Prediction NOX Emissions for High Speed DI Diesel Engine Based on PSOBP

SHENG-LAN TAN, RONG YANG, XIAO-HU YANG, JUN-MING HUANG, ZHI-HAO FAN

Abstract


In this paper, BP neural network (BPNN) is studied to model and predict NOX emission of direct injection diesel engine. The model selects four parameters as input, namely rotation speed, load, exhaust temperature and fuel-air ratio. Through testing, it is concluded that the prediction performance of BPNN model will be greatly affected by the initial weight and threshold, so that the prediction accuracy of the model is not high. In order to reduce the influence of initial weight and threshold on the prediction performance of BPNN model, this paper adopts Particle Swarm Optimization (PSO) algorithm to optimize the initial weight and threshold of BPNN, and establishes the corresponding prediction model of NOX emission of diesel engine. The results show that the prediction model of BPNN optimized by PSO algorithm can effectively reduce the influence of initial weight and threshold on BPNN and make the prediction results of the model more reliable. In particular, when PSO adopts non-linear dynamic weight strategy and synchronous learning factor strategy, the prediction performance of BPNN model is more provided with reliability.

Keywords


Diesel engine, NOX emissions, Prediction model, PSO, BP.Text


DOI
10.12783/dteees/peems2019/33926

Full Text:

PDF

Refbacks

  • There are currently no refbacks.