Model predictive control (MPC) has shown great potential in improving building performance and saving energy. However, after over 20 years of research, it is yet to be adopted by the industry. The ...difficulty of obtaining a sufficient control-oriented model is one major factor that hinders the application. In particular, what data is required to build the model and what control performance can be expected with a certain model remain unclear. This study attempts to uncover the underlying reasons and guide future research to tackle the challenges. It starts by clarifying a finer categorization of past studies with respect to both modeling methods and modeling purposes. An extended Level of Detail (LoD) framework is proposed to quantify the data usage in each study. Accordingly, meta-analyses are conducted to compare the data requirements of different modeling categories. The criteria and approaches for model performance evaluation are summarized and classified into validation and verification methods, followed by a discussion about the relationship between the model and control performance. The critical review provides new perspectives on the data requirements and performance evaluation of control-oriented models. Ultimately, the paper concludes with five directions for future research to bridge the gaps between data requirements, model performance, and control performance.
•A critical review to promote the MPC application in buildings.•An extended level of detail approach to quantify data usage in control-oriented model.•Data requirements decided by modeling methods, purposes and building systems.•Model complexity and data availability to be balanced for control-oriented modeling.•Bridging model and control performance is essential for commercial application.
In this paper, a semi-empirical Lithium-iron phosphate-graphite battery aging model is identified over data mimicking actual cycling conditions that a hybrid electric vehicle battery encounters under ...real driving scenarios. The aging model is then used to construct the severity factor map, used to characterize relative aging of the battery under different operating conditions. This is used as a battery degradation criterion within a multi-objective optimization problem where battery aging minimization is to be achieved along with fuel consumption minimization. The method proposed is general and can be applied to other battery chemistry as well as different vehicular applications. Finally, simulations conducted using a hybrid electric vehicle simulator show how the two modeling tools developed in this paper, i.e., the severity factor map and the aging model, can be effectively used in a multi-objective optimization problem to predict and control battery degradation.
•Battery aging model for hybrid electric vehicles using real driving conditions data.•Development of a modelling tool to assess battery degradation for real time optimization.•Development of an energy management strategy to minimize battery degradation.•Simulation results from hybrid electric vehicle simulator.
Predictive models serving as virtual sensors for online optimization and feedback control of diesel engines is gaining increasing attention. However, existing prediction models fall short in ...simultaneously achieving high prediction accuracy and fast computational speed. In this paper, a novel machine learning algorithm called deep autoencoding support vector regression (DASVR) was proposed, which combines the powerful non-linear feature extraction capability of artificial neural network (ANN) with the good adaptability of support vector regression (SVR) to low-dimensional input spaces. ANN-based autoencoder is firstly employed to extract features from the original high-dimensional input space, forming a low-dimensional latent variable space. SVR is then employed to perform non-linear mapping between latent variables and target output variables. Experiments were conducted on a 16-cylinder marine diesel engine under different load, rail pressure, and injection timing conditions. Prediction models for fuel consumption, NOx, PM, HC, and PM emissions were established based on experimental data and DASVR algorithm. The maximum error of DASVR-based models for the five desired output variables is less than 3.8 % under different operating conditions. DAVSR-based predictive models outperform conventional ANN-based and SVR-based models in terms of both prediction accuracy and real-time capability, and can theoretically meet the real-time requirement below 3000 r/min.
•Proposal for a novel machine learning algorithm - DASVR.•Reduce the dimensionality of input space by encoder-decoder neural network.•Map the latent variable space to the output space by SVR.•DASVR outperforms ANN and SVR in both accuracy and speed.•Detailed evaluation of prediction accuracy and speed for both BSFC and emissions.
