Spelling suggestions: "subject:"forklift"" "subject:"forklifts""
41 |
Development of a suspension system for a forklift cabin : A design project at Kalmar innovation centreSchäring, Isabelle, Forssell, Johan January 2024 (has links)
A suspension system is to be made to increase driver comfort and ergonomics. This kind of system is widely used in a variety of vehicles, such as tractors and trucks, but is not yet standard for forklifts. The purpose of the case study is to design a fully mechanical, passive suspension system that can be retrofitted to a particular forklift without any cabin or chassis modification. The main focus will be on the linkages used to limit unwanted movement of the cabin. A concept utilizing 2 sets of parallel linkages is chosen to continue develop with the application of design for environment and design for manufacturing. This manages to reduce the total number of component sneeded for the system down to 33 from 43 in the first draft of the concept. The design needs to be dimensioned and further tested before a prototype is built. If the system then is implemented, it will have lasting effects on the forklift industry and the ergonomics of operators.
|
42 |
Interface Development for Semi-Autonomous Trucks : Visual and Auditory FeedbackEriksson, Frida, Andersson, Märta January 2015 (has links)
Vehicles are becoming increasingly autonomous, as automotive industries are investing in innovative technology. Therefore the technology becomes more available and affordable, making it possible for Toyota Material Handling Europe (TMHE) to introduce automated features in their trucks. Vehicles that have a forward collision warning system, and thus are partly autonomous, are involved in less accidents than those without. In manufacturing industries there is currently a problem with truck collisions and an automated solution might be a suitable way to prevent these. When implementing an automation device, human machine interaction and user-friendliness are aspects to keep in mind during the development. The thesis concerns how autonomous features can assist the truck driver, and how to provide the driver with intuitive feedback. The purpose was to ensure the drivers’ and surrounding personnel’s safety as well as increase the productivity. Research was performed regarding in what situation an assisting device is needed and how to communicate information in an intuitive manner to help the driver in this situation. A conceptual interface was developed that allows communication between the driver and a future all-knowing system, that tracks all objects and personnel in a warehouse. The drivers have had a central role in the process. The observations were performed in the TMHE warehouse to identify situations. The most perilous and frequent situation was when drivers need to focus both in the fork and drive wheel directions simultaneously. This either puts the surroundings or the driver in danger. A conceptual interface was developed to help the driver in this situation. This resulted in a concept implementable in both current and future trucks, to harmonise the solution and ensure a safe warehouse environment. A lo-fi prototype was constructed and evaluated iteratively with drivers to ensure the quality and usability of the concept. The resulting feedback solution consists of sounds from speakers mounted in the headrest and a display interface with warning symbols. The sounds are directional to notify the driver if the danger is to the left or right behind his back. If the danger is only semi-close, the driver receives a warning, but if it is very close, the truck is stopped autonomously. The symbols appear on the display simultaneously as the sounds are heard, to provide further feedback. Additionally, an Autonomous Positioning feature has been developed, that consists of symbols and buttons on the display interface, as well as an alert sound from the display to indicate the system’s activation and deactivation. Safety is enhanced since neither personnel nor trucks are in risk of collision when implementing the concept. As the concept helps the driver position the truck effortlessly towards the pallet the productivity is also improved.
