Prompted by the desire to increase the industrial applicability range of self-tuning control, the objective of this work has been to extend the standard linear self-tuning framework to facilitate the design of self-tuning controllers for bilinear systems. Bilinear systems form a well structured class of non-linear systems within which linear systems coexist as a special subclass. They are, therefore, appropriate for modelling a wider range of processes and plant than the restrictive, yet convenient, linear model structures since such models are valid both within the linear subregion and beyond. In addition to extending the self-tuning framework for bilinear systems another significant contribution of the Thesis is the introduction of a cautious least squares estimation procedure which also enhances the existing linear self-tuning schemes.

Over the last decade, considerable effort has been devoted to the implementation and analysis of self-tuning controllers on systems which are assumed to be represented exactly by linear dynamical models. In this thesis we examine the robustness of the self-tuning controller, when applied to systems consisting of a nominal linear plant which may have linear or nonlinear perturbations. Robust stability is the primary criterion and most of the results are for the Clarke-Gawthrop version of the self-tuning controller. Conditions are derived for the robust stability of the adaptively controlled system in terms of the design choices available to the engineer setting up the self-tuning controller. These are strong stability results in that they are in terms of both 1₂ and 1_∞ stability. The results are shown to be applicable to the general delay case and in the presence of non-zero mean disturbances. Preliminary results are also obtained for the robust stability of the explicit self-tuning controller.

Thesis (Ph.D.) -- McMaster University, 2002. / Includes bibliographical references (leaves 246-253). Also available via World Wide Web.

Thesis (Ph.D.)--University of Wyoming, 2007. / Title from PDF title page (viewed on June 22, 2009). Includes bibliographical references (p. 120-127).

Thesis (Ph. D.)--Virginia Polytechnic Institute and State University, 1993. / Vita. Abstract. Includes bibliographical references (leaves 152-156). Also available via the Internet.

In this thesis, a Reinforcement Learning (RL) method called Sarsa is used to dynamically tune a PI-controller for a Continuous Stirred Tank Heater (CSTH) experimental setup. The proposed approach uses an approximate model to train the RL agent in the simulation environment before implementation on the real plant. This is done in order to help the RL agent initially start from a reasonably stable policy. Learning without any information about the dynamics of the process is not practically feasible due to the great amount of data (time) that the RL algorithm requires and safety issues. The process in this thesis is modeled with a First Order Plus Time Delay (FOPTD) transfer function, because almost all of the chemical processes can be sufficiently represented by this class of transfer functions. The presence of a delay term in this type of transfer functions makes them inherently more complicated models for RL methods. RL methods should be combined with generalization techniques to handle the continuous state space. Here, parameterized quadratic function approximation compounded with k-nearest neighborhood function approximation is used for the regions close and far from the origin, respectively. Applying each of these generalization methods separately has some disadvantages, hence their combination is used to overcome these flaws. The proposed RL-based PI-controller is initially trained in the simulation environment. Thereafter, the policy of the simulation-based RL agent is used as the starting policy of the RL agent during implementation on the experimental setup. As a result of the existing plant-model mismatch, the performance of the RL-based PI-controller using this primary policy is not as good as the simulationresults; however, training on the real plant results in a significant improvement in this performance. On the other hand, the IMC-tuned PI-controllers, which are the most commonly used feedback controllers are also compared and they also degrade because of the inevitable plant-model mismatch. To improve the performance of these IMC-tuned PI-controllers, re-tuning of these controllers based on a more precise model of the process is necessary. The experimental tests are carried out for the cases of set-point tracking and disturbance rejection. In both cases, the successful adaptability of the RL-based PI-controller is clearly evident. Finally, in the case of a disturbance entering the process, the performance of the proposed model-free self-tuning PI-controller degrades more, when compared to the existing IMC controllers. However, the adaptability of the RL-based PI- controller provides a good solution to this problem. After being trained to handle disturbances in the process, an improved control policy is obtained, which is able to successfully return the output to the set-point. / Process Control

No description available.

Zugl.: Freiburg (Breisgau), Univ., Diss., 2005

Techn. Universiẗat, Diss., 2005--Darmstadt.

[pt] Com o aumento da complexidade das aplicações de banco de dados a tarefa de sintonizar os sistemas de banco de dados para melhorar os tempos de resposta das consultas submetidas tem se tornado mais difícil. O uso de visões materializadas é uma forma de melhorar o desempenho do acesso aos dados. Trata-se de um objeto de um banco de dados relacional que armazena um conjunto de dados pré-calculados resultante de uma consulta. Esta dissertação propõe a criação automática de visões materializadas. Para tal utiliza uma arquitetura de auto-sintonia não-intrusiva, isto é, sem modificar o código do SGBD. São propostas métricas para estimar os custos envolvendo visões materializadas permitindo que se determine seu benefício efetivo. São apresentadas heurísticas para a tomada de decisão automática quanto à criação de visões materializadas em função de uma determinada carga de trabalho. São apresentados testes e simulações com o benchmark TPC-H na presença do SGBD MS SQL Server 2008. / [en] As database applications become more complex, tuning database systems in order to improve query response times also become harder. One could consider materialized views, a relational database object that stores data resulting from specific queries, to obtain better performances. This dissertation proposes the automatic creation of materialized views. A non-intrusive architecture is used in order to keep DBMS source codes unchanged. There is a need to estimate the creation cost and the benefits obtained from the presence of the materialized views. Heuristics are proposed to help with the automatic decision upon creation of these materialized views for a given workload. Simulation tests for TPC-H benchmark and MS SQL Server DBMS are presented.