezoelectric actuators Due to their integrated structure. Modelling the kinematic behaviour of these mechanisms has challenges due to their continuous structure and elastic deformation. This article presents a structural matrix-based method called the elastostatic method for static modelling of compliant mechanisms. The innovation of elastostatic modelling reduces calculations by approximating rotation and small displacement. The main goal of this research is to design and optimize the rhombus flexible mechanism using elastostatic modelling. This mechanism is optimized in such a way that, in addition to positioning, it has high magnification and low input stiffness. The rhombus mechanism has an integrated and simple structure and is used for micron positioning and piezoelectric actuator amplification. In this research, the rhombus mechanism has been modeled using the elastostatic method, and its dimensions have been optimized according to the parameters of the mechanism; For this purpose, it is necessary to check the modeling error. The modelling error is compared with simulation in finite element software and experimental results. The results show that the modelling used to design the rhombus mechanism has a 1.5% error compared to experimental results. Compliant mechanisms are designed and used for precise positioning and amplification of piezoelectric actuators Due to their integrated structure. Modelling the kinematic behaviour of these mechanisms has challenges due to their continuous structure and elastic deformation. This article presents a structural matrix-based method called the elastostatic method for static modelling of compliant mechanisms. The innovation of elastostatic modelling reduces calculations by approximating rotation and small displacement. The main goal of this research is to design and optimize the rhombus flexible mechanism using elastostatic modelling. This mechanism is optimized in such a way that, in addition to positioning, it has high magnification and low input stiffness. The rhombus mechanism has an integrated and simple structure and is used for micron positioning and piezoelectric actuator amplification. In this research, the rhombus mechanism has been modeled using the elastostatic method, and its dimensions have been optimized according to the parameters of the mechanism; For this purpose, it is necessary to check the modeling error. The modelling error is compared with simulation in finite element software and experimental results. The results show that the modelling used to design the rhombus mechanism has a 1.5% error compared to experimental results.

https://mej.aut.ac.ir/article_5517_1c2fa825c63064feab0f5c7962d5fbcf.pdf

Amirkabir University of Technology Amirkabir Journal of Mechanical Engineering 2008-6032 56 4 2024 06 21 Theoretic and Experimental Fatigue Analysis of Off-axis Unidirectional Rubbery Composites Using Nonlinear Life Prediction Model Theoretic and Experimental Fatigue Analysis of Off-axis Unidirectional Rubbery Composites Using Nonlinear Life Prediction Model 567 594 5519 10.22060/mej.2024.23129.7720 FA Majid Jamali School of Mechanical Engineering, Iran University of Science and Technology, Tehran, 16846-13114, Iran Bijan Mohammadi School of Mechanical Engineering, Iran University of Science and Technology, Tehran, 16846-13114, Iran. Mahmood Mehrdad Shokrieh School of Mechanical Engineering, Iran University of Science and Technology, Tehran, 16846-13114, Iran. Journal Article 2024 04 23 The aim of this article is analyzing the fatigue behavior of off-axis unidirectional rubbery composites under uniaxial tension-tension cyclic loading based on the developed damage-entropy model. The main advantage of the damage-entropy model is that it accounts for the viscoelastic property and temperature increase during the fatigue loading conditions. The off-axis rubbery composite lay-ups exhibit a nonlinear stress-strain response similar to the rubber matrix. Hence, the Newton-Raphson method is employed to capture the nonlinear behaviour of rubbery composites in this study. The failure criterion in the damage-entropy model is based on the fracture fatigue entropy value. To characterize the longitudinal, transverse, and in-plane shear behaviour of rubbery composites, static and fatigue experimental tests on different lay-ups are conducted. Moreover, the damage energy, the energy dissipation due to viscoelastic behaviour and the heat transfer to the environment during the fatigue loading will be calculated. Furthermore, the experimental results of [45]4 lay-up are utilized to validate the developed damage-entropy model. Finally, the experimental and modelling results of hysteresis energy, temperature change, and fatigue life of steel-cord rubber composite [45]4 lay-up for different stress levels subjected to stress ratio 0.1 and 1 Hz frequency, are compared. The comparison between the analytical results and experiments indicates the capabilities of the present model. The aim of this article is analyzing the fatigue behavior of off-axis unidirectional rubbery composites under uniaxial tension-tension cyclic loading based on the developed damage-entropy model. The main advantage of the damage-entropy model is that it accounts for the viscoelastic property and temperature increase during the fatigue loading conditions. The off-axis rubbery composite lay-ups exhibit a nonlinear stress-strain response similar to the rubber matrix. Hence, the Newton-Raphson method is employed to capture the nonlinear behaviour of rubbery composites in this study. The failure criterion in the damage-entropy model is based on the fracture fatigue entropy value. To characterize the longitudinal, transverse, and in-plane shear behaviour of rubbery composites, static and fatigue experimental tests on different lay-ups are conducted. Moreover, the damage energy, the energy dissipation due to viscoelastic behaviour and the heat transfer to the environment during the fatigue loading will be calculated. Furthermore, the experimental results of [45]4 lay-up are utilized to validate the developed damage-entropy model. Finally, the experimental and modelling results of hysteresis energy, temperature change, and fatigue life of steel-cord rubber composite [45]4 lay-up for different stress levels subjected to stress ratio 0.1 and 1 Hz frequency, are compared. The comparison between the analytical results and experiments indicates the capabilities of the present model.

