Despite the vast amount of delivered theoretical results, regarding the topic of controller design, more than 90% of the controllers used in industry (petro-chemical, pulp and paper, steel, mining, etc) are of PID type (P, PI, PII, PD). This shows the importance of progressing in the elaboration of methods that consider restricted complexity controllers for practical applications, and that are computationally simple. Iterative Feedback Tuning (IFT) stands out as a new solution that takes into account both constraints. It belongs to the family of model-free controller tuning methods. It was developed at Cesame in the nineties and, since then, many real applications of IFT have been reported. This algorithm minimizes a cost function by means of a stochastic gradient descent scheme. In spite of the fact that the method has had an unexpected success in the tuning of real processes, a number of issues had not been fully covered yet. This thesis focuses on two aspects of this set of uncovered theoretical points: the convergence rate of the algorithm and a robust estimation of its gradient. Optimal prefilters, left as a degree of freedom for the user in the first formulation of IFT, are computed at each experiment. Their application allows a reduction in the covariance of the gradient estimate. Depending on what particular aspect the user is interested in improving, one optimal prefilter is selected. Monte-Carlo simulations have shown an enhancement with regards to a constant prefilter. A flexible arm set-up mounted in our robotics laboratory is used as a test bed to compare a model-based controller design algorithm with a model-free controller tuning method. The comparison is performed with some specifications defined beforehand. The same set-up plus a couple of air-jets serves as a tester for our theoretical results, when the rejection of a perturbation is the ultimate objective. Both cases have confirmed the predicted good behaviour offered by IFT.

Métodos de controle baseado em dados utilizam dados de operação da planta para minimizar algum critério de desempenho, tipicamente quadrático, sem que seja necessário o conhecimento de um modelo da planta. O Iterative Feedback Tuning (IFT) é um destes métodos, utilizando dados obtidos da operação da planta em malha fechada para gerar uma estimativa não-polarizada do gradiente do critério de desempenho, que é então minimizado através de algum variante do método de Newton. Alguns resultados da literatura tratam da aplicação do IFT em sistemas não-lineares, sempre sob a condição de diferenciabilidade da não-linearidade (linearização em torno do ponto de operação). Um exemplo prático da aplicação do método na sintonia do controlador de um servomotor com folga é o único resultado encontrado. Este trabalho investiga, através de casos simulados e práticos, a aplicabilidade do método IFT em sistemas com não-linearidades não-diferenciáveis comumente encontradas em sistemas mecânicos, como a folga e a zona-morta, e a saturação, presente em todo e qualquer sistema físico. / Data based control methods use data collected from a plant’s “regular” operation to minimize a performance criterion, typically in quadratic form, without any information about a system’s model. Iterative Feedback Tuning (IFT) is one such method that uses closed-loop operating data to generate an unbiased estimate of the cost criterion gradient, which is then minimized with some form of the Newton method. Results in the literature for IFT applied to non-linear systems require that the non-linearity is differentiable (basically, linearizing the system around an operating point). One practical application of the algorithm to tune a controller for a servomotor system with backlash is the only result found. This work investigates, using simulated and practical examples, the suitability of the use of IFT in systems with non-differentiable non-linearities commonly found in mechanical systems, such as backlash and dead-zone, and saturation, found in every physical system.

Métodos de controle baseado em dados utilizam dados de operação da planta para minimizar algum critério de desempenho, tipicamente quadrático, sem que seja necessário o conhecimento de um modelo da planta. O Iterative Feedback Tuning (IFT) é um destes métodos, utilizando dados obtidos da operação da planta em malha fechada para gerar uma estimativa não-polarizada do gradiente do critério de desempenho, que é então minimizado através de algum variante do método de Newton. Alguns resultados da literatura tratam da aplicação do IFT em sistemas não-lineares, sempre sob a condição de diferenciabilidade da não-linearidade (linearização em torno do ponto de operação). Um exemplo prático da aplicação do método na sintonia do controlador de um servomotor com folga é o único resultado encontrado. Este trabalho investiga, através de casos simulados e práticos, a aplicabilidade do método IFT em sistemas com não-linearidades não-diferenciáveis comumente encontradas em sistemas mecânicos, como a folga e a zona-morta, e a saturação, presente em todo e qualquer sistema físico. / Data based control methods use data collected from a plant’s “regular” operation to minimize a performance criterion, typically in quadratic form, without any information about a system’s model. Iterative Feedback Tuning (IFT) is one such method that uses closed-loop operating data to generate an unbiased estimate of the cost criterion gradient, which is then minimized with some form of the Newton method. Results in the literature for IFT applied to non-linear systems require that the non-linearity is differentiable (basically, linearizing the system around an operating point). One practical application of the algorithm to tune a controller for a servomotor system with backlash is the only result found. This work investigates, using simulated and practical examples, the suitability of the use of IFT in systems with non-differentiable non-linearities commonly found in mechanical systems, such as backlash and dead-zone, and saturation, found in every physical system.

