the 18th World Congress The International Federation of Automatic Control, Milano (Italy), August 28 - September 2, 2011

The deployment of Renewable Energy (RE)-based generation has experienced a sustained global growth in the recent decades, driven by many countries' interest in reducing greenhouse gas emissions and dependence on fossil fuel for electricity generation. This trend is also observed in remote off-grid systems (isolated microgrids), where local communities, in an attempt to reduce fossil fuel dependency and associated economic and environmental costs, and to increase availability of electricity, are favouring the installation of RE-based generation. This practice has posed several challenges to the operation of such systems, due to the intermittent and hard-to-predict nature of RE sources. In particular, this thesis addresses the problem of reliable and economic dispatch of isolated microgrids, also known as the energy management problem, considering the uncertain nature of those RE sources, as well as loads. Isolated microgrids feature characteristics similar to those of distribution systems, in terms of unbalanced power flows, significant voltage drops and high power losses. For this reason, detailed three-phase mathematical models of the microgrid system and components are presented here, in order to account for the impact of unbalanced system conditions on the optimal operation of the microgrid. Also, simplified three-phase models of Distributed Energy Resources (DERs) are developed to reduce the level of complexity in small units that have limited impact on the optimal operation of the system, thus reducing the number of equations and variables of the problem. The proposed mathematical models are then used to formulate a novel energy management problem for isolated microgrids, as a deterministic, multi-period, Mixed-Integer Nonlinear Programming (MINLP) problem. The multi-period formulation allows for a proper management of energy storage resources and multi-period constraints associated with the commitment decisions of DERs. In order to obtain solutions of the energy management problem in reasonable computational times for real-time, realistic applications, and to address the uncertainty issues, the proposed MINLP formulation is decomposed into a Mixed-Integer Linear Programming (MILP) problem, and a Nonlinear programming (NLP) problem, in the context of a Model Predictive Control (MPC) approach. The MILP formulation determines the unit commitment decisions of DERs using a simplified model of the network, whereas the NLP formulation calculates the detailed three-phase dispatch of the units, knowing the commitment status. A feedback signal is generated by the NLP if additional units are required to correct reactive power problems in the microgrid, triggering a new calculation MINLP problem. The proposed decomposition and calculation routines are used to design a new deterministic Energy Management System (EMS) based on the MPC approach to handle uncertainties; hence, the proposed deterministic EMS is able to handle multi-period constraints, and account for the impact of future system conditions in the current operation of the microgrid. In the proposed methodology, uncertainty associated with the load and RE-based generation is indirectly considered in the EMS by continuously updating the optimal dispatch solution (with a given time-step), based on the most updated information available from suitable forecasting systems. For a more direct modelling of uncertainty in the problem formulation, the MILP part of the energy management problem is re-formulated as a two-stage Stochastic Programming (SP) problem. The proposed novel SP formulation considers that uncertainty can be properly modelled using a finite set of scenarios, which are generated using both a statistical ensembles scenario generation technique and historical data. Using the proposed SP formulation of the MILP problem, the deterministic EMS design is adjusted to produce a novel stochastic EMS. The proposed EMS design is tested in a large, realistic, medium-voltage isolated microgrid test system. For the deterministic case, the results demonstrate the important connection between the microgrid's imbalance, reactive power requirements and optimal dispatch, justifying the need for detailed three-phase models for EMS applications in isolated microgrids. For the stochastic studies, the results show the advantages of using a stochastic MILP formulation to account for uncertainties associated with RE sources, and optimally accommodate system reserves. The computational times in all simulated cases show the feasibility of applying the proposed techniques to real-time, autonomous dispatch of isolated microgrids with variable RE sources.

