No description available.

Relays and fuses are used in a multitude of applications to deliver and control current to its consumers. To improve the controllability of both the fuse and relay function, they can be moved to software. This will not only increase the number of available features but can also give live feedback of the systems current state. For an electronic fuse to work the software needs to be able to measure the amount of current flowing through the circuit. For this a voltage drop can be measured across a series resistor. This can be implemented in several ways and will because of this utilize the resources of the controlling microcontroller differently. This thesis focuses on the software implementation of two current measuring circuits and will characterize the behavior of these in terms of making it simple to compare their advantages and disadvantages. Any hardware related aspect found to have an impact on the result will be taken into account as well. Two implementations, using two different integrated circuits, are tested against each other. MAX4172 is fully analog while the INA226 is half digital. A shorter study of current sensing in general is made but the foundation of the study is the datasheets and characteristics of the components chosen. In this thesis, both the hardware and the software is designed and built which end with gathering data in two different electronic fuse scenarios. The thesis concludes that the two implementations as electronic fuse is at least a factor of 10 faster than the traditional breaking fuse. It also concludes that both implementations have their advantages and either may be the right solution depending on the intended end application. The fully analog solution is generally faster but is more susceptible to external noise and bad hardware design whereas the half digital is slower but more forgiving and thus shows of more precise measurements out of the box. / Relä och säkringar används i flera olika applikationer för att leverera och kontrollera strömmen till förbrukaren. Genom att förflytta denna uppgift till mjukvara kan detta förbättras genom att ge utökad funktionalitet samt realtidsövervakning av systemet. För att en mjukvara ska kunna kontrollera strömmen i en krets måste denna kunna mätas. Detta kan göras genom att mäta spänningsfallet över ett motstånd kopplat i serie med förbrukaren. Hur man väljer att implementera denna metod avgör vilka kravs som ställs på den kontrollerande mikrokontrollern. Denna studie kommer fokusera på mjukvaruimplementationen av två olika kretsar för strömmätning och har för avsikt att karakterisera dessa för att skapa en tydlig jämförelse av dess för- och nackdelar. De aspekter av hårdvaruimplementationen som påverkar detta resultat kommer också diskuteras. Två implementationer, bestående av två olika integrerade kretsar, ställs mot varandra. MAX4172 är en helt analog lösning medans INA226 är delvis digital. En kortare studie av strömmätning kommer utföras men grunden av denna studie består i huvudsak av datablad och karakteristik av de utvalda komponenterna. Både mjukvara och hårdvara är utvecklad och tillverkad för att kunna utföra test och inhämta den data som krävs för analysen. Studien visar att båda lösningar är åtminstone en faktor 10 snabbare än en traditionell säkring. Den visar också att en lösning inte är bättre än den andra i alla aspekter. Båda implementationerna kan vara den rätta lösningen beroende på de krav den tilltänkta produkten har. Den helt analoga lösningen är generellt snabbare men lider av en större känslighet mot störningar vilket ställer högre krav på hårdvarudesignen. Den delvis digitala lösningen är långsammare men betydligt mer förlåtande i dess implementation och har också större noggrannhet som resultat.

