Sök i SwePub databas

  Utökad sökning

Träfflista för sökning "db:Swepub ;access:(free);mspu:(licentiatethesis);srt2:(1990-1994);srt2:(1993)"

Sökning: db:Swepub > Fritt online > Licentiatavhandling > (1990-1994) > (1993)

  • Resultat 1-10 av 16
Sortera/gruppera träfflistan
  • Andersson, Kent (författare)
  • Preparation and Characterisation of Sputtered Titanium- and Zirconium Nitride Optical Films
  • 1993
  • Licentiatavhandling (övrigt vetenskapligt)abstract
    • Multilayered interference coatings based on titanium- and zirconium nitride and designed for solar control have been prepared using reactive d c magnetron sputtering. Preparation effects and degradation mechanisms were investigated. It was shown that the quality of the nitride strongly depends on the degree of crystallinity in the underlying oxide. It has been shown that the nitride layer partly oxidizes as the top oxide layer is deposited. The degradation is enhanced with temperature. A thin sacrificial layer of aluminium deposited between successive depositions of nitride and oxide is shown to improve the optical performance of the coating as preparedm as well as after accelerated ageing tests.The optical properties of opaque and semitransparent films of zirconium nitride have been studied. A thorough investigation of the influence of composition, deposition rate, substrate temperature and film thickness on the optical response of the film was performed. Both photometric and ellipsometric methods were used to determine thicknesses and the optical constants at wavelengths ranging from 0.23 to 25 μm. The resulting values of n and k, in the wavelength intervals where these independent methods are applicable, have been shown to agree extremely well. The results so far indicate an even larger potential for zirconium nitride based solar control coatings as compared to the titanium nitride based.Access to optical constants derived from films of zirconium nitride of variable quality made multilayer modelling a powerful tool in the design and analysis of solar control coatings.
  • Hedbrant, Johan, 1959- (författare)
  • Särimner a computer model for diabetes education
  • 1993
  • Licentiatavhandling (övrigt vetenskapligt)abstract
    • In diabetes mellitus pancreas fails to produce insulin enough to maintain tissue utilisation of blood glucose. To maintain a normalised metabolism many diabetic patients have to provide insulin from injections or an infusion pump. The disease affects some percent of western population.The insulin provided by the patient has to be adjusted to food and physical exercise. This may be achieved by following rules from diabetes professionals, but also by self learning how food, insulin and physical exercise interact and influence the blood glucose. It seems that the patients who have this intuitive knowledge of the interactions are the ones who manages their diabetes treatment best and have the highest quality of life. The learning have, however, been performed by trial and error - to the price of severe inconvenience and bad metabolic control.A combination of computer science and diabetes physiology has resulted in the computer program Särimner. It is created to give the user a possibility to experiment with diabetes treatment. Food, insulin and physical exercise may be varied and the blood glucose is calculated. Sarimner provides a short-cut to obtaining knowledge since it allows experiments, stimulates discussions and let the user by himself formulate and test hypotheses regarding experienced problems. Since Särimner may be adjusted to look more alike an individual, the experiments are driven by the user's own curiosity. He may experiment with situations of importance to himself and finally make himself the expert of his own treatment situation.The way Särimner is designed, allows interested users to get "under the skin" of the model and study details in the physiological processes. This transparency makes it possible to search for explanations to treatment phenomena. One drawback with the model is that it is quite complex and requires some knowledge from a user with the ambition to understand all the processes.To measure the impact of Särimner training, 11 diabetic teenagers were evaluated with respect to metabolic control, emotional adjustment, focus of control, self-esteem and ability to discuss treatment phenomena. No control group was possible to recruit.The results indicated that the education had been useful for some individuals. They increased their knowledge and ability to discuss treatment situations, their sense of control over the diabetes treatment, their self esteem and furthermore Särimner education may have caused a reduction of diabetes related stress. However, an increased level of diabetes related guilt did occur in some individuals, possibly due to either increased knowledge or a more internalised focus of control.The models ability to look alike reality is depending on for which purpose it is used. Even though it would be theoretically possible to fit the model to an individual, such an experiment would not be performable in reality since Särimner requires input data from the physiology which is impossible to measure. The properties of the model are, however, adequate for illustrating several treatment situations on a phenomenological level.
  • Jeppsson, Ulf (författare)
  • On the Verifiability of the Activated Sludge System Dynamics
  • 1993
  • Licentiatavhandling (övrigt vetenskapligt)abstract
    • Wastewater treatment processes are inherently dynamic because of variations in the influent flow rate, concentration, and composition. The adaptive behaviour of the microorganisms further emphasizes this fact. Mathematical models and computer simulations are essential to describe, predict, and control the complicated interactions of the processes. Any attempt to model all details of the various reaction mechanisms are, however, destined to fail due to lacking knowledge and the extreme complexity required for such models. A reduced order dynamic model for an activated sludge process performing carbonaceous removal, nitrification, and denitrification is presented herein. The identifiability of the model is investigated using both off-line and on-line methods and its dynamic behaviour is verified against simulations of a recognized model - the IAWPRC Activated Sludge Model No. 1. The required data for the identification algorithms is based on directly measurable real time data. The simplified model may serve as a tool for predicting the dynamic behaviour of an activated sludge process since the parameters under varying operating conditions can be tracked on-line. The model is aimed for operation and control purposes as an integral part of a hierarchical control structure.
  • Landelius, Tomas (författare)
  • Behavior Representation by Growing a Learning Tree
  • 1993
  • Licentiatavhandling (övrigt vetenskapligt)abstract
    • The work presented in this thesis is based on the basic idea of learning by reinforcement, within the theory of behaviorism. The reason for this choice is the generality of such an approach, especially that the reinforcement learning paradigm allows systems to be designed which can improve their behavior beyond that of their teacher. The role of the teacher is to define the reinforcement function, which acts as a description of the problem the machine is to solve.Learning is considered to be a bootstrapping procedure. Fragmented past experience, of what to do when performing well, is used for response generation. The new response, in its turn, adds more information to the system about the environment. Gained knowledge is represented by a behavior probability density function. This density function is approximated with a number of normal distributions which are stored in the nodes of a binary tree. The tree structure is grown by applying a recursive algorithm to the stored stimuli-response combinations, called decisions. By considering both the response and the stimulus, the system is able to bring meaning to structures in the input signal. The recursive algorithm is first applied to the whole set of stored decisions. A mean decision vector and a covariance matrix are calculated and stored in the root node. The decision space is then partitioned into two halves across the direction of maximal data variation. This procedure is now repeated recursively for each of the two halves of the decision space, forming a binary tree with mean vectors and covariance matrices in its nodes.The tree is the system's guide to response generation. Given a stimulus, the system searches for responses likely to result in highly reinforced decisions. This is accomplished by treating the sum of the normal distributions in the leaves as distribution describing the behavior of the system. The sum of normal distributions, with the current stimulus held fixed, is finally used for random generation of the response.This procedure makes it possible for the system to have several equally plausible responses to one stimulus. Not applying maximum likelihood principles will make the system more explorative and reduce its risk of being trapped in local minima.The performance and complexity of the learning tree is investigated and compared to some well known alternative methods. Presented are also some simple, yet principally important, experiments verifying the behavior of the proposed algorithm.
Skapa referenser, mejla, bekava och länka
  • Resultat 1-10 av 16
pil uppåt Stäng

Kopiera och spara länken för att återkomma till aktuell vy