| 11. |
|
|
| 12. |
|
|
| 13. |
|
|
| 14. |
- Alriksson, Peter
(författare)
-
State Estimation for Distributed and Hybrid Systems
- 2008
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- This thesis deals with two aspects of recursive state estimation: distributed estimation and estimation for hybrid systems. In the first part, an approximate distributed Kalman filter is developed. Nodes update their state estimates by linearly combining local measurements and estimates from their neighbors. This scheme allows nodes to save energy, thus prolonging their lifetime, compared to centralized information processing. The algorithm is evaluated experimentally as part of an ultrasound based positioning system. The first part also contains an example of a sensor-actuator network, where a mobile robot navigates using both local sensors and information from a sensor network. This system was implemented using a component-based framework. The second part develops, a recursive joint maximum a posteriori state estimation scheme for Markov jump linear systems. The estimation problem is reformulated as dynamic programming and then approximated using so called relaxed dynamic programming. This allows the otherwise exponential complexity to be kept at manageable levels. Approximate dynamic programming is also used to develop a sensor scheduling algorithm for linear systems. The algorithm produces an offline schedule that when used together with a Kalman filter minimizes the estimation error covariance.
|
|
| 15. |
- Bengtsson, Johan
(författare)
-
Closed-Loop Control of HCCI Engine Dynamics
- 2004
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The topic of the thesis is control of Homogeneous Charge Compression Ignition (HCCI) engine dynamics. HCCI offers a potential to combine high efficiency with very low emissions. In order to fulfill the potential benefits, closed-loop control is needed. The thesis discusses sensors, feedback signals and actuators for closed-loop control of the HCCI combustion. Closed-loop control of the HCCI combustion using ion current is demonstrated. Models of the HCCI dynamics suitable for purposes of control design are presented. It is shown that low-order models are sufficient to describe the HCCI dynamics. Models of HCCI combustion have been determined both by system identification and by physical modeling. Different methods for characterizing and controlling the HCCI combustion are outlined and demonstrated. In cases where the combustion phasing in a six-cylinder heavy-duty engine was controlled, either by a Variable Valve Actuation system using the inlet valve or a dual-fuel system, results are presented. Combustion phasing is a limiting factor of the load control and emission control performance. A system where control of HCCI on a cycle-to-cycle basis is outlined and cylinder individual cycle-to-cycle control on a six-cylinder heavy duty engine is presented. Various control strategies are compared. Model-based control, such as LQG and Model Predictive Control MPC, and PID control are shown to give satisfactory controller performance. An MPC controller is proposed as a solution to the problem of load-torque control with simultaneous minimization of the fuel consumption and emissions, while satisfying the constraints on cylinder pressure.
|
|
| 16. |
|
|
| 17. |
- Cervin, Anton
(författare)
-
Integrated Control and Real-Time Scheduling
- 2003
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The topic of the thesis is codesign of flexible real-time control systems. Integrating control theory and real-time scheduling theory, it is possible to achieve higher resource utilization and better control performance. The integration requires new tools for analysis, design, and implementation. The problem of scheduling the individual parts of a control algorithm is studied. It is shown how subtask scheduling can reduce the input-output latency in a set of control tasks. Deadline assignment under different scheduling policies is considered. A feedback scheduling architecture for control tasks is introduced. The scheduler uses feedback from execution-time measurements and feedforward from workload changes to adjust the sampling periods of a set of control tasks so that the combined performance of the controllers is optimized. The Control Server, a novel computational model for real-time control tasks, is presented. The model combines time-triggered I/O with dynamic, reservation-based task scheduling. The model provides short input-output latencies and minimal jitter for the controllers. It also allows control tasks to be treated as scalable real-time components with predictable performance. Two MATLAB-based toolboxes for analysis and simulation of real-time control systems have been developed. The Jitterbug toolbox evaluates a quadratic cost function for a linear control system with timing variations. The tool makes it possible to investigate the impact of delay, jitter, lost samples, etc., on control performance. The TrueTime toolbox facilitates detailed cosimulation of distributed real-time control systems. The scheduling and execution of control tasks is simulated in parallel with the network communication and the continuous process dynamics.
