“As one of the important basic equipments of smart grid, smart meter carries the tasks of collection, metering and transmission of raw electric energy data. In actual operational process, the network of provincial power companies of mainland China reflects the equipment failure is mostly related to component failure. This project establishes a power metering equipment failure-component failure correlation model to quickly locate the failed components, through the equipment failure phenomenon to quickly locate a number of components with the highest correlation, which is of great significance for the dismantling and sorting work of the equipment. On the other hand, as different testing organisations and manufacturers will test the key components in power metering equipment, resulting in multi-dimensional data variability in the testing of similar indexes, the data obtained cannot be used interchangeably, causing increased workload and waste of testing resources. This project takes electrolytic capacitors and batteries in power metering equipment as an example, establishes mathematical models through the average fitting method and the neural network fitting method, analyses the multidimensional data variability of index testing to achieve the normalization of multidimensional similar data, and makes suggestions for the standardisation of testing methods based on the project.”
Category: Postgraduate
The output power of large-scale power electronic systems, such microgrids and electric vehicles, has now expanded from the kW level to the MW level due to the growing industrialization of the world. The loss of a high output power system can be decreased by raising the bus voltage of the power electronics system. This project investigates the state-space modelling and control system of a three-level interleaved boost converter with coupled Inductors in an effort to fulfill both high output voltage and high output power requirements and to maximize the power density of the converter. In this project, a simplified modelling approach is proposed for the circuit, along with designing a dual closed loop controller, and the modelling results are verified by simulation based on the PLECS platform. Finally, a prototype of 1MW three-level Boost converter is built to verify the effectiveness of the simplified modelling approach and the dual closed-loop control method with open-loop and closed-loop experiments. The dual-loop PI control system can ensure its stable operation, and the simplified modeling approach suggested in this project can decrease the laborious state-space derivation calculation process, providing a research foundation for the use of increased output power.
Emerging DC loads such as datacenters and Electric Vehicle (EV) charging will account for nearly 70% electricity consumption in the UK and China by 2050. For datacenters, more than 7% efficiency increase and 6% equipment cost are achieved via DC systems. Due to the exponential demand for Machine Learning (ML), the next generation datacenters will largely adopt GPUs. However, high power consumption for GPUs has placed a great challenge for power supplies because of large current (＞200 A) at low voltage. Multiple power converters have to be used for and located in dedicated compartments in the server cubicle, which will introduce lengthy cabling and complex system causing significant losses. This project aims to develop a new topology called Mixed Analogue and Digital (MAD) circuit, which will be adopted for 48V to 12V and 48V to 4V conversions by using Gallium Nitride (GaN) devices. The final applicable prototype is to achieve high energy efficiency (more than 95%), high power density (more than 3000 W/in3) and high reliability of DC networks for datacenters with lower costs, which aligns with the sustainable energy theme.
This study examines the performance of a 13kW high-power inductive power transfer system utilising a hybrid core structure with novel nanocrystalline ribbon cores and nanocrystalline flake ribbons. This structure incorporates innovative nanocrystalline ribbon cores and nanocrystalline flake ribbons. Traditional laminated nanocrystalline ribbon cores suffer extensive edge losses due to the high flux density focused on the edge and considerable eddy current losses in the side wall. These factors can potentially lead to partial thermal overload. To address this problem, the proposed solution involves using nanocrystalline flake ribbons as a shielding material along the edge while keeping the nanocrystalline ribbon core as the primary magnetic connector. The performance of the hybrid core structure is assessed across varying power levels, reaching up to 13.8 kW. Experimental outcomes show an almost 2% increase in peak efficiency compared to the ferrite DMR44 and a 1% increase over standalone nanocrystalline ribbon cores, pushing the peak DC-DC efficiency to over 96%. Furthermore, at a 6.6 kW output power, the temperature increase following a 2-hour operation is significantly lowered to a maximum temperature of 76.5 °C with the suggested hybrid core, as opposed to 96.4 °C with the ferrite shield and 110.6 °C without any edge shield. The design emphasizes using nanocrystalline materials in inductive power transfer systems to enhance both efficiency and thermal performance.
This project utilizes blockchain technology to create a tool that could monitor, verify and report carbon emissions for business entities. After the Paris agreement and COP26, leaders had agreed and set ambitious carbon neutrality targets, which require close monitoring, data disclosure, and regular reviews. However, it is not an easy task for the industry as there are several challenges: poor data quality & management, different interpretations in carbon auditing guidelines, and a lack of resources for verification and validation. Thus, our team has researched and developed a prototype that can keep tracking, verifying, and reporting carbon emissions in particularly for real-estate sector seamlessly with blockchain and deep learning technology. The result shows that the tool could work well and greatly enhance efficiency, accuracy, and reliability with a fully digitalized solution. The project has been recognised in international competitions, including Innoverse in US and International Invention Innovation Competition in Canada.
As DC loads like electric vehicle chargers and DC renewable power sources like solar PV panels become more prevalent in distribution networks, the need for converters to convert DC power from the DC source to AC and vice versa for the DC loads is increasing. However, these converters are costly and result in power loss during the conversion process. To minimize the power loss and the converter cost, DC power sources should directly supply DC to DC loads, while AC power sources such as transformers powered by utility grids should supply AC directly to AC loads. This project develops new devices to facilitate this direct supply of AC and DC, reducing the need for costly converters and minimizing power loss. This innovative distribution method can speed up global decarbonization efforts by promoting the use of electric vehicle chargers and DC renewable power sources.