Inpipes produktion av liners sker idag till stor del manuellt. I syfte att minska variansen på produktkvalitet, råvaror och öka processens hastighet undersöktes möjligheterna att elektroniskt mäta utbuktningarna som uppstod i processen. Därefter undersöktes processens egenskaper inför att använda mätvärdena i en passande reglerstruktur.  Efter en urvalsprocess i samråd med Inpipe och iLogik ansågs två av fem mätmetoder intressanta att gå vidare med i projektet. Utifrån dem togs beslut att undersöka hur beröringsfria givare med ultraljud och optisk teknik presterade på att mäta utbuktningarnas höjd. Mätningar utfördes både i testmiljö och i fabrik med laser, ultraljudsgivare och manuellt. Resultat visade att ultraljudsgivaren hade störst precision där optiska givaren tenderade att ge ett konstant mätfel. Processens utmanande egenskaper inför en reglering av bulorna fastslogs till att främst innefatta långsamma hartsflöden, varierande produktlängd och ytvariationer. Trots utmaningar som vågor på produktens yta samt ultraljudsgivarens mätavvikelse på lutande produkt resulterade arbetet i att ultraljudsgivaren bör fungera tillfredställande till att mäta utbuktningarnas höjdförändringar och implementeras i en reglering. Utifrån processens egenskaper och inom ramen för denna studie anses framkoppling som en lämpad reglerstruktur, däremot krävs det fortsatta undersökningar för att bekräfta eller dementera det. / Inpipes production of liners is today substantially done manually, in order to reduce the variance of product quality, raw material and production speed, the possibility to electronically measure the swell of the product in the process was examined. The property of the industrial process was also examined to find a suitable regulation structure.  A presentation of different methods of measurement was conducted, Inpipe chose two of them as specially interesting. It included measurement of ultrasound and optical laser sensors and decision to examine their precision on the liners swell was made. The measurement was done in controlled environment on a product dummy and in factory on the real liner. The ultrasound sensor proved to give the highest precision while the laser sensor indicated to measure a constant offset. The most challenging properties of the industrial process was determined to be slow resin flows, varying product length and surface variation on the product. Despite challenges as wave on the liners surface and the ultrasounds sensors inability to measure the liners slope, the project led to recommend the ultrasound sensor for measure the height of the liners swell and use the value in a regulation of the process. In theory feed forward regulation seems to suit the properties of the industrial process, but further work must be done to really confirm or deny that claim.

In recent years, machine learning applications have gained great attention in the wind power industry. Among these, artificial neural networks have been utilized to predict the fatigue loads of wind turbine components such as rotor blades. However, the limited number of contributions and differences in the used databases give rise to several questions which this study has aimed to answer. Therefore, in this study, 5-min SCADA data from the Lillgrund wind farm has been used to train two feed-forward neural networks to predict the fatigue loads at the blade root in flapwise and edgewise directions in the shape of damage equivalent loads.The contribution of different features to the model’s performance is evaluated. In the absence of met mast measurements, mesoscale NEWA data are utilized to present the free flow condition. Also, the effect of wake condition on the model’s accuracy is examined. Besides, the generalization ability of the model trained on data points from one or multiple turbines on other turbines within the farm is investigated. The results show that the best accuracy was achieved for a model with 34 features, 5 hidden layers with 100 neurons in each hidden layer for the flapwise direction. For the edgewise direction, the best model has 54 features, 6 hidden layers, and 125 neurons in each hidden layer.For a model trained and tested on the same turbine, mean absolute percentage errors (MAPE) of 0.78% and 9.31% are achieved for the flapwise and edgewise directions, respectively. The seen difference is argued to be a result of not having enough data points throughout the range of edgewise moments. The use of NEWA data has been shown to improve the model’s accuracy by 10% for MAPE values, relatively. Training the model under different wake conditions did not improve the model showing that the wake effects are captured through the input features to some extent. Generalization of the model trained on data points from one turbine resulted in poor results in the flapwise direction. It was shown that using data points from multiple turbines can improve the model’s accuracy to predict loading on other turbines.