|
43 |
Produktionsflödesanalys : Förstudie av nya produktionsflöden för golvprodukter / Productions flow analysis : Preliminary study of new production flows for floor productsSvensson, Joel, Ideström, Gabriel January 2019 (has links)
Examensarbetet genomfördes i samarbete med Tarkett AB i Ronneby med syfte att konstruera ett nytt produktionsflöde vilket ska minska stopp i flödet samt nå en produktionsökning på 15%. Det nya flödet ska sedan planeras i den givna hallen. Leanprinciper är grunden för det nya flödet och metoder som värdeflödesanalys och ledningsverktyg används för att göra en nulägesbeskrivning. Information samlades in genom observationer som utfördes genom “att gå längs flödet”, intervjuer och teorin kring lean för att analysera flödet. Värdeflödesanalys gjordes på det befintliga flödet för att sedan kunna skapa framtida flöden. Idégenerering utfördes för att få fram olika koncept till nya alternativa flöden. Dessa koncept har sedan utvärderats genom verktyg och med generell bedömning. Ett koncept har valts ut att gå vidare med för fortsatt analys och försök att planera i den givna hallen för att få en förståelse hur flödet skulle fungera med de begränsningar som finns. En värdeflödesanalys gjordes för det befintliga flödet för att få överblick och ta fram värdehöjande och icke värdehöjande aktiviteter. Flödet begränsas av produktionsstopp på grund av aktiviteter som är icke värdehöjande och en flaskhals i produktionen som begränsar hastigheten på hela flödet. Alternativa flöden har tagits fram. Det flöde som valdes är ett lådflöde där materialet fylls i lådor för att sedan skickas vidare till maskinen som producerar golv av materialet. Lådorna kan fyllas 15 % snabbare men flaskhalsen kan fortfarande ej ta emot denna ökning. Lösningen blir kontinuerligt flöde på alla processer som det går att ha ett kontinuerligt flöde och ett buffertlager som förbrukas med jämna mellanrum där material i lådor minskar produktionsstopp eftersom lådorna ej är direkt kopplade till resten av linjen och därmed inte blir direkt påverkad utan fortsatt tillverkning kan ske. Flaskhalsen kan inte reduceras, men med lådflöde kan buffertlager användas. Med detta kan material från extrudern (råmaterial) som egentligen skulle gå till att fylla lådor skickas mot en annan linje. Detta leder till att två linjer kör samtidigt och här finns produktionsökningen. Genom att undvika produktionsstopp som orsakas av det befintliga flödet kan det ge en produktionsökning på närmare 8 % och möjlighet att köra flera linjer samtidigt kan ge ytterligare produktionsökning men är svår att uppskatta. En fortsatt analys behöver genomföras av flödet innan det kan implementeras i verkligheten. / This bachelor thesis was made with the support from Tarkett AB, Ronneby. The purpose of this task was to construct a new production flow, which should minimize the stop time and also increase the production ratio with 15%. The basics for lean principles were used in order to come up with the new production flow. Also methods like value flow analysis and the seven management tools were used to make a status report of today's production flow. Information was gathered from observations, which came from inspections of the machines that are meant to be used in the new flow, interviews and also theory about lean production. A value stream mapping was made on the already existing production flow, in order to create new ideas for the new production flow. These ideas were then evaluated with tools and general assessment, where the best concept was chosen to move on with further analysis. The next step was to plan the placement of machines and plan the forklift traffic at the given area. This will gave an understanding on how the new flow could work with the known limitations. The result shows a value flow analysis for the already existing production flow, which gives an overview on how to collect value-added and non-valueadded activities. The flow is restricted by stoppages due to the non-valueadded activities and bottlenecks. Alternative flows have been considered in the result, but the chosen flow is called box flow. With this concept the material will be filled in big boxes that can hold up to 1.1 tons of granulate. These boxes will then be stored in a warehouse until it’s collected and sent towards the machine which will use the material to produce the given amount of floor mat. The boxes can be filled 15 % faster, but the bottleneck will still not be able to keep up with this production increase. This was solved with a continuous flow on all processes and a buffer stock that will need to be continuously consumed from time to time. The bottleneck won’t increase in production speed, but with the box flow a buffer stock can be used. With this flow the material from extrusion (raw material) can be sent to other production lines. Which leads to two lines running simultaneously and here is the production increase. By avoiding production stops that are inflicted by the existing production flow, a production increase that is closer to 8 % can be performed and a possibility to increase it even further if more production lines can run at the same time. However, the production increase when taking the other production lines to account is hard to estimate. A continued analysis needs to be carried through to implement the new flow at Tarkett.
