Motivated by engineering applications, we address bounded steady-state optimal control of linear dynamical systems undergoing steady-state bandlimited periodic oscillations. The optimization can be cast as a minimization problem by expressing the state and the input as finite Fourier series expansions, and using the expansions coefficients as parameters to be optimized. With this parametrization, we address linear quadratic problems involving periodic bandlimited dynamics by using quadratic minimization with parametric time-dependent constraints. We hence investigate the implications of a discretization of linear continuous time constraints and propose an algorithm that provides a feasible suboptimal solution whose cost is arbitrarily close to the optimal cost for the original constrained steady-state problem. Finally, we discuss practical case studies that can be effectively tackled with the proposed framework, including optimal control of DC/AC power converters, and optimal energy harvesting from pulsating mechanical energy sources.
{"title":"Quadratic Constrained Periodic Optimization for Bandlimited Linear Systems Via the Fourier-Based Method","authors":"G. Moretti, L. Zaccarian, F. Blanchini","doi":"10.1115/1.4049541","DOIUrl":"https://doi.org/10.1115/1.4049541","url":null,"abstract":"\u0000 Motivated by engineering applications, we address bounded steady-state optimal control of linear dynamical systems undergoing steady-state bandlimited periodic oscillations. The optimization can be cast as a minimization problem by expressing the state and the input as finite Fourier series expansions, and using the expansions coefficients as parameters to be optimized. With this parametrization, we address linear quadratic problems involving periodic bandlimited dynamics by using quadratic minimization with parametric time-dependent constraints. We hence investigate the implications of a discretization of linear continuous time constraints and propose an algorithm that provides a feasible suboptimal solution whose cost is arbitrarily close to the optimal cost for the original constrained steady-state problem. Finally, we discuss practical case studies that can be effectively tackled with the proposed framework, including optimal control of DC/AC power converters, and optimal energy harvesting from pulsating mechanical energy sources.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"27 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91073467","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The combustion phasing of spark ignition (SI) engines is traditionally regulated with map-based spark timing (SPKT) control. The calibration of these maps is time-consuming for SI engines with a high number of control actuators. This paper proposes three online SPKT optimization algorithms that can utilize control-oriented semiphysics-based combustion models making the SPKT control algorithm more adaptive to different engine designs. These three SPKT optimizers do not require model inversion and derivative information. These methods also preserve the dependence between combustion phasing, knock, and coefficient of variation (COV) of indicated mean effective pressure (IMEP) models to avoid evaluating combustion models multiple times within one iteration. The two-phase and constraint relaxation methods are derived from direct search optimization theories. The recursive least square (RLS) polynomial fitting method can be considered as a virtual extreme seeking (ES) process that converts the original “black” box nonlinear constrained optimization into the solution of three low-order polynomial equations. Although these three online SPKT optimization approaches have unique properties making them preferable with certain types of combustion models, simulation and test results show that all of them can find the optimal SPKT with less than 10 evaluations of the combustion models. This fact makes it possible to implement the proposed model-based SPKT control strategy in future engine control units (ECUs).