Developing and improving upon a good empirical model for an engine can be time-consuming and costly. The goal of this thesis has been to evaluate data-driven modelling, specifically neural networks, to see how well it can handle training for some static models like the mass flow of air into the cylinder, mean effective pressure and pump mean effective pressure but also for transient modelling, specifically the exhaust gas temperature. These models are evaluated against the classical empirical models to see if neural networks are a viable modelling option. This is done with five different types of neural networks which are trained. These are the feed-forward neural network, Nonlinear autoregressive exogenous model network, layer recurrent network, long short term memory network and gated recurrent network.The inputs were determined by looking at more simple physical models but also looking at the covariance to determine the usefulness of the input. If the calculation time is small for the specific network, the neural network structure is tested and optimized by training many networks and finding the median/mean result for that specific test.The result has shown that the static models are handled very well by the most simple feed-forward network. For the exhaust temperature, both NARX and Layer recurrent network could predict and handle it well giving results very close to the empirical models and could be a viable option for transient modelling, on the other hand, Long short term memory, gated recurrent network and the feed-forward network had trouble predicting the exhaust gas temperature and returned bad results while training.

No description available.

No description available.

Les peignes de fréquences optiques, issus de diodes à blocage de modes, font partie des candidats potentiels pour les réseaux de transmission à multiplexage en longueurs d’onde (WDM). Cependant, les modes composant leur peigne, exhibent généralement des largeurs de raie optiques relativement élevées ( 1-100 MHz), rendant ainsi incompatible leur utilisation sur un réseau WDM employant des formats de modulation avancés d’ordre supérieur. Cette thèse étudie, une solution pour palier à cette limitation. La technique utilisée, dite d’asservissement à correction aval hétérodyne, effectue un traitement du flux lumineux en sortie du laser sans agir sur ce dernier, permettant de réduire le bruit de fréquence présent sur chacune des raies et par conséquent leur largeur de raie optique. Dans une première approche, la technique est appliquée à un laser mono-fréquence. Cela a permis d’une part de valider son fonctionnement et d’autre part d’identifier les limites intrinsèques du dispositif expérimental mis en place. Ainsi, nous démontrons que le niveau de bruit de fréquence minimum permis par notre système, correspond à un spectre optique de largeur de raie optique instantanée de 50 Hz et une largeur de raie de 1,6 kHz pour un temps d’observation de 10 ms. La technique est par la suite appliquée à une diode à blocage de modes actif. Le peigne de fréquences optiques ainsi généré, est composé de 21 modes, ayant tous une largeur de raie optique intrinsèque inférieure à 7 kHz, dont 9 modes sont sub-kHz. Pour un temps d’observation du spectre optique de 10 ms, ces modes exhibent tous une largeur de raie d’environ 37 kHz. Nous démontrons ainsi l’impact de la gigue d’impulsions sur les performances de la technique et nous soulignons l’intérêt d’une telle cohérence, pour le domaine des télécommunications optiques cohérentes (transmissions WDM cohérentes de type m-QAM avec des constellations d’ordre élevé, compatibles avec des débits multi-Tbit/s par raie). En dernier lieu, nous abordons une seconde technique consistant à pré-stabiliser la fréquence d’un laser par asservissement en boucle fermée. Elle repose sur l’utilisation d’un interféromètre à fibre déséquilibré comme référence pour réduire le bruit de fréquence d’un laser, situé particulièrement en basses fréquences. Appliquée à un laser mono-fréquence, elle a permis de réduire son bruit de fréquence technique conduisant ainsi à une nette amélioration de sa largeur de raie intégrée sur 3 ms, de 224 kHz à 37 kHz. Ce premier résultat représente un bon support vers l’exploration du potentiel des diodes à blocage de modes pour des applications métrologiques. / Optical frequency combs obtained from mode-locked laser diodes are potential candidates for WDM networks. However, their lines exhibit usually a broad optical linewidth ( 1-100 MHz). Thus their use is incompatible for high order modulation formats WDM based systems. This thesis investigates one solution to overcome this limitation. It consists of using a feed-forward heterodyne technique to reduce the frequency noise of each comb-line and consequently their optical linewidths. In a first approach, the technique is applied to a single-mode laser. This allowed us to validate its proper working and to identify the intrinsic limits of the experimental device set up. The latter analysis enabled us to reveal that the minimum achievable frequency noise level by our system, corresponds to a 50 Hz intrinsic optical linewidth spectrum and a 1,6 kHz optical linewidth based on 10 ms observation time. This technique is then applied to an actively mode-locked laser diode demonstrating, at our system output, a 21-line optical frequency comb with intrinsic optical linewidths reduced to below 7 kHz. It is worth noting that 9 among them, exhibit sub-kHz linewidths. For an observation time of 10 ms, all lines share the same optical linewidth, almost equal to 37 kHz. We thus show that the timing jitter impacts the technique performances. We also highlight the relevance of such coherence level for coherent optical communication. Lastly, we study a laser frequency pre-stabilization technique based on a locking to an unbalanced fiber interferometer. When applied to a single-mode laser, the technique showed a reduction of its technical frequency noise, thus leading to a clear improvement of its integrated optical linewidth from 224 kHz to 37 kHz for 3 ms observation time. This first result provides a good support towards the exploration of mode-locked laser diodes potential for metrological applications.

