Application of PSO-BP Neural Network in NTP for Diesel Engine Denitration Experiment

Abstract

Artificial neural network is an empirical model that imitates the function of biological neural network. BP neural network is a multi-layer money-box feedback neural system. PSO is an evolutionary computing technology, adopting PSO-BP neural network to achieve faster convergence and more accurate prediction. The removal of NOx from diesel engine exhaust as the main power unit of ships is a hot research topic at present. In this paper, a simulated diesel engine distribution system is established, and the denitrification of simulated exhaust gas is studied by non-thermal plasma (NTP) generated by corona discharge. The NOx concentration at the exit of NTP system was predicted by PSO-BP neural network, and the NOX concentration at the exit of NTP system was compared with the NOX concentration at the exit of NTP system by the analysis, which verified the accuracy of the experiment.