Electrical and Electronic Engineering - Theses
Permanent URI for this collection
Search Results
1 - 4 of 4
-
ItemModel Predictive Controller Tuning by Machine Learning and Ordinal Optimisation( 2021)While for the past several decades model predictive control (MPC) has been an established control strategy in chemical process industries, more recently there has been increased collaboration in MPC research between academia and automotive companies. Despite the promising work thus far, one particular challenge facing the widespread adoption of MPC in the automotive industry is the increased calibration requirement. The focus of the research in this thesis is to develop methods towards reducing the calibration effort in designing and implementing MPC in practice. The research is tailored by application to offline tuning of quadratic-cost MPC for an automotive diesel air-path, to address the limited time-availability to perform online tuning experiments. Human preferences can be influential in automotive engine controller tuning. Some earlier work has proposed a machine learning controller tuning framework (MLCTF), which learns preferences from numeric data labelled by human experts, and as such, these learned preferences can be replicated in automated offline tuning. Work done in this thesis extends this capability by allowing for preferences to be learned from pairwise comparison data, with monotonicity constraints in the features. Two methods are proposed to address this: 1) an algorithm based around Gaussian process regression; and 2) a Bayesian estimation procedure using a Dirichlet prior. These methods are successfully demonstrated in learning monotonicity-constrained utility functions in time-domain features from data consisting of pairwise rankings for diesel air-path trajectories. The MLCTF also constitutes a plant model, yet there will typically be some uncertainty in an engine model, especially if it has been identified from data collected with a limited amount of experimentation time. To address this, an active learning framework is proposed for selection of the next operating points in the design of experiments, for identifying linear parameter-varying systems. The approach is based on exploiting the probabilistic features of Gaussian process regression to quantify the overall model uncertainty across locally identified models, resulting in a flexible methodology which accommodates for various techniques to be applied for estimation of local linear models and their corresponding uncertainty. The framework is applied to the identification of a diesel engine air-path model, and it is demonstrated that measures of model uncertainty can be quantified and subsequently reduced. To make the most of the limited availability for online tuning experiments, an ordinal optimisation (OO) approach is proposed, which seeks to ensure that offline tuned controllers can perform acceptably well, once tested online with the physical system. Via the use of copula models, an OO problem is formulated to be compatible with the tuning of controllers over an uncountable search space, such as quadratic-cost MPC. In particular, results are obtained which formally characterise the copula dependence conditions required for the OO success probability to be non-decreasing in the number of offline controllers sampled during OO. A gain-scheduled MPC architecture was designed for the diesel air-path, and implemented on an engine control unit (ECU). The aforementioned non-decreasing properties of the OO success probability are then specialised to tuning gain-scheduled controller architectures. Informed by these developments, the MPC architecture was firstly tuned offline via OO, and then tested online with an experimental diesel engine test rig, over various engine drive-cycles. In the experimental results, it was found that some offline tuned controllers outperformed a manually tuned baseline MPC, the latter which has comparable performance to proprietary production controllers. Upon additional manual tuning online, the performance of the offline tuned controllers could also be further refined, which illustrates how offline tuning via OO may complement online tuning approaches. Lastly, using an analytic lower bound developed for OO under a Gaussian copula model, a sequential learning algorithm is developed to address a probabilistically robust offline controller tuning problem. The algorithm is formally proven to yield a controller which meets a specified probabilistic performance specification, assuming that the underlying copula is not too unfavourably far from a Gaussian copula. It is demonstrated in a simulation study that the algorithm is able to successfully tune a single controller to meet a desired performance threshold, even in the presence of probabilistic uncertainty in the diesel engine model. This is applied to two case studies: 1) `hot-starting' an online tuning procedure; and 2) tuning for uncertainty inherent across a fleet of vehicles.
-
ItemA Model-based Approach for High Performance Motion Control in Industrial Machines( 2020)Industrial robotics typically consider laser/water cutting, grinding, etc. Within these machines, the motion controller is responsible for the positioning of the end effector. The performance of the motion controller directly influences the quality of the resulting product as tolerance/accuracy are surrogates for machining quality. This is particularly relevant in tracking and contouring applications when the system has structural flexibility, and no direct feedback measurement of the end-effector position is available. Traditional control architectures in machining are unable to explicitly bound tracking and/or contouring errors, and conservative operation is used to ensure satisfactory performance of the overall system. Bounding errors without unduly compromising machine throughput requires advanced control algorithms. The development of such algorithms is the focus of this thesis. Although numerous control methods are proposed, the proportional integral derivative (PID) based cascaded control is still the most prevalent in the industry. Based on this fact, the research starts by objectively assessing the tracking control performance on a single-axis industrial platform. The results provide practitioners with an in-depth understanding of the benefits and limitations of existing control algorithms as well as the motivation to consider advanced controllers as alternatives to the PID-based approach. For the single-axis tracking problem, this research proposes a model predictive based approach that guarantees a desired level of tracking error is met for the cases where the structure is flexible and the end-effector position is estimated. To achieve this, a robust control invariant set is estimated using a computationally tractable algorithm and incorporated into the problem formulation. The applicability of the proposed approach is successfully demonstrated via simulation and experiments conducted on a commercial single-axis system. In terms of biaxial applications, the dual-drive gantry machines are widely used in industry for manufacturing. However, the non-synchronised movement of the dual drive may lead to deterioration in contouring accuracy. In this research, we propose two model predictive based control architectures based on the switched linear time invariant control-oriented models, that is able to guarantee a two-dimensional contouring tolerance in the presence of uncertainty arising from imperfect drive synchronisation. The performance and computational tractability of the proposed approach are demonstrated using high fidelity simulation and experiment.
