http://theses.ncl.ac.uk/jspui/handle/10443/90 Sun, 10 May 2026 04:03:41 GMT 2026-05-10T04:03:41Z http://theses.ncl.ac.uk/jspui/handle/10443/6758 Title: Machine learning applications for building energy analytics Authors: Khalil, Mohamad Abstract: As smart meters in buildings continue to be implemented worldwide, an unprecedented volume of time series data sets related to building energy consumption and performance have been collected. However, traditional analytics models, and physics based approaches used in the building sector face challenges in effectively handling and managing time series data from smart meters. This is mainly because of its inherent variation, rapid velocity, and the need for advanced data preprocessing methods to handle it. This thesis proposes a range of data-driven frameworks and models to address several key challenges in the domain of building energy consumption and performance. The primary focus revolves around three key components: 1) Predicting the presence of occupants in buildings through the application of pre trained data-driven models, this thesis proposes a novel transfer learning framework. This framework leverages past knowledge from similar domains, enabling efficient adaptation to new occupancy prediction tasks and boosting accuracy, especially in scenarios with scarce training data. 2) Examining the effectiveness of employing global forecasting models under the context of building energy consumption, rather than relying on a singular forecasting model. Unlike single forecasting methods, the proposed global forecasting models can simultaneously learn from multiple time series associated with building energy consumption. This helps uncover hidden connections in smart metering data, improving transfer performance and knowledge across various forecasting tasks. 3) Unravelling the underlying properties of energy consumption through the analysis of time series features, this thesis presents a forecastability framework tailored to meet this specific need. The framework relies on a feature matrix extracted through interpretable time series feature extraction techniques. Through a supervised learning, the framework is trained to establish a mapping between the extracted features and the target label of interest. This enables the prediction of how forecastable a given time series is within the context of energy consumption. Keywords: Machine Learning, Forecasting Building Energy Consumption, Smart metering. Description: Ph. D. Thesis. Wed, 01 Jan 2025 00:00:00 GMT http://theses.ncl.ac.uk/jspui/handle/10443/6758 2025-01-01T00:00:00Z http://theses.ncl.ac.uk/jspui/handle/10443/5121 Title: Software package applications for designing rail freight interchanges Authors: David, Raphael Kling Abstract: Rail freight transport has a crucial role to play in the economy, delivering significant reductions in logistics costs, pollution, and congestion. Typically, the conventional architecture and layout of the rail freight interchange constrain the capacity and performance of the whole railway system. A well-designed rail freight interchange can enhance the system performance by maximizing vehicle usage and minimizing last mile distribution cost. Therefore, the study of rail freight interchange operation is considered crucial to understand how to increase and improve the attractiveness for rail freight transport. This thesis uses game engines to develop software packages that are used for the design of new rail freight interchanges, considering multistakeholder decisions drivers. A novel and modular approach has been applied with the purpose of developing and deploying simulation tools that can be used by multiple stakeholders to: -Understand the impact of multiple-criteria decision analysis on rail freight interchange layouts; -Use a genetic algorithm to identify the most suitable components of the future interchange to be designed, considering the multi-stakeholders’ priorities; - Quickly enable the design of a wide variety of rail freight interchanges from the information selected by a decision maker in a computer-based userfriendly interface. This research has proposed a framework for software development. Three case studies are used to illustrate adaptability of a number of applications for different scenarios. The findings of the research contribute to a better understanding of the impacts of the multiple stakeholder’s decisions on rail freight interchange designs. Key words: Rail Freight Interchanges, Multi stakeholders decision, genetic algorithm Description: Ph.D. Thesis Wed, 01 Jan 2020 00:00:00 GMT http://theses.ncl.ac.uk/jspui/handle/10443/5121 2020-01-01T00:00:00Z http://theses.ncl.ac.uk/jspui/handle/10443/4468 Title: The use of linear motor technology to increase capacity in conventional railway systems Authors: Powell, Jonathan Peter Abstract: Wheel/rail adhesion is an important constraint on the design and operation of conventional railways. The research question considered for this thesis is whether linear motor technology can improve the performance of railway systems by reducing the dependence of tractive and braking effort on the available wheel/rail adhesion. The two principal contributions of the research are an analysis of the influence of several different linear motor technologies on the capacity of conventional railways, and the development of a new design concept for train braking (named LEMUR – Linear Electromagnetic Machine Using Rails). Multi-train simulation of three different railway networks was used to investigate the capacity benefits and energy consumption of the LEMUR concept, along with four other existing or proposed implementations of linear induction motor technology with the running rail used as the secondary component of the motor. A model of each network was built using OpenTrack software, and Monte Carlo simulation with pseudorandom distributions of initial delays to train services was carried out