|
44 |
Estimativa do estado de carga de baterias em robôs móveis autônomos / Battery state of charge estimation in autonomous mobile robotsOliveira, Marcelo Manoel de 19 April 2013 (has links)
Cada vez mais robôs móveis autônomos estão sendo utilizados em diversas tarefas e em ambientes com elevado risco para atividades humanas que a paralisação de suas atividades podem gerar outros riscos, perdas e elevados custos. Assim, o estado de carga (SOC) de sistemas de baterias em robôs móveis autônomos é um parâmetro importante na prevenção de uma falha primária nessa aplicação, a ausência de energia. Este trabalho apresenta os métodos existentes na literatura para a determinação do estado de carga de baterias e as tecnologias de baterias disponíveis utilizadas em robôs móveis autônomos ou veículos autônomos guiados. A partir desses estudos foi desenvolvido um modelo de medida, baseado no modelo combinado e foram realizados testes de bancadas para levantamento dos parâmetros e características de três modelos de células de baterias: Lítio Polímero (Li-PO), Níquel-Cádmio (NiCd) e Lítio-Ferro-Polímero (LiFePO4). Com esses parâmetros, aplicou-se o método de estimativa de carga baseado na técnica do Filtro de Kalman Estendido (EKF). Através dos testes, analisou-se comparativamente a resposta do método proposto e a resposta do método OCV e a capacidade de carga real. / Autonomous mobile robots have being increasingly used in various tasks, environments and activities of high risk to human that the stoppage of its activities may generate other risks, losses and high costs. Thus the state of charge (SOC) of battery systems in autonomous mobile robots, is an important parameter to prevent a primary failure in this application, the lack of energy. The paper presents the existing methods in the literature to determine the battery state of charge and battery commercial technologies available used in an autonomous mobile robot or autonomous guided vehicle, from these studies a measurement model based on combined model was developed and testing benches for three cells models on Lithium Polymer Battery (Li-PO), Nickel Cadmium (NiCd) and lithium-iron-Polymer (LiFePO4) batteries were performed for lifting the parameters and apply the battery state of charge method based on the Extended Kalman Filter (EKF) technique. The tests were analyzed in order to observe the comparatively response of the proposed method, the OCV method and Real charge capacity.
|
45 |
Estimativa do estado de carga de baterias em robôs móveis autônomos / Battery state of charge estimation in autonomous mobile robotsMarcelo Manoel de Oliveira 19 April 2013 (has links)
Cada vez mais robôs móveis autônomos estão sendo utilizados em diversas tarefas e em ambientes com elevado risco para atividades humanas que a paralisação de suas atividades podem gerar outros riscos, perdas e elevados custos. Assim, o estado de carga (SOC) de sistemas de baterias em robôs móveis autônomos é um parâmetro importante na prevenção de uma falha primária nessa aplicação, a ausência de energia. Este trabalho apresenta os métodos existentes na literatura para a determinação do estado de carga de baterias e as tecnologias de baterias disponíveis utilizadas em robôs móveis autônomos ou veículos autônomos guiados. A partir desses estudos foi desenvolvido um modelo de medida, baseado no modelo combinado e foram realizados testes de bancadas para levantamento dos parâmetros e características de três modelos de células de baterias: Lítio Polímero (Li-PO), Níquel-Cádmio (NiCd) e Lítio-Ferro-Polímero (LiFePO4). Com esses parâmetros, aplicou-se o método de estimativa de carga baseado na técnica do Filtro de Kalman Estendido (EKF). Através dos testes, analisou-se comparativamente a resposta do método proposto e a resposta do método OCV e a capacidade de carga real. / Autonomous mobile robots have being increasingly used in various tasks, environments and activities of high risk to human that the stoppage of its activities may generate other risks, losses and high costs. Thus the state of charge (SOC) of battery systems in autonomous mobile robots, is an important parameter to prevent a primary failure in this application, the lack of energy. The paper presents the existing methods in the literature to determine the battery state of charge and battery commercial technologies available used in an autonomous mobile robot or autonomous guided vehicle, from these studies a measurement model based on combined model was developed and testing benches for three cells models on Lithium Polymer Battery (Li-PO), Nickel Cadmium (NiCd) and lithium-iron-Polymer (LiFePO4) batteries were performed for lifting the parameters and apply the battery state of charge method based on the Extended Kalman Filter (EKF) technique. The tests were analyzed in order to observe the comparatively response of the proposed method, the OCV method and Real charge capacity.
