The Netherlands boasts an extensive road network that requires meticulous maintenance and preservation. Dutch asphalt pavements are assessed through functional properties such as mix stiffness and resistance to fatigue. However, current testing practice for these properties is intensive in time and resources, leading to the exploration of alternative methods for performance prediction. Recently, Artificial Intelligence (AI) has emerged as a tool for performance prediction in pavement engineering. Despite its potential, the application of AI is constrained by its limited interpretability and inconsistency with known physical laws. To enhance consistency and interpretability in AI predictive models, Physics Informed AI (PIAI) emerges as a promising approach.

This research develops a PIAI framework for physics infusion in pavement performance predictions. This infusion is accomplished through a Physics-Informed Loss Function balancing data and physics components in model training. The data component assures model predictions approximate the targets, whereas the physics component enforces a subset of features to follow a preset physical model. These components are also present during feature selection, where the physical model is used to guide the inclusion of important features in a PIAI model.

Using the developed framework, this research presents two PIAI prediction models based on the NL-LAB datasets. These models infuse homogenization theory and energy dissipation theory to enhance interpretability and consistency in stiffness and fatigue predictions. The results obtained on both models suggest that physics infusion is feasible without compromising prediction accuracy, balancing physical and statistical knowledge when predicting pavement performance. These findings also indicate that the PIAI framework is a promising approach for infusing physics into AI prediction models. Physics infusion can potentially enhance the acceptance and trust of AI within the pavement engineering community. Furthermore, the developed framework has the potential to accelerate pavement performance assessments by reducing the need for extensive material testing. Its flexibility also supports the incorporation of new physical models, fostering innovation and sustainability in pavement engineering.

The motivation for this research is rooted in the expected increase in demand for international train trips within the EU railway network, driven by policies promoting sustainable transportation solutions. Notable initiatives include the EU’s Green Deal objectives and the commitment to achieve climate neutrality by 2050. Additionally, significant investments in rail infrastructure, such as the Trans-European Transport Network, the introduction of high-speed rail for international travel, and the reduction of barriers to promote international train travel, are contributing factors. This surge in demand requires a reliable railway network and improved management of temporary capacity reductions, particularly during maintenance operations at cross-border sections. Cross-border rail transport is essential for this strategy but faces significant challenges, including technical barriers, regulatory discrepancies, and coordination inefficiencies among infrastructure managers from different countries.

Despite its importance, academic literature on cross-border maintenance strategies is notably limited or non-existent. This thesis begins by describing the current situation of railway cross-border sections to identify the main challenges and existing coordination practices. It then proceeds to analyse these challenges through expert opinions gathered via a questionnaire. By highlighting the key issues and potential solutions, this research aims to fill the gap in the literature and provide a comprehensive framework for improving cross-border railway maintenance coordination.

This thesis also develops a digital and cooperative framework (DCF) to enhance cross-border maintenance decision support within the railway system of the European Union (EU), addressing the cross-border challenges holistically. It emphasizes the introduction of digital twin technology to meet the urgent need for infrastructure digitalization and coordination between the authorities and infrastructure managers in different countries, enabling the cooperative optimisation of the capacity of railway networks. The DCF consists of three components: the establishment of a European Railway Entity, the creation of a European Railway Forum, and the implementation of digital twin (DT) technology. The European Railway Entity aims to centralize coordination, streamline decision-making, and enforce standardized regulations across member states. The European Railway Forum focuses on fostering collaboration and knowledge sharing among stakeholders, facilitating continuous improvement in maintenance practices. The DT technology offers real-time data visualization, predictive maintenance, and advanced analytics to optimize maintenance operations and enhance infrastructure reliability, enabling an evidence-based decision-making process.

Each of the three components of the DCF is evaluated for its potential benefits and implementation challenges. The European Railway Entity can enhance the coordination and standardization of maintenance operations but may face resistance due to its hierarchical structure and potential conflicts with infrastructure managers whose national interests might be compromised for the sake of overall system performance. The European Railway Forum is cost-effective and practical, promoting voluntary data sharing and continuous improvement; however, suggestions and implementation involve complex procedures that require trust and the management of key confidential information. DT offers the greatest potential for innovation and future-readiness, but it demands a significant initial investment, robust data security measures, and the design of complex mechanisms to enable coordination between different DT platforms and protocols across countries.

