The importance of balance between exploration and exploitation plays a crucial role while solving combinatorial optimization problems. This balance is reached by two general techniques: by using an appropriate problem solver and by setting its proper parameters. Both problems were widely studied in the past and the research process continues up until now. The latest studies in the field of automated machine learning propose merging both problems, solving them at design time, and later strengthening the results at runtime. To the best of our knowledge, the generalized approach for solving the parameter setting problem in heuristic solvers has not yet been proposed. Therefore, the concept of merging heuristic selection and parameter control have not been introduced. In this thesis, we propose an approach for generic parameter control in meta-heuristics by means of reinforcement learning (RL). Making a step further, we suggest a technique for merging the heuristic selection and parameter control problems and solving them at runtime using RL-based hyper-heuristic. The evaluation of the proposed parameter control technique on a symmetric traveling salesman problem (TSP) revealed its applicability by reaching the performance of tuned in online and used in isolation underlying meta-heuristic. Our approach provides the results on par with the best underlying heuristics with tuned parameters.:1 Introduction 1 1.1 Motivation 1 1.2 Research objective 2 1.3 Solution overview 2 2 Background and RelatedWork Analysis 3 2.1 Optimization Problems and their Solvers 3 2.2 Heuristic Solvers for Optimization Problems 9 2.3 Setting Algorithm Parameters 19 2.4 Combined Algorithm Selection and Hyper-Parameter Tuning Problem 27 2.5 Conclusion on Background and Related Work Analysis 28 3 Online Selection Hyper-Heuristic with Generic Parameter Control 31 3.1 Combined Parameter Control and Algorithm Selection Problem 31 3.2 Search Space Structure 32 3.3 Parameter Prediction Process 34 3.4 Low-Level Heuristics 35 3.5 Conclusion of Concept 36 4 Implementation Details 37 4.2 Search Space 40 4.3 Prediction Process 43 4.4 Low Level Heuristics 48 4.5 Conclusion 52 5 Evaluation 55 5.1 Optimization Problem 55 5.2 Environment Setup 56 5.3 Meta-heuristics Tuning 56 5.4 Concept Evaluation 60 5.5 Analysis of HH-PC Settings 74 5.6 Conclusion 79 6 Conclusion 81 7 FutureWork 83 7.1 Prediction Process 83 7.2 Search Space 84 7.3 Evaluations and Benchmarks 84 Bibliography 87 A Evaluation Results 99 A.1 Results in Figures 99 A.2 Results in numbers 105

Optimization is omnipresent in our world. Its numerous applications spread from industrial cases, such as logistics, construction management or production planning; to the private sphere, filled with problems of selecting daycare or vacation planning. In this thesis, we concentrate on expensive black-box optimization (EBBO) problems, a subset of optimization problems (OPs), which are characterized by an expensive cost of evaluating an objective function. Such OPs are reoccurring in various domains, being known as: hyperpameter optimization in machine learning, performance configuration optimization or parameter tuning in search-based software engineering, simulation optimization in operations research, meta-optimization or parameter tuning in the optimization domain itself. High diversity of domains introduces a plethora of solving approaches, which adhere to a similar structure and workflow, but differ in details. The software frameworks stemming from different areas possess only partially intersecting manageability points, i.e., lack manageability. In this thesis, we argue that the lack of manageability in EBBO is a major problem, which leads to underachieving optimization quality. The goal of this thesis is to study the role of manageability in EBBO and to investigate whether improving the manageability of EBBO frameworks increases optimization quality. To reach this goal, we appeal to software product line engineering (SPLE), a methodology for developing highly-manageable software systems. Based on the foundations of SPLE, we introduce a novel framework for EBBO called BRISE. It offers: 1) a loosely-coupled software architecture, separating concerns of the experiment designer and the developer of EBBO strategies; 2) a full coverage of all EBBO problem types; and 3) a context-aware variability model, which captures the experiment-designer-defined OP with the content model; and manageability points including their variants and constraints with the cardinality-based feature model. High manageability of the introduced BRISE framework enables us: 1) to extend the framework with novel efficient strategies, such as adaptive repetition management; and 2) to introduce novel EBBO mechanisms, such as multi-objective compositional surrogate modeling, dynamic sampling and hierarchical surrogate modeling. The evaluation of the novel approaches with a set of case studies, including: the WFG benchmark for multi-objective optimization, combined selection and parameter control of meta-heuristics, and energy optimization; demonstrated their superiority over the state-of-the-art competitors. Thus, it supports the research hypothesis of this thesis: Improving manageability of an EBBO framework enables to increase optimization quality.

