On uncertainty and robustness in evolutionary optimization-based multicriterio decision-making

Programa de Doctorado: Sistemas Inteligentes y Aplicaciones Numéricas en Ingeniería

Engineering knowledge-based variance-reduction simulation and G-dominance for structural frame robust optimization

[EN] This paper proposes the incorporation of engineering knowledge through both (a) advanced state-of-the-art preference handling decision-making tools integrated in multiobjective evolutionary algorithms and (b) engineering knowledge-based variance reduction simulation as enhancing tools for the robust optimum design of structural frames taking uncertainties into consideration in the design variables.The simultaneous minimization of the constrained weight (adding structuralweight and average distribution of constraint violations) on the one hand and the standard deviation of the distribution of constraint violation on the other are handled with multiobjective optimization-based evolutionary computation in two different multiobjective algorithms. The optimum design values of the deterministic structural problem in question are proposed as a reference point (the aspiration level) in reference-point-based evolutionary multiobjective algorithms (here g-dominance is used). Results including

T-spline parameterization of 2D geometries based on the meccano method with a new T-mesh optimization algorithm

[EN]We present a new method, based on the idea of the meccano method and a novel T-mesh optimization procedure, to construct a T-spline parameterization of 2D geometries for the application of isogeometric analysis. The proposed method only demands a boundary representation of the geometry as input data. The algorithm obtains, as a result, high quality parametric transformation between 2D objects and the parametric domain, the unit square. First, we define a parametric mapping between the input boundary of the object and the boundary of the parametric domain. Then, we build a T-mesh adapted to the geometric singularities of the domain in order to preserve the features of the object boundary with a desired tolerance…

Optimization of RNA-based transgene expression by targeting Protein Kinase R

The aim of this thesis was to establish a method for repeated transfection of in vitro transcribed RNA (IVT-RNA) leading to a sustained protein expression lasting for days or even weeks. Once transfected cells recognize IVT-RNA as "non-self" and initiate defense pathways leading to an upregulated interferon (IFN) response and stalled translation. In this work Protein Kinase R (PKR) was identified as the main effector molecule mediating this cellular response. We assessed four strategies to inhibit PKR and the IFN response: A small molecule PKR inhibitor enhanced protein expression and hampered the induction of IFN-transcripts, but had to be excluded due to cytotoxicity. A siRNA mediated PKR knockdown and the overexpression of a kinase inactive PKR mutant elevated the protein expression, but the down-regulation of the IFN response was insufficient. The co-transfer of the viral inhibitors of PKR and the IFN response was most successful. The use of E3, K3 and B18R co-transfection enabled repeated IVT-RNA-based transfection of human fibroblasts. Thus, the developed protocol allows a continuous IVT-RNA encoded protein expression of proteins, which could be the basis for the generation of induced pluripotent stem cells (iPS) for several therapeutic applications in regenerative medicine or drug research.

Modelling, simulation and optimization of maintenance considerations on condition based maintenance

Globalization has increased the pressure on organizations and companies to operate in the most efficient and economic way. This tendency promotes that companies concentrate more and more on their core businesses, outsource less profitable departments and services to reduce costs. By contrast to earlier times, companies are highly specialized and have a low real net output ratio. For being able to provide the consumers with the right products, those companies have to collaborate with other suppliers and form large supply chains. An effect of large supply chains is the deficiency of high stocks and stockholding costs. This fact has lead to the rapid spread of Just-in-Time logistic concepts aimed minimizing stock by simultaneous high availability of products. Those concurring goals, minimizing stock by simultaneous high product availability, claim for high availability of the production systems in the way that an incoming order can immediately processed. Besides of design aspects and the quality of the production system, maintenance has a strong impact on production system availability. In the last decades, there has been many attempts to create maintenance models for availability optimization. Most of them concentrated on the availability aspect only without incorporating further aspects as logistics and profitability of the overall system. However, production system operator’s main intention is to optimize the profitability of the production system and not the availability of the production system. Thus, classic models, limited to represent and optimize maintenance strategies under the light of availability, fail. A novel approach, incorporating all financial impacting processes of and around a production system, is needed. The proposed model is subdivided into three parts, maintenance module, production module and connection module. This subdivision provides easy maintainability and simple extendability. Within those modules, all aspect of production process are modeled. Main part of the work lies in the extended maintenance and failure module that offers a representation of different maintenance strategies but also incorporates the effect of over-maintaining and failed maintenance (maintenance induced failures). Order release and seizing of the production system are modeled in the production part. Due to computational power limitation, it was not possible to run the simulation and the optimization with the fully developed production model. Thus, the production model was reduced to a black-box without higher degree of details.