{"title":"Online Spark Timing Optimization With Complex High-Fidelity Combustion Phasing, Knock, and Coefficient of Variation of IMEP Models","authors":"Qilun Zhu, R. Prucka, Shu Wang, Michael J Prucka","doi":"10.1115/1.4049733","DOIUrl":"https://doi.org/10.1115/1.4049733","url":null,"abstract":"\u0000 The combustion phasing of spark ignition (SI) engines is traditionally regulated with map-based spark timing (SPKT) control. The calibration of these maps is time-consuming for SI engines with a high number of control actuators. This paper proposes three online SPKT optimization algorithms that can utilize control-oriented semiphysics-based combustion models making the SPKT control algorithm more adaptive to different engine designs. These three SPKT optimizers do not require model inversion and derivative information. These methods also preserve the dependence between combustion phasing, knock, and coefficient of variation (COV) of indicated mean effective pressure (IMEP) models to avoid evaluating combustion models multiple times within one iteration. The two-phase and constraint relaxation methods are derived from direct search optimization theories. The recursive least square (RLS) polynomial fitting method can be considered as a virtual extreme seeking (ES) process that converts the original “black” box nonlinear constrained optimization into the solution of three low-order polynomial equations. Although these three online SPKT optimization approaches have unique properties making them preferable with certain types of combustion models, simulation and test results show that all of them can find the optimal SPKT with less than 10 evaluations of the combustion models. This fact makes it possible to implement the proposed model-based SPKT control strategy in future engine control units (ECUs).","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"5 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78314098","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Xiaosong Hu, Yang Xin, F. Feng, Kailong Liu, Xianke Lin
Accurate prediction of the remaining useful life (RUL) of lithium-ion batteries can improve the durability, reliability, and maintainability of battery system operation in electric vehicles. To achieve high-accuracy RUL predictions, it is necessary to develop an effective method for long-term nonlinear degradation prediction and quantify the uncertainty of the prediction results. To this end, this paper proposes a hybrid approach for lithium-ion battery RUL prediction based on particle filter (PF) and long short-term memory (LSTM) neural network. First, based on the training set, the model parameters are iteratively updated using the PF algorithm. Second, the LSTM model parameters are obtained using the training set. The mean and standard deviation in the prediction stage are obtained through Monte Carlo (MC) dropout. Finally, the mean value predicted by MC-dropout is used as the measurement for the PF in the prediction phase, the standard deviation represents the uncertainty of the prediction result, and the mean and standard deviation are integrated into the measurement equation of the model. The experimental results show that the proposed hybrid approach has better prediction accuracy than the PF, LSTM algorithm, and two other types of hybrid approaches. The hybrid approach can obtain a narrower confidence interval.
{"title":"A Particle Filter and Long Short-Term Memory Fusion Technique for Lithium-Ion Battery Remaining Useful Life Prediction","authors":"Xiaosong Hu, Yang Xin, F. Feng, Kailong Liu, Xianke Lin","doi":"10.1115/1.4049234","DOIUrl":"https://doi.org/10.1115/1.4049234","url":null,"abstract":"\u0000 Accurate prediction of the remaining useful life (RUL) of lithium-ion batteries can improve the durability, reliability, and maintainability of battery system operation in electric vehicles. To achieve high-accuracy RUL predictions, it is necessary to develop an effective method for long-term nonlinear degradation prediction and quantify the uncertainty of the prediction results. To this end, this paper proposes a hybrid approach for lithium-ion battery RUL prediction based on particle filter (PF) and long short-term memory (LSTM) neural network. First, based on the training set, the model parameters are iteratively updated using the PF algorithm. Second, the LSTM model parameters are obtained using the training set. The mean and standard deviation in the prediction stage are obtained through Monte Carlo (MC) dropout. Finally, the mean value predicted by MC-dropout is used as the measurement for the PF in the prediction phase, the standard deviation represents the uncertainty of the prediction result, and the mean and standard deviation are integrated into the measurement equation of the model. The experimental results show that the proposed hybrid approach has better prediction accuracy than the PF, LSTM algorithm, and two other types of hybrid approaches. The hybrid approach can obtain a narrower confidence interval.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"4 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81926004","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jae-Hyeon Park, Karmvir Singh Phogat, Whimin Kim, D. Chang
In this article, we devise a variant of the extended Kalman filter that can be generally applied to systems on manifolds with simplicity and low computational cost. We extend a given system on a manifold to an ambient open set in Euclidean space and modify the system such that the extended system is transversely stable on the manifold. Then, we apply the standard extended Kalman filter derived in Euclidean space to the modified dynamics. This method is efficient in terms of computation and accurate in comparison with the standard extended Kalman filter. It has the merit that we can apply various Kalman filters derived in Euclidean space including extended Kalman filters for state estimation for systems defined on manifolds. The proposed method is successfully applied to the rigid body attitude dynamics whose configuration space is the special orthogonal group in three dimensions.