The determination of the pre-arcing time of a fuse given a specific current is crucial during the selection process if the fuse is to be effective for the given application. Often the data supplied by the manufacturer is enough to determine this time, however this is not always the case. For this purpose a fuse model of a diazed fuse has been developed in order to simulate the pre-arcing times in the range of 2 ms to 2 s. The model is implemented in MATLAB and simulates the pre-arcing time for a given current together with fuse parameters such as the materials and the geometry of the fuse. The model has been tested experimentally by running specific currents designed to represent a wide variety of currents which a fuse might be subjected to. The tests were done on 10A gG fuses for which the current through the fuse was logged and then used as input to the fuse model, the simulated time and the measured time was then compared. The simulated times are in general faster than the measured pre-arcing times, this is mainly due to the rate of change of thermal conductivity of the sand with respect to temperature being approximated in conjunction with thermal radiation not being taken into consideration. More testing needs to be done to validate the model for different current ratings and fuse-element designs. / Att bestämma smälttiden för en säkring som utsätts för en given ström är viktigt för att kunna avgöra om dess prestanda är tillräcklig för en viss applikation. I de flesta fall är det möjligt att använda tillverkarens data tillsammans med vågformen för att bestämma smälttiden, men det kan även inträffa att detta inte är möjligt. För dessa fall har en modell av en diazed säkring utvecklas för att simulera smälttiden inom 2 ms till 2 s. Modellen har implementerats i MATLAB och beräknar smälltiden för en given ström och parametrar för säkringen. Modellen har testats experimentellt genom att säkringar har utsatts för en mängd olika strömmar som representerar de olika strömmar som säkringen kan utsättas för. Testerna gjordes på 10 A gG säkringar där strömmen genom säkringen loggades för att sedan användas som ingångsvärde till modellen, tiderna för den uppmätta smälttiden och den beräknades smälltiden jämfördes sedan. De simulerade smälttiderna är lägre jämfört med de uppmätta, detta beror dels på att förändringen av värmeledningsförmågan med temperaturökning hos sanden är okänd och är uppskattad från referenser och dels på att värmestrålning inte har ingått i modellen. Mer tester behöver utföras för att validera modellen mot fler strömstyrkor och designer. / Kandidatexjobb i elektroteknik 2020, KTH, Stockholm

In this dissertation, we introduce a novel text representation method mainly used for text classification purpose. The presented representation method is initially based on a variety of closeness relationships between pairs of words in text passages within the entire corpus. This representation is then used as the basis for our multi-level lightweight ontological representation method (TOR-FUSE), in which documents are represented based on their contexts and the goal of the learning task. The method is unlike the traditional representation methods, in which all the documents are represented solely based on the constituent words of the documents, and are totally isolated from the goal that they are represented for. We believe choosing the correct granularity of representation features is an important aspect of text classification. Interpreting data in a more general dimensional space, with fewer dimensions, can convey more discriminative knowledge and decrease the level of learning perplexity. The multi-level model allows data interpretation in a more conceptual space, rather than only containing scattered words occurring in texts. It aims to perform the extraction of the knowledge tailored for the classification task by automatic creation of a lightweight ontological hierarchy of representations. In the last step, we will train a tailored ensemble learner over a stack of representations at different conceptual granularities. The final result is a mapping and a weighting of the targeted concept of the original learning task, over a stack of representations and granular conceptual elements of its different levels (hierarchical mapping instead of linear mapping over a vector). Finally the entire algorithm is applied to a variety of general text classification tasks, and the performance is evaluated in comparison with well-known algorithms.

