Abstract: The paper focuses on a blade-type conveyor belt cleaner operated by a dual direct-acting three-way proportional pressure-reducing valve-controlled (PPRV) asymmetric hydraulic cylinder. It proposed an adaptive control algorithm based on PI (Prandtl-Ishlinskii) inverse feedforward compensation to mitigate the pressure hysteresis, dead zone and gain nonlinearity inherent in the PPRVs. The primary aim was to ensure precise control of the blade's driving torque in a conveyor belt cleaner. The algorithm was designed to maintain consistent pressure against the conveyor belt and ensure effective cleaning throughout the blade's entire lifecycle. Through online identification of PI model parameters using a recursive least squares method with forgetting factors, the inverse model was resolved, enabling feedforward compensation for linear active disturbance rejection control based on the hysteresis characteristics of the PPRVs. Simulation and experimental results demonstrate the effectiveness of the proposed algorithm in reducing PPRV hysteresis, dead zone and gain nonlinearity, thereby improving the system's dynamic response. Compared to the PID algorithm, the proposed method exhibits a 65% increase in dynamic response, with blade driving torque error controlled within ±2 N·m in the torque range of 100～700 N·m.