Machine Learning-Based Method for Energy Economy Driver Assistance

Strict emission regulations have significantly driven researchers to enhance vehicles' energy efficiency, leading to advancements in automotive technology. To conduct tests in a controlled environment, driving simulators have become the optimal choice for their high repeatability, reduced development costs, and decreased overall workload associated with real-world testing. This paper proposes a novel method for real-time identification of energy-intensive driving behaviors. The method employs the Iterative Density-Based Spatial Clustering of Applications with Noise (I-DBSCAN) algorithm to classify driving styles based on key performance indicators such as energy efficiency and safety. To detect energy-efficient driving behaviors, the study uses a Bayesian optimization-based Long Short-Term Memory neural network (LSTM) and a Random Forest (RF) pattern recognition model. These methods are validated using SCANeR™ Studio in an urban driving environment. The research results demonstrate that the model excels in accurately identifying energy-efficient driving behaviors, with an F-score reaching up to 0.992, show-casing significant potential in promoting vehicle energy savings and sustainable driving practices, thereby paving the way for more sustainable transportation solutions.