Model predictive control (MPC) is a multivariable control algorithm commonly used when tracking a reference trajectory is the primary goal. It minimizes the steady-state tracking error, increases the closed-loop bandwidth, and enables the controller to track a reference signal, the major requirements for nanopositioning applications. These capabilities motivated this research on improving the positioning performance of the piezoelectric tube scanner (PTS) in an atomic force microscope (AFM) to achieve better imaging at high scanning speed. This paper surveys the effectiveness of an MPC controller for AFM imaging by considering its different features and comparing its results with those obtained from the in-built proportional integral (PI) controller.
We consider the optimal flows of intra-day gas supply for the production and pipeline network for Adelaide, Melbourne and Sydney. We model the demand in the network of three major cities as a system of connected three state Markov chains subject to shortfall, pipeline capacity and production capacity constraints. A comparison with a static optimisation model is provided to show the improved capacity for the analysis of demand satisfaction in this model.
A variable-gain model-independent control law is proposed to solve the problem of rendezvous to a leader for a directed network of Euler-Lagrange agents. A sufficiency condition for stability is developed, requiring centralised design of the two control gains. Compared to existing results which use constant-gain model-independent controllers for directed networks, our work has two key differences. Firstly, the damping term may begin at zero, and increases only if rendezvous has not been achieved. Constant-gain controllers use conservative gain values which negatively impact convergence speed. Secondly, we introduce a novel method of analysing the Lyapunov derivative, which provides useful and unique insight into analysis of variable-gain controllers for multi-agent systems. Simulations are provided to show the effectiveness of the controller.
A Network control system (NCS) is a control system in which communication between subsystems takes place over a digital network. Numerous results exist in the literature on modelling, analysis and design of NCSs in the presence of specific communication constraints such as packet dropouts, delays, data rates, quantization, etc. However, when analysing NCSs implemented on real physical networks, the existing results are based on restrictive assumptions. We consider NCSs over WirelessHART, the first international standard for industrial process control. With the goal of closing the gap between theory and practice, we propose for the first time a hybrid control-oriented model of WirelessHART NCSs under source routing configuration. Moreover, asymptotic and exponential stability results are presented under reasonable conditions.
In this paper, a self-tuning linear quadratic supervisory regulator using a large-signal state estimator for a diesel driven generator set is proposed. The regulator improves operational efficiency, in comparison to current implementations, by (i) automating the initial tuning process and (ii) enabling automated retuning capabilities. Utilizing a first principles-based nonlinear model detailed in [1], the procedure is demonstrated through simulations after real system measurements have been used for parameter identification. The regulator is able to suppress load-induced variations successfully throughout the operating range of the diesel generator.
This paper proposes a fault detection algorithm for the roll dynamic model of a generic delta-wing aircraft. It is assumed that the system model has some noise/uncertainties and the measured roll angle and roll rate used by the control law are faulty/under attack. The proposed fault detection algorithm employs a fault detection estimator and adaptive thresholds to detect the occurrence of a fault in the sensor measurements. Simulation results are presented to show the performance of the proposed algorithm.
The compass-gait biped model is a unique, yet simple model, often used as a test bench for new control strategies in the field of legged locomotion. This paper presents a novel continuous mathematical model for compass-gait biped robots, modeling the discontinuities in impact with a nonlinear function. Natural oscillation is defined as a free response under the damping compensation to achieve persistent oscillation. The phenomenon of natural oscillation was previously exploited in the linear dynamics of multi-segmental locomotor, and it is now extended to this humanoid system with the contact force from the ground modelled with a nonlinear bump function.
This papers considers parameter estimation of a constant gain system with only binary measurements of the input and output signals available. It is assumed that the thresholds of the binary quantizers can be adapted. Depending on the prior information about the stochastic nature of the input signal and the presence of output noise, several estimation schemes based on empirical measures and stochastic approximation algorithms are proposed and analysed. It is shown that the algorithms converge almost surely to the true gain of the system, and the algorithms are illustrated in simulation examples.
With the penetration of distributed solar photovoltaic (PV) generations, problems of degraded power quality and voltage oscillations are caused by unpredictable solar energy in residential microgrid. To alleviate such problems, the employment of distributed battery energy storage (BES) has been proposed and widely studied. However, with the present low renewable feed-in tariffs in many countries, microgrid users are discouraged to install expensive BES if they are not allowed to make energy-trading with the BES. In this paper, a residential microgrid power market is proposed to encourage the energy-trading of users with BES. As excessive energy-trading can exacerbate rather than alleviating the afore-mentioned problems associated with intermittent solar energy, a dissipativity-based distributed control approach is developed to coordinate individual users to achieve desirable microgrid-wide performance.
We propose a dissipativity based robust stability analysis and control design for plantwide chemical processes. A plantwide process is represented as a network of process units (subsystems) interconnected via mass and energy flows (e.g., material recycle and energy integration) and controlled by a network of controllers. In this analysis, the dynamical properties of process units are captured by their dissipativity. The plantwide stability is determined based on the dissipativity of each process unit and the network topology. To perform robust stability analysis, the process units are represented using LPV models which accommodate their uncertainties and a method is developed to determine the dissipativity (the storage function and the supply rate) of process units as a function of the scheduling variable. By polytopic LPV modelling, a dynamical system can be represented as the linear combination of the models at convex vertices. This allows the proposed dissipativity to be determined based on the dissipativity of each model at vertices. This plantwide analysis is based on the dissipativity of the LPV models of process units rather than the model for the entire plant and as such it very scalable. In this work, the results include: (1) determining dissipativity for LPV systems (2) robust analysis on the effects of uncertainties in process units on the stability of the plantwide system (3) robust control design for plantwide systems.
