In this article, a model to aid the performance monitoring of operational diesel engines is developed. This pertains mainly to the industrial settings. Feed-forward and modular neural network-based models predict the the specific fuel consumption in the direct injection four-stroke diesel engine. The engine monitoring programs are limited by the fact that engine performance maps are difficult and time consuming to develop. The current models cannot develop specific engine characteristics. Here, a neural network-based predictive model is developed. The parameters like the engine speed and torque are required by it. The five known engine parameters viz. the rated power, rated and minimum specific fuel consumption bore and stroke should be supplied.
the operational and design parameters affecting engine fuel consumption are to be reviewed in the engine design. factors that scale the engine performance are used for analysis. Network architecture and learning rate parameters are optimized. A genetic algorithm-based global search coupled with a locally adaptive learning algorithm for weight optimization is used. The neural network test responses are validated and stratical verification of the Network training errors is also done.

Get the report here:
http://ir.dut.ac.za/bitstream/handle/103...sequence=1