Offset-Free Model Predictive Control of a Heat Pump

This work presents an offset-free model predictive control (OF-MPC) design for energy-efficient control of a heat pump. A model developed from a combination of first-principles and empirical components with parameters estimated using real heat pump data is used as a test bed for the implementation of the MPC design. Open-loop system dynamics are examined first to design and implement the appropriate control structure. Next, a linear model is identified using appropriate step test simulations on the detailed heat pump model. Subsequently, a model predictive controller formulation, which eliminates the tradeoff between tracking and energy objectives, is designed. The MPC includes an augmented model (including disturbance states) and an associated Luenberger observer to estimate the disturbance (plant-model mismatch at steady state). Simulation results subject to realistic disturbances and measurement noise demonstrate that energy savings can be achieved (anywhere between 1.7% and 1.9%), while preserving a s...

Glycemia Regulation Considering Offset-free MPC with Pulse Inputs and Parameter Variations

Trajectory tracking control of a VTOL unmanned aerial vehicle using offset-free tracking MPC

Flexible operation of coal-fired power plants is becoming increasingly necessary for successful integration of large-scale renewable power generation into the power grid. The maximum ramp rate and the number of load cycles are generally limited by the thermal stress experienced by the boiler pressure parts, turbine metallurgy and creep and fatigue of critical thick-walled components Main steam temperature is a critical operating parameter that must be controlled within acceptable limits for safe operation. Main steam temperature deviation beyond acceptable limit has impact on boiler pressure parts and turbine material of construction due to creep and fatigue effect. Base load operating units do not require steep ramp rate and hence recommended ramping rates are kept low within the safe operating zone in comparison to the flexible operation of the units with wide range load change width. Thermal stresses are caused by the temperature changes inside the thick-walled components and turbine steam admission parameters. Hence, the quality of main steam temperature control plays a vital role in flexible operation of the coal fired units. Conventional cascaded PID temperature control loop architecture performs well at steady state condition within a limited variation of load change at low ramp rate but it acts slowly and performs poorly at transient operating conditions of flexible operation of the boiler turbine with wide range load variation and load cycle with high ramp rate and remains far from rated conditions. In this paper, a Multi-Input Multi-Output (MIMO) Non-linear Model Predictive Control (MPC) design for regulation of the main steam temperature of a Once-Through supercritical Boiler is proposed. The controller is based on a non-linear dynamic model which incorporates dynamics of the variables of interest. It has the capability to operate effectively across a wide load range while maintaining main steam temperature within acceptable limits. A notable advancement in this design of MPC is the incorporation of coal flow demand and feedwater flow demand as additional control inputs alongside primary and secondary spray flows. In simulation test cases, the MPC controller demonstrates satisfactory performance and computational efficiency.

In this paper, constrained model predictive control (MPC) designs for a class of structured uncertain time-delay systems have been developed, where state delays as well as a random input delay have been taken into account. By replacing the control strategy with a freer one, an improved MPC design with larger initial feasible region has been developed. Furthermore, by analyzing the obtained algorithm, some useful properties of the solutions to the MPC optimization problem have been established. With these properties, a variant of the obtained algorithm with dramatically reduced number of online optimizing variables and hence the online computational burden has also been developed. However, the control performance degrades slightly. The two algorithms haven been proved to stabilize the closed loop system in the mean square sense and to guarantee the satisfaction of constraints. Finally a numeric example is given to illustrate the proposed results.

In this paper, a model predictive control (MPC) design method is proposed for Hammerstein systems with multivariable nonlinearities. Iterative inversion is introduced such that classical linear MPC design can be used. The condition for which the numerical inversion is guaranteed to converge is derived. From this condition, the input space at which the nonlinear function is invertible can be obtained, which poses extra input constraints for MPC design. The overall constraints are transformed and approximated by a set of linear constraints by means of convex hull construction. Illustrative examples are presented to demonstrate the effectiveness of the proposed control method.