{"title":"Transversely Stable Extended Kalman Filters for Systems on Manifolds in Euclidean Spaces","authors":"Jae-Hyeon Park, Karmvir Singh Phogat, Whimin Kim, D. Chang","doi":"10.1115/1.4049540","DOIUrl":"https://doi.org/10.1115/1.4049540","url":null,"abstract":"\u0000 In this article, we devise a variant of the extended Kalman filter that can be generally applied to systems on manifolds with simplicity and low computational cost. We extend a given system on a manifold to an ambient open set in Euclidean space and modify the system such that the extended system is transversely stable on the manifold. Then, we apply the standard extended Kalman filter derived in Euclidean space to the modified dynamics. This method is efficient in terms of computation and accurate in comparison with the standard extended Kalman filter. It has the merit that we can apply various Kalman filters derived in Euclidean space including extended Kalman filters for state estimation for systems defined on manifolds. The proposed method is successfully applied to the rigid body attitude dynamics whose configuration space is the special orthogonal group in three dimensions.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"18 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78095941","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Energy harvesting vibration absorbers (EHVAs) represent a novel type of vibration absorbers where the dissipated energy is harnessed in the absorber system. Conventional optimization-based methods can be utilized for optimal design of EHVAs, but this usually involves in iterative design procedures, particularly for approaching performance limits. In this note, a visualization technique is proposed. The problem of existence and uniqueness solutions is addressed; the intimate relationship between energy harvesting and vibration suppression performances is disclosed; and the fundamental issue of determining performance limit with this visualized method is solved. These features form solid contributions of the current proposal over those optimization-based design methods. The corresponding design procedures are illustrated and the claims are further validated through real-time simulations to the optimal design of EHVAs.
{"title":"Optimal Design for Energy Harvesting Vibration Absorbers","authors":"Jiqiang Wang","doi":"10.1115/1.4049235","DOIUrl":"https://doi.org/10.1115/1.4049235","url":null,"abstract":"\u0000 Energy harvesting vibration absorbers (EHVAs) represent a novel type of vibration absorbers where the dissipated energy is harnessed in the absorber system. Conventional optimization-based methods can be utilized for optimal design of EHVAs, but this usually involves in iterative design procedures, particularly for approaching performance limits. In this note, a visualization technique is proposed. The problem of existence and uniqueness solutions is addressed; the intimate relationship between energy harvesting and vibration suppression performances is disclosed; and the fundamental issue of determining performance limit with this visualized method is solved. These features form solid contributions of the current proposal over those optimization-based design methods. The corresponding design procedures are illustrated and the claims are further validated through real-time simulations to the optimal design of EHVAs.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"13 1 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90743704","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Seong Hyeon Hong, Jackson Cornelius, Yi Wang, K. Pant
This paper presents a new artificial neural network (ANN)-based system model that concatenates an optimized artificial neural network (OANN) and a neural network compensator (NNC) in series to capture temporally varying system dynamics caused by slow-paced degradation/anomaly. The OANN comprises a complex, fully connected multilayer perceptron, trained offline using nominal, anomaly free data, and remains unchanged during online operation. Its hyperparameters are selected using genetic algorithm-based meta-optimization. The compact NNC is updated continuously online using collected sensor data to capture the variations in system dynamics, rectify the OANN prediction, and eventually minimize the discrepancy between the OANN-predicted and actual response. The combined OANN–NNC model then reconfigures the model predictive control (MPC) online to alleviate disturbances. Through numerical simulation using an unmanned quadrotor as an example, the proposed model demonstrates salient capabilities to mitigate anomalies introduced to the system while maintaining control performance. We compare the OANN–NNC with other online modeling techniques (adaptive ANN and multinetwork model), showing it outperforms them in reference tracking of altitude control by at least 0.5 m and yaw control by 1 deg. Moreover, its robustness is confirmed by the MPC consistency regardless of anomaly presence, eliminating the need for additional model management during online operation.