Vid samverkansbroar fästes tvärförband mellan huvudbalkar med funktion att leda laster ut mot huvudbalkarna. Dessa kan fästas via s.k. livavstyvningsplåtar mot huvudbalkens liv samt flänsar. Denna infästning kan ske med svetsförband. Vid ett antal fall har utmattningssprickor uppkommit i svetsanslutningen mellan livavstyvningsplåten samt överfläns på huvudbalk. Tydliga riktlinjer finns ej i dagsläget för vilken typ av svetsförband som bör nyttjas vid uppförande av benämnd konstruktionsanslutning. Här genomförs en jämförande studie av kälsvetsförband mot stumsvetsförband, med samt utan svetsutbredning. Då risk för sprickinitiering till följd av utmattning i kälsvetsförbandet studerades beaktades risk för brott i såväl svetstå som svetsrot. Den aktuella studien ämnade ge svar på huruvida en korrelation mellan geometriska parametrar såsom avstånd mellan huvudbalkar samt konsollängd på farbana kunde kopplas till rekommendation gällande vilket svetsutförande som borde nyttjas vid olika dimensioner av dessa båda geometrier. Vidare ämnade studien utröna om eventuella gränsvärden för val av svetsanslutningar kunde fastställas. För framtagande av data nyttjades BRIGADE PLUS, ett beräkningsprogram som nyttjar FEM-modellering. Här användes ett uppfört projekt, Viadukt Söder, beläget i Östersund, som grundmodell för parameterstudien i BRIGADE PLUS. Detta var projekterat av ELU Konsult, på uppdrag av Trafikverket. Beräkningsmodellen bestod av en begränsad del av den aktuella bron, där utmattningslastfordon 3 i SS EN 1993-2 användes för att via influenslinjer återspegla trafiklast på bron och därmed finna värsta lastpositioner. Dessa positioner nyttjades sedermera för beräkning av påverkan från Utmattningslastfordon 3 på den studerade detaljen i form av spänningsamplitud. Analys av dessa spänningsamplituder omräknades sedermera till ett antal lastcykler till brott som sedan ställdes mot de riktlinjer som återfinns i SS EN 1993-1-9:2005. Här nyttjades de förbandsklasser som enligt nationella normer ses som gällande. Resultatet påvisade en exponentiell minskning av antal lastcykler till brott, med avseende på utmattning, då brogeometrin utökades. Utmärkande här var att risk för brott i livavstyvningsplåten klart översteg risk för brott i överflänsen, som i sin tur näst intill kunde bortses ifrån. Den geometriska parameter som blev styrande var konsollängd på farbana. Tydlig differens i livslängd mätt i antal cykler gick att utläsa för de olika svetsförbanden, där styrande faktor blev de respektive förbandsklasserna vid analys av sprickinitiering. Gränsvärden för då antal cykler till brott understeg gränsvärden enligt SS EN 1993-1-9:2005 kopplat till de geometriska parametrarna, gick att fastställa för respektive svetsförband. Kälsvetsens kapacitet med avseende på utmattning understeg vida stumsvetsens kapacitet. / For composite bridges, transverse bracings are attached between main girders with the function of distributing loads between the main girders. These can be attached via so-called web stiffeners against the main girders web and flanges. This attachment can be done via welded joints. In the event of emergence, fatigue cracks have arisen in the welding connection between the web stiffener and the top flange of the main girder. Clear guidelines do not exist at present for the type of welded joints that should be used in the construction of the termed design connection. Here, a comparative study is carried out between fillet welded joints and butt welded joints, with flat and inclined welds. When the risk of crack initiation as a result of fatigue in the fillet weld was studied, the risk of breakage in both toe and root was considered. The current study intended to provide answers as to whether a correlation between geometric parameters such as distance between main girders and cantilever length of the deck slab could be linked to recommendation regarding which welding design should be used at different dimensions of these two properties. Furthermore, the study intended to determine whether any limit values for the selection of welding connections could be determined. For the production of data, BRIGADE PLUS, a calculation program that utilizes FEM modeling, was used. Here, a constructed project, Viadukt Söder, situated in Östersund, was used as the basic model for the parameter study in BRIGADE PLUS. This was designed by ELU Konsult, on behalf of the Swedish Transport Administration. The calculation model consisted of a limited part of the current bridge, where Fatigue Load Vehicle 3 in SS EN 1993-2 was used to reflect traffic loads on the bridge via influenza lines and thereby find