Professor Julian Dunne

This project is being funded by  the ESPRC and Jaguar Cars to create a robust real-time cylinder-pressure reconstruction method for gasoline engines. Direct cylinder pressure measurement is not generally  cost-effective or durable on production engines. A  successful outcome will enable IC engines in general to  be controlled more efficiently in a cost-effective way – thereby producing less CO2. 

The research is trying to find the best adaptive configuration of neural networks using simultaneous measurements of crank-kinematics and knock-sensor output. These networks are being tested on real engine data.