Machine tools are essential components of modern manufacturing. They are com posed of various mechanical, hydraulic, and electrical systems such as the spindle, tool changer, cooling system, and the linear and rotary feed drives. Due to their com plexity, high cost, and importance to the manufacturing process it is recommended to implement some sort of condition monitoring and predictive maintenance to ensure that they remain reliable and high performing. One way of potentially implement ing predictive maintenance and condition monitoring is digital twins. Digital twins enable the real-time, accurate, and complex modeling and monitoring of mechanical systems. They utilize data collected from the system to constantly update their mod els which can be used for monitoring of the systems state and future predictions. This work presents a digital twin workbench of a machine tool feed drive. The workbench enables the collection and analysis of large, varied, high-frequency data which can be used to construct a digital twin of the feed drive. A digital twin can enable many other useful functionalities. Some of these functionalities include condition moni toring, modeling, control, visualization, and simulation. These functionalities can enable maximum asset performance and are key in implementing effective predictive maintenance. The main contributions of this work are the following: The design and iv construction of a machine tool feed drive which implements a novel external distur bance force method. A new method of fault detection in ball screws using interacting multiple models which was shown to provide accurate estimates of levels of preloads in a ball screw driven feed drive. A digital twin based modeling strategy and analysis of the data generated by the system including system modeling and observations on modeling difficulties. / Thesis / Master of Applied Science (MASc) / Digital twins enable the real-time, accurate, and complex modeling and monitoring of mechanical systems. Machine tools are essential components of modern manufac turing. They are composed of various mechanical, hydraulic, and electrical systems such as the spindle, tool changer, cooling system, and linear and rotary feed drives. This work presents the design of a workbench of a machine tool linear feed drive, a fault detection strategy, and a digital twin modeling solution. The workbench enables the collection and analysis of large, varied, high-frequency data which can be used to construct a digital twin of the feed drive. A digital twin can enable many other useful functionalities. Some of these functionalities include condition monitoring, modeling, control, visualization, and simulation. These functionalities can enable maximum asset performance and are key in implementing effective predictive maintenance.

A digital twin project is launched by the Integrative Systems Biology (ISB) research team and led by Gunnar Cedersund. The digital twin project is based on biological models of physiological processes, that can interact and be tailored for a specific person. However, the digital twin can currently not analyse movements of a human body. In this master thesis, the aim was to create a useful pipeline that expands the digital twin project with biomechanical modelling of movements, and also visualises the twins by letting the concept take human form. The biomechanical analysis was done in the software OpenSim, where the movements of a motion captured dance were analysed. To generate a simulation of the motion with an acceptable error in a reasonable computation time, a musculoskeletal model was created in OpenSim and scaled to best fit the anthropometry of the dancer. Then, the motion was estimated with an optimised procedure by using the scaled model and the motion capture data. The Root-Mean Squared (RMS) error of the estimated dance with accuracy 10-6 was 2.39 cm. In this thesis, the torque in each joint for the dance motion was estimated. The loads and muscle forces can also be estimated in OpenSim. One useful application is for calculating energy consumption. In order to calculate muscle forces, external forces needs to be measured while recording motion capture. This is something that will be focused on in the future, when continuing with this project. The visualisation of the digital twins were made in Unreal Engine with MetaHuman avatars. The dance recorded in motion capture, were applied to the avatars in order to make them dance. The recorded dance was the same for both OpenSim and Unreal Engine, so the dance could both be viewed and analysed. In conclusion, we have added a new feature to the existing digital twin technology: movements and simulation of the musculoskeletal system. This new feature can in the future be used for both medical purposes such as movement-based rehabilitation as well as for integration into dance performances.

A rapidly growing interest in augmented and virtual reality within industrial areas such as manufacturing, quality control, and fleet monitoring has been seen in the last couple of years. This technology shift could bring a new era to the industry sector in the near future. This study aims to evaluate if using virtual reality can be a more efficient way of monitoring lots of data than a traditional monitor based solution or not. In this study, a virtual reality application has been created in order to provides a virtual environment where operators can access and monitor their assets, which a proof-of-concept digital model represents. The digital model presents information about the components from a physical asset’s current state and status. This VR application was then evaluated in an A/B test against an existing monitor-based dashboard application. The A/B test was conducted with 10 participants performing 11 different tasks. The results show that VR technology could be a promising solution for operating and monitoring fleet unit assets, with an overall improvement in the efficiency of 17% for all of the participants.