The proposed DCF is discussed in two case studies. The first case study consists of a generic case of a bridge for railways between two countries focusing on the implementation and assessing potential benefits and limitations of DT as a tool of the DCF. The second case study addresses the cross-border situation in the Netherlands, proposing an implementation plan and evaluating potential advantages and disadvantages. Additionally, the Emmerich–Oberhausen maintenance project on the cross-border section between the Netherlands and Germany is analysed. This case study highlights the benefits, and challenges of implementing the proposed DCF.

In conclusion, the proposed digital and cooperative framework, integrating the European Railway Entity, the European Railway Forum, and DT technology, aims to support maintenance decisions and address the current challenges in cross-border railway maintenance. This integrated framework seeks to improve operational efficiency, increase network reliability, and support the sustainability goals of the EU, ultimately making the railway system more attractive to users.

Smartphones have become widely accessible and advanced, pocket-sized computers capable of sensing, sending, and processing an ever-increasing variety of data. Modern phones are, for example, equipped with sensors such as accelerometers and GPS-sensors. Meanwhile, an increase in rail traffic leads to an increase in the need of having a reliable and comfortable railway system, an objective which can be achieved through additional monitoring. Through mobile crowdsensing, the smartphone data from passengers could potentially provide an additional and almost continuous information source on the railway system health state.

This thesis explores to what extent smartphone accelerometer measurements on board of in-service passenger trains can be used to monitor railway track quality. The thesis contains a literature review and a survey with European infrastructure managers. They identify that repeatability, as a result of operational variance, is a key research gap and challenge in the implementation of this technology. The thesis assesses accelerometer characteristics from a set of contemporary smartphones through laboratory tests. The tests uncover limitations and heterogeneity in the sensing capabilities such as sampling frequencies, low-pass filters and eigenfrequencies of the devices. The thesis also makes use of a case study, in which different smartphones are placed in various positions on different in-service passenger trains, running over the same tracks in the Netherlands. The vertical acceleration signals are analysed and compared. The vehicle speed and the position of the smartphone within the car body are found to significantly influence both the frequency content and vertical acceleration amplitudes measured. Finally, the different signals are related back to a Track Quality Index based on the standard deviation of the Longitudinal Level D1 track parameter. A degree of consistency in identifying locations with a low relative track quality is displayed, albeit with a significant degree of variance due to the previously mentioned factors.

An issue of utmost significance constitutes the maintenance of engineering systems exposed to corrosive environments, e.g. coastal and marine environments, highly acidic environments, etc. The most beneficial sequence of maintenance decisions, i.e. the one that corresponds to the minimum maintenance cost, can be sought as the solution to an optimization problem. Owing to the high complexity of this sequential decision optimization problem, traditional methods such as thresholdbased approaches, fail to arrive at an optimal strategy, while at the same time the commonly used offline knowledge about the environment can not capture efficiently the stochastic way in which an engineering system deteriorates. Over the last few years, Deep Reinforcement Learning (DRL) has been proven a promising tool to tackle such problems, being often limited though by the dimensionality curse and the implications caused by large state and action spaces, an issue which leads to simplifications like their discretization. Bayesian principles and model updating are the most widely used tools to model accurately systems with high uncertainty, by incorporating data acquired through monitoring devices and thus improving the knowledge about the stochastic system.

This research proposes an integrated framework that aims to determine an optimal sequence of maintenance decisions over the lifespan of deteriorating engineering systems, combining the aforementioned core concepts of Deep Reinforcement Learning (DRL) and Bayesian Model Updating (BMU). More specifically, it investigates different Deep Reinforcement Learning (DRL) algorithms, namely Double Deep Q-Network (DDQN), Advantage Actor Critic (A2C), and Proximal Policy Optimization (PPO), while the updating of the uncertain parameters is performed through sampling, i.e. No-U-Turn Sampler (NUTS). All these tools will be first applied to elementary problems for the sake of verification and validation, while the culmination of this research is the application of the framework on a more realistic and complicated, multi-component structure. The obtained results are compared with benchmark performances to properly showcase the efficiency and the weaknesses of the tool.