|
46 |
Exponering av luftföroreningar : -vid arbete i hamnmagasin / Occupational exposure to air pollution : -at port warehouseNordgren, Susanne January 2015 (has links)
Syftet med examensarbetet var att översiktligt undersöka vilka hälsofarliga luftföroreningarsom medarbetarna vid hamnmagasinen eventuellt exponeras föroch huruvida något hygieniskt gränsvärde riskerade att överskridas samt att seom de skyddsåtgärder som vidtagits kan förbättras. Resultatet från studien visaratt det finns vissa exponeringsrisker för de som arbetar vid hamnen både frångods, dieselavgaser samt rester av gasbehandlingsmedel i containrar.Arbetsmomenten där truckföraren befinner sig inne i trucken bedöms somrelativt skyddat både från partiklar och gaser, men det finns andra arbetsuppgifterdär arbetaren inte är lika skyddad. Mätningarna av kvävedioxid visadeatt exponeringen av dieselavgaser för den medarbetare som öppnadecontainrarna inte översteg några hygieniska gränsvärden, under de dagar sommätningarna genomfördes, men var något högre än för truckföraren som satt ihytten. Huruvida exponeringen av damm från gods riskerar att utvecklas tillhälsoproblem är omöjligt att avgöra utan en grundlig riskbedömning, därdammätningar kan ge en fingervisning om hur riskfylld situationen är för desom eventuellt exponeras.Organisationen bör implementera säkrare rutiner och genomföra åtgärder sommotiverar medarbetarna att använda befintlig skyddsutrustning. Det gäller intebara för exponering av dieselavgaser och damm från lossning och lastning avgods utan även vid öppnandet av containrar där rester av gasbehandlingsmedelkan finnas kvar i containern och där några ämnen misstänkts varacancerframkallande.Slutsatser: Genomföra en grundlig riskbedömning och kartlägga möjliga hälsoriskerkring luftföroreningar. Upprätta skriftliga arbetsbeskrivningar där eventuella risker föreligger. Installera ventilation i magasinen som styrs av halten kvävedioxid för attsäkerställa att höga halter av dieselavgaser inte uppstår. Behov av motivationshöjande insatser kring säkerhetskultur ochanvändningen av skyddsutrustning. Implementera fungerande rutiner som följer lagstiftningen förhanteringen av damm och gaser som gäller för cancerframkallandeämnen. / The purpose of this thesis was to examine the risk of hazardous air pollutantsthat employees at the harbor may be exposed to and assess whether anyexposure limits might be exceeded, and to review if the security measures taken,can be improved. The results of the study show that there is some risk ofexposure for those working in the harbor from both the cargo, diesel exhaustand residues from fumigants in the containers. The operations when the truckdriver is inside the forklift is considered relatively protected from both particlesand gases, but there are other tasks where the worker is not as protected.Measurements of nitrogen dioxide showed that exposure to diesel exhaust forthe employee who opened the containers during the days that the measurementswere performed did not exceed some critical values, but was slightly higherthan for the driver who was sitting in the forklift. Whether the exposure to dust,from the cargo, constitutes a risk for health effects is impossible to determinewithout a thorough risk assessment, where measurements of dust can give anindication of how risky the situation is for the exposed workers.The organization should implement safer practices and implementing measuresthat motivates employees to use existing protective equipment. This applies notonly for exposure to diesel exhaust and dust from loading and unloading ofgoods, but also at the opening the doors on the containers in which residues offumigants, in which some are suspected to be carcinogenic, can remain in thecontainer.Conclusions: Conduct a thorough risk assessment and identify potential health riskswith air pollution. Establish written work instructions where potential hazards exist. Install ventilation in the warehouses, which is controlled by the levels ofnitrogen dioxide, to ensure that high levels of diesel exhaust does notoccur. Need to increase motivation for safety and the use of protectiveequipment. Implement procedures to comply with the legislation for the managementof dust and gases that applies to carcinogens substances.
