Electrical and Electronic Engineering - Research Publications
Permanent URI for this collection
Search Results
1 - 10 of 17
-
ItemMulti-agent gradient climbing via extremum seeking control(IFAC - International Federation of Automatic Control, 2014)A unified framework based on discrete-time gradient-based extremum seeking control is proposed to localise an extremum of an unknown scalar field distribution using a group of equipped with sensors. The controller utilises estimates of gradients of the field from local dithering sensor measurements collected by the mobile agents. It is assumed that distributed coordination which ensures uniform asymptotic stability with respect to a prescribed formation of the agents is employed. The framework is useful in that a broad range of nonlinear programming algorithms can be combined with a wide class of cooperative control laws to perform extreme source seeking. Semi-global practical asymptotically stable convergence to local extrema is established in the presence of bounded field sampling noise.
-
ItemPDE Battery Model Simplification for Charging Strategy Evaluation(IEEE Press, 2015)A safe, fast charging strategy is desired in the utilisation of rechargeable Lithium-ion batteries. Traditionally, experimental methods are used in exploring and evaluating new strategies, but these require extensive time and cost. This paper aims to establish a model-based system for quick and accurate evaluation of charging strategies. Starting from a nonlinear coupled partial differential equation (PDE) battery model that accurately captures system dynamics, simplification techniques are conducted based on the identification of separable time scales within the states. By pertinent use of a singular perturbation approach, a PDE model simplification framework containing families of simplified battery models is established. All assumptions are explicitly stated and shown to enable families of simplified models to be rigorously justified. An evaluation procedure synthesised from the simplified models and averaging theory is proposed. This procedure is implemented on several typical battery charging strategies. The benefits relative to simulation on other higher order models are assessed in terms of computational efficiency and accuracy and demonstrate significant computational savings are possible with the proposed approach.
-
ItemPDE Battery Model Simplification for SOC and SOH Estimator Design(IEEE, 2015)Accurate knowledge of the battery state-of-charge (SOC) and state-of-health (SOH) is critical for optimal and safe utilisation of the battery. Although the battery system dynamics contain electrochemical, thermal, electrical, and ageing phenomena, most state estimators resort to equivalent circuit models (ECM). These models are often not accurate and are problematic for SOC estimation during an extended range of operations and do not address SOH dynamics. In this paper, starting from an initial high-fidelity Lithium-ion (Li-ion) battery model consisting of a set of partial differential equations (PDE), a recently proposed framework for PDE battery model simplification is employed and one of these obtained models is used for battery state estimation. Model order reduction techniques are then constructively applied to the simplified PDE battery model and resulted in a computationally efficient ordinary differential equation (ODE) model. Based on this obtained ODE model, an extended Kalman filter (EKF) is designed for the estimation of both SOC and SOH. Simulations over 20 cycles show the designed estimator is capable of simultaneously estimating the battery's SOC in each electrode and SOH.
-
ItemSimplification Techniques for PDE Based Li-Ion Battery Models(IEEE, 2015-12-14)Battery systems are becoming increasingly prevalent as a source of power for applications across domains from consumer electronics to automotive, due to a range of factors such as portability and environmental considerations. The relatively high cost of batteries leads to a natural tradeoff in their use to ensure the lifetime of the battery is not unduly compromised while still delivering good performance. Similar tradeoffs have been successfully dealt with in other systems using model based control and estimation techniques, and this motivates their use for battery systems. Complicating this process is the complex nature of the physics-based models describing the operation of a battery cell, as these consist of a large number of partial differential equations spanning multiple, coupled domains. This second paper of the tutorial session will briefly review the existing physics-based battery models, and introduce recent approaches that have been used to develop simplified models based on the original high-fidelity model. The assumptions underpinning the model simplification will be presented and discussed.
-
ItemGaussian Processes with Monotonicity Constraints for Preference Learning from Pairwise Comparisons(IEEE, 2018)In preference learning, it is beneficial to incorporate monotonicity constraints for learning utility functions when there is prior knowledge of monotonicity. We present a novel method for learning utility functions with monotonicity constraints using Gaussian process regression. Data is provided in the form of pairwise comparisons between items. Using conditions on monotonicity for the predictive function, an algorithm is proposed which uses the weighted average between prior linear and maximum a posteriori (MAP) utility estimates. This algorithm is formally shown to guarantee monotonicity of the learned utility function in the dimensions desired. The algorithm is tested in a Monte Carlo simulation case study, in which the results suggest that the learned utility by the proposed algorithm performs better in prediction than the standalone linear estimate, and enforces monotonicity unlike the MAP estimate.