Development of a model-based transient calibration process for diesel engine electronic control module tables – Part 2: modelling and optimization

This is the second part of a study investigating a model-based transient calibration process for diesel engines. The first part addressed the data requirements and data processing required for empirical transient emission and torque models. The current work focuses on modelling and optimization. The unexpected result of this investigation is that when trained on transient data, simple regression models perform better than more powerful methods such as neural networks or localized regression. This result has been attributed to extrapolation over data that have estimated rather than measured transient air-handling parameters. The challenges of detecting and preventing extrapolation using statistical methods that work well with steady-state data have been explained. The concept of constraining the distribution of statistical leverage relative to the distribution of the starting solution to prevent extrapolation during the optimization process has been proposed and demonstrated. Separate from the issue of extrapolation is preventing the search from being quasi-static. Second-order linear dynamic constraint models have been proposed to prevent the search from returning solutions that are feasible if each point were run at steady state, but which are unrealistic in a transient sense. Dynamic constraint models translate commanded parameters to actually achieved parameters that then feed into the transient emission and torque models. Combined model inaccuracies have been used to adjust the optimized solutions. To frame the optimization problem within reasonable dimensionality, the coefficients of commanded surfaces that approximate engine tables are adjusted during search iterations, each of which involves simulating the entire transient cycle. The resulting strategy, different from the corresponding manual calibration strategy and resulting in lower emissions and efficiency, is intended to improve rather than replace the manual calibration process.

AGENT BASED, DISTRIBUTED CONTROL STRATEGIES AND OPTIMIZATION OF PLUG-IN ELECTRIC VEHICLES ON SMART/MICROGRID ARCHITECTURES

This thesis will present strategies for the use of plug-in electric vehicles on smart and microgrids. MATLAB is used as the design tool for all models and simulations. First, a scenario will be explored using the dispatchable loads of electric vehicles to stabilize a microgrid with a high penetration of renewable power generation. Grid components for a microgrid with 50% photovoltaic solar production will be sized through an optimization routine to maintain storage system, load, and vehicle states over a 24-hour period. The findings of this portion are that the dispatchable loads can be used to guard against unpredictable losses in renewable generation output. Second, the use of distributed control strategies for the charging of electric vehicles utilizing an agent-based approach on a smart grid will be studied. The vehicles are regarded as additional loads to a primary forecasted load and use information transfer with the grid to make their charging decisions. Three lightweight control strategies and their effects on the power grid will be presented. The findings are that the charging behavior and peak loads on the grid can be reduced through the use of distributed control strategies.

Dynamic Optimization of SLA-Based Services Scaling Rules

Current advanced cloud infrastructure management solutions allow scheduling actions for dynamically changing the number of running virtual machines (VMs). This approach, however, does not guarantee that the scheduled number of VMs will properly handle the actual user generated workload, especially if the user utilization patterns will change. We propose using a dynamically generated scaling model for the VMs containing the services of the distributed applications, which is able to react to the variations in the number of application users. We answer the following question: How to dynamically decide how many services of each type are needed in order to handle a larger workload within the same time constraints? We describe a mechanism for dynamically composing the SLAs for controlling the scaling of distributed services by combining data analysis mechanisms with application benchmarking using multiple VM configurations. Based on processing of multiple application benchmarks generated data sets we discover a set of service monitoring metrics able to predict critical Service Level Agreement (SLA) parameters. By combining this set of predictor metrics with a heuristic for selecting the appropriate scaling-out paths for the services of distributed applications, we show how SLA scaling rules can be inferred and then used for controlling the runtime scale-in and scale-out of distributed services. We validate our architecture and models by performing scaling experiments with a distributed application representative for the enterprise class of information systems. We show how dynamically generated SLAs can be successfully used for controlling the management of distributed services scaling.

A benchmark of kriging-based infill criteria for noisy optimization

Responses of many real-world problems can only be evaluated perturbed by noise. In order to make an efficient optimization of these problems possible, intelligent optimization strategies successfully coping with noisy evaluations are required. In this article, a comprehensive review of existing kriging-based methods for the optimization of noisy functions is provided. In summary, ten methods for choosing the sequential samples are described using a unified formalism. They are compared on analytical benchmark problems, whereby the usual assumption of homoscedastic Gaussian noise made in the underlying models is meet. Different problem configurations (noise level, maximum number of observations, initial number of observations) and setups (covariance functions, budget, initial sample size) are considered. It is found that the choices of the initial sample size and the covariance function are not critical. The choice of the method, however, can result in significant differences in the performance. In particular, the three most intuitive criteria are found as poor alternatives. Although no criterion is found consistently more efficient than the others, two specialized methods appear more robust on average.