Model-based robot controllers require customized control-oriented models, involving expert knowledge and trial and error. Remarkably, the Koopman operator enables the control-oriented model ...identification through the input-output mapping set, breaking through the barriers of the customization services. However, in recent years, research on Koopman-based robot control has mostly focused on lifting function construction, deviating from the original intention of improving the controller performance. Thus, we propose a robot controller autogeneration framework using the Bayesian-based Koopman operator, significantly releasing labor and eliminating the design obstacle. First, we introduce the Koopman-based system identification method and offer the basic lifting function design criteria. Then, a Bayesian-based optimization strategy with resource allocation is designed, which allows for the simultaneous optimization of the lifting function and the controller. Next, taking model-predictive control (MPC) as an example, a mission-oriented controller autogeneration framework is developed. Simulation and experimental results indicate that, under various robots and data sources, the proposed framework can effectively generate the robot controllers and perform with a far greater level of mission accuracy than the unoptimized Koopman-based MPC. Meanwhile, the proposed technique exhibits an obvious compensation effect against disturbances, demonstrating its practicability in robot control.
Magneto-rheological dampers are an effective technology to control the damping coefficient of a semi-active suspension. Most of the contributions in literature propose damper models to be used in ...simulation, or as damping force virtual sensors in control applications. Typically, phenomenological models or complex black-box approaches, relying on Neural Networks, are employed. In this work, we propose a semi-active MR model based on a Hammerstein–Wiener scheme, meant not only for force estimation but also – in a more genuinely control-oriented perspective – to be proactively used in the suspension controller design. Despite being a black-box model, each component is shown to serve for the characterization of a specific feature of the MR damper, and its identification is done thanks to an ad hoc design of experiments. In particular, the Wiener part of the model is shown to be essential for the proper modelling of the magnetization dynamics of the magneto-rheological fluid, which usually is a neglected aspect in control-oriented models. The proposed scheme is validated on a testbench using realistic road solicitations.
The thermal management system (TMS) in electric vehicles (EVs) is a comprehensive system that integrates an air conditioning system for the cabin, a temperature control system for the battery, and a ...cooling system for the motor. The currently used PI control strategy can only meet the basic TMS functions and cause high energy consumption. In this paper, we present a novel model predictive control (MPC) strategy for the TMS to optimize operational performance in real time. Different from the independent PI control for the individual components, MPC can predict future operation conditions and provide the optimal operating inputs in advance. A complete control-oriented model for MPC is developed, and the MPC strategy is designed to minimize the total power consumption of the TMS under the control-oriented model and constraints. The evaluation is carried out under several cases including the fixed ambient temperature, realistic ambient temperature, and different vehicle speeds. The results showed that the novel MPC strategy saved energy consumption by 5.9%–10.3% in these cases when compared to the PI strategy, demonstrating the effectiveness and feasibility of the proposed MPC control.
•Proposed a model predictive control strategy for thermal management systems in EVs.•Developed a control-oriented model based on data-driven methods and physical principles.•Minimizes total power consumption while maintaining cabin and battery temperatures.•Results demonstrate the effectiveness of the MPC strategy under various conditions.
Control of laser power to improve part quality is critical for fabrication of complex components via Laser Powder Bed Fusion (LPBF) additive manufacturing (AM) processes. If the laser power is too ...low, it will result in a small melt pool and lack of fusion; on the other hand, if the laser power is too high, it will result in keyhole and material evaporation. This paper examines a model-based feed-forward control for laser power in LPBF to improve build quality by avoiding the onset of keyhole formation or reducing over-melting. First, an analytical, control-oriented model on the dynamics of melt-pool cross-sectional area in scanning a multi-track part was developed, and then a nonlinear inverse-dynamics controller was designed to adjust laser power such that the melt-pool cross-sectional area can be regulated to a constant set point during the build process. The resulting control trajectory on laser power from the simulated closed-loop controller was then implemented in a LPBF process as a feed-forward (FF) controller for laser power. Multiple bead-on-plate samples of Inconel 625, with different number of tracks and track lengths, were then built on an EOSINT M 280 AM system to evaluate the performance of the resulting FF-Analytic controller. Experimental results demonstrated that the proposed FF-Analytic control of laser power was able to avoid the onset of keyhole formation that occurred under a constant laser power for certain samples. Furthermore, the proposed FF-Analytic control was demonstrated to have significantly reduced over-melting at the returning ends of the laser scan path in scanning a multi-track part compared to applying a constant laser power, albeit with some over-compensation due to modeling imperfection. Overall, the proposed FF-Analytic control of laser power had 23–40% lower average error rate than applying a constant laser power in regulating the melt-pool cross-sectional area to a constant reference value, in terms of measurements of cross-sections at track ends.
