Vehicular CO2 Emission Prediction Using LSTM Network

Climate change has become one of the most pressing challenges confronting the global community today. Governments worldwide are now implementing measures to address this issue, recognizing that if left unchecked, it could become irreversible and result in catastrophic consequences, potentially including the extinction of humanity. Greenhouse gas concentration, particularly CO2, serves as a significant contributor to climate change. Vehicular emissions, a major source of CO2 emissions, play a substantial role in this regard. While specialized sensors can be utilized to monitor such emissions, their scalability and effectiveness are limited. To tackle this problem, the proposed solution offers an efficient, feasible, and scalable system for monitoring vehicular CO2 emissions. This solution employs a Long Short-Term Memory (LSTM) network and has been trained and evaluated using openly accessible data based upon On-Board Diagnostics II (OBD-II). A comparison between the model introduced in this study and an avant-garde solution has also been presented in this chapter. The suggested solution is cloud-deployable, with dongles based on IoT installed on vehicular end. These dongles gather sensor readings from vehicles and channel them to cloud, where the model runs and provides real-time predictions of the vehicle’s CO2 emissions.