https://mej.aut.ac.ir/article_5519_9790cda261599bb19e940b5dad1d7333.pdf

Amirkabir University of Technology Amirkabir Journal of Mechanical Engineering 2008-6032 56 4 2024 06 21 Highway decision-making strategy for autonomous vehicle for overtaking maneuver using deep reinforcement learning (DRL) method Highway decision-making strategy for autonomous vehicle for overtaking maneuver using deep reinforcement learning (DRL) method 595 620 5525 10.22060/mej.2024.22682.7659 FA Ali Rizehvandi Faculty of Mechanical Engineering, K.N.Toosi University of Technology, Tehran, Iran Shahram Azadi Faculty of Mechanical Engineering, K.N.Toosi University of Technology, Tehran, Iran Journal Article 2023 09 30 Automated driving represents a novel technology aimed at reducing traffic accidents and enhancing driving efficiency. This research introduces a deep reinforcement learning (DRL) approach for autonomous vehicles, focusing on overtaking scenarios on highways. Initially, a highway traffic environment is established, to guide the agent through surrounding vehicles both efficiently and safely. A hierarchical control framework is outlined to manage high-level driving decisions alongside low-level control aspects like car speed and acceleration. Subsequently, a specialized DRL-based method known as Deep Deterministic Policy Gradient (DDPG) is employed to devise decision-making strategies on the highway. The DDPG offers continuous action space exploration, making it suitable for tasks like autonomous driving where actions are not discrete. Unlike DQN, it can handle high-dimensional action spaces more effectively, enhancing its applicability in complex environments like highway driving. The efficacy of the DDPG algorithm is compared to that of the DQN algorithm, with subsequent evaluation of the results. Simulation outcomes demonstrate that the DDPG algorithm adeptly handles highway driving tasks with efficiency and safety. The study underscores the potential of DRL techniques, particularly the DDPG approach, in advancing the capabilities of autonomous vehicles and improving their performance in complex driving scenarios. Automated driving represents a novel technology aimed at reducing traffic accidents and enhancing driving efficiency. This research introduces a deep reinforcement learning (DRL) approach for autonomous vehicles, focusing on overtaking scenarios on highways. Initially, a highway traffic environment is established, to guide the agent through surrounding vehicles both efficiently and safely. A hierarchical control framework is outlined to manage high-level driving decisions alongside low-level control aspects like car speed and acceleration. Subsequently, a specialized DRL-based method known as Deep Deterministic Policy Gradient (DDPG) is employed to devise decision-making strategies on the highway. The DDPG offers continuous action space exploration, making it suitable for tasks like autonomous driving where actions are not discrete. Unlike DQN, it can handle high-dimensional action spaces more effectively, enhancing its applicability in complex environments like highway driving. The efficacy of the DDPG algorithm is compared to that of the DQN algorithm, with subsequent evaluation of the results. Simulation outcomes demonstrate that the DDPG algorithm adeptly handles highway driving tasks with efficiency and safety. The study underscores the potential of DRL techniques, particularly the DDPG approach, in advancing the capabilities of autonomous vehicles and improving their performance in complex driving scenarios.

https://mej.aut.ac.ir/article_5525_ebe69d8b20c008ae7927d02acf046126.pdf