|
|
| 18. |
- Cescon, Marzia
(författare)
-
Modeling and Prediction in Diabetes Physiology
- 2013
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Diabetes is a group of metabolic diseases characterized by the inability of the organism to autonomously regulate the blood glucose levels. It requires continuing medical care to prevent acute complications and to reduce the risk of long-term complications. Inadequate glucose control is associated with damage, dysfunction and failure of various organs. The management of the disease is non trivial and demanding. With today’s standards of current diabetes care, good glucose regulation needs constant attention and decision-making by the individuals with diabetes. Empowering the patients with a decision support system would, therefore, improve their quality of life without additional burdens nor replacing human expertise. This thesis investigates the use of data-driven techniques to the purpose of glucose metabolism modeling and short-term blood-glucose predictions in Type I Diabetes Mellitus (T1DM). The goal was to use models and predictors in an advisory tool able to produce personalized short-term blood glucose predictions and on-the-spot decision making concerning the most adequate choice of insulin delivery, meal intake and exercise, to help diabetic subjects maintaining glycemia as close to normal as possible. The approaches taken to describe the glucose metabolism were discrete-time and continuous-time models on input-output form and statespace form, while the blood glucose short-term predictors, i.e., up to 120 minutes ahead, used ARX-, ARMAX- and subspace-based prediction.
|
|
| 19. |
- de Maré, Lena
(författare)
-
Feeding Strategies Based on Probing Control for E. coli and V. cholerae Cultivations
- 2006
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The recombinant DNA techniques have made it possible to produce many different proteins for a wide range of applications. The vector encoding for the recombinant protein is often inserted in the bacterium E. coli as it is a well studied and a well-known organism. To achieve a high productivity, it is important to reach a high cell density. This is obtained through fed-batch operation where the nutrient, usually glucose, is added continuously. The feeding strategy should be designed to avoid starvation and overfeeding. This is not an easy task as on-line measurements of key variables are normally not available. A probing feeding strategy using the measurements of the standard dissolved oxygen sensor is developed by M. Åkesson. The key idea is to superimpose pulses on the feed-rate and make use of the responses in the dissolved oxygen in a feedback algorithm. Also, when the maximum transfer capacity of the reactor is reached the feed-rate is decreased in order to keep the reactor working under aerobic conditions. The strategy has been successfully implemented in cultivations with different E. coli strains and on different scales. The probing feeding strategy is further developed in many aspects in this thesis. Firstly, it is implemented with good results also in V. cholerae cultivations, which shows how general the strategy is. Also, a model of a bio-reactor operating in fed-batch mode is presented and verified. The effect on the tuning rules of the probing controller is investigated. The probing feeding strategy is further improved for a more efficient product synthesis. This new fermentation technique manipulates the temperature when the maximum oxygen transfer capacity of the reactor is reached. The strategy consists of a mid-ranging controller structure and a modified probing controller. It is analysed and evaluated in experiments and simulations. Furthermore, some E. coli production strains need additions of amino acids or complex media besides the carbon nutrient to grow and produce the recombinant protein. The probing control concept is therefore extended in order to handle these situations. Feeding strategies for dual feeding of amino acids or complex media and glucose are developed.
|
|
| 20. |
- Dressler, Isolde
(författare)
-
Modeling and Control of Stiff Robots for Flexible Manufacturing
- 2012
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- To survive on a global market, small and medium size enterprises (SMEs) require affordable and competitive industrial automation for high quality flexible manufacturing. This thesis contributes to the development of robot concepts that fit the needs of SMEs. A major part of the thesis deals with the modeling of the three degree-of-freedom (DOF) Gantry-Tau parallel kinematic robot, which has the potential to fulfill the requirements on accuracy, mechanical stiffness and conceptual flexibility of a robot for SMEs. Additionally, concepts that aid the SMEs to achieve the required accuracy and a more intuitive robot operation were developed and evaluated. The modeling of the Gantry-Tau robot includes both kinematic and dynamic modeling. Based on the nominal kinematic model, kinematic error models were developed, as well as kinematics for the F1-type Gantry-Tau, a Gantry-Tau architecture extended to 6~DOF. The modeling was evaluated in kinematic calibration experiments. A rigid body model was derived and identified, including friction in the actuators. As noticeable flexible behaviour was observed, the compliance dynamics were identified by black box modeling. Kinematic calibration was not only considered for evaluation of the kinematic models developed, but it was also studied how to automize the kinematic calibration procedure, so that it can be executed by non-expert SME staff after a possible geometric reconfiguration of the robot. In the search of affordable, accurate and reusable measurement devices for kinematic calibration in SMEs, the usage of camera vision for kinematic calibration was evaluated. To make the programming of a robot trajectory fast and intuitive, lead-through programming was recently introduced. A new concept for lead-through programming in contact situations is proposed in this thesis, where two force sensors are used. While the first sensor is used for guiding the robot, the second force sensor measures the tool force, which can prevent damage of the tool or workpiece and can help to keep a steady contact between tool and surface. The concept was demonstrated in two example applications. A possibility to improve the performance for a repeatedly executed motion is iterative learning control (ILC). An ILC algorithm is evaluated on the Gantry-Tau robot, which uses an estimate of the tool motion, based on measurements from an accelerometer mounted at the end-effector plate and in addition to measurements on the motor side. The performance of the tool motion was shown to be considerably improved compared to the case when only motor side measurements are used.
|
|
