Abstract:

Hybrid electric vehicles' emissions worsen during cold start, making energy management strategies a cost-effective solution. However, traditional cold-start emission models, based on statistics and experience, are challenging to predict accurately and require significant experimental cost and time. Accurate models are needed for developing energy management strategies for hybrid vehicles. The use of neural networks for cold start emission prediction of hybrid electric vehicles has been widely used because of its low cost and high prediction accuracy. In this paper, the genetic algorithm is used to optimize the weights and thresholds of the neural network, and a neural network cold-start emission prediction model for hybrid electric vehicles is built. The authors trained the neural network using the high temperature (40 °C), normal temperature (23 °C) and low temperature (−20 °C) cold start data of hybrid vehicles obtained in the World Light Vehicle Test Code (WLTC), studied the cold start prediction results under the different cold start temperature, studied the cold start prediction results under the different input parameters, studied the cold start prediction results of the neural network optimized by the genetic algorithm. Under different input parameters and cold start temperature, the GA-BP algorithm can accurately predict the emissions of carbon monoxide (CO), carbon dioxide (CO2), hydrocarbons (THC), nitrogen oxides (NOx) and particulate matter number (PN) under hybrid electric vehicles.