The Netherlands is a northwestern European country. The location makes the Netherlands an important entry for trade to the European continent resulting in high traffic intensities [1]. The Netherlands has one of the densest road networks in the world, having over 139.000km of roadways [2, 3]. The understanding of the link between the functional properties and the behavior of the road is crucial [4]. However, predicting such properties is still challenging, and expectations differ from the results found in the field [5, 4]. Data-driven approaches have been part of pavement engineering for decades and have shown to be a powerful tool for performance prediction [6]. This thesis aims to develop a machine learning framework for predicting stiffness, fatigue resistance, resistance to permanent deformation, and water sensitivity. Along with sensitivity analysis to understand the deviations from expectation. The objective is to develop a framework to understand the applicability of machine learning tools and the impacts of the composition parameters into the mix. The models developed applied three different machine learning tools for regression: Support Vector Machine, Random Forests, and Gradient Boosting. The models were compared to a statistical model to validate the work. The statistical approach was a Multiple Linear Regression. All developed models used the database from the NL-LAB project. The machine learning models were compared, and the best-performing proceeded to a sensitivity analysis. The sensitivity analysis used SHAP values, which derive from the Games Theory and have shown to be powerful tools for complex model interpretability [7]. The models had good accuracy prediction. For most properties, machine learning had a higher performance than the statistical model. Gradient Boosting performed the best from the machine learning tools and was selected to finalize the research. The sensitivity analysis had good results, confirming some of the expectations and setting a precedent for researching others.

Pavement structures are assessed based on its functional and structural capacity in order to evaluate the safety of its use. Structural pavement evaluation is often carried out by road authorities or consulting companies that use non-destructive testing (NDT) methods to assess the remaining life and structural capacity of a road structure at project or network level. In Western Europe and in most parts of the world, the most common equipment used as part of the NDT method is the use of the Falling Weight Deflectometer (FWD). The FWD is able to capture deflection values that can be used for the backcalculation of moduli and to determine remaining life of a structure by mechanistic-empirical equations.
In the Netherlands, like in many other countries, motorways have an increased number traffic flow based on high population densities accumulating in major cities. Due to this, pavement evaluation with stationary equipment such as the FWD can become relatively expensive due to the disruption of traffic. To solve this problem, research institutions and consulting companies have developed different versions of continuous evaluation equipment that are able to perform the structural analysis in a similar way to the FWD. In 2018, Dynatest® launched the Rapid Pavement Tester (RPT or RAPTOR) which aims to perform functional and structural evaluation of road networks at traffic speed. Continuous evaluation devices such as the RAPTOR are in constant development in order to achieve the quality and guaranty of use that equipment such as the FWD have in the pavement engineering industry. Currently this type of device uses technology that is less accurate than the sensors (Geophones) used in FWD equipment. This becomes a significant impediment in the evaluation of high stiffness pavement structures as the calculated deflections are in the lower end of the spectrum. This research aims to find a methodology that can be used in order to find limitation stiffness parameter values for which it is viable to use continuous evaluation equipment. Additionally, this research aims to find a method that can be used in the pavement engineering industry and research that is able to aid the data collection process of pavement layer information by means of machine learning tools. This was a problem faced during the elaboration of this research as the data collected to perform the analysis was incomplete in terms of layer thicknesses information. This process is carried by means of an artificial neural network (ANN) that is able to predict layer thicknesses and moduli based on deflection values and deflection parameters that obtained with the FWD.
The analysis of this research is carried by comparing deflection values in different road networks collected with the FWD and the RAPTOR under similar weather conditions, where the predicted layer information is used to assess the cut-off values and limitations that the current version of the RAPTOR has when compared to the FWD. From the presented results it was found that the asphalt layer modulus showed the highest correlation to the limitation values where the RAPTOR is able to present reliable results when compared to the FWD. Additionally, promising results were found in the use of ANN method to predict missing layer information which are assumed to improve with a specific build in the ANN architecture.

There is an urgent need for technical innovations in the construction industry to meet the European Parliament’s qualifications for reducing the energy footprint of buildings. This thesis is devoted to conducting research on control strategies that optimally manage passive energy sources, e.g., natural ventilation and solar irradiance, in combination with an active energy source. The main objective is to develop a control strategy that is both on-line applicable and maximizes the performance of the building energy management system in terms of passive fraction of energy while maintaining indoor thermal comfort. Energy-saving model predictive control (MPC) structures have been extensively researched in the literature on building energy management systems. These studies usually focus on the energy distribution in multi-zoned buildings and rarely consider optimal control of a single or multiple passive energy sources. In addition, most studied building models are based on general buildings and are established by means of simulation software tools. A more experimental study has not yet been conducted on optimal energy management systems for a building that is maximized in harvesting passive energy. This work investigates five MPC strategies as a way to optimize the operations of four solar blinds, a thermal chimney, and an active energy source. In these strategies, linear and nonlinear MPC are considered in the forms of centralized and hierarchical architectures. White-box modeling and linearization methods are adopted to develop the required linear and nonlinear building models. Thereafter, the proposed modeling methodology is validated by using experimental data. The hierarchical MPC architecture that considers a hybrid structure with a linear MPC agent for solar blind operations, a nonlinear MPC tracker for ventilation, a linear Kalman filter, and separated state-update loops appears to be the best-performing strategy. This control structure is also applied in a case study, in which it is tested on experimental data from the real-case office building, which is controlled by a rule-based control structure. The results show that the developed control structure is able to outperform the rule-based controller in terms of minimizing energy consumption and maintaining thermal comfort.

