Abstract: Specialized vehicles in mining often use heavy-duty solid rubber tires or polyurethane-filled tires, which exhibit significantly different tire-road contact mechanisms and operating environments compared to traditional road vehicles. Consequently, conventional tire models are difficult to apply directly, and the complex nonlinear characteristics of these vehicles complicate the acquisition of tire forces. To address this, a reduced-order cascading sliding mode observer for articulated special mining vehicles is proposed, enabling accurate estimation of longitudinal and lateral tire forces with lower costs and reduced computational resource consumption. The eight-degree-of-freedom dynamic model of the articulated vehicle is organized and decoupled to describe its dynamic characteristics. Based on this, a reduced-order cascade sliding mode observer is designed using a sliding mode control algorithm, incorporating a saturation function to reduce chattering. Typical operational scenarios for mining articulated transport vehicles are designed, including smooth step circular steering and alternating folding-and-unfolding for dynamic steering experiments. Validation experiments are conducted on an articulated vehicle test platform, where relevant operational data are collected to reconstruct and estimate tire forces in the Matlab/Simulink environment, followed by comparative analysis. The root mean square error (E_RMS) and root mean square percentage error (E_RMS,P) metrics are introduced to evaluate the precision of the tire force cascading observer. Results indicate that the sliding mode-based reduced-order cascading observer features fast convergence, with the E_RMS,P of tire force estimates under both conditions remaining within 15%, while the E_RMS,P of lateral force estimates for the rear vehicle not exceeding 8%, effectively capturing and reprodu cing the dynamic variations in tire forces.