{"title":"Optimized Artificial Neural Network Model and Compensator in Model Predictive Control for Anomaly Mitigation","authors":"Seong Hyeon Hong, Jackson Cornelius, Yi Wang, K. Pant","doi":"10.1115/1.4049130","DOIUrl":"https://doi.org/10.1115/1.4049130","url":null,"abstract":"\u0000 This paper presents a new artificial neural network (ANN)-based system model that concatenates an optimized artificial neural network (OANN) and a neural network compensator (NNC) in series to capture temporally varying system dynamics caused by slow-paced degradation/anomaly. The OANN comprises a complex, fully connected multilayer perceptron, trained offline using nominal, anomaly free data, and remains unchanged during online operation. Its hyperparameters are selected using genetic algorithm-based meta-optimization. The compact NNC is updated continuously online using collected sensor data to capture the variations in system dynamics, rectify the OANN prediction, and eventually minimize the discrepancy between the OANN-predicted and actual response. The combined OANN–NNC model then reconfigures the model predictive control (MPC) online to alleviate disturbances. Through numerical simulation using an unmanned quadrotor as an example, the proposed model demonstrates salient capabilities to mitigate anomalies introduced to the system while maintaining control performance. We compare the OANN–NNC with other online modeling techniques (adaptive ANN and multinetwork model), showing it outperforms them in reference tracking of altitude control by at least 0.5 m and yaw control by 1 deg. Moreover, its robustness is confirmed by the MPC consistency regardless of anomaly presence, eliminating the need for additional model management during online operation.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"12 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84276433","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Andrew A. Stanley, A. Amini, C. Glick, Nathan S. Usevitch, Yigit Menguc, Sean Keller
Resistor–capacitor (RC) response time models for pressurizing and depressurizing a pneumatic capacitor (mass accumulator) through a resistor (flow restriction) comprise a framework to systematically analyze complex fluidic circuits. A model for pneumatic resistance is derived from a combination of fundamental fluid mechanics and experimental results. Models describing compressible fluid capacitance are derived from thermodynamic first principles and validated experimentally. The models are combined to derive the ordinary differential equations that describe the RC dynamics. These equations are solved analytically for rigid capacitors and numerically for deformable capacitors to generate pressure response curves as a function of time. The dynamic pressurization and depressurization response times to reach 63.2% (or 1−e−1) of exponential decay are validated in simple pneumatic circuits with combinations of flow restrictions ranging from 100 μm to 1 mm in diameter, source pressures ranging from 5 to 200 kPa, and capacitor volumes of 0.5 to 16 mL. Our RC models predict the response times, which range from a few milliseconds to multiple seconds depending on the combination, with a coefficient of determination of r2=0.983. The utility of the models is demonstrated in a multicomponent fluidic circuit to find the optimal diameter of tubing between a three-way electromechanical valve and a pneumatic capacitor to minimize the response time for the changing pressure in the capacitor. These lumped-parameter models represent foundational blocks upon which timing models of pneumatic circuits can be built for a variety of applications from soft robotics and industrial automation to high-speed microfluidics.
{"title":"Lumped-Parameter Response Time Models for Pneumatic Circuit Dynamics","authors":"Andrew A. Stanley, A. Amini, C. Glick, Nathan S. Usevitch, Yigit Menguc, Sean Keller","doi":"10.1115/1.4049009","DOIUrl":"https://doi.org/10.1115/1.4049009","url":null,"abstract":"\u0000 Resistor–capacitor (RC) response time models for pressurizing and depressurizing a pneumatic capacitor (mass accumulator) through a resistor (flow restriction) comprise a framework to systematically analyze complex fluidic circuits. A model for pneumatic resistance is derived from a combination of fundamental fluid mechanics and experimental results. Models describing compressible fluid capacitance are derived from thermodynamic first principles and validated experimentally. The models are combined to derive the ordinary differential equations that describe the RC dynamics. These equations are solved analytically for rigid capacitors and numerically for deformable capacitors to generate pressure response curves as a function of time. The dynamic pressurization and depressurization response times to reach 63.2% (or 1−e−1) of exponential decay are validated in simple pneumatic circuits with combinations of flow restrictions ranging from 100 μm to 1 mm in diameter, source pressures ranging from 5 to 200 kPa, and capacitor volumes of 0.5 to 16 mL. Our RC models predict the response times, which range from a few milliseconds to multiple seconds depending on the combination, with a coefficient of determination of r2=0.983. The utility of the models is demonstrated in a multicomponent fluidic circuit to find the optimal diameter of tubing between a three-way electromechanical valve and a pneumatic capacitor to minimize the response time for the changing pressure in the capacitor. These lumped-parameter models represent foundational blocks upon which timing models of pneumatic circuits can be built for a variety of applications from soft robotics and industrial automation to high-speed microfluidics.