This dissertation presents the theory, design, implementation and successful deployment of a cognitive engine decision algorithm by which a cognitive radio-equipped mobile robot may adapt its motion and radio parameters through multi-objective optimization. This provides a proof-of-concept prototype cognitive system that is aware of its environment, its user's needs, and the rules governing its operation. It is to take intelligent action based on this awareness to optimize its performance across both the mobility and radio domains while learning from experience and responding intelligently to ongoing environmental mission changes. The prototype combines the key features of cognitive radios and autonomous vehicles into a single package whose behavior integrates the essential features of both. The use case for this research is a scenario where a small unmanned aerial vehicle (UAV) is traversing a nominally cyclic or repeating flight path (an â •orbitâ •) seeking to observe targets and where possible avoid hostile agents. As the UAV traverses the path, it experiences varying RF effects, including multipath propagation and terrain shadowing. The goal is to provide the capability for the UAV to learn the flight path with respect both to motion and RF characteristics and modify radio parameters and flight characteristics proactively to optimize performance. Using sensor fusion techniques to develop situational awareness, the UAV should be able to adapt its motion or communication based on knowledge of (but not limited to) physical location, radio performance, and channel conditions. Using sensor information from RF and mobility domains, the UAV uses the mission objectives and its knowledge of the world to decide on a course of action. The UAV develops and executes a multi-domain action; action that crosses domains, such as changing RF power and increasing its speed. This research is based on a simple observation, namely that cognitive radios and autonomous vehicles perform similar tasks, albeit in different domains. Both analyze their environment, make and execute a decision, evaluate the result (learn from experience), and repeat as required. This observation led directly to the creation of a single intelligent agent combining cognitive radio and autonomous vehicle intelligence with the ability to leverage flexibility in the radio frequency (RF) and motion domains. Using a single intelligent agent to optimize decision making across both mobility and radio domains is unified multi-domain decision making (UMDDM). / Ph. D.

No / Clustering is an essential research problem which has received considerable attention in the research community for decades. It is a challenge because there is no unique solution that fits all problems and satisfies all applications. We target to get the most appropriate clustering solution for a given application domain. In other words, clustering algorithms in general need prior specification of the number of clus- ters, and this is hard even for domain experts to estimate especially in a dynamic environment where the data changes and/or become available incrementally. In this paper, we described and analyze the effec- tiveness of a robust clustering algorithm which integrates multi-objective genetic algorithm into a frame- work capable of producing alternative clustering solutions; it is called Multi-objective K-Means Genetic Algorithm (MOKGA). We investigate its application for clustering a variety of datasets, including micro- array gene expression data. The reported results are promising. Though we concentrate on gene expres- sion and mostly cancer data, the proposed approach is general enough and works equally to cluster other datasets as demonstrated by the two datasets Iris and Ruspini. After running MOKGA, a pareto-optimal front is obtained, and gives the optimal number of clusters as a solution set. The achieved clustering results are then analyzed and validated under several cluster validity techniques proposed in the litera- ture. As a result, the optimal clusters are ranked for each validity index. We apply majority voting to decide on the most appropriate set of validity indexes applicable to every tested dataset. The proposed clustering approach is tested by conducting experiments using seven well cited benchmark data sets. The obtained results are compared with those reported in the literature to demonstrate the applicability and effectiveness of the proposed approach.

京都大学 / 新制・課程博士 / 博士(情報学) / 甲第24264号 / 情博第808号 / 新制||情||136(附属図書館) / 京都大学大学院情報学研究科数理工学専攻 / (主査)教授 山下 信雄, 准教授 福田 秀美, 教授 太田 快人 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

The complete design process for a new nuclear power plant concept is costly, long, complicated, and the work is generally split between several specialized groups. These design groups separately do their best to design the portion of the reactor that falls in their expertise according to the design criteria before passing the design to the subsequent design group. Ultimately, the work of each design group is combined, with significant iteration between groups striving to facilitate the integration of each of the heavily interdependent systems. Such complex interaction between experts leads to three significant problems: (1) the issues associated with knowledge management, (2) the lack of design optimization, and (3) the failure to discover the hidden interdependencies between different design parameters that may exist. Some prior work has been accomplished in both developing common frame of reference (CFR) support systems to aid in the design process and applying optimization to nuclear system design.The purpose of this work is to use multi-objective optimization to address the second and third problems above on a small subset of reactor design scenarios. Multi-objective optimization generates several design optima in the form of a Pareto front, which portrays the optimal trade-off between design objectives. As a major part of this work, a system design optimization tool is created, namely the Optimization and Preference Tool for the Improvement of Nuclear Systems (OPTIONS). The OPTIONS tool is initially applied to several individual nuclear systems: the power conversion system (PCS) of the Integral, Inherently Safe Light Water Reactor (I²S-LWR), the Kalina cycle being proposed as the PCS for a LWR, the PERCS (or Passive Endothermic Reaction Cooling System), and the core loop of the Zion plant. Initial sensitivity analysis work and the application of the Non-dominated Sorting Particle Swarm Optimization (NSPSO) method provides a Pareto front of design optima for the PCS of the I²S-LWR, while bringing to light some hidden pressure interdependencies for generating steam using a flash drum. A desire to try many new PCS configurations leads to the development of an original multi-objective optimization method, namely the Mixed-Integer Non-dominated Sorting Genetic Algorithm (MI-NSGA). With this method, the OPTIONS tool provides a novel and improved Pareto front with additional optimal PCS configurations. Then, the simpler NSGA method is used to optimize the Kalina cycle, the PERCS, and the Zion core loop, providing each problem with improved designs and important objective trade-off information. Finally, the OPTIONS tool uses the MI-NSGA method to optimize the integration of three systems (Zion core loop, PERCS, and Rankine cycle PCS) while increasing efficiency, decreasing costs, and improving performance. In addition, the tool is outfitted to receive user preference input to improve the convergence of the optimization to a Pareto front.