Modeling, estimation, and control in large-scale urban road networks with remaining travel distance dynamics

Modelling And Predictive Control Of Inverted Pendulum

An industrial case study on the combined identification and offset-free control of a chemical process

ABSTRACTThis paper presents a nonlinear model predictive control (MPC) formulation for obstacle avoidance in high-speed, large-size autono-mous ground vehicles (AGVs) with high centre of gravity (CoG) that operate in unstructured environments, such as military vehicles. The term ‘unstructured’ in this context denotes that there are no lanes or traffic rules to follow. Existing MPC formulations for passenger vehicles in structured environments do not readily apply to this context. Thus, a new nonlinear MPC formulation is developed to navigate an AGV from its initial position to a target position at high-speed safely. First, a new cost function formulation is used that aims to find the shortest path to the target position, since no reference trajectory exists in unstructured environments. Second, a region partitioning approach is used in conjunction with a multi-phase optimal control formulation to accommodate the complicated forms the obstacle-free region can assume due to the presence of multiple obstacles in the prediction horizon in an unstructured environment. Third, the no-wheel-lift-off condition, which is the major dynamical safety concern for high-speed, high-CoG AGVs, is ensured by limiting the steering angle within a range obtained offline using a 14 degrees-of-freedom vehicle dynamics model. Thus, a safe, high-speed navigation is enabled in an unstructured environment. Simulations of an AGV approaching multiple obstacles are provided to demonstrate the effectiveness of the algorithm.

Although Model Predictive Control (MPC) has been widely investigated for energy efficient climate control of buildings, most prior works have neglected humidity in the problem formulation and performance evaluations. A climate control algorithm that ignores humidity cannot be used in practice, especially in hot-humid climates. Apart from the discomfort of occupants, high humidity over long periods will lead to issues such as mold growth, adversely impacting occupant health. In this paper, we provide an MPC formulation that explicitly accounts for humidity constraints in a principled manner. We show how to construct data-driven low order models of a cooling and dehumidifying coil that can be used in the MPC formulation. The resulting controller's performance is tested in simulation using a plant that differs significantly from the model used by the optimizer. In spite of the large plant model mismatch, the proposed MPC controller performs well. Humidity constraints are seen to be active for a large part of the day, especially in the summer. MPC formulations that ignore humidity would lead to a poor indoor climate in these situations.

This paper presents a remote design of model predictive control (MPC) system with model diagnosis and model adjustment. The design incorporates the model diagnosis and model adjustment into MPC. Furthermore, the MPC, model diagnosis and model adjustment are integrated in the remote controllable MPC system. The technologies of .NET, XML-RPC, Ajax asynchronous interaction and plugin are adopted. To enhance the MPC operating performance, the diagnosis module and adjustment module are designed as independent plugins of MPC. Implementations of the design will benefit the practical application of MPC as well as promoting the acceptability and usage of MPC in industry.

To overcome the convective time delay issue for active closed-loop flow control, a model based predictive control algorithm is analyzed. From a controls perspective, forms of internal model control or model predictive control can be used to accommodate and minimize the effect of systems with pure time delays. Moreover, the Smith predictor is a commonly employed control technique to negate the pure time delay in a closed-loop system. This form of model predictive control is applied to the asymmetric vortex problem of an axisymmetric forebody (specifically a von Karman ogive with fineness ratio of 3.5). The full-order Navier-Stokes equations are numerically solved on the forebody at a high angle of attack and provide the plant process. Small port and starboard blowing patches are used to introduce fluidic disturbances at the nose of the ogive to augment the global flow state and produce a deterministic vortex state. As the active flow control technique exploits a convective instability, a convective time delay exists. Linear-time-invariant and non-linear time-invariant models are developed from the open-loop dynamics. A Smith predictor is employed within the full order CFD simulation. The results of the predictive control are compared to open-loop and model-free closed-loop behavior. It is shown that the predictive control developed in this paper while very suitable for control of this type of flow, is very sensitive to modeling uncertainties.

Conventional feedback controllers (of PID type) are used to control two fundamental physical state of a batch fermentation process, namely pH and temperature. To improve the efficient operation of the batch process, the set points of PID controllers are updated using an online optimization technique. Using multiple model approach, the whole operating range of the fermentation process is approximated with local linear models. The optimization problem is solved for each local linear model using a linear model predictive control (MPC) design. Each local linear MPC design computes one local optimum trajectory for pH and one for temperature. Local optimum trajectories are multiplied by their relative interpolation functions and the summation of these gives the global optimum trajectories which are used as set points for PID controllers. At every given time instant the entire optimization is repeated giving new set points for the PID controllers. Thus a relative simple method is capable for optimal control of a very complex nonlinear batch fermentation process using simple PID controllers.

Model Predictive Control (MPC) is used mainly for specific handling of constraints. This paper presents about the design of dynamic linear controller for a binary distillation column. The design is based on MPC, which is based on prediction of control variable. We have used Ogunnaike and ray 3×3 process and Doukas and Luyben 4×4 for the distillation column. We have implemented manipulated and controlled variable for 3×3 process using MPC. We have also analyzed the manipulated and controlled variables for 4×4 process using MPC.