Scenario-aware resource allocation for electric construction transport vehicles using multi-source fusion
Abstract
Electric power infrastructure construction sites require transport vehicles to operate under changing terrain, payload, road condition, energy price, and task demand. Conventional experience-based dispatch cannot promptly adjust vehicle scale, drive mode, or charging strategy, causing low utilization, high operating cost, and energy waste. This paper proposes a scenario-aware resource allocation method for electric construction transport vehicles. Multi-source features from construction tasks, roads, vehicle states, energy prices, weather, and charging availability are fused and processed by an attention-enhanced CNN-LSTM scenario recognizer. A rolling NSGA-II optimizer then selects vehicle deployment scale, drive mode, and energy replenishment plans under transport efficiency, cost, energy, and empty-mileage objectives. For performance evaluation, a reference-informed numerical case is constructed using settings and performance ranges summarized from recent IEEE studies on traffic prediction, EV charging, commercial electric fleet scheduling, and mobile charging task allocation. Under this evaluation setting, the framework achieves 95.7% scenario-recognition accuracy, 25.6% transport-efficiency improvement, 18.3% operating-cost reduction, 22.4% energy-consumption reduction, and 20.1% empty-mileage reduction. The proposed framework converts a project-level planning concept into a reproducible computational architecture for intelligent dispatch in power infrastructure construction.