Large-Scale Digital Twins Notification System (LSDTNS) monitors a Digital Twin (DT) cluster for a predefined critical state, and once it detects such a state, it sends a Notification Event (NE) to a predefined recipient. Additionally, the time from producing the DT's Complex Event (CE) to sending an alarm has to be less than a predefined deadline. However, addressing scalability and multi-objectives, such as deployment cost, resource utilization, and meeting the deadline, on top of process scheduling, presents a complex challenge. Therefore, this thesis presents a complex methodology consisting of three contributions that address system scalability, multi-objectivity and scheduling of CE processes using Reinforcement Learning (RL). The first contribution proposes the IoT Notification System Architecture based on a micro-service-based notification methodology that allows for running and seamlessly switching between various CE reasoning algorithms. Our proposed IoT Notification System architecture addresses the scalability issue in state-of-the-art CE Recognition systems. The second contribution proposes a novel methodology for multi-objective optimization for cloud provisioning (MOOP). MOOP is the first work dealing with multi-optimization objectives for microservice notification applications, where the notification load is variable and depends on the results of previous microservices subtasks. MOOP provides a multi-objective mathematical cloud resource deployment model and demonstrates effectiveness through the case study. Finally, the thesis presents a Scheduler for large-scale Critical Notification applications based on a Deep Reinforcement Learning (SCN-DRL) scheduling approach for LSDTNS using RL. SCN-DRL is the first work dealing with multi-objective optimization for critical microservice notification applications using RL. During the performance evaluation, SCN-DRL demonstrates better performance than state-of-the-art heuristics. SCN-DRL shows steady performance when the notification workload increases from 10% to 90%. In addition, SCN-DRL, tested with three neural networks, shows that it is resilient to sudden container resources drop by 10%. Such resilience to resource container failures is an important attribute of a distributed system.

The agile creation of 3D CAD platforms (3D CAD models that can be configured to obtain a family of Products) has become an important factor for increasing competitiveness of organizations that create discrete products. Design Automation (DA) is a powerful tool that can be used for speeding up and optimizing the design process of those 3D CAD platforms. Nonetheless, for effectively applying DA on the development of 3D CAD platforms it is desirable to count on tools that address the three fundamental hurdles that are also obstructing the wide adoption of DA in practice. These hurdles are the lack of identification of DA opportunities, the absence of performance metrics, and the absence of methods for continuous improvement. This dissertation contributes a set of methods and tools to incrementally improve the process for creating 3D CAD platforms towards increased automation. The tools proposed include the development of a Metrics Framework (MF) for assessing the capabilities of an organization for creating 3D CAD platforms; a method for increasing the organizational capabilities for creating 3D CAD platforms, and a method for identifying optimal improvement efforts for creating 3D CAD platforms in a multi-objective scenario

This research study focuses on solving the deployment planning problem for UAV-BSs using Multi-Objective Evolutionary Algorithms (MOEAs). The main research objectives encompass gridbased modelling of the target area, investigating evolution parameters, and evaluating algorithm performance in diverse deployment scenarios. Cost, coverage, and interference are considered as objectives along with specific constraints to generate optimal deployment plans. The solution incorporates objective decision support for selecting the best solution among the Pareto front. The research also accounts for parameter initialization and UAV network heterogeneity. Through comprehensive evaluations, the proposed solution demonstrates computational efficiency and the ability to generate satisfactory deployment plans. The study recommends using NonDominated Sorting Genetic Algorithm-II (NSGA-II) for optimal performance. The research also incorporates a fitness approximation technique to reduce computational time while maintaining solution quality. The findings provide valuable insights and recommendations for efficient and balanced deployment planning. However, the research acknowledges limitations and suggests future enhancements. Overall, this research contributes to the field by establishing a foundation for robust and practical deployment plans, guiding future advancements. Future research should focus on addressing identified limitations to enhance applicability and effectiveness in real-world deployment scenarios.