This paper investigates modeling and control of an underactuated limit-cycle walker for escaping from the quagmire. First, we introduce an underactuated rimless wheel model and develop the equations of motion and collision according to the phase sequence. Second, we design a method for generating a few steps walking motion based on collisionless limit cycle walking. The robot is then controlled from the initial static state to the terminal one including the impulsive effects and the influence of the reaction force from the quagmire. The fundamental properties of the generated motion are discussed through numerical simulations.
Micro-air vehicles (MAVs) are envisaged to spend a large portion of their mission within urban environments, which in general are rich in large obstacles (both natural and man-made). These obstacles can be a hindrance to MAV flight, but also have the potential to generate orographic updrafts when wind impinges on them. In theory, MAVs can exploit these updrafts in order to conserve power. However, finding, navigating between, and utilizing these updrafts is a significant challenge. We explore three aspects of this urban soaring challenge: updraft prediction and sensing, path-planning, and control. In an effort to predict urban updrafts, large-scale computational fluid dynamics (CFD) simulations of various environments have been performed. These are then combined with real-time flow field data from several multi-hole pressure probes attached to the MAV to produce better estimates of the current updraft field. The CFD results are used in large-scale path-planning through the use of a randomized planning algorithm to plan energy-efficient paths through known environments. Finally, a demonstration of "wind-hovering" in an orographic updraft using a simplified trajectory determination algorithm and control system is presented. Our vision is an autonomous platform that utilizes a database of flows around canonical shapes, together with a map, and feedback from flow sensors, to effectively navigate between urban soaring locations and maintain prolonged soaring flight.
A game-theoretic approach to load scheduling is explored within the context of constrained discrete-time linear time-invariant dynamical systems. Rigidity of individual load inputs dictates that it is only possible to translate the corresponding signals back-and-forth in time in order to form a feasible schedule. A game is devised to encode the trade-off between user sensitivities to shifts from the requested load and the need to respect constraints on the dynamic response of the system over a planning horizon. The optimal schedule is an equilibrium of the game for an appropriately selected pricing scheme. As such, learning algorithms can be used to devise distributed algorithms for recovering the optimal schedule.
It has been shown that Laguerre functions successfully reduce the number of decision variables in model predictive control of linear time invariant (LTI) systems. In this paper we extend the use of Laguerre functions to deal with linear time-varying (LTV) systems in the model predictive control framework. It is shown that Laguerre functions enable us to reduce the number of decision variables significantly and with a satisfactory performance. An example illustrating the unconstrained and constrained cases is provided.
This paper presents an application of feedback control system for relative performance management in virtualized software system. The system dynamics are characterized in a block-oriented nonlinear system identification. The models are estimated in Hammerstein-Wiener structure. The advantage of this approach is the elimination of nonlinearity impact to the performance management by consolidating nonlinear compensators in the feedback control loop. PI controller parameters are designed based on the step response curve generated from Frequency Sampling Filter functions. The experiments conducted in a two-class of virtual machines environment have shown the control system robustness with a great performance stability in the presence of unpredictable disturbances.
In heterogeneous multi-agent systems, a common assumption is that the agents are able to measure their individual output, which only depends on their individual state. In our recent work on observer-based synchronization, we have relaxed this assumption by introducing distributed observers, which enable synchronization only relying on relative measurements, and thereby guarantee ℋ∞-type performance with respect to disturbances. One assumption however, is that every agent possesses models of its neighbours. This paper presents a novel method, where no exact model on the neighbours is needed.
This paper exploits some recent results on monotonic tracking control to solve tracking problems in the presence of polytopic constraints on the state and input functions. The control architecture involved is a simple state feedback control with a feedforward action from the reference. The proposed technique does not require an online optimisation procedure, thus reducing considerably the computational burden. Indeed, under suitable assumptions, the tracking problem can be recast as a quadratic programming problem that can be easily solved offline. When applicable, this method represents an advantage over well established techniques such as model predictive control.
We design a state feedback control law to adaptively stabilize a 2 × 2 linear hyperbolic system of partial differential equations (PDEs) with uncertain in-domain coupling coefficients. We do this by combining an identifier with a backstepping-based controller and show boundedness and integrability in the ℒ 2 -sense of all signals in the closed loop system. The theory is demonstrated in a simulation.
Frequency interval controllability and observability gramian matrices are important in order to understand the characteristics of systems which are inherently frequency dependent. Obtaining these frequency interval controllability and observability gramian matrices requires solving a pair of Lyapunov equations. However for certain systems these Lyapunov equations are not solvable. In addition the eigenvalues of the product of the frequency interval controllability and observability gramians may also be complex numbers and therefore these gramians are not applicable to used in the context of model reduction. To overcome these issues, generalized frequency interval controllability and observability gramians are introduced in this paper and the applicability of these generalized gramians to be used in model reduction is demonstrated.
This paper is concerned with leader-following consensus seeking in heterogeneous nonlinear multi-agent systems in networked environments. The key idea is to develop a network-based leader-following consensus protocol with heterogeneous gain matrix for each follower agent while addressing the effect of network-induced delay. The Lyapunov-Krasovskii approach is utilized to design the network-based consensus controller gain, which results in a sufficient condition that guarantees exponential convergence of the leader-following consensus error to a bounded region. The workability of the proposed leader-following consensus algorithm is demonstrated by an illustrative example of networked Chua's circuits.