•A control-oriented model for a vehicle driveline with DCT is developed with practical aspects.•A torque observer is designed to estimate torque through both clutches and the drive shaft in a ...DCT.•The observer is based on the control-oriented model and uses data available on production cars.•The effectiveness of the developed model and the torque observer is demonstrated by experiments.•The proposed methods of modeling and torque estimation are useful for accurate powertrain controls.
Over the years, dual-clutch transmission (DCT) has demonstrated its higher efficiency and superior shift performances over other types of transmissions, and has been increasingly used in modern mass-produced vehicles. However, due to the absence of the smoothing effects of torque converters, vehicles with DCT are easily exposed to driveline oscillations that lead to poor driving quality, especially during gear shifts. Therefore, torque transfer through the driveline should be controlled with great care by two clutches and engine to achieve the DCT's outstanding performance. The main obstacle to the accurate torque control is its lack of adequate sensors in production vehicles. Thus, the objectives of this paper are two-fold. First, a control-oriented model with practical concerns is implemented for DCT drivelines, aiming to accurately describe the powertrain oscillations that should be suppressed by the torque control. Secondly, a real-time torque monitoring strategy based on the proposed model is suggested to deal with the absence of torque sensors. The primary task of the torque estimator is to concurrently estimate the torque transmitted through both clutches and drive shaft by using only readily available data from production cars. The developed torque estimator is verified through multiple experiments under various driving conditions.
The open‐cathode design with simultaneously integrated air supply and coolant flow systems can significantly simplify the structure of proton exchange membrane fuel cell (PEMFC). Typically, the ...performance of open‐cathode PEMFC system is highly influenced by temperature and oxygen excess ratio which can be regulated by the air flow fan. The axial air flow fan is the only actuator which is responsible for the oxygen supply to the cathode and cooling air at the same time. In this study, oxygen mass flow rate across cathode channels is regarded as the only control variable of the open‐cathode FC system. Air flow is regulated based on a set of optimal oxygen excess ratios in different stack currents for power optimization and overheating protection. However, the air flow rate is determined through a sensor installed to measure the air flow rate in the cathode channels. The control strategy simultaneously focuses on performance optimization with overheating protection and air starvation prevention, whereas the fuel cell operates its full range. The efficacy of the proposed method and control is verified through experiments with a small open‐cathode fuel cell system with a real‐time platform of NI (National Instruments) devices.
In this paper, a control-oriented model of a solid oxide fuel cell system is formulated and analyzed in detail. First, a lumped model based on first principle laws is formulated and tuned using ...experimental data coming from a real solid oxide fuel cell system test bench. The model calibration is carried out based on an optimization approach to minimize the error between the experimental data and the model one. To systematically analyze the system behavior, an equilibrium point analysis is formulated and developed. The analysis results show the maximum steady-state electrical power under each constant stack temperature. This will allow to appropriately select operation points during the system operation. Secondly, Lyapunov's theory is used to characterize the local stability of the equilibrium points. The results show that the equilibrium points are locally stable. Besides, comparison between the initial nonlinear model with the linearized model is performed to show the efficacy of the linearised model analysis. Finally, the frequency response of the linearized model is performed. This analysis provides key information about control system design in order to efficiently operate the solid oxide fuel cell system.
•Formulation, calibration and validation of a lumped SOFC model.•Steady-state characterization of the SOFC system model.•Local stability and step response analysis of the SOFC system based on the model.•Frequency response analysis of the SOFC model. Suggestions to design controllers.