Project overview
This project applied extant machine learning algorithms, developed for large-scale image datasets, on a long (3.5 year) DTS dataset (along with all relevant cable and environmental data) from three interconnectors linking Jersey to France. The specific aims of this project were to:
1. undertake dimensionality reduction and simple compression techniques to reduce computation time without significant information loss.
2. apply standard signal processing techniques to filter out negligible fluctuations.
3. with this reduced and filtered dataset, transform the problem into one of pattern recognition to enable the application of a range of techniques including: Variational Autoencoders and WaveNet.
4. use these tools to identify both spatial and temporal signals within the data and relate to known electrical (load, cable electrical) and environmental (ambient temperature; sediment thermal property; depth of cover) drivers.
1. undertake dimensionality reduction and simple compression techniques to reduce computation time without significant information loss.
2. apply standard signal processing techniques to filter out negligible fluctuations.
3. with this reduced and filtered dataset, transform the problem into one of pattern recognition to enable the application of a range of techniques including: Variational Autoencoders and WaveNet.
4. use these tools to identify both spatial and temporal signals within the data and relate to known electrical (load, cable electrical) and environmental (ambient temperature; sediment thermal property; depth of cover) drivers.
