786 research outputs found
Optimal Sub-References for Setpoint Tracking: A Multi-level MPC Approach
No full textWe propose a novel method to improve the convergence performance of model predictive control (MPC) for setpoint tracking, by introducing sub-references within a multilevel MPC structure. In some cases, MPC is implemented with a short prediction horizon due to limited on-line computation capacity, which could lead to deteriorated dynamic performance. The introduced multi-level optimization method can generate proper sub-references for the MPC setpoint tracking problem, and efficiently improve the dynamic performance. In the higher level a specific performance criterion is taken as the objective, while explicit MPC is utilized in the lower level to represent the control input. The generated sub-references are then used in MPC for the real system with prediction horizon restrictions. Setpoint-tracking MPC for linear systems is used to illustrate the approach throughout the paper. Numerical simulations show that MPC with sub-references significantly improves the convergence performance compared with regular MPC with the same prediction horizon. Thus, it can be concluded that MPC with sub-references has a high potential to tackle more complicated control problems with limited computation capacity.Transport and PlanningControl & SimulationDelft Center for Systems and Contro
Combined MPC and reinforcement learning for traffic signal control in urban traffic networks
No full textIn general, the performance of model-based controllers cannot be guaranteed under model uncertainties or disturbances, while learning-based controllers require an extensively sufficient training process to perform well. These issues especially hold for large-scale nonlinear systems such as urban traffic networks. In this paper, a new framework is proposed by combining model predictive control (MPC) and reinforcement learning (RL) to provide desired performance for urban traffic networks even during the learning process, despite model uncertainties and disturbances. MPC and RL complement each other very well, since MPC provides a sub-optimal and constraint-satisfying control input while RL provides adaptive control laws and can handle uncertainties and disturbances. The resulting combined framework is applied for traffic signal control (TSC) of an urban traffic network. A case study is carried out to compare the performance of the proposed framework and other baseline controllers. Results show that the proposed combined framework outperforms conventional control methods under system uncertainties, in terms of reducing traffic congestion. Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Team Bart De SchutterControl & SimulationDelft Center for Systems and Contro
Solar Forecasting Requirements for Buildings MPC
No full textAbstractModel Predictive Control (MPC) is a key issue to deal with Net Zero Energy Buildings (NZEBs) and Communities. The application of an MPC scheme in buildings requires an accurate building model and a weather forecast.How accurate needs to be a weather forecast is a common question for MPC applications.In this work a comfort tracking problem is solved through a receding horizon MPC Scheme. To take into account both solar gains and thermal inertia a second order state space model is assumed for a generic building. The Scheme is applied for Almería's climate (South Spain)to a generic building for two different seasons of the year.The impact of forecast accuracy on comfort tracking performance is assessed through a pseudorandom heteroskedastic function. The result is generalized for a community of buildings.It is shown that solar radiation forecast uncertainty has the bigger impact on the MPC performance and a methodology to make a quantitative evaluation of the forecast requirements is provided
Learning safety in model-based Reinforcement Learning using MPC and Gaussian Processes
No full textThis paper proposes a method to encourage safety in Model Predictive Control (MPC)-based Reinforcement Learning (RL) via Gaussian Process (GP) regression. The framework consists of 1) a parametric MPC scheme that is employed as model-based controller with approximate knowledge on the real system's dynamics, 2) an episodic RL algorithm tasked with adjusting the MPC parametrization in order to increase its performance, and 3) GP regressors used to estimate, directly from data, constraints on the MPC parameters capable of predicting, up to some probability, whether the parametrization is likely to yield a safe or unsafe policy. These constraints are then enforced onto the RL updates in an effort to enhance the learning method with a probabilistic safety mechanism. Compared to other recent publications combining safe RL with MPC, our method does not require further assumptions on, e.g., the prediction model in order to retain computational tractability. We illustrate the results of our method in a numerical example on the control of a quadrotor drone in a safety-critical environment.Team Azita DabiriDelft Center for Systems and Contro
MPC-based COLREGS Compliant Collision Avoidance for a Multi-Vessel Ship-Towing System