BACKGROUND - The development of Industry 4.0 increases the opportunities to both automate and digitize processes in the manufacturing industry. The steel industry has been around for many years, which means firmly anchored operations and both manual- and automated processes. To make better decisions, identify bottlenecks, and test new functions without having to stop the production, a digital model of the process can be helpful. Furthermore, with the rapid development of technology, digital models can be further developed into digital twins. A digital twin should be able to handle the communication between the physical- and digital world automatically and analyze data to make decisions in the process. RESEARCH QUESTIONS What are the challenges of developing a digital model representing a production line within a global steel manufacturing company? What opportunities could a digital model of a production line entail, and how could Industry 4.0 technologies create opportunities to further develop the digital model into a digital twin? METHODS - In this project, both a literature- and case study have been carried out. During the literature study, techniques that can be used to develop the digital model further have been investigated. During the case study, a digital model of a Quench Line was developed to gather practical experience of what it can mean to create a digital model of a manufacturing process within a steel manufacturing company. The model has been developed in MATLAB/Simulink. RESULTS - The most significant challenges when developing digital flow simulation models identified in this project were data management/access, handling variations, verifying the model, andlack of knowledge linked to digital models in general. The opportunities identified and confirmed in this project were that the model could be used to carry out new logistics planning, bottleneck analyses, and test new machine implementations. To further develop the digital model into a digital twin, Industry 4.0 technologies will be crucial. The technologies that will be useful are the Internet of Things, Artificial Intelligence, Machine Learning, Cloud Computing, and Big Data.

Digitaliseringens framfart har frambringat ett värdefullt verktyg inom förvaltningen – digital tvilling. Ett sätt att digitalt avbilda ett objekt, en fysisk motpart med information om objektets egenskaper, användning mm. Etableringen inom andra branscher har mer eller mindre varit påtaglig. Däremot har utvecklingen specifikt inom fastighetsbranschen stagnerat en aning gentemot övriga. Rapporten syftar därmed till att undersöka inte bara de eventuella utmaningar, utan även de potentiella möjligheter som digital tvilling kan tänkas medföra inom fastighetsbranschen. Utöver det har en undersökning om definitionsgrunden för digital tvilling genomförts för att ytterligare avgränsa arbetet. Som ett tillägg har andra frågeställningar, relaterat till implementeringen av digital tvilling, beaktats.  Det som kan konkluderas beträffande arbetet, handlar främst om de incitament som krävs för att implementera digital tvilling, de utmaningar och möjligheter som finns förestående samt vilka åtgärder som krävs för att konceptet ska bli applicerbart i underhållsplanen. Resultatet från denna information har uthämtats från litteraturstudie och inte minst intervjuer. Där slutsatsen som kan dras är hur arbetet har effektiviserats och medfört till bland annat kostnadsbesparingar och effektivare arbetssätt i förvaltningen. Utmaningar fanns dock vad gäller att hålla sig uppdaterat beträffande informationen i underlaget och nya arbetssätt med att göra den digitala tvillingen lätthanterlig för berörda aktörer. Slutligen krävs det även engagemang och att sakkunniga inom området är mottaglig för den nya tekniken som har presenterats, både för att implementeringen av digital tvilling ska initieras och för att den inte minst ska bli användbar i underhållsplanen. / The rise of digitalization has produced a valuable tool in property management– digital twin. A way to digitally create and depict an object, a physical prototype with information about the object's properties, usage, etc. The implementation in other industries has more or less been succesful. However, the development, particularly in the real estate industry, have stagnated somewhat compared to forementioned industries. The report thus aims to examine and study the potential challenges and opportunitites that digital twin may bring in the real estate industry. In addition, a study on the definition basis for digital twin has been carried out to further delimit the study. As an addition, other research questions related to the implementation of digital twin have been taken into account. What can be concluded regarding the work is mainly about the incentives required to implement digital twin, the challenges and opportunities that are imminent and what measures are required for the concept to be applicable in the maintenance plan. The results of this information have been obtained from literature studies and interviews. Where the conclusion that can be drawn is how the work has been streamlined and brought, among other things, cost savings and more efficient working methods in property management. However, there were challenges in keeping up to date with the information in the basis material and new ways of working to make the digital twin easy to manage for stakeholders. Finally, it also requires commitment and that experts in the field are receptive to the new technology that has been presented, both for the implementation of digital twin to be initiated and for it to be useful in the maintenance plan.