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"50 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79328613","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Artificial muscles (AMs) traditionally rely on pneumatic sources of fluid power. The use of hydraulics can increase the power and force to weight and volume ratios of AM actuators. This paper develops a control-centric third-order single-input single-output (SISO) lumped-parameter dynamic model and sliding mode position controller based on Filippov's principle of equivalent dynamics for a braided hydraulic artificial muscle (HAM) actuator. The model predicts the nonlinear behavior of the HAM free contraction and captures the fluid and actuator nonlinear dynamic interactions in addition to the braid deformation. Model simulations are compared to experimental results for quasi-static pressurization, isometric pressurization, and open-loop square wave commands at 0.25, 0.5, and 1 Hz. Experiments of sine wave tracking at 0.25, 0.5, and 1 Hz and continuous square wave tracking at 0.067 Hz are conducted using a sliding mode controller (SMC) derived from the model. The SMC achieves a steady-state error of 6 μm at multiple setpoints within the actuator's 17 mm stroke. Compared to a proportional-integral-derivative (PID) controller, the SMC root-mean-square (RMS) error, mean error, and absolute maximum error are reduced on average by 53%, 61%, and 44%, respectively, demonstrating the benefit of model-based approaches for controlling HAMs.
{"title":"Theoretical Control-Centric Modeling for Precision Model-Based Sliding Mode Control of a Hydraulic Artificial Muscle Actuator","authors":"Jonathon E. Slightam, M. Nagurka","doi":"10.1115/1.4049565","DOIUrl":"https://doi.org/10.1115/1.4049565","url":null,"abstract":"\u0000 Artificial muscles (AMs) traditionally rely on pneumatic sources of fluid power. The use of hydraulics can increase the power and force to weight and volume ratios of AM actuators. This paper develops a control-centric third-order single-input single-output (SISO) lumped-parameter dynamic model and sliding mode position controller based on Filippov's principle of equivalent dynamics for a braided hydraulic artificial muscle (HAM) actuator. The model predicts the nonlinear behavior of the HAM free contraction and captures the fluid and actuator nonlinear dynamic interactions in addition to the braid deformation. Model simulations are compared to experimental results for quasi-static pressurization, isometric pressurization, and open-loop square wave commands at 0.25, 0.5, and 1 Hz. Experiments of sine wave tracking at 0.25, 0.5, and 1 Hz and continuous square wave tracking at 0.067 Hz are conducted using a sliding mode controller (SMC) derived from the model. The SMC achieves a steady-state error of 6 μm at multiple setpoints within the actuator's 17 mm stroke. Compared to a proportional-integral-derivative (PID) controller, the SMC root-mean-square (RMS) error, mean error, and absolute maximum error are reduced on average by 53%, 61%, and 44%, respectively, demonstrating the benefit of model-based approaches for controlling HAMs.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"154 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73157620","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Proton exchange membrane (PEM) electrolyzer can produce gases at the pressure suitable for direct storage into metal hydride cylinders, bypassing compressors, and other auxiliary components. For direct storage into metal hydride containers, hydrogen gas's pressure and flowrate must be well controlled. However, the PEM electrolyzer's time-variant and nonlinear dynamics call for an adaptive control to maintain its output performance. Therefore, in this paper, a model-free relay-feedback autotuning approach is proposed to tune a proportional-integral (PI) controller online. The controller determines the voltage supply to the electrolyzer to track a certain current setpoint, which corresponds to a constant hydrogen production rate. A gain scheduling approach is developed to pick up the right controller gain at different setpoints, minimizing the tuning frequency. A self-assessment algorithm is developed to determine the situations where the autotuner should activate to update the PI parameters, thus allowing the control system to be tuned autonomously. The autotuning PI control is successfully tested with a PEM electrolyzer setup. Experiment results showed that autotuner with gain scheduling could tune the controller parameters producing a desired transient behavior and is adaptive to the variations in setpoint and operating conditions.
