LLM-ENHANCED REAL-TIME FREIGHT RATE PREDICTION AND DYNAMIC PRICING FRAMEWORK FOR INTELLIGENT TRUCKING SYSTEMS

Freight rate prediction and dynamic pricing are critical components of modern logistics systems, directly influencing profitability, operational efficiency, and market competitiveness. Traditional pricing models in the trucking industry rely on historical averages and rule-based heuristics, which are insufficient for capturing real-time market volatility, demand-supply fluctuations, and contextual factors such as route conditions, fuel prices, and broker behavior. This paper proposes a hybrid AI framework that integrates Large Language Models (LLMs), time-series forecasting models, and Retrieval-Augmented Generation (RAG) to enable real-time freight rate prediction and intelligent pricing decisions. The system combines structured numerical prediction with unstructured contextual reasoning, allowing it to interpret market signals such as broker messages, load descriptions, and external economic indicators. The proposed architecture introduces a dual-layer model: (1) a machine learning-based predictive layer for quantitative freight rate estimation, and (2) an LLM-based contextual reasoning layer that adjusts predictions based on real-world operational intelligence. Additionally, a dynamic decision agent evaluates whether to accept, reject, or negotiate freight offers based on predicted profitability. The framework is designed for integration into intelligent trucking systems and dispatching platforms such as modern AI-driven logistics assistants. The study provides a conceptual architecture, algorithmic formulation, and evaluation framework for future empirical validation.