Green light optimal speed advisory (GLOSA) systems require reliable estimates of signal switching times to improve vehicle energy/fuel efficiency. Deployment of successful infrastructure to vehicle communication requires Signal Phase and Timing (SPaT) messages to be populated with most likely estimates of switching times and confidence levels in these estimates. Obtaining these estimates is difficult for actuated signals where the length of each green indication changes to accommodate varying traffic conditions and pedestrian requests. This dissertation explores the different ways in which predictions can be made for the most likely switching times. Data are gathered from six intersections along the Gallows Road corridor in Northern Virginia. The application of long-short term memory neural networks for obtaining predictions is explored for one of the intersections. Different loss functions are tried for the purpose of prediction and a new loss function is devised. Mean absolute percentage error is found to be the best loss function in the short-term predictions. Mean squared error is the best for long-term predictions and the proposed loss function balances both well. The amount of historical data needed to make a single accurate prediction is assessed. The assessment concludes that the short-term prediction is accurate with only a 3 to 10 second time window in the past as long as the training dataset is large enough. Long term prediction, however, is better with a larger past time window. The robustness of LSTM models to different demand levels is then assessed utilizing the unique scenario created by the COVID-19 pandemic stay-at-home order. The study shows that the models are robust to the changing demands and while regularization does not really affect their robustness, L1 and L2 regularization can improve the overall prediction performance. An ensemble approach is used considering the use of transformers for SPaT prediction for the first time across the six intersections. Transformers are shown to outperform other models including LSTM. The ensemble provides a valuable metric to show the certainty level in each of the predictions through the level of consensus of the models. Finally, a hybrid approach integrating deep learning and controller logic is proposed by predicting actuations separately and using a digital twin to replicate SPaT information. The approach is proven to be the best approach with 58% less mean absolute error than other approaches. Overall, this dissertation provides a holistic methodology for predicting SPaT and the certainty level associated with it tailored to the existing technology and communication needs. / Doctor of Philosophy / Automated and connected vehicles waste a lot of fuel and energy to stop and go at traffic signals. The ideal case is for them to be able to know when the traffic signal turns green ahead of time and plan to reach the intersection by the time it is green, so they do not have to stop. Not having to stop can save up to 40 percent of the gas used at the intersection. This is a difficult task because the green time is not fixed. It has a minimum and maximum setting, and it keeps extending the green every time a new vehicle arrives. While this is good for adapting to traffic, it makes it difficult to know exactly when the traffic signal turns green to reach the intersection at that time. In this dissertation, different models to know ahead of time when the traffic signal will change are used. A model is chosen known as long-short term memory neural network (LSTM), which is a way to recognize how the traffic signal is expected to behave in the future from its past behavior. The point is to reduce the errors in the predictions. The first thing is to look at the loss function, which is how the model deals with error. It is found that the best thing is to take the average of the absolute value of the error as a percentage of the prediction if the prediction is that traffic signal will change soon. If it is a longer time until the traffic signal changes, the best way is to take the average of the square of the error. Finally, another function is introduced to balance between both. The second thing explored is how far back in time data was needed to be given to the model to predict accurately. For predictions of less than 20 seconds in the future, only 3 to 10 seconds in the past are needed. For predictions further in the future, looking further back can be useful. The third thing explored was how these models would do after rare events like COVID-19 pandemic. It was found that even though much fewer cars were passing through the intersections, the models still had low errors. Techniques were used to reduce the model reliance on specific data known as regularization techniques. This did not help the models to do better after COVID, but two techniques known as L1 and L2 regularization improved overall performance. The study was then expanded to include 6 intersections and used three additional models in addition to LSTM. One of these models, known as transformers, has never been used before for this problem and was shown to make better predictions than other models. The consensus between the models, which is how many of the models agree on the prediction, was used as a measure for certainty in the prediction. It was proven to be a good indicator. An approach is then introduced that combines the knowledge of the traffic signal controller logic with the powerful predictions of machine learning models. This is done by making a computer program that replicates the logic of the traffic signal controller known as a digital twin. Machine learning models are then used to predict vehicle arrivals. The program is then run using the predicted arrivals to provide a replication of the signal timing. This approach is found to be the best approach with 58 percent less error than the other approaches. Overall, this dissertation provides an end-to-end solution that uses real data generated from intersections to predict the time to green and estimate the certainty in prediction that can help automated and connected vehicles be more fuel efficient.