|
47 |
Development of Methodology for Finite Element Simulation of Overhead Guard Impact Test / Utveckling av metodik för finita elementsimulering av skyddstak utsatt för fallprovHallén, Axel, Hjorth, Jacob January 2022 (has links)
Forklifts that are capable of lifting heavy loads and reaching high lift heights are required by stan-dards to have an overhead guard to protect the operator from falling objects. The same standardsspecify a standardized procedure for testing the strength of these overhead guards. The test in-volves dropping ten 45 kg wooden cubes and a heavy timber load onto the overhead guard. Thesedestructive tests are time-consuming and expensive, and it is the purpose of this master’s thesis todevelop a methodology for simulating this kind of test using the finite element method with a largedisplacements, explicit scheme using the solver RADIOSS by Altair. This was achieved by firstdesigning, constructing, and testing a physical prototype of an overhead guard to use as a referencefor a finite element methodology to be validated against. The work has also included tensile testingof the overhead guard material, and this was done both to obtain material data from the sametype of material as the prototype, and to get Johnson-Cook material parameters, which are hardto come by in the literature. Next, a basic finite element model was created which showed a verylarge discrepancy compared to the physical test results. An extensive investigation into aspectssurrounding finite element modeling and material modeling was undertaken, and resulted in a fi-nal model which overestimated the displacements by about 40 % only. The remaining inaccuracyis believed to mostly stem from inadequate strain-rate sensitivity data, caused by limitations inavailable resources for material testing.
|
48 |
Vätgasdrivna arbetsmaskiners tekniska mognad : Dagens etableringsmöjligheter och potentiella tillämpningar i Gävleborgs framtida vätgassamhälleLärkfors, Selinn, Svedlund, Carolina January 2021 (has links)
Förbränning av fossila bränslen är den största orsaken till ökade växthusgasutsläpp i atmosfären som i sin tur ligger till grund för klimatförändringarna. Europeiska Unionen uttrycker klimatförändringarna som ett existentiellt hot och har som mål att Europa ska bli en klimatneutral kontinent till år 2050. Arbetsmaskiner, som vanligtvis drivs på diesel, är ett energikrävande fordonsslag vid användning som år 2016 stod för 6 % av Sveriges totala växthusgasutsläpp. Vätgas kan vara ett alternativt drivmedel till diesel eller andra fossila drivmedel för att minska de utsläpp som arbetsmaskiner ger upphov till. Syftet med studien är att uppmärksamma arbetsmaskiners roll i en omställning till vätgasdrift samt sprida kunskap om vad en omställning skulle innebära för dagens användare av arbetsmaskiner. Detta görs genom att belysa möjlig etablering i nutid av fem utvalda vätgasdrivna arbetsmaskiner samt gestalta dem i ett framtida vätgassamhälle. De arbetsmaskiner som inkluderas i studien är hjullastare, pistmaskin, sopmaskin, traktor och motviktstruck. Metoden består av en förenklad litteraturöversyn i kombination med personlig kommunikation samt en anpassning av verktyget Technology Readiness Level (TRL) i syfte att bedöma den tekniska mognaden. Resultatet visar att det i dagsläget finns etableringsmöjligheter i olika former för samtliga arbetsmaskiner i regionen baserat på TRL. Vätgasdrivna motviktstruckar och sopmaskiner är båda tillräckligt mogna tekniker för att etableras direkt på marknaden genom inköp. Vätgasdrivna hjullastare, traktorer och pistmaskiner är fortfarande under utveckling och kan därför etableras i regionen genom forskning