Es planteja la fonamentació d'un model integrat de formació humana, desenvolupament i intervenció organitzacional en un context pedagògic socioeducatiu. S'explicita un nucli d'intel·ligibilitat epistemològic, (Gergen, 1996) i es desenvolupen els seus continguts metateòrics, teòrics i metodològics Per a situar el model en una perspectiva metateòrica constructivista i construccionista social, s'examinen dos canvis considerats rellevants en les explicacions epistemològiques de la ciència que donen suport a aquesta perspectiva: el gir lingüístic i el que els biòlegs biocibernètics de la modernitat anomenen gir ontològic. Aquests fonaments científics del constructivisme, es recullen en el disseny del model que es fa càrrec de tres explicacions plantejades per aquests girs així com de les seves conseqüències. El disseny del model assumeix aquestes distincions i estableix un creuament amb els desenvolupaments post-analítics del gir lingüístic en la comprensió performativa del llenguatge, per a la concepció de l'organització com una xarxa de converses. Assumeix el caràcter emergent de la conducta lingüística en l'acoblament amb el medi. aquest acoblament estructural s'entén enactivament mitjançant l'ús recursiu del llenguatge, per a generar compromisos vers un aprenentatge que integri les persones al sistema. El disseny reencuadra la investigació-acció utilitzant eines de cibernètica de primer ordre en una cibernètica de segon ordre, mitjançant converses recurrents per articular les metes de l'organització amb l'espai relacional. Incorpora metodologies de primera persona performatives i comunicacionals com eines per integrar la dimensió dels valors d'un mateix amb les metes de l'organització amb l'objectiu de reproduir-la mitjançant la integració social. Opera mitjançant el treball en grup per tal que les converses es situïn en la segona persona, (jo-altres), per tal de possibilitar l'emergència de la primera persona.Assumeix la distinció de les dues dimensiones del ésser cognitiu (Varela 2000): una d'enllaç mitjançant el llenguatge per la coderiva del sistema amb el medi, i una dimensió d'excedent de significació, és a dir, d'interpretació per a la reformulació de la experiència en el llenguatge (narrativa), que conservi aquest acoblament amb el medi. Per tal de reduir les pertorbacions que provoca el trencament de l'acoblament emergeix la interpretació com l'alternativa per a generar possibilitats d'acció; passar d'un estat present amb trencaments per fer emergir un estat desitjat com a futur viable. En aquesta escletxa que s'obre en l'espai relacional entre l'estat present i l'estat desitjat pren sentit l'ús d'aquestes eines de conversa per a accedir a un rerefons amb mirada compartida. L'estat desitjat acoblat al llenguatge opera com un feed-forward positiu per a la construcció de sentit. S'ajuda a més d'una visió binocular del canvi, analitzant dos series categorials de pensament en la modernitat. Finalment es descriu l'aplicació i els resultats del Model en una organització de serveis de salut. / Se plantea la fundamentación de un modelo integrado de formación humana, capacitación e intervención organizacional en un contexto pedagógico socio educativo. Explicita un núcleo de inteligibilidad epistemológico, (Gergen, 1996) desarrollando sus contenidos metateoricos, teóricos y metodológico Para situar el modelo en una perspectiva metateórico constructivista y construccionista social, se examinan dos cambios considerados relevantes en las explicaciones epistemológicas de la ciencia que dan sustento a esta perspectiva: el giro lingüístico y el llamado por los biólogos biocibernéticos giro ontológico de la modernidad. Estos fundamentos científicos del constructivismo contenidos en este último, se recogen en el diseño del modelo, que se hace cargo de tres explicaciones que dejan planteadas este giro y de las consecuencias que de ellas surgen para su formalización.- El diseño del modelo asume estas distinciones y realiza un cruce con los desarrollos post analíticos del giro lingüístico en la comprensión performativa del lenguaje, para la concepción de la organización como una red de conversaciones. Asume el carácter emergente de la conducta lingüística en el acoplamiento con el medio. Este acoplamiento estructural se entiende enactivamente mediante el uso recursivo del lenguaje, para generar compromisos para un aprendizaje que integre las personas al sistema. El diseño es un Modelo autopoiético que metodológicamente reencuadra la investigación-acción usando herramientas de cibernética de primer orden en una cibernética de segundo orden, mediante conversaciones recurrentes para articular las metas de la organización con el espacio relacional. Incorpora metodologías de primera persona performativas comunicacionales como herramientas para integrar la dimensión valórica del si mismo con las metas de la organización, para reproducirla mediante integración social. Opera mediante trabajo grupal para que las conversaciones se sitúen en la segunda persona, (yo-otros), para posibilitar la emergencia de la primera persona.Asume la distinción de las dos dimensiones del ser cognitivo (Varela 2000), una de enlace mediante el lenguaje para la coderiva del sistema con el medio, y una dimensión de excedente de significación, esto es, de interpretación para la reformulación de la experiencia en el lenguaje (narrativa), que conserve ese acoplamiento con el medio. Para reducir las perturbaciones que provocan el quiebre del acoplamiento emerge la interpretación como la alternativa para generar posibilidades de acción, pasar de un estado presente con quiebres para hacer emerger un estado deseado como futuro viable. En esa brecha que se abre en el espacio relacional entre el estado presente y el estado deseado hace sentido el uso de estas herramientas conversacionales para acceder a un trasfondo con mirada compartida. El estado deseado acoplado al lenguaje opera como un feed forward positivo para la construcción de sentido. Se apoya para ello, además, en una visión binocular del cambio, analizando dos series categoriales de pensamiento en la modernidad. Finalmente se describe la aplicación y resultados del Modelo en una organización de servicios de salud. / The foundation for an integrated model of human formation, training and organizational intervention within a socio-educational pedagogic context is presented here. It explicits an epistemological intelligibility core (Gergen, 1996) developing its theoretical, metatheoretical and methodological contents. To place the model into a metatheoretical social constructivism and social constructionist perspective, two relevant considered shifts in epistemological explanations of science that supports this perspective are reviewed: the linguistic turn and the so called by biocybernetical biologists ontological turn of modernity. These scientific foundations of constructivism contained in this turn, are brought into the model design that takes charge of three explanations put forward by the turn and the consequences arising from them for its formalization.The design of the model assumes these distinctions and performs a crossing with the post analytical developments of the linguistic turn in the performative comprehension of language, for conceiving the organization as a network of conversations. It assumes the emerging character of linguistic behavior in coupling to the environment. This structural coupling is enactively understood through the recursive use of language, to generate commitments for a learning that integrates persons to the system.The design is an autopoietical model that methodologically reframes research-action using tools of first order cybernetics within a second order cybernetics, through recurrent conversations to articulate organization's goals with the relational space. It incorporates communicational first person performative methodologies as tools to integrate the valoric dimension of the self with the organization's goals, to reproduce it through social integration. It operates through group work placing conversations in second person (I - others), for making the emergence of the first person possible.It assumes the distinction of the two dimensions of the cognitive being (Varela, 2000), one as a link of language for the co-derive of the system with the environment and one as signification surplus, that is, of interpretation for the reformulation of the experience in language (narrative), that keep that coupling with the environment.Interpretation emerges as the alternative for generating action possibilities to reduce perturbations that provoke the breaking of coupling, to go through a present state with breakings to an emergent wished state as a viable future. In this breach opened in the relational space between the present state and the wished state the use of these conversational tools makes sense to access a background with shared vision. The wished state coupled to language operates as a positive feed forward for constructing meaning. It also supports itself on a binocular vision of change, analyzing two categorical series of thought in modernity.Finally, the application and results of the model are described within a health services organization.

