Offshore floating photovoltaics (OFPVs) show great potential for future renewable energy generation by complementing existing wind energy. To maximize power output, it is essential to accurately predict the response of these structures under wave-induced forces.

Although finite-element fluid-structure interaction (FE-FSI) models can simulate the response of very large floating structures (VLFSs) with high accuracy, they are computationally expensive and time-intensive. To address this challenge, this thesis develops and optimizes three surrogate deep learning models: a Convolutional Neural Network (CNN), a Long Short-Term Memory (LSTM) network, and a hybrid CNN-LSTM model. An iterative trial-and-error approach was employed to fine-tune the hyperparameters and improve model performance.

The CNN model accurately captured spatial features but struggled with generalization across varying sea states. The LSTM model captured temporal patterns well, but had lower accuracy and longer training times. The hybrid CNN-LSTM model combined the strengths of both approaches, achieving a low weighted absolute percentage error (WAPE) of 0.98 %. All models performed well under typical wave conditions but exhibited reduced accuracy in extreme scenarios, particularly for low-amplitude tilts. A sensitivity analysis indicated that these errors were likely caused by insufficient representation of extreme sea states in the training data.

The hybrid CNN-LSTM model was selected as the model that most accurately predicts the response of VLFSs due to its balanced performance, minimal prediction time, and generalization capability across typical operating conditions. While the model shows high reliability for typical sea states, its performance declines for extreme conditions, such as extremely large waves or very small tilts. Despite reduced reliability in extreme conditions, the model provides a promising tool for real-time OFPV response prediction and preliminary studies.

The Overhead Catenary Line System (OCLS) is crucial for continuous electrical supply to trains in electrified railway networks. This article outlines the research conducted on the exploration of the sustainable design space of the Overhead Contact Line support structure on the main Dutch rail network. The research began with a comprehensive review of the literature on OCLS support structures, covering publications from 2013 to 2022. A multi-objective optimisation model is developed to explore sustainable design possibilities. Further result analyses elucidate the defining parameters and variables of the support structure design. The findings presented in this
research serve as a basis for more in-depth research on the discussion surrounding existing regulations and specifications and sustainability within the rail branch.

Through improved plastic waste separation EU recycling goals can be reached and environmental economic advantages can be unlocked. To help with this endeavour, this research explores dynamic separation efficiency determination and waste stream characterization through near infrared (NIR) separation unit and belt weigher data in a plastic waste sorting plant in Scandinavia. For the showcasing of these concepts, the goal was to predict the product quality of the high-quality (HQ) agglomeration line, using data of the first NIR-scanner in the agglomeration line as prediction input. In the agglomeration line two NIR-scanners are connected in series to ensure high-quality separation of the material. Through the NIR-scanners, material specific area flow data is available and through the belt weighers mass flow input to each NIR-scanner is provided. Quality criteria are weight shares of PO (target material) and PVC (main contaminant). Difficulties arose, as the material-specific mass flow is needed for quality determination but only the total mass flow is provided. This was addressed by modelling area densities using a linear regression model, with belt weigher and NIR-scanner data as input. Using the calculated area densities, the material-specific mass flow was determined. For validation, summed material flows were compared with belt weigher data, yielding a mean absolute error (MAE) of 141 [kg/h] and a mean relative error (MRE) of 3%. The separation efficiency was determined through an XGBoost model, to predict material-specific area flow of the second NIR-scanner. Results were a MAE of 50.02 [m²/h] and an MRE of 1.1% for the total area flow. The final separation step could not be validated, as no NIR-analyser is present behind the second NIR-scanner. Therefore, separation efficiencies from the previous separator were transferred. Joining all three concepts the weight share of PO and PVC could be predicted with a MAE of 0.36% and 0.007%. For the joint outcome, greater uncertainty contribution was ascertained for the area densities compared to the XGBoost application. Future research is recommended for separation efficiency determination of the last separation step and for improved modelling of the area densities.

This thesis aims to investigate the feasibility of developing a successful unsupervised Structural Health Monitoring (SHM) methodology to detect damage in structures, specifically bridges. Detecting damage, especially in its earliest stages, is challenging, thus prompting the need for robust and effective methods. The success of such a methodology could lead to timely inspections and interventions, resulting in significant economic benefits and preventing further damage, including potential failure of the structure in use.

The approach involves a literature review to establish relevant background knowledge and useful concepts. From this, a methodology is developed utilizing unsupervised machine learning, specifically Gaussian Mixture Models (GMM), to identify abnormal behavior indicative of structural damage.

A Finite Element Method (FEM) model of a simple bridge is created and monitored over a three-year period, serving as a testing ground for the methodology and a primary source for data generation. Temperature data and its effects on the natural frequencies of the bridge model are used to establish a baseline for normal or healthy behavior. Synthetic damage, such as settlement and stiffness reduction, is then introduced to the model to create anomalies or abnormal behavior. The developed methodology is tested using three case studies, each with varying types of synthetic damage. By using both the healthy and unhealthy data generated from the model, the healthy behavior of the bridge is captured using GMM. The model then progressively incorporates unhealthy data into the proposed anomaly detection algorithm. The algorithm evaluates the likelihood of each incoming data point of belonging within the healthy distribution, resulting in data points being classified as either healthy or flagged as abnormal.

The case studies presented in this research underscore the efficacy of the proposed anomaly detection approach. In scenarios involving sudden or abrupt damage, the algorithm swiftly and accurately labels abnormal points. For gradual damage scenarios, such as settlement, the algorithm consistently identifies abnormal points, with the rate of abnormal point detection accelerating over time. This detection rate is contrasted with the rate of erroneous abnormal point labeling when processing an exclusively healthy data set through the anomaly detection algorithm. This comparison reveals a higher rate of abnormal point identification when actual damage is present, affirming the effectiveness of the unsupervised SHM methodology in pinpointing abnormal behavior within the modeled bridge structure.

The railway transportation system is a critical infrastructure that supports the functionality of a country through ways like the transportation of people from point to point to carry out their daily activities i.e., work, study, and leisure; the timely transportation of freight within a country or cross-borders, contributing to economic development, etc., thus it has to be effective in its operation and service provision. Normal operation can get affected when railway systems such as trains, signalling systems, etc. fail to operate, railway infrastructures are damaged, occurrences of collisions or derailment. These events result in inconveniences caused to railway users, loss in service confidence and decline in revenue. Hence, railway organisations continuously put in effort to reduce the occurrence of railway disruptions that are within their means to prevent. However, there have been reports on railway disruptions caused by events that are uncontrollable within the means of the organisations. The railway transportation system faces continuous ‘threats’ despite the continuous effort put in by railway organisations to provide mitigation measures. This thesis has classified these ‘threats’ as expected and unexpected. Expected ‘threats’ refer to events whereby their potential of disrupting the operation of the railway transportation system is anticipated, thus mitigation measures can be implemented in advance so as to eliminate or prevent the event from happening. On the other hand, unexpected ‘threats’ refer to events whereby the occurrences cannot be controlled and sometimes it can be unpredictable. The objective of this thesis is to study how the resilience of a railway organisation, in view of the occurrences of 2 unexpected external events (i.e., climate change and cyber-attack), can be assessed by incorporating the elements of the 4R Resilience Framework to its operation and management. The assessment work is carried out through a semi-qualitative tool created by this thesis which adopts a systematic approach by making use of a 5-stages cycle framework.