An autoclave is an advanced machine that sterilizes objects using high-pressure and high heat, with water steam as the medium. Autoclaves are frequently found in hospitals and other places where sterility is required. This project aims to meet the company's need to create a digital twin (DT) of an autoclave. The purpose is to test the control unit that controls the physical autoclave by exposing the DT to the same program as a real autoclave. A DT is a virtual model of a physical system, and in this project, it represented the autoclave and its sensors. The model was programmed in a graphical programming language NI LabView, with the same input and output signals as a real autoclave. The model was based on data-driven logic rather than physical based logic. From a real autoclave run where signals were recorded every second, conclusions could be drawn about how much each unique analog signal changed in combination with other signals through interpolation. The interpolation of the analog signals was used to capture the characteristics of these signals.  For validation, the DT was loaded into a Hardware In the Loop (HIL) system that simulates the autoclave with the DT but retains the control unit from the autoclave, which is the unit the company wants to test. The developed DT was tested against three goals describing how closely the values should align over an accumulated time. The results were compared each second between the real run and the DT run. The data-driven DT model met one of the three goals set, however, the DT model's characteristics resembled those of the real run, making the model useful as the control system does not interrupt the simulation for disallowed or deviant values. / En autoklav är en avancerad maskin som rengör objekt till en steril nivå med ett högt tryck och hög temperatur, där vattenånga används som medium. Autoklaven har ett vanligt förekommande på sjukhus och andra platser där sterilitet är ett krav. Det här projektet går ut på att möta företagets behov av att skapa en digital tvilling (DT) av en autoklav. Syftet är att testa kontrollenheten som styr den fysiska autoklaven genom att en DT ska utsättas för samma programkörning som en verklig autoklav. En DT är en virtuell modell av ett fysiskt system och i detta projekt var autoklaven och sensorerna i maskinen en DT. Modellen programmerades i ett grafiskt programmeringsspråk, NI LabVIEW med samma in- och utsignaler som en verklig autoklav. Modellen utgår ifrån en datadriven metod och inte en fysikalisk formulerad logik. Datan samlades in från en körning av en verklig autoklav, där signalerna sparades varje sekund. Slutsatser för hur mycket varje unik analog signal förändrades i kombination med övriga signaler kunde dras med hjälp av interpolering. Interpoleringen av de analoga signalera kunde användas för att fånga deras karakteristik.  För validering integrerades DT i ett Hardware In the Loop (HIL) system som hjälper till att simulera autoklaven. HIL-systemet har kontrollenheten kvar från autoklaven som är den enhet företaget vill utföra tester på. Den framtagna DT testades mot tre mål som beskriver hur nära värdena skall ligga under en ackumulerad tid. Resultatet jämfördes för varje sekund mellan den verkliga och DT körningen. Den datadrivna DT modellen uppfyllde 1 av 3 mål som ställdes, men DT modellens karakteristik efterliknade den från verkliga körningen vilket gör modellen användbar då kontrollsystemet inte avbryter simuleringen för ej tillåtna eller avvikande värden.

The overall goal of this work is to create an open-source digital twin of the wire arc directed energy deposition process using robot operating system 2 for interpass temperature regulation of a maraging steel alloy. This framework takes a novel approach to regulating the interpass temperatures by using in-situational infrared camera data and a closed loop feedback control that is enabled by robot operating system 2. This is the first implementation of robot operating system 2 for wire arc directed energy deposition and this framework outlines a sensor and machine agnostic approach for creating a digital twin of this additive manufacturing process. In-situ control of the welding process is conducted on a maraging steel alloy demonstrating interpass temperature regulation leads to improved as-built surface roughness and more consistent as-built hardness. An evaluation of three distinct weld modes: Pulsed MIG, CMT MIX, and CMT Universal and two primary process parameters: travel speed and wire feed speed was conducted to identify suitable process windows for welding the maraging alloy. Single track welds for each parameter and weld mode combination were produced and evaluated against current weld bead metrics in the literature. Non destructive profilometry and destructive characterization were performed on the single track welds to evaluate geometric features like wetting angle, dilution percentage, and cross sectional area. In addition, the role of material feed rate on heat input and the cross sectional area was examined in relation to the as-built hardness. The robot operating system 2 digital twin provides a visualization environment to monitor and record real time data from a variety of sensors including robot position, weld data, and thermal camera images. Point cloud data is visualized, in real time, to provide insight to the captured weld meta data. Capturing in-situ data from the wire arc directed energy deposition process is critical to establishing an improved understanding of the process for parameter optimization, tool path planning, with both required to build repeatable, quality components. This work presents an open-source method to capture multi-modal data into a shared environment for improved data capture, data sharing, data synchronization, and data visualization. This digital twin provides users enhanced process control capabilities and greater flexibility by utilizing the robot operating system 2 as a middleware to provide interoperability between sensors and machines.