Abnormal electromagnetic wave propagation characterized by negative group velocity and consequently negative group delay (NGD) has been observed in certain materials as well as in artificially built structures. Within finite frequency intervals where an NGD phenomenon is observed, higher frequency components of the applied waveform are propagated with phase advancement, not delay, relative to the lower frequency components. These media have found use in many applications that require positive delay compensation and an engineered phase characteristic, such as eliminating phase variation with frequency in phase shifters, beam-squint minimization in phased array antenna systems, size reduction of feed-forward amplifiers and others. The three principal questions this thesis addresses are: can a generic formulation for artificial NGD structures based on electric circuit resonators be developed; is it possible to derive a quantitative functional relationship (asymptotic limit) between the maximum achievable NGD and the identified trade-off quantity (out-of-band gain); and, can a microwave circuit exhibiting a fully loss-compensated NGD propagation in both directions be designed and implemented? A generic frequency-domain formulation of artificial NGD structures based on electric circuit resonators is developed and characterized by three parameters, namely center frequency, bandwidth and the out-of-band gain. The developed formulation is validated through several topologies reported in the literature. The trade-off relationship between the achievable NGD on one hand, and the out-of-band gain on the other, is identified. The out-of-band gain is shown to be proportional to transient amplitudes when waveforms with defined “turn on/off” times are propagated through an NGD medium. An asymptotic limit for achievable NGD as a function of the out-of-band gain is derived for multi-stage resonator-based NGD circuits as well as for an optimally engineered linear causal NGD medium. Passive NGD media exhibit loss which can be compensated for via active elements. However, active elements are unilateral in nature and therefore do not allow propagation in both directions. A bilateral gain-compensated circuit is designed and implemented, which overcomes this problem by employing a dual-amplifier configuration while preserving the overall circuit stability.