Au sein d'un convertisseur la défaillance d'un composant de puissance est un événement critique tant par le risque d'explosion du boitier et sa propagation au sein du système (forte énergie stockée dans l'alimentation) que par l'interruption de service qui en découle (systèmes embarqués, systèmes de production en flux tendu). Ce mémoire de thèse propose un ensemble de solutions nouvelles portant sur la problématique de l'isolement ultime de défauts "derniers secours" d'une part, et sur la connexion "automatique" et à faible "coût" d'un circuit en secours. L'objectif de cette approche globale est de concilier "sécurité électrique absolue" et "continuité de service" pour les systèmes de conversion sensibles devant être intégrés au coeur des applications critiques. Le premier chapitre et son annexe permettent de rappeler les causes et les conséquences des défauts internes au sein d'une structure de base d'onduleur à deux niveaux de tension, laquelle fait l'objet de nombreuses simulations de modes dégradés en configuration variateur de vitesse sur machine synchrone. De cette analyse il ressort qu'une structure d'isolement symétrique à deux voies couplées, placées sur les pôles du bus DC, à déclenchement spontané (fusible) et/ou commande (rupteur), est à même de sécuriser toutes les mailles du circuit, de façon modulaire et non intrusive. / In all these traditional industries, or in more sensitive sectors and high technology, it appears that the safe operation of power systems becomes a critical and strategic area essential. In the area of application that focuses, design dependability and now rests primarily on an approach to reliability of components used, the use of close protection, monitoring alarms and management stop / reset / recovery. In our view, this approach is incomplete quickly when electrical safety and absolute continuity of a permanent mission should be carried out simultaneously in the presence of an internal failure of sensitive functions for low and medium power (eg, orders and bodies actuation of vehicles) or highly critical (nuclear). In this area, topologies and failure modes are at the heart of the problem. In this paper, we will focus primarily on the inverters and choppers structures at two levels of voltage (single-cell arm, <1kV), with simple configuration and multiphase parallel, although the concepts are presented, as examples, partially extrapolated to the structures of three voltage levels (arms multicellular) and rectifier (low-frequency phase control and high-frequency switching PWM). We will highlight the need to limit the intensity of these failures and to electrically isolate the defective cell and symmetrically of this inverter by multipole devices, passive or spontaneous breaking mixed cut ordered in the form of fuses integrated and distributed of multi-channel passive isolators, to imagine and develop. We will show that this process of isolation of the last backup is needed to connect, form series or parallel to the defective cell, a cell rescue in passive redundancy. The cell structure backup connection pooled by spontaneous (automatic) is especially promising as detailed in our eyes because of its simplicity and its integrability. Next, we present the isolation technologies fuse (not included, miniatures, CMS and multilayer chip fuse), their characteristics, their current limitations and operating in a switching cell test. A methodology and design of symmetrical two-way fuse (dual-fuse) on FR4 PCB - Copper will be presented in Comsol ™ and evaluated initially in static thermal IR. A passive two-way switch, relatively original material for integrating energy embedded in FR4 substrate, will be presented and fully dimensioned plans on electrical, thermal and mechanical also using finite element simulations in Comsol ™. Another aspect of exploratory analysis, mainly experimental, or to characterize the failure modes of bullets and casings ultimate power compared between the technologies of encapsulation by epoxy resin (standard discrete case) and a silicone gel (module) is provided under conditions of stress controlled and reproducible. This step is necessary to characterize the resistive mode of a chip based on faulty stresses and stability over time of the residual strength according to the nature of the encapsulant, ie the very sustainability of this failure mode. A mixed-encapsulant resin - gel will be presented and characterized, providing an excellent compromise for medium power applications. Positive results and little known today, will allow us to exploit in the next chapter, this property of stable ohmic mode of the chip failed in a structure to aid automated connection series interesting. In the end, we will detail the demonstrator prototype and introduced to the context with which we will validate the isolation structures and prototypes fuses the property of stable ohmic mode highlighted in the aspect of technological analysis of selected devices. These results allow us to refine the solutions adopted for specifications and guide future management strategy of defects whether internal or external to the topology. Supervisor digital device - sensor for the detection and reconfiguration of internal control orders will be assessed.

In this dissertation, we introduce a novel text representation method mainly used for text classification purpose. The presented representation method is initially based on a variety of closeness relationships between pairs of words in text passages within the entire corpus. This representation is then used as the basis for our multi-level lightweight ontological representation method (TOR-FUSE), in which documents are represented based on their contexts and the goal of the learning task. The method is unlike the traditional representation methods, in which all the documents are represented solely based on the constituent words of the documents, and are totally isolated from the goal that they are represented for. We believe choosing the correct granularity of representation features is an important aspect of text classification. Interpreting data in a more general dimensional space, with fewer dimensions, can convey more discriminative knowledge and decrease the level of learning perplexity. The multi-level model allows data interpretation in a more conceptual space, rather than only containing scattered words occurring in texts. It aims to perform the extraction of the knowledge tailored for the classification task by automatic creation of a lightweight ontological hierarchy of representations. In the last step, we will train a tailored ensemble learner over a stack of representations at different conceptual granularities. The final result is a mapping and a weighting of the targeted concept of the original learning task, over a stack of representations and granular conceptual elements of its different levels (hierarchical mapping instead of linear mapping over a vector). Finally the entire algorithm is applied to a variety of general text classification tasks, and the performance is evaluated in comparison with well-known algorithms.

