Traditional trains, with steel rails and wheels, have been around for centuries. The very good energy efficiency of steel-wheeled trains is one of the reasons that they are one of the more promising transportation modes in the future, where energy conservation becomes more and more important. Furthermore, electric trains are able to efficiently use sustainable power. They usually power their on-board motors by a sliding contact, with an overhead catenary wire or a 'third' power rail. However, trains using a linear motor have also emerged. These have several advantages: the amount of moving parts is reduced and the amount of traction force available is independent of the slipperiness of the tracks. Some systems still use a 'third rail' for power transfer however, which is a sensitive moving part and limits the train speed. Furthermore, arcs can cause damage under reduced air pressure. This may be eliminated using a linear doubly-fed induction machine (DFIM), which may be able to simultaneously propel and provide power to the train. The latter can be used to charge an on-board battery, which can provide the energy to keep the velocity constant between the stations, where active tracks may be omitted to save costs.

In this thesis, the use of a DFIM for a train application has been reviewed. A linear DFIM has been designed in a previous work \cite{Becetti2021}, whose parameters are used in this thesis to simulate the concept. A controller is to be designed which allows the prescribed use of simultaneous propelling and charging. This thesis explores the requirements and possible controller candidates for the train application, which is followed by simulation and laboratory testing.

The application for a vehicle drive provides unique objectives and restrictions. One of the most restricting elements is that preferably no trackside (stator) quantities should be required for the control of the DFIM. After all, requiring trackside quantities would require a critical very dependable communication link between the track and vehicle. For this reason, estimation methods have been selected to control the vehicle with only vehicle-side measurements.

Three main categories of controllers have been identified: field-oriented control, direct torque control, and model predictive control. It has been found that each controller family has its advantages and disadvantages, and the best controller depends on the final system design of the train system. Field oriented controllers have the lowest current ripple, and can therefore be very predictable and efficient. Direct torque control is best used if it becomes very important to keep the final algorithm very simple. A model predictive controller excels if additional constraints are required, such as when using a multilevel inverter that requires balancing. This thesis provides a selection that can be used to make a decision in a later stage.

The operational principle of the train with the DFIM has been verified with a simplified laboratory test. A rotary DFIM has been powered with a fixed grid frequency-powered stator and an inverter-controlled rotor. Using a simplified control algorithm, the experiments have verified that the DFIM can be used to charge the vehicle at a high efficiency of up to 85\%, and that the DFIM can indeed be used to simultaneously charge the vehicle and provide a tractive force. Additionally, a Simulink-simulation is performed where the performance of the linear DFIM design is validated. It has been concluded that the DFIM is an attractive candidate for use with a linear motor-driven train since it can indeed fulfil the roles of both motor and charger simultaneously and efficiently. ... Read more

In this thesis, the implementation of a passive, chipless, frequency coded Radio-Frequency Identification (RFID) tag for bedload transport studies is proposed. The proposed tag will be deployed in the semi-arid Río Colorado river, Bolivia with the aim to develop quantitative sediment transport models that relate transport to grain size. The designed tag is an open-loop resonator with a fragment-loading structure, that has an op- timised configuration based on a Multiobjective Evolutionary Algorithm based on Decomposition combined with Enhanced Genetic Operators (MOEA/D-GO). The designed RFID tag can ideally reach a size of 4 by 4 millimetres with a maximum calculated reading range of 1.3 meters, and operates in the ultra wide band from 3 to 7 gigahertz. Numerous simulations on the tags were run to verify their properties. The tags proved to have a good directivity, quality factor and radio cross section on its resonant frequency. The tags could reach resonance frequencies as low as 2.9 gigahertz and quality factors as high as 130. The proof of concept on a Printed Circuit Board with an FR-4 substrate results in a tag of 6.4 by 3.4 millimetres. Unfortunately, these properties could not yet be verified by measuremen ... Read more