{"title":"Auto-Tuning Control of Proton Exchange Membrane Water Electrolyzer With Self-Assessment and Gain Scheduling","authors":"A. Keow, Zheng Chen","doi":"10.1115/1.4049365","DOIUrl":"https://doi.org/10.1115/1.4049365","url":null,"abstract":"\u0000 Proton exchange membrane (PEM) electrolyzer can produce gases at the pressure suitable for direct storage into metal hydride cylinders, bypassing compressors, and other auxiliary components. For direct storage into metal hydride containers, hydrogen gas's pressure and flowrate must be well controlled. However, the PEM electrolyzer's time-variant and nonlinear dynamics call for an adaptive control to maintain its output performance. Therefore, in this paper, a model-free relay-feedback autotuning approach is proposed to tune a proportional-integral (PI) controller online. The controller determines the voltage supply to the electrolyzer to track a certain current setpoint, which corresponds to a constant hydrogen production rate. A gain scheduling approach is developed to pick up the right controller gain at different setpoints, minimizing the tuning frequency. A self-assessment algorithm is developed to determine the situations where the autotuner should activate to update the PI parameters, thus allowing the control system to be tuned autonomously. The autotuning PI control is successfully tested with a PEM electrolyzer setup. Experiment results showed that autotuner with gain scheduling could tune the controller parameters producing a desired transient behavior and is adaptive to the variations in setpoint and operating conditions.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"8 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75967574","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper presents and experimentally evaluates a nested combined plant and controller optimization (co-design) strategy that is applicable to complex systems that require extensive simulations or experiments to evaluate performance. The proposed implementation leverages principles from Gaussian process (GP) modeling to simultaneously characterize performance and uncertainty over the design space within each loop of the co-design framework. Specifically, the outer loop uses a GP model and batch Bayesian optimization to generate a batch of candidate plant designs. The inner loop utilizes recursive GP modeling and a statistically driven adaptation procedure to optimize control parameters for each candidate plant design in real-time, during each experiment. The characterizations of uncertainty made available through the GP models are used to reduce both the plant and control parameter design space as the process proceeds, and the optimization process is terminated once sufficient design space reduction has been achieved. The process is validated in this work on a lab-scale experimental platform for characterizing the flight dynamics and control of an airborne wind energy (AWE) system. The proposed co-design process converges to a design space that is less than 8% of the original design space and results in more than a 50% increase in performance.
{"title":"Gaussian Process-Driven, Nested Experimental Co-Design: Theoretical Framework and Application to an Airborne Wind Energy System","authors":"Joe Deese, P. Tkacik, C. Vermillion","doi":"10.1115/1.4049011","DOIUrl":"https://doi.org/10.1115/1.4049011","url":null,"abstract":"\u0000 This paper presents and experimentally evaluates a nested combined plant and controller optimization (co-design) strategy that is applicable to complex systems that require extensive simulations or experiments to evaluate performance. The proposed implementation leverages principles from Gaussian process (GP) modeling to simultaneously characterize performance and uncertainty over the design space within each loop of the co-design framework. Specifically, the outer loop uses a GP model and batch Bayesian optimization to generate a batch of candidate plant designs. The inner loop utilizes recursive GP modeling and a statistically driven adaptation procedure to optimize control parameters for each candidate plant design in real-time, during each experiment. The characterizations of uncertainty made available through the GP models are used to reduce both the plant and control parameter design space as the process proceeds, and the optimization process is terminated once sufficient design space reduction has been achieved. The process is validated in this work on a lab-scale experimental platform for characterizing the flight dynamics and control of an airborne wind energy (AWE) system. The proposed co-design process converges to a design space that is less than 8% of the original design space and results in more than a 50% increase in performance.","PeriodicalId":54846,"journal":{"name":"Journal of Dynamic Systems Measurement and Control-Transactions of the Asme","volume":"71 1","pages":""},"PeriodicalIF":1.7,"publicationDate":"2021-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88491428","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：