Abnormal electromagnetic wave propagation characterized by negative group velocity and consequently negative group delay (NGD) has been observed in certain materials as well as in artificially built structures. Within finite frequency intervals where an NGD phenomenon is observed, higher frequency components of the applied waveform are propagated with phase advancement, not delay, relative to the lower frequency components. These media have found use in many applications that require positive delay compensation and an engineered phase characteristic, such as eliminating phase variation with frequency in phase shifters, beam-squint minimization in phased array antenna systems, size reduction of feed-forward amplifiers and others. The three principal questions this thesis addresses are: can a generic formulation for artificial NGD structures based on electric circuit resonators be developed; is it possible to derive a quantitative functional relationship (asymptotic limit) between the maximum achievable NGD and the identified trade-off quantity (out-of-band gain); and, can a microwave circuit exhibiting a fully loss-compensated NGD propagation in both directions be designed and implemented? A generic frequency-domain formulation of artificial NGD structures based on electric circuit resonators is developed and characterized by three parameters, namely center frequency, bandwidth and the out-of-band gain. The developed formulation is validated through several topologies reported in the literature. The trade-off relationship between the achievable NGD on one hand, and the out-of-band gain on the other, is identified. The out-of-band gain is shown to be proportional to transient amplitudes when waveforms with defined “turn on/off” times are propagated through an NGD medium. An asymptotic limit for achievable NGD as a function of the out-of-band gain is derived for multi-stage resonator-based NGD circuits as well as for an optimally engineered linear causal NGD medium. Passive NGD media exhibit loss which can be compensated for via active elements. However, active elements are unilateral in nature and therefore do not allow propagation in both directions. A bilateral gain-compensated circuit is designed and implemented, which overcomes this problem by employing a dual-amplifier configuration while preserving the overall circuit stability.

Increased oil prices and environmental issues have increased a need of loweringthe emissions from and the fuel consumption in heavy construction machines. Anatural solution to these issues is a lowered input power through downsizing ofthe engine. This implies a demand on higher transmission efficiency, in order tominimize the intrusion on vehicle performance. More specifically, alternatives tothe conventional torque converter found in heavier applications today, must beinvestigated. One important part of this is the task of controlling the transmissionwithout jeopardising the advantages associated with the torque converter, such asrobustness and controllability.In this thesis, an alternative transmission concept for a backhoe loader is investigated.The studied concept is referred to as a 2-mode Jarchow power-splittransmission, where a mechanical path is added to a hydrostatic transmission inorder to increase transmission efficiency. The concept is evaluated in computerbased simulations as well as in hardware-in-the-loop simulations, where a physicalhydrostatic transmission is exposed for the loads caused by the vehicle duringvarying conditions. The loads are in turn simulated according to developed modelsof the mechanical parts of the vehicle drive line.In total, the investigated concept can be used instead of the torque converterconcept, if the hydrostatic transmission is properly controlled. The results alsoshow that there is a high possibility that the combustion engine in the backhoeloader can be downsized from 64 kW to 55 kW, which would further increase thefuel savings and reduce the emissions.