to compare train movements under the influence of the delays typically encountered during day-to-day operation. An indication of the improvements in railway capacity possible with different linear motor technology options was then derived from these simulations. The results of the experiments indicate that the LEMUR concept provided the greatest increase in capacity and the lowest energy consumption of the five linear motor technology options tested. Although the limitations of the study do introduce some uncertainty into the precise values of capacity and energy consumption obtained, the experimental methods were considered sufficiently robust for this conclusion to remain valid. The most promising application in the study was suburban passenger services that are part of busy mixed-traffic networks. Here, the capacity benefits of the LEMUR concept appear to show sufficient promise to justify further development and application. Description: PhD Thesis Fri, 01 Jan 2016 00:00:00 GMT http://theses.ncl.ac.uk/jspui/handle/10443/4468 2016-01-01T00:00:00Z http://theses.ncl.ac.uk/jspui/handle/10443/4429 Title: Design of an intelligent embedded system for condition monitoring of an industrial robot Authors: Jaber, Alaa Abdulhady Abstract: Industrial robots have long been used in production systems in order to improve productivity, quality and safety in automated manufacturing processes. There are significant implications for operator safety in the event of a robot malfunction or failure, and an unforeseen robot stoppage, due to different reasons, has the potential to cause an interruption in the entire production line, resulting in economic and production losses. Condition monitoring (CM) is a type of maintenance inspection technique by which an operational asset is monitored and the data obtained is analysed to detect signs of degradation, diagnose the causes of faults and thus reduce maintenance costs. So, the main focus of this research is to design and develop an online, intelligent CM system based on wireless embedded technology to detect and diagnose the most common faults in the transmission systems (gears and bearings) of the industrial robot joints using vibration signal analysis. To this end an old, but operational, PUMA 560 robot was utilized to synthesize a number of different transmission faults in one of the joints (3 - elbow), such as backlash between the gear pair, gear tooth and bearing faults. A two-stage condition monitoring algorithm is proposed for robot health assessment, incorporating fault detection and fault diagnosis. Signal processing techniques play a significant role in building any condition monitoring system, in order to determine fault-symptom relationships, and detect abnormalities in robot health. Fault detection stage is based on time-domain signal analysis and a statistical control chart (SCC) technique. For accurate fault diagnosis in the second stage, a novel implementation of a time-frequency signal analysis technique based on the discrete wavelet transform (DWT) is adopted. In this technique, vibration signals are decomposed into eight levels of wavelet coefficients and statistical features, such as standard deviation, kurtosis and skewness, are obtained at each level and analysed to extract the most salient feature related to faults; the artificial neural network (ANN) is then used for fault classification. A data acquisition system based on National Instruments (NI) software and hardware was initially developed for preliminary robot vibration analysis and feature extraction. The transmission faults induced in the robot can change the captured vibration spectra, and the robot’s natural frequencies were established using experimental modal analysis, and also the fundamental fault frequencies for the gear transmission and bearings were obtained and utilized for preliminary robot condition monitoring. In addition to simulation of different levels of backlash fault, gear tooth and bearing faults which have not been previously investigated in industrial robots, with several levels of ii severity, were successfully simulated and detected in the robot’s joint transmission. The vibration features extracted, which are related to the robot healthy state and different fault types, using the data acquisition system were subsequently used in building the SCC and ANN, which were trained using part of the measured data set that represents the robot operating range. Another set of data, not used within the training stage, was then utilized for validation. The results indicate the successful detection and diagnosis of faults using the key extracted parameters. A wireless embedded system based on the ZigBee communication protocol was designed for the application of the proposed CM algorithm in real-time, using an Arduino DUE as the core of the wireless sensor unit attached on the robot arm. A Texas Instruments digital signal processor (TMS320C6713 DSK board) was used as the base station of the wireless system on which the robot’s fault diagnosis algorithm is run. To implement the two stages of the proposed CM algorithm on the designed embedded system, software based on the C programming language has been developed. To demonstrate the reliability of the designed wireless CM system, experimental validations were performed, and high reliability was shown in the detection and diagnosis of several seeded faults in the robot. Optimistically, the established wireless embedded system could be envisaged for fault detection and diagnostics on any type of rotating machine, with the monitoring system realized using vibration signal analysis. Furthermore, with some modifications to the system’s hardware and software, different CM techniques such as acoustic emission (AE) analysis or motor current signature analysis (MCSA), can be applied. Description: PhD Thesis Fri, 01 Jan 2016 00:00:00 GMT http://theses.ncl.ac.uk/jspui/handle/10443/4429 2016-01-01T00:00:00Z