Beamlet based direct aperture optimization for MERT using a photon MLC

PURPOSE A beamlet based direct aperture optimization (DAO) for modulated electron radiotherapy (MERT) using photon multileaf collimator (pMLC) shaped electron fields is developed and investigated. METHODS The Swiss Monte Carlo Plan (SMCP) allows the calculation of dose distributions for pMLC shaped electron beams. SMCP is interfaced with the Eclipse TPS (Varian Medical Systems, Palo Alto, CA) which can thus be included into the inverse treatment planning process for MERT. This process starts with the import of a CT-scan into Eclipse, the contouring of the target and the organs at risk (OARs), and the choice of the initial electron beam directions. For each electron beam, the number of apertures, their energy, and initial shape are defined. Furthermore, the DAO requires dose-volume constraints for the structures contoured. In order to carry out the DAO efficiently, the initial electron beams are divided into a grid of beamlets. For each of those, the dose distribution is precalculated using a modified electron beam model, resulting in a dose list for each beamlet and energy. Then the DAO is carried out, leading to a set of optimal apertures and corresponding weights. These optimal apertures are now converted into pMLC shaped segments and the dose calculation for each segment is performed. For these dose distributions, a weight optimization process is launched in order to minimize the differences between the dose distribution using the optimal apertures and the pMLC segments. Finally, a deliverable dose distribution for the MERT plan is obtained and loaded back into Eclipse for evaluation. For an idealized water phantom geometry, a MERT treatment plan is created and compared to the plan obtained using a previously developed forward planning strategy. Further, MERT treatment plans for three clinical situations (breast, chest wall, and parotid metastasis of a squamous cell skin carcinoma) are created using the developed inverse planning strategy. The MERT plans are compared to clinical standard treatment plans using photon beams and the differences between the optimal and the deliverable dose distributions are determined. RESULTS For the idealized water phantom geometry, the inversely optimized MERT plan is able to obtain the same PTV coverage, but with an improved OAR sparing compared to the forwardly optimized plan. Regarding the right-sided breast case, the MERT plan is able to reduce the lung volume receiving more than 30% of the prescribed dose and the mean lung dose compared to the standard plan. However, the standard plan leads to a better homogeneity within the CTV. The results for the left-sided thorax wall are similar but also the dose to the heart is reduced comparing MERT to the standard treatment plan. For the parotid case, MERT leads to lower doses for almost all OARs but to a less homogeneous dose distribution for the PTV when compared to a standard plan. For all cases, the weight optimization successfully minimized the differences between the optimal and the deliverable dose distribution. CONCLUSIONS A beamlet based DAO using multiple beam angles is implemented and successfully tested for an idealized water phantom geometry and clinical situations.

Optimization of Optical Follow-up Strategies Based on Covariance Analysis

An in-depth study, using simulations and covariance analysis, is performed to identify the optimal sequence of observations to obtain the most accurate orbit propagation. The accuracy of the results of an orbit determination/ improvement process depends on: tracklet length, number of observations, type of orbit, astrometric error, time interval between tracklets and observation geometry. The latter depends on the position of the object along its orbit and the location of the observing station. This covariance analysis aims to optimize the observation strategy taking into account the influence of the orbit shape, of the relative object-observer geometry and the interval between observations.

Optimization of TRPV6 Calcium Channel Inhibitors Using a 3D Ligand-Based Virtual Screening Method

Herein, we report the discovery of the first potent and selective inhibitor of TRPV6, a calcium channel overexpressed in breast and prostate cancer, and its use to test the effect of blocking TRPV6-mediated Ca2+-influx on cell growth. The inhibitor was discovered through a computational method, xLOS, a 3D-shape and pharmacophore similarity algorithm, a type of ligand-based virtual screening (LBVS) method described briefly here. Starting with a single weakly active seed molecule, two successive rounds of LBVS followed by optimization by chemical synthesis led to a selective molecule with 0.3 μM inhibition of TRPV6. The ability of xLOS to identify different scaffolds early in LBVS was essential to success. The xLOS method may be generally useful to develop tool compounds for poorly characterized targets.