Advances in hardware design have made wheeled mobile robots (WMRs) exceptionally mobile. To fully exploit this mobility, WMR planning, control, and estimation systems require motion models that are fast and accurate. Much of the published theory on WMR modeling is limited to 2D or kinematics, but 3D dynamic (or force-driven) models are required when traversing challenging terrain, executing aggressive maneuvers, and manipulating heavy payloads. This thesis advances the state of the art in both the formulation and calibration of WMR models We present novel WMR model formulations that are high-fidelity, general, modular, and fast. We provide a general method to derive 3D velocity kinematics for any WMR joint configuration. Using this method, we obtain constraints on wheel ground contact point velocities for our differential algebraic equation (DAE)-based models. Our “stabilized DAE” kinematics formulation enables constrained, drift free motion prediction on rough terrain. We also enhance the kinematics to predict nonzero wheel slip in a principled way based on gravitational, inertial, and dissipative forces. Unlike ordinary differential equation (ODE)-based dynamic models which can be very stiff, our constrained dynamics formulation permits large integration steps without compromising stability. Some alternatives like Open Dynamics Engine also use constraints, but can only approximate Coulomb friction at contacts. In contrast, we can enforce realistic, nonlinear models of wheel-terrain interaction (e.g. empirical models for pneumatic tires, terramechanics-based models) using a novel force-balance optimization technique. Simulation tests show our kinematic and dynamic models to be more functional, stable, and efficient than common alternatives. Simulations run 1K-10K faster than real time on an ordinary PC, even while predicting articulated motion on rough terrain and enforcing realistic wheel-terrain interaction models. In addition, we present a novel Integrated Prediction Error Minimization (IPEM) method to calibrate model parameters that is general, convenient, online, and evaluative. Ordinarily system dynamics are calibrated by minimizing the error of instantaneous output predictions. IPEM instead forms predictions by integrating the system dynamics over an interval; benefits include reduced sensing requirements, better observability, and accuracy over a longer horizon. In addition to calibrating out systematic errors, we simultaneously calibrate a model of stochastic error propagation to quantify the uncertainty of motion predictions. Experimental results on multiple platforms and terrain types show that parameter estimates converge quickly during online calibration, and uncertainty is well characterized. Under normal conditions, our enhanced kinematic model can predict nonzero wheel slip as accurately as a full dynamic model for a fraction of the computation cost. Finally, odometry is greatly improved when using IPEM vs. manual calibration, and when using 3D vs. 2D kinematics. To facilitate their use, we have released open source MATLAB and C++ libraries implementing the model formulation and calibration methods in this thesis.

A typical wastewater effluent of a chemical process can contain several strong acids/bases, weak acids/bases as well as their salts. They must be neutralized before being discharged to the environment in order to protect aquatic life and human welfare. However, neutralization process is highly non&ndash / linear and has time&ndash / varying characteristics. Therefore, the control of pH is a challenging problem where advanced control strategies are often considered. In this study, the aim is to design a pH control system that will be capable of controlling the pH-value of a plant waste-water effluent stream having unknown acids with unknown concentrations using an on&ndash / line identification procedure. A Model Predictive Controller, MPC, and a Fuzzy Logic Controller, FLC, are designed and used in a laboratory scale pH neutralization system. The characteristic of the upstream flow is obtained by a small identification reactor which has ten times faster dynamics and which is working parallel to actual neutralization tank. In the control strategy, steady&ndash / state titration curve of the process stream is obtained using the data collected in terms of pH value from the response of the identification reactor to a pulse input in base flow rate and using the simulated response of the identification reactor for the same input. After obtaining the steady&ndash / state titration curve, it is used in the design of a Proportional&ndash / Integral, PI, and of an Adaptive Model Predictive Controller, AMPC. On the other hand, identification reactor is not used in the FLC scheme. The performances of the designed controllers are tested mainly for disturbance rejection, set&ndash / point tracking and robustness issues theoretically and experimentally. The superiority of the FLC is verified.

In this thesis, the walking of a humanoid system is simulated applying an optimal predictive control algorithm. The simulation is built using Matlab and Simulink softwares. Four separate physical models are developed to represent the single support and the double support phases of a full gait cycle. The models are three dimensional and their properties are analogous to the human&rsquo / s. In this connection, the foot models in the double support phases include an additional joint which connects the toe to the foot. The kinematic relationships concerning the physical models are formulated recursively and the dynamic models are obtained using the Newton &ndash / Euler formulation. The computed torque method is utilized at the level of joints. In the double support phase, the redundancy problem is solved by the optimization of the actuating torques. The command accelerations required to control the gait are obtained by applying an optimal predictive control law. The introduced humanoid walker achieves a sustainable gait by tuning the optimization and prediction parameters. The control algorithm manages the tracking of the predefined walking pattern with easily realizable joint accelerations. The simulation is capable of producing all the reaction forces, reaction moments and the values of the other variables. During these computations, a three dimensional view of the humanoid walker is animated simultaneously. As a result of this study, a suitable simulation structure is obtained to test and improve the mechanical systems which perform bipedal locomotion. The modular nature of the simulation structure developed in this study allows testing the performance of alternative control laws as well.

This dissertation investigates a new design for a magnetostrictive actuator that employs individually controlled coils distributed axially along the magnetostrictive rod. As a quantitative goal, the objective is to show that the multi-coil actuator can operate effectively at frequencies as high as 10,000 Hz with 900 N force and 50 microns of displacement. Conventional, single coil actuators with the same parameters for force and displacement develop significant attenuation in their response at frequencies above the first longitudinal vibration resonance at about 2750 Hz. The goal of the research is to investigate whether multiple coils can effectively increase the frequency range a least four times the range of conventional magnetostrictive actuators. This document derives a new mathematical model of the actuator that represents the spatial distributions of magnetic field and vibration, devises a control design that takes advantage of the multiple inputs to control the displacement of the actuator while consuming minimum electrical power, and describes a prototype multi-coil actuator and experimental system developed to test the idea. The simulations of the multi-coil actuator and control design demonstrate successful transient operation of the actuator over the targeted frequency range with feasible levels of input power and current. Experimental tests of the design, although limited by a computer sampling rate less than 10,000 Hz, are able to validate the predictions of the developed model of the actuator and reproduce the simulated control performance within the constraints of the experimental system.