och ytterligare utveckling eller först efter produktlansering på marknaden. De utvalda vätgasdrivna arbetsmaskinerna kan verka i ett framtida vätgassamhälle på liknande vis som motsvarande fossildrivna arbetsmaskiner gör idag med fördelar som mindre miljöpåverkan, vibrationer och buller som i sin tur medför mindre underhållsbehov med tillhörande kostnader. Resultatet visar att det finns möjligheter redan idag att påskynda en förändring, inte bara i Gävleborg utan i hela Sverige. De utvalda arbetsmaskinerna är verksamma inom branscher som motsvarar stora delar av det svenska näringslivet, en omställning till vätgasdrift kan därför ha en stor betydelse för en nationell reducering av växthusgasutsläpp. Det finns stor potential, särskilt för aktörer inom industrin, att upprätta egen vätgasproduktion och därmed bli självförsörjande på vätgas som drivmedel till arbetsmaskiner. Detta kan bli en av de radikala förändringar som enligt EU behövs för att uppnå målet om att åstadkomma klimatneutralitet år 2050. / Burning of fossil fuels is the biggest cause of increased greenhouse gas (GHG) emissions into the atmosphere, which in turn leads to climate change. The European Union expresses climate change as an existential threat and aims to make Europe a climate-neutral continent by 2050. Non-road machinery vehicles (NMV), which are usually powered by diesel, are an energy-intensive type of vehicle during usage that in 2016 accounted for 6% of Sweden's total GHG emissions. Hydrogen can be an alternative fuel to diesel or other fossil fuels in order to reduce the emissions that originate from NMVs. The aim of this study is to draw attention to the role of NMVs in a conversion to hydrogen operation and to spread knowledge about what a conversion would mean for today's users of work machines. This is done by highlighting possible establishment in the present of five selected hydrogen powered NMVs and illustrate them in a future hydrogen society. The NMVs included in the study are wheel loaders, snow groomers, street sweepers, tractors and counterbalanced forklifts. The method consists of a simplified literature review in combination with personal communication and an adaptation of the tool Technology Readiness Level (TRL) in order to assess technical maturity. The results show that there are establishment opportunities in the present in various forms for all NMVs in Gävleborg based on TRL. Hydrogen powered counterbalanced forklifts and street sweepers are both sufficiently mature technologies to be established directly in the market through purchasing. Hydrogen powered wheel loaders, tractors and snow groomers are still under development and can therefore be established in the region through research and further development or after product launch. The selected hydrogen powered NMVs can operate in a future hydrogen society similarly to fossil-fueled NMVs but with benefits such as less environmental impact, vibrations and noise which in turn entails less maintenance needs with associated costs. There are opportunities already today to accelerate a change, not only in Gävleborg but throughout Sweden. The selected NMVs are active in industries that correspond to large parts of the Swedish business community, a transition to hydrogen operation can therefore be of great importance for a national reduction of GHG emissions. There is great potential, especially for players in the industry, to establish their own hydrogen production and thereby become self-sufficient in hydrogen intended for NMVs. This could be one of the radical changes that, according to the EU, are needed to achieve the goal of accomplishing climate-neutrality by 2050.