the worst load positions. These positions were subsequently used for calculating the influence of Fatigue Load Vehicle 3 on the studied part in the form of stress amplitude. Analysis of these stress amplitudes was subsequently recalculated to a number of load cycles to offenses that were then set against the guidelines found in SS EN 1993-1-9: 2005. Here, the fatigue resistance values were used which according to national standards are regarded as valid. The result showed an exponential decrease in the number of load cycles to fracture, with respect to fatigue, as the bridge geometry was increased. The characteristic feature here was that the risk of breakage in the web stiffener clearly exceeded the risk of breakage in the upper flange, which in turn could almost be disregarded. The geometric parameter that became guiding was cantilever length on the deck. Clear difference in service life measured in number of cycles could be read out for the different welded joints, where the controlling factor became the respective fatigue resistance values when analyzing crack initiation. Limit values for when the number of cycles to break below the limit values according to SS EN 1993-1-9:2005 linked to the geometric parameters, could be determined for each welded joint. The capacity of the fillet weld with respect to fatigue is far below the capacity of the butt weld.

BIM (Building Information Modeling) är idag ett vedertaget begrepp inom byggprojektering.En viktig styrka med BIM är möjligheten till koordinering mellan samtligaaktörer i projektet. BIM-verktygen bygger på parametrisering vilket innebär att modellensobjekt har en stor uppsättning data kopplad till sig. Objektens egenskaperkan ändras genom att justera parameterdatan. Parametriseringen medför också attdata relativt enkelt kan nås, extraheras och styras av externa program. Det är i dettasyfte som implementeringen av grafisk programmering fyller en viktig funktion. Medhjälp av grafisk programmering kan flera operationer i BIM-arbetet automatiseras.Därmed finns potentialen att spara värdefull tid för byggprojektören. Autodesk Revit är ett av de verktyg som dominerar BIM-marknaden. I Revit kananvändaren designa och dokumentera ett byggprojekt i alla dess faser. Sedan någraår tillbaka finns det ett tillägg till programmet som heter Dynamo. Dynamo tillåteranvändaren att styra Revit med hjälp av skript i en användarvänlig grafisk miljö.I denna uppsats redogörs för Dynamos möjligheter och begränsningar applicerat påväletablerade arbetsmoment inom byggprojektering. Här redogörs för optimeringspotentialenoch lämpliga användningsområden. Slutsatser som dragits grundar sig ihög grad på en empirisk undersökning där ett antal grafiska skript har skapats ochutvärderats. Resultatet visar på att fl era vitala moment i byggprojektörens vardagliga arbetsflödekan effektiviseras med hjälp av införandet av grafisk programmering. Med effektiviseringavses i detta fall primärt tidsbesparingar som bidrar till ökad ekonomiskhållbarhet för byggprojekteringsföretaget. / BIM (Building Information Modeling) is today a proven concept in construction design.An important strength with BIM is the possibility of coordination between allactors in the project. The BIM tools are based on parameterization, which meansthat the model's object has a large set of data attached to it. The properties ofthe objects can be changed by adjusting the parameter data. Parameterization alsomeans that data can be easily accessed, extracted and controlled by external applications.It is for this purpose that the implementation of graphic programming playsan important role. With the help of graphical programming, several operations inthe BIM work can be automated. Thus, there is the potential to save valuable timefor the construction projector. Autodesk Revit is one of the tools that dominates the BIM market. In Revit, theuser can design and document a construction project in all its phases. For a fewyears back, there is an addition to the program called Dynamo. Dynamo allows theuser to control Revit using scripts in a user-friendly graphical environment.This essay explains Dynamo's possibilities and limitations applied to work processesin building design. It describes the optimization potential and appropriate applications.Conclusions are mainly drawn based upon the generation and evaluation ofgraphical scripts. The result shows that several vital moments in the construction projector's dailywork ow can be rationalized through the introduction of graphic programming. Rationalizationmeans time savings that contribute to increased economic sustainabilityfor the construction company.

