In-cylinder soot concentration measurement by Neural Network Two Colour technique (NNTC) on a GDI engine

In the present study, a new method named Neural Network Two Colour (NNTC) has been developed and applied to a turbocharged gasoline direct injection engine to directly analyze, within combustion chamber, most sooting conditions, by varying rail pressure and start of injection. The engine cylinder head was modified to create two sealed optical accesses for endoscope lighting and visualization, allowing image acquisition directly into combustion chamber. After an optical calibration of the system, several combustion images were acquired on 16 engine operating points with 8 levels of rail pressure and start of injection. The images were post-processed by the NNTC in order to calculate instantaneous soot production for each investigated operating point. The results demonstrated that the NNTC technique can be used as accurate method to detect soot production directly within combustion chamber. Moreover, by means of this method, it is possible to compute soot production in specific areas of combustion chamber.