Abdelrahman completed his MEng (Hons) in Mechanical Engineering with Honours in 2019 at City, University of London followed by an MSc in Racing Engine Systems at Oxford Brookes University in 2020. During his time of studies, Abdu was highly involved in IMechE Formula Student, being an active member in City Racing's ICE Team and Oxford Brookes Racing's Electric Team. Starting the second year of his undergraduate degree as a design co-ordinator, he took on positions of Powertrain Lead and Team Lead in his final years at City. In addition to timekeeping, procurement, team integration and ensuring deliverables are met, part of his responsibilities included design, optimization of Intake and Exhaust Systems and Integration of a small-scale Electric Supercharger. At Oxford Brookes Racing, his responsibilities included design, development and integration of OBR20's epicyclic gearbox. He spent his summers working alongside industry engineers at Mercedes-Benz, Daytona Motorsport and Honda Motorcycles where he gained invaluable experience in automotive engine design, manufacturing, and maintenance. His interest in Digital Systems, Control and Optimization developed while working on his MSc thesis at Brookes, investigating Machine Learning Algorithms to Predict and Reduce Emissions in a GDI Engine with Positive Valve Overlap for Racing Applications. Within AAPS-CDT, Abdu sees a valuable opportunity to play a major role in tackling the rising sustainability issues and help provide a cleaner future for successive generations.
Abdelrahman's PhD will address the shortcomings associated with the conventional map-based controller design and calibration practices used in powertrain development. It will provide novel, futuristic, non-linear physical causality predictive modelling and experimental approaches for system identification to conclude a real-time capable control system.
The context of research
In machine learning, regression is often considered a black box methodology used commonly for identification of suitable functions from a hypothesis. Its primary aim is to estimate a predicted function which leads to minimal expected error on future data, and not necessarily to benefit the overall understanding of the derived output-input relationship governed by the physical interpretation of the system. Consequently, such modelling applications are often highly constrained, allowing training solely for thinly dispersed linear models. To this day, the task of forming intelligent algorithms capable of nonlinear interpretations through physical system observation has received very little consideration, it nevertheless forms the foundation of the field of system identification. Typically, approaches for nonlinear system identification include autoregressively modelling time evolution or utilising Volterra series multidimensional convolution integrals. This project will address the shortcomings associated with the conventional map-based controller design and calibration practices used in powertrain development. It will provide novel, futuristic, non-linear physical causality predictive modelling and experimental approaches for system identification to conclude a real-time capable control system through supervised neural network-based machine learning algorithms for system optimization. The project will be undertaken in collaboration with Koenigsegg Automotive AB and Freevalve AB on the novel cam-less engine technology, Freevalve, enabling major efficiency and power improvements for future powertrains. It will enable the full utilization of Freevalve’s potential and the reduction of harmful gaseous and particulate matter emissions, putting the technology in a market-leading position ready for large-scale implementation.
Its aims and objectivesThe aim of this research is to utilize model-based machine learning algorithms to optimize the real-time multi-variable transient control strategy of a Freevalve-equipped Koenigsegg engine. The project will combine state-of-the-art theoretical modelling approaches, classification and regression predictive algorithms calibrated with dynamic experimental data, to understand and characterise a Freevalve-equipped engine’s cycle-by-cycle behaviour under transient operating conditions to better inform controller design for the purpose of training a deep neural network to produce a real-time capable control model. To achieve the aim of this project, the following objectives have been tentatively drafted:
To review existing data and models of the Freevalve engine to support calibration.
To use existing engine calibration test data, Bath know-how about high output engine combustion, and literature findings to improve and correlate the current combustion model.
To utilise classification and regression machine learning predictive algorithms to assist in physical subsystem (turbocharging, fuel injection, heat transfer, etc.) identification for high fidelity engine performance model.
To build toxic emissions neural network models using existing test data for transient engine emissions prediction.
The validated and calibrated GT model as a result of (b), (c), (d) will serve as a virtual testbed for engine-focused controller optimization work.
To build statistical models to develop optimal control strategies for the aforementioned physical subsystems (Freevalve, injection, EGR, turbocharger, etc)
To validate the control strategies in Hardware in the Loop (HiL) testbed and deploy on vehicle with appropriate real-time hardware and software for clock speed control.
Its potential applications and benefits
Academic: The outcomes of this research will provide engine researchers and engineers with novel AI-based modelling approaches and experimental methods for real-time engine and controller optimization. Corresponding to forthcoming 4th industrial revolution, mainly focusing on increased connectivity, automation, machine learning, and real-time data processing, groups working on digital systems, optimization, and integration will highly benefit from the analysis and novel interpretation of data outlined as one of the core research aims of this project. Furthermore, groups working on improving cold-start and warm-up behaviours as well as fuel efficiency and lean-burn technologies would have particular interest in the generated modelling and experimental techniques of this project for providing a refined and simplistic methodology for optimization.
Industrial: The outcomes of this project will directly impact the mass roll-out of Freevalve technology to mainstream manufacturers and reduce time and cost outlays currently associated with a largely experimental calibration process. It will serve major assistance to distinguish Freevalve cam-less engine technology as a one-ff product capable of real-time prediction and optimization through offline training. A further prospective iteration once this project is completed could incorporate Freevalve as an online-training capable technology, gathering data from infrastructure and other connected vehicles to perform live adaptation and optimization tasks using cloud computing.
© Copyright 2024 AAPS CDT, Centre for Doctoral Training in Advanced Automotive Propulsion Systems at the University of Bath