This thesis investigates the questions of where to grasp and how to grasp a given object with an articulated robotic grasping device. To this end, aspects of grasp synthesis and hand motion planning and control are investigated. Grasp synthesis is the process of determining a palm pose with respect to the target object, as well as a hand joint configuration and/or grasp contact points such that a successful grasp execution is allowed. Existing methods tackling the grasp synthesis problem can be categorized in analytical and empirical approaches. Analytical approaches are based on geometric, kinematic and/or dynamic formulations, whereas empirical methods aim at mimicking human strategies.An overarching idea throughout this thesis is to circumvent the curse of dimensionality, which is inherent in high-dimensional planning problems, by incorporating empirical data in analytical approaches. To this end, tools from the field of constrained optimization are used (i) to synthesize grasp families based on available prototype grasps, (ii) to incorporate heuristics capturing human grasp strategies in the grasp synthesis process and (iii) to encode demonstrated grasp motions in primitive motion controllers.The first contribution is related to the computation and analysis of grasp families which are represented as Independent Contact Regions (ICR) on a target object’s surface. To this end, the well-known concept of the Grasp Wrench Space for a single grasp is extended to be applicable for a set of grasps. Applications of ICR include grasp qualification by capturing the robustness of a grasp to position inaccuracies and the visual guidance of a demonstrator in a teleoperating scenario. In the second main contribution of this thesis, it is shown how to reduce the grasp solution space during the synthesis process by accounting for human approach strategies. This is achieved by imposing appropriate constraints to a corresponding optimization problem. A third contribution in this dissertation is made to reactive motion planning. Here, primitive controllers are synthesized by estimating the free parameters of corresponding dynamical systems from multiple demonstrated trajectories. The approach is evaluated on an anthropomorphic robot hand/arm platform. Also, an extension to a Model Predictive Control (MPC) scheme is presented which allows to incorporate state constraints for auxiliary tasks such as obstacle avoidance.

Thesis (MScEng)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: Cash can be seen as a fast moving consumer good. Approaching cash as inventory within the ATM cash management environment of a South African retail bank, provided the opportunity to apply well known industrial engineering techniques to the financial industry. This led to the application of forecasting, inventory management, operational research and simulation methods. A forecasting model is designed to address the multiple seasonalities and calendar day effects that is prevalent in the demand for cash. Special days, e.g. paydays, lead to an increase in demand for cash. The weekday on which the special day falls will also influence the demand. The multiplicative Holt-Winters method is combined with an improvised distribution method to determine the demand for cash for the region and per ATM. Reordering points are calculated and simulated to form an understanding of the effect this will have on the ATM network. Direct replenishment and the traveling salesman problem is applied and simulated to determine the difference in using one or the other. Various simulation models are build to test the operational and financial impact when certain variables are amended. It is evident that more work is required to determine the optimal combination of variable values, i.e. forecasting frequency, aggregate forecasting or individual forecasting, reorder levels, loading levels, lead times, cash swap or cash add, and the type of transportation method. Each one of these are a science in itself and cannot be seen (calculated) in isolation from the other as a change in one can affect the overall operational efficiency and costs of the ATM network. The thesis proves that significant cost savings is possible, compared to the current set-up, when applying industrial engineering techniques to a geographical ATM network within South Africa. / AFRIKAANSE OPSOMMING: Kontant kan gesien word as vinnig bewegende verbruikersgoedere. Deur kontant te benader as voorraad binne die ATM kontant bestuur omgewing van ’n Suid Afrikaanse kleinhandelsbank, het dit die geleentheid geskep om bekende bedryfsingenieurstegnieke toe te pas in die finansiële industrie. Dit het gelei tot die toepassing van vooruitskatting, voorraadbestuur, operasionele navorsing en simulasie metodes. ’n Vooruitskattingsmodel is ontwerp om die verskeie seisoenaliteite en kalenderdae effekte wat deel uitmaak van die vraag na kontant aan te spreek. Spesiale dae, bv. betaaldae, lei tot ’n toename in die vraag na kontant. Die weeksdag waarop die spesiale dag voorkom sal ook ’n invloed hê op die vraag. Die multiplikatiewe Holt-Winters metode is gekombineer met ’n geïmproviseerde verspreidingsmetode om die vraag na kontant vir die streek en per ATM the bepaal. Bestellingsvlakke is bereken en gesimuleer om ’n prentjie te skep van die invloed wat dit op die ATM netwerk sal hê. Direkte hervulling en die handelsreisigerprobleem is toegepas en gesimuleer om die verskille te bepaal tussen die gebruik van of die een of die ander. Veskeie simulasie modelle is gebou om die operasionele en finansiële impak te toets, wanneer sekere veranderlikes aangepas word. Dit is duidelik dat meer werk nodig is om die optimale kombinasie van veranderlike waardes te bepaal, bv. vooruitskatting frekwensie, totale vooruiskatting of individuele vooruitskatting, bestellingsvlakke, leitye, kontant omruiling of kontant byvoeging, en die tipe vervoermetode. Elkeen van hierdie is ’n wetenskap op sy eie en kan nie in isolasie gesien en bereken word nie, want ’n verandering van een se waarde kan die hele operasionele doeltreffendheid en kostes van die ATM netwerk beïnvloed.