The topic of multi-objective algorithms has been researched for many years, where hundreds of multi-objective algorithms have been developed. With the field of search-based software engineering attracting use-cases, more research on which algorithms are fitting the area is still lacking. Comparing algorithms is key to fully understand what properties of multi-objective algorithms can bring benefit to an application area. In this case, search-based software testing. Three multi-objective algorithms—AGE-II, SIBEA and MOEA/D—are selected for implementation in the Evosuite tool to compare against the two popular algorithms NSGA-II and SPEA2, together with a random-search baseline. The algorithms create test cases for 100 randomly selected Java classes in the SF100 benchmark suite over a maximum of 1000 generations. The objectives of the algorithms are to maximize four coverage criteria—banch, line, method, and statement. The benchmarking shows that SPEA2 completes themost goals on average at 67.04%, also having the most stable results at ±20.28 goals for the ten executions. Tests generated by the random-search baseline has the highest branch, line and statement coverage, with NSGA-II generating tests with the best method coverage. Neither AGE-II or SIBEA could compete with the algorithms already implemented in Evosuite. The MOEA/D implementation could not be completed. More implementations and benchmarks of multi-objective algorithms are needed to find concrete links between technique to goals and coverage amount, especially surrounding the random-search baseline being a top performer.

Nowadays, it has become crucial to transform company production processes in order to reduce carbon emissions. Therefore, the company LKAB is working to make the production process of iron ore pellets fossil-free. In particular, this thesis project focuses on the pellet induration stage and addresses the mathematical optimization of the process. In particular, the idea is to combine state-of-the-art optimization algorithms with simulation software. This thesis aims to address the problem of multi-objective optimization within the context of simulation-based scenarios, i.e., the aim is to study Simulation-Based Multi-Objective Optimization Problems. The primary focus of this thesis is to thoroughly investigate and compare three different algorithms applied to two distinct problem formulations. By doing so, we aim to gain valuable insights into the suitability of different approaches and evaluate the algorithms' performance in achieving the desired objectives. / Nuförtiden har det blivit mycket viktigt att förändra företagens produktionsprocesser för att minska koldioxidutsläppen. Företaget LKAB arbetar för att göra produktionsprocessen av järnmalmspellets fossilfri. Detta examensarbete fokuserar på pelletiseringsfasen och behandlar den matematiska optimeringen av processen. Mer specifikt, tanken är att kombinera toppmoderna optimeringsalgoritmer med simuleringsprogramvara. Syftet med examensarbetet är att studera problem med flermålsoptimering inom ramen för simuleringsbaserade scenarier, dvs. syftet är att studera ett simuleringsbaserat flermålsoptimeringsproblem. Det primära fokuset i examnesarbetet är att grundligt undersöka och jämföra tre olika algoritmer som tillämpas på två distinkta problemformuleringar. Genom att göra detta vill vi få värdefull insikt om lämpligheten av de olika strategierna och utvärdera algoritmernas prestanda för att uppnå de önskade målen.

The global prominence of energy-efficient retrofit in the context of aging properties has garnered noteworthy attention. This surge in interest can be attributed to several advantages, encompassing economically viable carbon dioxide (CO₂) emissions reduction, diminished energy expenditures, and improved indoor air quality. Passive retrofits, such as thermal insulation and fenestration improvement, and active retrofits, such as heating setpoint temperature optimization, offer great potential for CO₂ reduction and energy savings. The central objective of this study is ascertaining the feasibility of attaining a 40% reduction in CO₂ emissions with the lowest cost and with constraints on heating setpoints temperature by finding optimal design parameters encompassing thermal insulation (including both single and double-layer), fenestration, and heating setpoint temperatures. This inquiry is substantiated through a case study of the Leblanc residence on the University of Ottawa campus. In pursuit of this objective, a thermal model of the Leblanc building was developed via EnergyPlus and subsequently subjected to a validation process following ASHRAE Guideline 14. After validation, an array of discrete optimization scenarios was executed using the NSGA-II model, facilitated by the JEPLUS+EA software. This approach aimed to identify the most suitable parameters for achieving optimal CO₂ reduction and cost outcomes. Notably, the results showcased 20 solutions, each boasting a reduction of 40% or more in CO₂ emissions and heating setpoint temperature higher than 18 °C. While the choice to prioritize either cost or CO₂ reduction remains at the user's discretion, four solutions have been discerned as the most effective. Furthermore, the findings suggest that implementing these optimal solutions can significantly decrease CO₂ emissions, ranging between 41.79% and 46.36%. The associated costs were also determined to fall within $36,262 to $57,934.