No full textCollision avoidance plays a vital role in autonomous vehicle systems. As the complexity and scale of missions increase, multi-vehicle systems are adopted in practice. However, there is limited research on collision avoidance of a physically interconnected multi-vessel system. This paper proposes a control scheme for tugboats to tow a ship in congested port areas ensuring collision avoidance that is compliant with COLREGS. The Model Predictive Control (MPC) strategy is used to optimize the towing angles, towing forces, and tugboats’ thruster forces and moment. The COLREGS rules are integrated into the ship reference system by altering predefined waypoints to guide the towing system in a safe and lawful way. By designing the cost function for the ship and tugboats in the MPC controller system, the proposed control scheme makes the ship-towing system stay away from the obstacles and follow the calculated waypoints, achieving collision avoidance. Simulation experiments indicate that the proposed method can deal with static and dynamic obstacle situations in complex water traffic environments, and the collision avoidance operations comply with the COLREGS rules.Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Transport Engineering and Logistic
Distributed MPC for Large Freeway Networks Using Alternating Optimization
No full textThe Model Predictive Control (MPC) framework has shown great potential for the control of Variable Speed Limits (VSLs) and Ramp Metering (RM) installations. However, the implementation to large freeway networks remains challenging. One major reason is that, by considering the VSLs to be discrete decision variables, an extremely difficult Mixed Integer Nonlinear Programming (MINLP) optimization problem has to be solved within every controller sampling interval. Consequently, many related papers relax the MINLP problems by considering the VSLs to be continuous variables. This paper proposes two novel MPC algorithms for coordinated control of discrete VSLs and continuous RM rates that do not make this relaxation. The proposed algorithms use a distributed control architecture and an alternating optimization scheme to relax the MINLP optimization problems but still consider the VSLs as discrete variables and, hence, offer a trade-off between computational complexity and system performance. The performance of the proposed algorithms is evaluated in a case study. The case study shows that relaxing the VSLs to be continuous variables with a distributed architecture results in a significant performance loss. Furthermore, both proposed algorithms have a lower computational complexity than the more conventional centralized approach and, as a result, they do manage to solve all optimization problems within the sampling intervals. Moreover, one of the proposed algorithms has a system performance that is remarkably similar to the optimal performance of the centralized approach.Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Team Bart De Schutte
Cooperative adaptive cruise control: Using information from multiple predecessors in combination with MPC
No full textCooperative adaptive cruise control (CACC) makes the vehicle follow its predecessor at a close but safe distance, and uses information received from other vehicles to accomplish this task. In literature and in practice, the control method mostly applied for CACC is proportional integral derivative (PID) control, possibly with some refinement for gear shifting or comfort. The control method called model predictive control (MPC) can also be used for CACC, and from literature it appears to be more promising than PID, because of its ability to anticipate future situations and to implement constraints directly into the control algorithm. MPC applies the first input of a control input sequence that optimises a performance index calculated from predicted system behaviour, based on a prediction model, subject to operational constraints, in a receding horizon approach. Furthermore, literature has shown that with PID the use of state information from the second predecessor or the platoon leader, in addition to the direct predecessor’s states, can improve the CACC performance. Therefore, in this thesis the approach of using such additional communicated information from either the second predecessor or the platoon leader is combined with the use of MPC as control method. The goal is to investigate whether any of these two configurations give an increase in performance compared with similar configurations with PID as control method, and compared with a more basic configuration that uses just the direct predecessor’s state information with either MPC or PID. Also, the possibly added value of using communicated predicted states, in addition to current states, with MPC is investigated. The CACC controllers are designed to control the throttle, the brakes, and the gears, subject to operational constraints on acceleration, velocity, and vehicle-to-vehicle distance. The PID-based CACC controller contains a proportional feedback of the errors in velocity, position, and acceleration, combined with an automatic transmission scheme, and the control input is restricted at time instants at which a constraint is (almost) violated. The MPC-based CACC controller at each time step minimises the expected errors in position and velocity and the corresponding input variation. The MPC prediction model is obtained by approximating a nonlinear vehicle model by a piecewise affine (PWA) model, and converting the MPC optimisation problem into a mixed integer linear programming (MILP) problem. In this project, tuning is done by applying simulated annealing for a scenario involving four CACC-controlled vehicles following a platoon leader. Then, the tuned controllers are implemented in a validation scenario comprising a larger platoon undergoing a longer simulation. The results from simulating this validation scenario show that the PID-based CACC controller has a low responsiveness, compared with MPC, and lets the first two vehicles crash. With MPC several peaks and oscillations in throttle/brake input and acceleration occur, and it is expected that with the MPC-based CACC controllers as designed and tuned here, string stability will not always be achieved for increasing platoon lengths. It is expected that properly retuning will result in better performing controllers. However, due to limited time, this retuning could not be performed within the scope of this project, and is therefore left as a recommendation. Therefore, only preliminary conclusions can be formulated, which are that MPC should be preferred over PID as a control method for CACC, because it is safer. Moreover, with MPC it should be preferred to, in addition to the current states of the direct predecessor, at least use the current states of the second predecessor and/or the predicted future states from the direct predecessor, in order to achieve better string stability.Delft Center for Systems and ControlMechanical, Maritime and Materials Engineerin