Made available in DSpace on 2014-12-17T14:08:39Z (GMT). No. of bitstreams: 1 LuizHGP_DISSERT.pdf: 1454316 bytes, checksum: 0866b81b5bfb98284278c13af6a47bdc (MD5) Previous issue date: 2009-08-07 / A pesquisa tem como objetivo desenvolver uma estrutura de controle preditivo neural, com o intuito de controlar um processo de pH, caracterizado por ser um sistema SISO (Single Input - Single Output). O controle de pH ? um processo de grande import?ncia na ind?stria petroqu?mica, onde se deseja manter constante o n?vel de acidez de um produto ou neutralizar o afluente de uma planta de tratamento de fluidos. O processo de controle de pH exige robustez do sistema de controle, pois este processo pode ter ganho est?tico e din?mica n?olineares. O controlador preditivo neural envolve duas outras teorias para o seu desenvolvimento, a primeira referente ao controle preditivo e a outra a redes neurais artificiais (RNA s). Este controlador pode ser dividido em dois blocos, um respons?vel pela identifica??o e outro pelo o c?lculo do sinal de controle. Para realizar a identifica??o neural ? utilizada uma RNA com arquitetura feedforward multicamadas com aprendizagem baseada na metodologia da Propaga??o Retroativa do Erro (Error Back Propagation). A partir de dados de entrada e sa?da da planta ? iniciado o treinamento offline da rede. Dessa forma, os pesos sin?pticos s?o ajustados e a rede est? apta para representar o sistema com a m?xima precis?o poss?vel. O modelo neural gerado ? usado para predizer as sa?das futuras do sistema, com isso o otimizador calcula uma s?rie de a??es de controle, atrav?s da minimiza??o de uma fun??o objetivo quadr?tica, fazendo com que a sa?da do processo siga um sinal de refer?ncia desejado. Foram desenvolvidos dois aplicativos, ambos na plataforma Builder C++, o primeiro realiza a identifica??o, via redes neurais e o segundo ? respons?vel pelo controle do processo. As ferramentas aqui implementadas e aplicadas s?o gen?ricas, ambas permitem a aplica??o da estrutura de controle a qualquer novo processo

In order to meet tight product quality specifications for chemical batch processes, it is vital to monitor and control product quality throughout the batch duration. However, the frequent lack of in situ sensors for continuous monitoring of batch product quality complicates the control problem and calls for novel control approaches. This thesis focuses on the study and application of multivariate statistical methods to control product quality in chemical batch processes. These multivariate statistical methods can be used to identify data-driven prediction models that can be integrated within a model predictive control (MPC) framework. The ideal MPC control strategy achieves end-product quality specifications by performing trajectory tracking during the batch operating time. However, due to the lack of in-situ sensors, measurements of product quality are usually obtained by laboratory assays and are, therefore, inherently intermittent. This thesis proposes a new approach to realise trajectory tracking control of batch product quality in those situations where only intermittent measurements are available. The scope of this methodology consists of: 1) the identification of a partial least squares (PLS) model that works as an estimator of product quality, 2) the transformation of the PLS model into a recursive formulation utilising a moving window technique, and 3) the incorporation of the recursive PLS model as a predictor into a standard MPC framework for tracking the desired trajectory of batch product quality. The structure of the recursive PLS model allows a straightforward incorporation of process constraints in the optimisation process. Additionally, a method to incorporate a nonlinear inner relation within the proposed PLS recursive model is introduced. This nonlinear inner relation is a combination of feedforward artificial neural networks (ANNs) and linear regression. Nonlinear models based on this method can predict product quality of highly nonlinear batch processes and can, therefore, be used within an MPC framework to control such processes. The use of linear regression in addition to ANNs within the PLS model reduces the risk of overfitting and also reduces the computational e↵ort of the optimisation carried out by the controller. The benefits of the proposed modelling and control methods are demonstrated using a number of simulated batch processes.