In current design of steel eccentrically braced frames (EBFs), the yielding link is coupled with the floor beam. This often results in oversized link elements, which leads to over-designed structures and foundations. In addition, the beams are expected to sustain significant damage through repeated inelastic deformations under design level earthquakes, and thus the structure may require extensive repair or need to be replaced. These drawbacks can be mitigated by designing EBFs with replaceable shear links. Two different replaceable link types with alternate section profiles, connection configurations, welding details and intermediate stiffener spacing were tested. A total of 13 cyclic quasi-static full-scale cyclic tests were performed, which included tests on eccentrically braced frames with the replaceable shear links, to study their inelastic seismic performance. The links exhibited a very good ductile behaviour, developing stable and repeatable yielding. Additional inelastic rotation capacity can be achieved with bolted replaceable links when allowing bolt bearing deformations to occur. The on-site replaceability of the link sections is confirmed even in the presence of residual deformations of 0.5% drift.

In current design of steel eccentrically braced frames (EBFs), the yielding link is coupled with the floor beam. This often results in oversized link elements, which leads to over-designed structures and foundations. In addition, the beams are expected to sustain significant damage through repeated inelastic deformations under design level earthquakes, and thus the structure may require extensive repair or need to be replaced. These drawbacks can be mitigated by designing EBFs with replaceable shear links. Two different replaceable link types with alternate section profiles, connection configurations, welding details and intermediate stiffener spacing were tested. A total of 13 cyclic quasi-static full-scale cyclic tests were performed, which included tests on eccentrically braced frames with the replaceable shear links, to study their inelastic seismic performance. The links exhibited a very good ductile behaviour, developing stable and repeatable yielding. Additional inelastic rotation capacity can be achieved with bolted replaceable links when allowing bolt bearing deformations to occur. The on-site replaceability of the link sections is confirmed even in the presence of residual deformations of 0.5% drift.

The market for virtual reality products has grown rapidly the last few years, and as the demand grows, the supply should naturally follow. More games, applications and also web sites for virtual reality are going to have to be designed and produced to support this demand. However, the guidelines and the literature for designing user interfaces specifically for virtual reality are few and often outdated. The purpose of this thesis is to help contribute to determine how to design user interfaces for virtual reality, when one is making selection with fuse-based clicks. In this thesis, a prototype for a TV-platform is designed and produced in WebVR, using the framework A-Frame. The prototype is then evaluated with usability tests and redesigned as according to an iterative design process. The evaluations showed that items were easily clicked by mistake and often unknowingly. Furthermore, the test user participants often failed to notice user interface objects if the objects were not placed in the centre areas of the user interface. The conclusion suggests that navigation paths within the user interface could mitigate accidental clicks, and auditory feedback could help users notice the accidental clicks. Moreover, between 700ms to 1000ms, but closer to 1000ms, could be an appropriate fuse time to facilitate ease of making selections at the cost of speed efficiency.

In this dissertation, we introduce a novel text representation method mainly used for text classification purpose. The presented representation method is initially based on a variety of closeness relationships between pairs of words in text passages within the entire corpus. This representation is then used as the basis for our multi-level lightweight ontological representation method (TOR-FUSE), in which documents are represented based on their contexts and the goal of the learning task. The method is unlike the traditional representation methods, in which all the documents are represented solely based on the constituent words of the documents, and are totally isolated from the goal that they are represented for. We believe choosing the correct granularity of representation features is an important aspect of text classification. Interpreting data in a more general dimensional space, with fewer dimensions, can convey more discriminative knowledge and decrease the level of learning perplexity. The multi-level model allows data interpretation in a more conceptual space, rather than only containing scattered words occurring in texts. It aims to perform the extraction of the knowledge tailored for the classification task by automatic creation of a lightweight ontological hierarchy of representations. In the last step, we will train a tailored ensemble learner over a stack of representations at different conceptual granularities. The final result is a mapping and a weighting of the targeted concept of the original learning task, over a stack of representations and granular conceptual elements of its different levels (hierarchical mapping instead of linear mapping over a vector). Finally the entire algorithm is applied to a variety of general text classification tasks, and the performance is evaluated in comparison with well-known algorithms.