A Comparative Study of Applying Active-Set and Interior Point Methods in MPC for Controlling Nonlinear pH Process
No full textA comparative study of Model Predictive Control (MPC) using active-set method and interior point methods is proposed as a control technique for highly non-linear pH process. The process is a strong acid-strong base system. A strong acid of hydrochloric acid (HCl) and a strong base of sodium hydroxide (NaOH) with the presence of buffer solution sodium bicarbonate (NaHCO3) are used in a neutralization process flowing into reactor. The non-linear pH neutralization model governed in this process is presented by multi-linear models. Performance of both controllers is studied by evaluating its ability of set-point tracking and disturbance-rejection. Besides, the optimization time is compared between these two methods; both MPC shows the similar performance with no overshoot, offset, and oscillation. However, the conventional active-set method gives a shorter control action time for small scale optimization problem compared to MPC using IPM method for pH control
Nonlinear control of electronic converters: Fast and optimal control using sampling-driven nonlinear MPC
No full textThis thesis considers the design of the sampling-driven nonlinear model predictive controller (SD-NMPC) for power electronic converters. Most MPC schemes that are used nowadays lack in the application for systems with high sampling frequencies because of their high computational time, especially when the system is nonlinear. This is, for example, the case for most of the power electronic systems. Recently, a new approach of MPC control is proposed which is based on sampling control inputs from the input space. In contrast with other MPC approaches, this approach is focused on suboptimal control instead of the optimal control in other MPC approaches which usually results in a less optimal control sequence, but, furthermore, it still benefits from all the advantages of the MPC control method. This thesis describes several efficient and fast methods to determine the offline part of the SD-NMPC controller, which are the local controller with corresponding DOA and the initial control sequence steering the system into this DOA, using only linear optimization tools. Among all these methods, we determined the best option for the buck-boost converter with resistive load and the VSI controlling a PMSM. Besides the offline part, we also improved the online part of the SD-NMPC control method which resulted in a fast and optimal controller for these systems. This method has shown all the benefits of the implementation of an MPC controller. Based on the results, we could conclude that the SD-NMPC control method is a fast and efficient way to apply an MPC method to power electronic converters.https://doi.org/10.4121/14223062.v1 Matlab code supplementary to this MSc thesisMechanical Engineering | Systems and Contro
A novel bi-level temporally-distributed MPC approach: An application to green urban mobility
No full textModel predictive control (MPC) has been widely used for traffic management, such as for minimizing the total time spent or the total emissions of vehicles. When long-term green urban mobility is considered including e.g. a constraint on the total yearly emissions, the optimization horizon of the MPC problem is significantly larger than the control sampling time, and thus the number of the variables that should be optimized per control time step becomes very large. For systems with dynamics that involve nonlinear, non-convex, and non-smooth functions, including urban traffic networks, this results in optimization problems that are computationally intractable in real time. In this paper, we propose a novel bi-level temporal distribution of such complex MPC optimization problems, and we develop two mathematically linked short-term and long-term MPC formulations with small and large control sampling times that will be solved together instead of the original complex optimization problem. The resulting bi-level control architecture is used to solve the two MPC formulations online for real-time control of urban traffic networks with the objective of long-term green mobility. In order to assess the performance of the bi-level control architecture, we perform a case study where a rough version of the model of the urban traffic flow, S-model, is used by the long-term MPC level to estimate the states of the urban traffic networks, and a detailed version of the model is used by the short-term MPC level. The results of the simulations prove the effectiveness (with respect to the objective of control, as well as computational efficiency) of the proposed bi-level MPC approach, compared to state-of-the-art control approaches.Control & SimulationTransport and PlanningTeam Bart De SchutterDelft Center for Systems and Contro
- …