Métodos de controle baseado em dados utilizam dados de operação da planta para minimizar algum critério de desempenho, tipicamente quadrático, sem que seja necessário o conhecimento de um modelo da planta. O Iterative Feedback Tuning (IFT) é um destes métodos, utilizando dados obtidos da operação da planta em malha fechada para gerar uma estimativa não-polarizada do gradiente do critério de desempenho, que é então minimizado através de algum variante do método de Newton. Alguns resultados da literatura tratam da aplicação do IFT em sistemas não-lineares, sempre sob a condição de diferenciabilidade da não-linearidade (linearização em torno do ponto de operação). Um exemplo prático da aplicação do método na sintonia do controlador de um servomotor com folga é o único resultado encontrado. Este trabalho investiga, através de casos simulados e práticos, a aplicabilidade do método IFT em sistemas com não-linearidades não-diferenciáveis comumente encontradas em sistemas mecânicos, como a folga e a zona-morta, e a saturação, presente em todo e qualquer sistema físico. / Data based control methods use data collected from a plant’s “regular” operation to minimize a performance criterion, typically in quadratic form, without any information about a system’s model. Iterative Feedback Tuning (IFT) is one such method that uses closed-loop operating data to generate an unbiased estimate of the cost criterion gradient, which is then minimized with some form of the Newton method. Results in the literature for IFT applied to non-linear systems require that the non-linearity is differentiable (basically, linearizing the system around an operating point). One practical application of the algorithm to tune a controller for a servomotor system with backlash is the only result found. This work investigates, using simulated and practical examples, the suitability of the use of IFT in systems with non-differentiable non-linearities commonly found in mechanical systems, such as backlash and dead-zone, and saturation, found in every physical system.

Moore’s law states that the number of transistors in an Integrated Circuit (IC) doubles every two years. Ever-increasing performance in mask writing machinery is therefore required being the first step in the manufacturing process. Many factors affect the quality of the end product, with the motion control system playing an important role. This thesis analyzes the performance of the motion controller for the positioning system in a mask writer application. The target motion in the X-axis in the mask writer is by design highly repetitive and predictable. As of today a feedforward-feedback controller is used, tuned for low deviation during writing. In this thesis it is found that the motion control can be improved by exploiting the repetitive nature of the motion task. Two iterative methods are explored, Iterative Feedback Tuning (IFT) and Iterative Learning Control (ILC). IFT is implemented as a parameter optimizing method for the existing Proportional-Integral-Derivative (PID) controller. Given suboptimal initial parameters the algorithm converges to a global minimum using a cost function to minimize total deviation and constraints on the maximum deviation. With the optimized parameter settings an improvement of a 31 % decrease in total deviation is seen compared to the default setting. ILC is implemented as a replacement to the current controller in an exposure motion. With the use of saved data from previous iterations the control signal is updated and refined to better suit the target motion. ILC is a promising method within high precision motion control by virtue of not needing a model of the system and its ability to suppress reoccurring disturbances. The algorithm achieves an improvement of a 94% decrease in total deviation during writing compared to the current controller. However, with this implementation long term stability is not guaranteed. A stable implementation of the algorithm tested on a test rig achieves an improvement of a 79.8% decrease in deviation during writing compared to the current feedforward-feedback controller. Additionally, correlations between parameter values of the current feedback controller and servo characteristics are analyzed to aid in the manual tuning process. Tuning the PID controller for fast rise time decreases the total deviation during writing. The derivative gain in the controller should be high to decrease the overshoot caused by the aggressive controller. This will induce some oscillations into the system, however not at the cost of performance as a result of the smooth motion during writing.