Abstract: As the core power equipment in modern agricultural production in China, the tractor plays a significant role in promoting sustainable agricultural development. During plowing operations, complex and variable field conditions often lead to frequent fluctuations in traction resistance, making it difficult for tractors to maintain highly efficient and energy-saving operating states. Traditional internal combustion engine tractors face inherent challenges such as carbon emission restrictions and thermal efficiency bottlenecks, while pure electric solutions are limited by current battery technology in terms of energy density and endurance, making it difficult to meet the demands of high-power continuous operation. To address these issues, this paper proposes a distributed hybrid electric tractor (DHET) architecture with decoupled power output. This architecture significantly improves system energy efficiency and operational adaptability through the decoupling characteristics of multiple power sources, but it also imposes higher requirements on the efficiency of energy allocation under dynamic working conditions. As a core technology for achieving high efficiency, energy savings, and reliable operation in hybrid electric tractors, the energy management strategy directly affects overall vehicle performance. The adaptive equivalent consumption minimization strategy (A-ECMS) is widely used in hybrid power systems due to its excellent fuel economy and optimization potential. However, the effectiveness of this algorithm highly depends on the dynamic calibration accuracy of the equivalent factor, and it suffers from limitations in adaptability to varying working conditions. Therefore, this study proposes an energy management strategy that integrates clustering optimization and A-ECMS. A hierarchical architecture is adopted to enhance the adaptability and fuel economy of the control strategy through multi-stage optimization. Specifically, to address the strong subjectivity of traditional working condition division methods, a K-means clustering algorithm is introduced to analyze standard cycles, with the silhouette coefficient (SC) used as an evaluation metric to determine the optimal number of clusters. Furthermore, based on the clustering results, dynamic programming (DP) is employed to solve for the optimal equivalent factor corresponding to each cluster. To overcome the inadequacy of a single equivalent factor in online control, a mapping model of the equivalent factor based on a continuous weight re-kernel function is constructed. This model enables dynamic matching of the equivalent factor according to real-time working condition features and offline cluster centers, thereby enhancing the adaptive adjustment capability of the A-ECMS algorithm. Hardware-in-the-loop (HIL) experimental results demonstrate that the proposed method can maintain the engine and motor operating within high-efficiency ranges under most working conditions. Under the tested conditions, the equivalent fuel consumption of this strategy is reduced by 6.54% compared to the adaptive equivalent consumption minimization strategy based on a PI controller, significantly improving the fuel economy of the hybrid electric tractor and indicating promising prospects for engineering applications.