Bayesian system identification, including parameter estimation and model selection, is widely used to infer partially known, unobservable parameters of the models of physical systems when measurement data is available. A common assumption in the Bayesian system identification literature is that the discrepancy between model predictions and measurements can be described as independent, identically distributed realizations from a univariate Gaussian distribution. However, the decreasing cost of sensors and monitoring systems leads to more frequent structural measurements in close proximity to each other (e.g. fiber optics and strain gauges). In such cases, dependency in modeling uncertainty could be significant, both in space and time, and the assumption of uncorrelated Gaussian error may lead to inaccurate parameter estimation.

The aim of this thesis is to explore how Bayesian system identification can be feasibly performed using large datasets when spatial and/or temporal dependence might be present and to assess the impact of considering this dependence. A pool of models, each assuming a different correlation structure, is defined and Bayesian inference is performed. In particular, stress measurements obtained on a steel road bridge are used to update the parameters of the corresponding FE model and the parameters of the correlation structure. The results are compared to a reference model where only measurements of the response peaks are used under the assumption of independence. Nested sampling is utilized to compute the evidence under each model and Bayesian model selection is applied. The question of efficiently performing system identification for large datasets (N > 10² for temporal dependencies and N > 10³ for combined spatial and temporal dependencies) is investigated, and a novel approach for efficiently calculating the exact log-likelihood is derived. An approximation based on the Fisher information matrix is used to efficiently calculate the information content of measurements.

It is found that the choice of correlation function can significantly affect the posterior distribution of the model prediction uncertainty. Additionally, it is shown that using large datasets and considering dependence makes it possible to perform system identification for a larger number of parameters compared to the reference model. The results of the case study indicate that using measurements from multiple sensors under combined spatial and temporal dependence and additive model prediction error yields reduced uncertainty in the posterior and up to 29% reduction of the posterior predictive credible interval range compared to the reference case. Furthermore, the efficiency of the proposed likelihood evaluation method is assessed. Using this method, exact calculation of the log-likelihood can be performed for >10⁶ points in under a second in the case of correlation in one dimension. For combined spatial and temporal correlation it is shown to be approximately 900 times faster than naive evaluation for a 64 by 64 grid of observations. The results of the case study indicate that the described approach can be feasibly applied to real-world structures and can potentially improve parameter estimation and reduce prediction uncertainty. These findings suggest that further research into the approach could yield improvements over current methods.

Soil is an important part of a railway track structure. It provides the foundation for the structure and distributes the imposed loads. Consequently, soil has an influence on the condition of a track structure and a low quality soil can ultimately cause failure of the track structure. To be able to include the condition of a soil in the assessment on track quality, the following research question is posed:
“What is the quantitative influence of soil on the track condition for the railway tracks between Utrecht and Den Bosch?”

To be able to answer the research question, first the influence of soil on track condition is distinguished. The most important failure types, progressive shear failure and excessive plastic deformation, are best characterised by the strength of the soil and the deformation of the tracks respectively. For these characteristics a method is developed, which assesses the track condition. For strength a bearing capacity approach is used and for deformations a beam-on-an-elastic-foundation model is used. The process of the method is illustrated by executing it for a specific trajectory, being the Utrecht-Den Bosch line in the Netherlands. To take into account the variability of different soils, the soil type and some specific soil parameters derived from Cone Penetration Tests are determined and applied in the calculations for the assessment.

From the results can be concluded that for the Utrecht-Den Bosch line, the track quality considering passenger trains will either deteriorate limitedly or not at all. This is the case for quality related to both strength and deformation. For cargo trains, the quality related to deformation is such that limited deterioration can be expected. For strength, however, a large number of sections has a condition varying between the failure level and the desired level, which means that significant deterioration is possible, though, there are sections with higher quality as well.