|
49 |
Contamination Level Detection of Hydraulic Pressure Filters in Forklifts : using only pump motor currents and load pressure measurementsSehlstedt, Robert, Sellén, Erik January 2022 (has links)
With the advent of Industry 4.0 and the Internet of Things, collecting data on Cyber-Physical systems has become the norm practice in large scale industries. By collectingrelevant data, it is possible to monitor the health status of whole systems or specificcomponents within them. Such practices allow for historical maintenance strategies suchas reactive maintenance or preventive maintenance to be phased out.In this thesis two separate algorithms are presented, both designed to identify contaminationlevels in the hydraulic pressure filters of forklifts. Furthermore, in contrast torelevant literature for similar applications only sensory data from the hydraulic pump’smotor current and hydraulic fluid pressure at the load was used. More specifically, theproposed algorithms are based on trends observed in the relationship between the measurementsand how it changes over time. The algorithms were evaluated on data fromfour forklifts used in Toyota’s factory. The forklifts had been collecting data while usedin production for over a year.The results indicate strong evidence that both algorithms can be used to detect degradationin the hydraulic system. This is especially true for one forklift where it was knownthat the damage at the time of replacement was substantial. However, it cannot be trulyestablished without further testing whether the algorithms detect degradation in the filteror pump.
|
50 |
A deep learning based anomaly detection pipeline for battery fleetsKhongbantabam, Nabakumar Singh January 2021 (has links)
This thesis proposes a deep learning anomaly detection pipeline to detect possible anomalies during the operation of a fleet of batteries and presents its development and evaluation. The pipeline employs sensors that connect to each battery in the fleet to remotely collect real-time measurements of their operating characteristics, such as voltage, current, and temperature. The deep learning based time-series anomaly detection model was developed using Variational Autoencoder (VAE) architecture that utilizes either Long Short-Term Memory (LSTM) or, its cousin, Gated Recurrent Unit (GRU) as the encoder and the decoder networks (LSTMVAE and GRUVAE). Both variants were evaluated against three well-known conventional anomaly detection algorithms Isolation Nearest Neighbour (iNNE), Isolation Forest (iForest), and kth Nearest Neighbour (k-NN) algorithms. All five models were trained using two variations in the training dataset (full-year dataset and partial recent dataset), producing a total of 10 different model variants. The models were trained using the unsupervised method and the results were evaluated using a test dataset consisting of a few known anomaly days in the past operation of the customer’s battery fleet. The results demonstrated that k-NN and GRUVAE performed close to each other, outperforming the rest of the models with a notable margin. LSTMVAE and iForest performed moderately, while the iNNE and iForest variant trained with the full dataset, performed the worst in the evaluation. A general observation also reveals that limiting the training dataset to only a recent period produces better results nearly consistently across all models. / Detta examensarbete föreslår en pipeline för djupinlärning av avvikelser för att upptäcka möjliga anomalier under driften av en flotta av batterier och presenterar dess utveckling och utvärdering. Rörledningen använder sensorer som ansluter till varje batteri i flottan för att på distans samla in realtidsmätningar av deras driftsegenskaper, såsom spänning, ström och temperatur. Den djupinlärningsbaserade tidsserieanomalidetekteringsmodellen utvecklades med VAE-arkitektur som använder antingen LSTM eller, dess kusin, GRU som kodare och avkodarnätverk (LSTMVAE och GRU) VAE). Båda varianterna utvärderades mot tre välkända konventionella anomalidetekteringsalgoritmer -iNNE, iForest och k-NN algoritmer. Alla fem modellerna tränades med hjälp av två varianter av träningsdatauppsättningen (helårsdatauppsättning och delvis färsk datauppsättning), vilket producerade totalt 10 olika modellvarianter. Modellerna tränades med den oövervakade metoden och resultaten utvärderades med hjälp av en testdatauppsättning bestående av några kända anomalidagar under tidigare drift av kundens batteriflotta. Resultaten visade att k-NN och GRUVAE presterade nära varandra och överträffade resten av modellerna med en anmärkningsvärd marginal. LSTMVAE och iForest presterade måttligt, medan varianten iNNE och iForest tränade med hela datasetet presterade sämst i utvärderingen. En allmän observation avslöjar också att en begränsning av träningsdatauppsättningen till endast en ny period ger bättre resultat nästan konsekvent över alla modeller.
|
Page generated in 0.0552 seconds