[pt] O Algoritmo Genético de Chaves Aleatórias Enviesadas (BRKGA) é uma metaheurística populacional utilizada na obtenção de soluções ótimas ou quase ótimas para problemas de otimização combinatória. A parametrização do algoritmo é crucial para garantir seu bom desempenho. Os valores dos parâmetros têm uma grande influência em determinar se uma boa solução será encontrada pelo algoritmo e se o processo de busca será eficiente. Uma maneira de resolver esse problema de configuração de parâmetros é por meio da abordagem de parametrização online (ou controle de parâmetros). A parametrização online permite que o algoritmo adapte os valores dos parâmetros de acordo com os diferentes estágios do processo de busca e acumule informações sobre o espaço de soluções nesse processo para usar as informações obtidas em estágios posteriores. Ele também libera o usuário da tarefa de definir as configurações dos parâmetros, resolvendo implicitamente o problema de configuração. Neste trabalho, avaliamos duas estratégias para implementar o controle de parâmetros no BRKGA. Nossa primeira abordagem foi adotar valores de parâmetros aleatórios para cada geração do BRKGA. A segunda abordagem foi incorporar os princípios adotados pelo irace, um método de parametrização do estado da arte, ao BRKGA. Ambas as estratégias foram avaliadas em três problemas clássicos de otimização (Problema de Permutação Flowshop, Problema de Cobertura de Conjuntos e Problema do Caixeiro Viajante) e levaram a resultados competitivos quando comparados ao algoritmo tunado. / [en] The Biased Random-Key Genetic Algorithm (BRKGA) is a populationbased metaheuristic applied to obtain optimal or near-optimal solutions to combinatorial problems. To ensure the good performance of this algorithm (and other metaheuristics in general), defining parameter settings is a crucial step. Parameter values have a great influence on determining whether a good solution will be found by the algorithm and whether the search process will be efficient. One way of tackling the parameter setting problem is through the parameter control (or online tuning) approach. Parameter control allows the algorithm to adapt parameter values according to different stages of the search process and to accumulate information on the fitness landscape during the search to use this information in later stages. It also releases the user from the task of defining parameter settings, implicitly solving the tuning problem. In this work, we evaluate two strategies to implement parameter control in BRKGA. Our first approach was adopting random parameter values for each of BRKGA s generations. The second approach was to introduce the principles adopted by Iterated Race, a state-of-the-art tuning method, to BRKGA. Both strategies were evaluated in three classical optimization problems (Flowshop Permutation Problem, Set Covering Problem, and the Traveling Salesman Problem) and led to competitive results when compared to the tuned algorithm.

Oscillating positive displacement pumps are used in many industrial sectors. Mechanical stroke generators / drives such as crank drive, spring-cam drive etc have reached a high technological level, but to which are set mechanical limits. Especially in the smallest dosing range <1 l/h at a pressure range from 100 … 400 bar considerable optimisation is still required with regard to precision and continuity / control range of the dosing flow. In order to expand the current application possibilities for the diaphragm metering pump technology, it is necessary to use new drive systems such as linear motor technology. A linear motor is an electric drive unit which transmits the oscillating delivery movement of the dosing pumps directly to the displacer (hydraulic piston, diaphragm, etc.) without any mechanics, so that highly dynamic movements can be carried out with maximum precision, an individual kinematic profile and a control range of 1:200. The examples of gas odorization and filling processes show how the linear motor drive can be used to technologically solve and even optimize the customer's process requirements. The linear motor pump can realize an integrated 3-parameter control and is therefore suitable for almost any kinematically solvable dosing task.

No description available.