An increasing load on the Dutch rail network requires new solutions for railway monitoring as increasing wear and tighter schedules limit the possibilities of regular maintenance. UAV (Unmanned Aerial Vehicle) based photogrammetry is such a solution which limits the presence of people near the railway and train hindrance. Photogrammetric feature extraction methods struggle with extracting features from the rusty rail and reflective roll band. Therefore line features are used to aid reconstruction. Based on SfM (Structure-from-Motion) camera parameters, a line-based reconstruction is created of two data-sets. a A line detection algorithm is used to find line segments in the images. Found segments are matched between pairs of images by pixel window cross-correlation and geometric boundaries. Matched segments are then compared to find sets of three matching images. Reconstruction is done by minimizing line reprojection error in an iterative least squares solution. Results show promise for manually matched segments, however unfavourable automatic matching results prevent large scale reconstruction. A case study reveals matching struggles with camera rotation between images, matching problem mitigation leaves much room for improvement.

Autonomous Underwater Vehicles (AUVs) are unmanned vehicles that give the opportunity to carry out lengthy and dangerous tasks autonomously. This is particularly useful for survey tasks, where the objective is to search the seafloor for objects. In this thesis work a planning system is developed that can plan a path for survey tasks, while considering environmental challenges such as communication limitations and location uncertainty. To compensate for location uncertainty, the planning system requires a higher level of abstraction compared to conventional path planning algorithms. For that reason, the planning problem is modelled in the PDDL, creating a powerful and flexible planning system which deals with the complex survey problem. Besides that, some additional planners are added to support the PDDL-planner and provide suitable plans for the AUV to carry out. The resulting plans are evaluated by simulation, showing that the planning system can successfully survey different scenarios. Besides that, the PDDL model is validated by means of the Event-B formal method, in order to obtain mathematical proofs of the validity of the planning model. The results are a step forward in achieving full autonomy of the AUVs. Besides that, a demonstration of the applicability of PDDL in real-world problems is given.

In this thesis, the applicability of tram-train in an urban region is studied. A three-step top-down method is used, to narrow down the solution space and determine whether a tram-train solution can be applied, is suitable and technically feasible.

Learning capabilities are a key requisite for an autonomous agent operating in dynamically changing and complex environments, where pre-programming is not anymore possible. Furthermore, it is essential to guarantee that the learning agent will act safely by considering its stability properties. In this thesis, novel conditions are proposed, aiming to examine stability of the learned dynamics for two important model classes; namely Rectified Linear Unit (ReLU) Deep Neural Networks (DNNs) and Locally Weighted Learning (LWL). For the former method, a theoretical and computational framework is developed by establishing an equivalence between ReLU DNN models and Piecewise Affine (PWA) systems. This allows to leverage well-known tools of PWA system analysis, and consequently compute, characterize equilibria and determine their region of attraction for ReLU DNNs. Due to their increased complexity, a structured search for appropriate stability conditions was performed for LWL methods until the optimal trade-off between conservativeness and computational efficiency was obtained. These stability conditions are given as Linear Matrix Inequality (LMI) problems and they consist the first stability results in literature for these two model classes. Their efficacy is assessed in numerical and real-world dynamical systems and it is shown that the proposed LMIs are not unreasonably conservative, as they can evaluate accurately the stability properties of these two representations. Finally, this work demonstrates how to formulate appropriate stability conditions for learning methods in a principled manner.

In this thesis, a method to monitor the health condition of railway crossings based on vibration data recorded by the accelerometers installed on the crossing is proposed. Due to various types of trains and other exogenous factors, responses obtained from accelerometers vary, even when the crossing has the same state condition. As a consequence, the degradation level is difficult to estimate. The method proposed in this paper uses Convolutional Neural Networks (CNN) algorithms for estimating the degradation level of the crossing and suppressing deviations caused by various inputs so that defects on the crossing can be detected in the early stage. The method is evaluated using a real-life dataset from a crossing located in Amsterdam. Different architectures are proposed and tested. With one of the architectures (ConvNet), the degradation level of the crossing can be estimated with minimum deviations (2.03 \% MSE). Other architectures performed in the range of 3.49 to 3.05 \% MSE. With the proposed methodology, defects on the crossing can be detected two months before the defects are spotted during the visual inspection.