This dissertation aims to provide critical methodological advancements for sensor fusion and physics-informed machine learning in metal additive manufacturing (MAM) to assist practitioners in detecting quality control structural anomalies. In MAM, there is an urgent need to improve knowledge of the internal layer fusion process and geometric variation occurring during the directed energy deposition processes. A core challenge lies in the cyclic heating process, which results in various structural abnormalities and deficiencies, reducing the reproducibility of manufactured components. Structural abnormalities include microstructural heterogeneities, porosity, deformation and distortion, and residual stresses. Data-driven monitoring in MAM is needed to capture process variability, but challenges arise due to the inability to capture the thermal history distribution process and structural changes below the surface due to limitations in in-situ data collection capabilities, physical domain knowledge integration, and multi-data and multi-physical data fusion. The research gaps in developing system-based generalizable artificial intelligence (AI) and machine learning (ML) to detect abnormalities are threefold. (1) Limited fusion of various types of sensor data without handcrafted selection of features. (2) There is a lack of physical domain knowledge integration for various systems, geometries, and materials. (3) It is essential to develop sensor and system integration platforms to enable a holistic view to make quality control predictions in the additive manufacturing process. In this dissertation, three studies utilize various data types and ML methodologies for predicting in-process anomalies. First, a complementary sensor fusion methodology joins thermal and ultrasonic image data capturing layer fusion and structural knowledge for layer-wise porosity segmentation. Secondly, a physics-informed data-driven methodology for joining thermal infrared image data with Goldak heat flux improves thermal history simulation and deformation detection. Lastly, a physics-informed machine learning methodology constrained by thermal physical functions utilizes in-process multi-modal monitoring data from a digital twin environment to predict distortion in the weld bead. This dissertation provides current practitioners with data-driven and physics-based interpolation methods, multi-modal sensor fusion, and anomaly detection insights trained and validated with three case studies.

Cooperative driving in a connected vehicle (CV) environment has received increasing attention over the years due to its ability to enhance driving safety and efficiency. Despite many efforts that have been made in this field, the role of human drivers and pedestrians is frequently omitted. It is important to consider them to develop cooperative driving algorithms that are intelligent and robust to incorporate any uncertainty brought by humans. In this dissertation, a framework of a multi-driver in-the-loop driving simulator and a pedestrian in-the-loop digital twin system is introduced. Three important topics in cooperative driving were investigated using the developed framework: the effects of human-machine interface (HMI) design for cooperative driving, vehicle-pedestrian interaction under occlusion scenarios, and multi-vehicle decision-making at weaving segments. In the first topic, three HMIs were designed for collaborative speed adaptation following the skills, rules, and knowledge (SRK) taxonomy. The HMI designs were tested using a multi-driver simulator, and the results showed that the graphic-based HMI improved cooperative driving performance and was preferred by the participants. In the second task, a Digital Twin framework for CV and pedestrian in-the-loop simulation was proposed based on Carla-Sumo Co-simulation and Cave automatic virtual environment (CAVE). The effects of Vehicle-Pedestrian (V2P) warning systems under occlusion scenarios were investigated for different connectivity and vehicle automation levels. In the third task, an edge-enhanced graph attention deep reinforcement learning algorithm was developed to aid autonomous vehicles in diverging at weaving segments. The results showed that the proposed algorithms outperformed existing models and performed well in real-world driving scenarios. The dissertation provides insights into developing safe and efficient cooperative driving algorithms and applying advanced simulation technologies to human-in-the-loop cooperative driving testing.