Abstract
This paper proposes a time series model based on wavelet transform and long short-term memory (LSTM) network to forecast vehicle emission. It implements the semi-supervised collaborative training regression to compensate missing emissions data. The accumulated carbon monoxide (CO), hydrocarbons (HC), and nitric oxide (NO) concentrations emitted by vehicles in different lanes per hour were taken to quantitatively characterize the vehicle emissions. The original time series of vehicle emission data, which may be highly variable, is decomposed into several lowly variable sub-series by wavelet transform. For each sub-series, an LSTM time series model is proposed to forecast vehicle emissions. More specifically, the inputs of that LSTM model are the weather variables, the driving variables of the concerned vehicle and historical emissions records while its output is the predicted accumulated concentrations of CO, HC, and NO. The three types of predicted concentrations of all sub-series are summed up, respectively, and produce the desired prediction of the total emission of each type. The proposed model is verified through real data which was collected between May 2017 and December 2017 at the multi-lane monitoring station of Baimiao South Road, Daxing District, Beijing, China. It confirms that our model based on wavelet transform and LSTM can efficiently improve the correlation coefficient (R) and the index of agreement (IA) against conventional models, such as ARIMA and wavelet-ARIMA model.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.