Mathematician in Digital Services
Mathematician in Digital Services
Dr. Yanfeng Liang
What is your current role?
I am a Mathematician in the Digital Services Team. An understanding of mathematics and its application is vital across all sectors, particularly with the data-rich environments of industry today. My expertise in data science and mathematical modelling helps to bring a new dimension to our Digital Services Team.
What is your academic/industrial background?
I obtained a Bachelor’s degree in Applied Mathematics at the University of Strathclyde. I then completed a PhD, also at the University of Strathclyde, on mathematical modelling in infectious diseases. The focus of this doctoral project was to use mathematical modelling to look at the dynamical behaviour of various infectious diseases. The models developed were used to predict the spread of infection under the influence of different environmental and demographic factors, the effectiveness of preventative measures, and the potential for eradication.
After completing my PhD, I worked for three years on British Council-funded Postdoctoral research focusing on mathematical modelling on the spread of both the Zika virus and Dengue. I collaborated with researchers in Brazil on the Zika virus. On the Dengue project, I worked in collaboration with governments, academia and industry in Malaysia, analysing the effectiveness of a new control measure in eliminating the number of dengue cases in Malaysia.
I hold an honorary research associate position at the University of Strathclyde, and I am a Member of the Institute of Mathematics and its Application.
What are your main areas of expertise?
My own areas of expertise include:
What are your current key projects and who are your key clients?
Here at TÜV SÜD National Engineering Laboratory, I am currently working on a BEIS-supported project where sand erosion data is gathered from Coriolis flow meters from different manufacturers. Using modelling techniques, we are able to predict the remaining useful life of flow meters when exposed to certain hours of sand erosion. Using data-driven models, we investigate how each Coriolis meter performs differently, despite being exposed to the same hours of erosion, in addition to predicting which meter is less susceptible to sand erosion.
I am also working on a BEIS-supported project examining diagnostic data generated from four ultrasonic flow meters from different manufacturers, performing fault diagnosis, and predicting the cause of drifts seen in variables. Using statistical models and machine learning models, this enables us to better understand the performance of the ultrasonic meters and enables condition-based monitoring.
What most excites/interests you about your role?
With the sheer volumes of data available in industry today, modelling can be used to find correlations between data that otherwise would be missed. The ability to work with a variety of datasets coming from different flow meters and under different operating conditions means that as a result, different modelling techniques can be applied, and you obtain different, interesting and useful insights. The flexibility in conducting research and improving the performance of models, as well as extending knowledge and apply this in different areas such as life sciences; the ability to use results from models to make an impact in real- life scenarios is extremely satisfying.
What future trends do you see developing in your area of work?
As the world is moving towards digitalisation, where data is produced in large volumes and fast speeds, having a good system to store data is simply not enough. More and more, industry will need to use advanced statistical models, along with machine-learning models, to analyse their data, in order to unlock the potential and the benefits that lie within it.
Along with the advancement in data modelling in flow measurement, a more reliable approach in calibrating equipment, namely condition-based monitoring can be used, which in return will improve the industry’s efficiency by ensuring errors are identify and rectified quickly.