-
ItemHierarchical model predictive control of an unmanned-aerial-vehicle based multitarget-multisensor data fusion system( 2011)There has been rapidly growing interest in the development of increased automation in the military in recent years. In particular, the number of unmanned aircraft and ground vehicles being put into use is rapidly increasing. Concurrently, there has been an associated increase in the amount of research being performed to develop increased autonomy, moving from relatively simple remotely controlled devices to autonomous systems that are able to operate in a sense-think-act paradigm, i.e., robots. An application of robotic technology that is of considerable military significance is that of detection and tracking of enemy assets. A key advantage of using autonomous vehicles in this application is that the locations and details of potential threats can be determined using relatively inexpensive unmanned vehicles with the operator of the system standing back at a safe distance. One example is the use of teams of unmanned aerial vehicles (UAVs) carrying passive direction-of-arrival sensors to detect and track enemy emitters such as radar-carrying platforms so as to enable reaction to the threats with other resources which could, say, include jammers or missile-carrying aircraft. In this thesis the problem of how to adaptively control the trajectories of UAVs in such an application in order to optimize performance in response to target measurements, while avoiding no-fly zones, is considered and a solution is developed. Because of the complexity of situations that are encountered, a major issue is how to formulate the problem in a manner which enables efficient computation of optimal behaviours for the platforms. In fact, an optimal solution cannot in practice be found by any physically implementable method. Hence, in this thesis an approach will be developed that enables implementation of a computationally feasible, albeit suboptimal solution, that takes into account both short-term and long-term goals. To this end, the problem will be addressed by developing a hierarchical control approach, incorporating an automated planner and a low-level (short-term) control algorithm. A key aim is to use a consistent mathematical framework that can be generalized to a range of optimal control problems. As a result, all components of the controllers that are developed are based on concepts from estimation theory, dynamic programming and optimal control, giving a mathematically coherent and scalable solution. To evaluate the effectiveness of the approach, first a controller is developed using an idealized UAV model and simulations are performed. Its performance is compared with a commonly used "myopic" control approach and found to give important improvements. Subsequently an improved planner is incorporated and tested, and then a version of the controller using a fixed-wing aircraft model for the UAVs is implemented. This version is also tested by simulation and found to perform successfully. Finally, a mathematical analysis of stability is commenced and significant headway made towards a stability proof.
-
ItemModelling and control design of river systems( 2011)Farming consumes a large amount of water usage and it is reported that large portion of this water is wasted through inefficient water distribution from river to farms. More efficient water distribution and preservation of environmental demands can be achieved through better control and decision support systems. In order to design better control and decision support systems, a river model is required. This model needs to be able to capture the relevant river dynamics and easy to be used for control design. Traditionally, the Saint Venant equations have been used to model river systems. These equations are nonlinear hyperbolic partial differential equation (PDE) and are solved numerically using Preissmann scheme. The simulated Saint Venant equations are compared against operational data from the Broken River, and it is found that the Saint Venant equations are accurate in representing the river systems. Through further study, it is found that a single segmentation, i.e. treating the river as one long stretch with uniform geometry is sufficiently accurate for representation of the river for the purpose of control design. For the representation of meandering river, the Saint Venant equations are as accurate a two-dimensional flow model. The nonlinearities in the Saint Venant equations are also investigated. From the nonlinearity test, it is found that the Saint Venant equations are approximately linear within an operating region. The Saint Venant equations are difficult to use for control design. An alternative model is therefore sought. Based on the operational data from the Broken River, simple time delay model (TDM) and integrator delay model (IDM) are proposed and estimated using system identification procedures. These models are found to be accurate in capturing the relevant dynamics of the river system. Furthermore, they are easy to use for control design. It is found that the time delay varies with the flow and hence controllers must be robust to variations in the time delay. A comparison between both TDM and IDM and the Saint Venant equations shows that they are as accurate as the Saint Venant equations within the operating range. The TDM and IDM are desirable as they are easier to be used for control design and decision support system. The TDM and IDM are used to design Model Predictive Control (MPC) to control the river system. The choice of using MPC is motivated by the fact that MPC handles constraints very well. Despite that, tuning the weights in the MPC cost function is not trivial. The methods of reverse engineering are used to obtain these weights. Building on the results of existing method of reverse engineering used in the literatures, two additional methods are developed. In addition, the design of MPC from scratch is also considered. A realistic year long simulations using both MPCs on the Broken River is carried out. The MPCs are compared with the current manual operation and a decentralised control configuration. The results show that with MPCs, significant water savings, improvement of water delivery service to the irrigators and the environmental demands satisfaction are achieved.