Holistic Condition Monitoring Methodology for Power System Assets using Machine Learning

Applications will remain open until a suitable candidate has been identified. Early applications are encouraged.

Fulfilling the complex and numerous requirements of modern power systems is becoming increasingly more challenging, one of the primary contributors being the ageing infrastructure. This not only impacts on reliability but also creates a major dilemma for system operators who need to choose between asset life extension, and replacement and modernisation. Testing, condition monitoring and diagnostics ensure that plant and equipment meet design and operational specifications, and allow operators to manage their assets timely and effectively. A variety of condition monitoring techniques are available for high voltage equipment such as partial discharge detection, vibration measurements, and thermal imaging. Each of these techniques has its benefits and drawbacks which in many cases can affect their ability to provide a holistic and accurate assessment of an asset’s condition if used individually. Moreover, their effective utilisation requires continuous collection and analysis of large volumes of data by engineers with specialist knowledge, often using proprietary instruments and software, something that can make their adoption and implementation challenging.

The research will aim to develop a holistic condition monitoring solution for power system apparatus, combining aspects of electrical and non-electrical techniques, able to provide automated condition diagnosis using artificial intelligence. It will initially involve the acquisition and analysis of data from small scale experiments in the laboratory to identify potential correlation between the different monitoring techniques as well as the capabilities of sensors and signal processing algorithms. The experience gained from the aforementioned investigation will form the basis for constructing the layers of deep neural networks to automate the condition diagnosis and reporting using machine learning. The data collected will be used for training the developed models so they can successfully identify incipient faults and categorise the condition of equipment based on severity without human intervention, and ultimately provide prognosis of asset life expectancy.

Applicants must have (or about to obtain) at least a 2:1 degree (or international equivalent) in a relevant science or engineering discipline (e.g. Electrical/Electronic Engineering, Computer Science). Programming experience in Python or MATLAB will be considered a benefit.

Start date: 1^st October 2022

The successful applicant will join the growing Energy and Power Research Group which has an excellent track record in innovative experimental and theoretical studies of the physics of switching arcs and in the development of condition monitoring tools and methodologies. The Group has 6 academic members and approximately 20 PhD students. The high-power test laboratory, specifically built for switching arc research, is unique among all UK universities. There are also smaller high voltage test facilities for partial discharge testing and a suite of diagnostic and measurement systems including high speed cameras, mass spectrometer, pressure/voltage/current sensors, and high-speed data acquisition systems.

For any enquiries please contact Dr Christos Zachariades on: C.Zachariades@liverpool.ac.uk

To apply for this opportunity please visit: https://www.liverpool.ac.uk/study/postgraduate-research/how-to-apply/ and click on the 'Ready to apply? Apply online' button, to start your application.