A data-driven framework for constructing representative operating cycles of high-horsepower tractors from field-road measurements

Accurate characterization of real-world operating behaviours of high-horsepower agricultural tractors is a fundamental challenge for subsequent modelling, analysis, and system design. The highly variable and multi-modal nature of field-road operations makes it difficult to construct operating cycles that faithfully represent measured dynamics while preserving statistical distributions and temporal correlation characteristics. This study proposes a data-driven framework for constructing representative operating cycles of high-horsepower tractors based on extensive field-road measurements. The framework integrates K-means clustering guided by multiple clustering validity indices, including the Davies–Bouldin index and Silhouette coefficient, with vehicle specific power per unit mileage as a physically informed descriptor. A reduced-dimensional feature space is established to characterize measured tractor dynamics, enabling the extraction of representative kinematic fragments from complex operational datasets. These fragments are then synthesized into a unified operating cycle that preserves the intrinsic statistical structure of the measured data. The representativeness of the constructed cycle is quantitatively evaluated using both characteristic parameter deviation and autocorrelation function (ACF) analysis. The average deviation of key kinematic features is 5.28%, while the root mean square error between the measured and synthesized ACF curves is only 0.0312, indicating strong consistency in both statistical distribution and temporal continuity. To further assess practical applicability, Cruise-based energy consumption simulations are conducted under three distinct load conditions. The predicted fuel consumption differs from experimental measurements by less than 10%, confirming that the synthesized cycle effectively reproduces real operating behaviour within acceptable measurement uncertainty. The proposed framework provides a systematic approach for measurement-based representation and validation of operating profiles of high-horsepower agricultural machinery, supporting further developments in modelling, evaluation, and control of agricultural vehicle systems.