Master thesis presentation Dinesh Krishnan & Niranjan Suresh, MPCAS

Abstract: We are living in a time where data is the new oil, where the industries are data-driven. This change was expedited by, the enormous amount of data that we are producing each and every second and the increase in computational power. At present the automotive sector is thriving in this data driven model by their ubiquitous need for common and industrial purposes and the various data they are collecting to improvise the sector as a whole. We can now see an automobile as not just a mechanical product but as a robot on wheels. Along with this data-driven model, the electrification of the automobiles has revolutionized the industry. In the electrification process the battery module is one of the key components that are powering the systems. So the battery modules must be used by the vehicle optimally to increase the life of the battery modules. To do this we must analyse the data from the battery modules for its efficient usage. But due to certain hardware issue or if the vehicle is out of range and it could not update the data we might loose data. This loss of data can obstruct the efficient usage of the data in machine learning models to optimize the system. \\

Though there are several methods to impute the missing data, for example there are statistical methods such as the Auto-regressive methods, they are limited by their time and the high cost of their computations. This thesis focuses on this problem and designing a neural network model for masked prediction of the Time series data. In this thesis, a Transformer Network is implemented for the masked prediction of the missing time series data. \\

In this thesis we have built the machine learning model from scratch after weighing several factors. The data on which the model is trained is generated by the vehicle and collected and then pre-processed to improve the quality of the data in accordance to the model of selection. The model developed here is a variation of the transformer model, this transformer is called the Time Series Transformer(TST) which predicts the missing values in the time series data. This model is then evaluated with suitable metrics in accordance to the model and the problem statement. The thesis aims to predict the missing values in-order to improve the quality of the data collected and its quality usage to improve the performance of the vehicle.

Supervisors: Deepak Guru Ganesan, Emil Johansson

Examiner: Mats Granath

Opponents: Apoorva Udayakumar, Aditya Padmanabhan Varma