-
ItemA machine learning approach for tuning model predictive controllers(IEEE, 2018-01-01)Many industrial domains are characterized by Multiple-Input-Multiple-Output (MIMO) systems for which an explicit relationship capturing the nontrivial trade-off between the competing objectives is not available. Human experts have the ability to implicitly learn such a relationship, which in turn enables them to tune the corresponding controller to achieve the desirable closed-loop performance. However, as the complexity of the MIMO system and/or the controller increase, so does the tuning time and the associated tuning cost. To reduce the tuning cost, a framework is proposed in which a machine learning method for approximating the human-learned cost function along with an optimization algorithm for optimizing it, and consequently tuning the controller, are employed. In this work the focus is on the tuning of Model Predictive Controllers (MPCs), given both the interest in their implementations across many industrial domains and the associated high degrees of freedom present in the corresponding tuning process. To demonstrate the proposed approach, simulation results for the tuning of an air path MPC controller in a diesel engine are presented.
-
ItemExtremum-Seeking-Based Adaptive Scan for Atomic Force Microscopy(IEEE, 2017)Improving the imaging speed in Atomic Force Microscopy (AFM) is of high interest due to its typically prolonged imaging duration. Conventionally, the line rate of the scan is fixed at a conservative value in order to ensure a safe tip-sample contact force even in the worst case of sample aspect ratio and linear scan speed. In this paper, an adaptive scan method is proposed to adapt the scan line rate based on the extremum-seeking control framework. A performance metric balancing both imaging speed and accuracy is proposed, and an extremum-seeking approach is designed to optimise the metric based on error feedback. Semi-global practical asymptotic stability (SGPAS) result is shown, and the proposed method is demonstrated via simulation.
-
ItemHybrid Extremum Seeking for Black-Box Optimization in Hybrid Plants: An Analytical Framework(IEEE, 2018-01-01)This paper presents an analytical framework to design and analyze hybrid extremum seeking controllers for plants with hybrid dynamics. The extremum seeking controllers are characterized by a hybrid dither generator, a hybrid Jacobian estimator, and a hybrid dynamic optimizer. This structure allows us to consider a family of novel extremum seeking controllers that have not been studied in the literature before. Moreover, the hybrid extremum seeking controllers can be applied to plants with hybrid dynamics generating well-defined response maps. A convergence result is established for the closed -loop system by using singular perturbation theory for hybrid dynamical systems with hybrid boundary layers.
-
ItemObserving the Slow States of General Singularly Perturbed Systems(IEEE, 2020)This paper studies the behaviour of observers for the slow states of a general singularly perturbed system - that is a singularly perturbed system which has boundary-layer solutions that do not necessarily converge to a slow manifold. The solutions of the boundary-layer system are allowed to exhibit persistent (e.g. oscillatory) steady-state behaviour which are averaged to obtain the dynamics of the approximate slow system. It is shown that if an observer has certain properties such as asymptotic stability of its error dynamics on average, then it is practically asymptotically stable for the original singularly perturbed system.
-
ItemA Framework for Reference Generation and Control in AFM Imaging(IEEE, 2016)The objective of constant-force contact mode Atomic Force Microscopy (AFM) is to obtain a topography estimate by scanning a probe over the surface along a designated path. This typically involves the design of XY reference scanning trajectories, a controller in the XY direction to track the trajectories, and a controller in the Z direction to regulate the tip-sample interaction force subject to perturbation of the unknown sample topography. Conventionally, the reference trajectories and the XY and Z controllers are designed and analysed separately, disregarding the underlying connections between the designs. The purpose of this paper is to provide a general framework for reference generation and XYZ actuation control for AFM imaging under this decoupled design structure. The framework provides precise conditions under which one can combine a large class of reference generators, actuators and controllers to achieve the convergence of XY tracking error and Z regulation error. A simple tuning guideline is also provided that balances precision and speed. In particular, it is shown that the ultimate bound on the errors can be made arbitrarily small and the spatial impact of any initial errors can be arbitrarily shortened by selecting a sufficiently small scan rate parameter.