A deep learning based encoder-decoder model for speed planning of autonomous electric truck platoons.

Area of Science:

• * Transportation Engineering
• * Artificial Intelligence
• * Sustainable Energy

Background:

• * Electric truck platooning is crucial for extending the range of electric vehicles in long-haul transport.
• * Optimizing platoon speed for energy efficiency is a significant challenge.
• * Data-driven solutions for truck platooning are limited, with first-principles approaches being difficult.

Purpose of the Study:

• * To develop a novel deep learning approach for optimizing autonomous electric truck platoon speed profiles.
• * To ensure platoon string stability while considering battery State of Charge (SOC), traffic, and road conditions.
• * To provide a framework for efficient long-haul electric truck route planning and operational decision-making.

Main Methods:

• * Utilized a sequence-to-sequence encoder-decoder deep learning model.
• * Trained the model on diverse highway drive cycles for long-haul applicability.
• * Performed hyperparameter tuning to select the most suitable model architecture.
• * Tested the framework by predicting drive cycles for various SOC profiles.

Main Results:

• * Demonstrated the model's capability to predict drive cycles and speed profiles for electric truck platoons.
• * Showcased the framework's effectiveness in considering multiple operational constraints.
• * Presented a case study on route prediction and its impact on policy decisions.

Conclusions:

• * The proposed deep learning framework enables efficient planning of feasible routes for electric trucks.
• * It effectively manages constraints including battery capacity, discharge rate, and charging infrastructure.
• * The approach ensures stable platooning and optimizes energy utilization for long-haul electric trucking.