Dual-Channel Thermal-Drift Modeling and Analytical Inverse Compensation for Tractor Hydraulic Output Power Measurement

During tractor hydraulic output power testing, the temperature rise induced by throttling loading causes simultaneous drift in the output characteristics of pressure sensors and turbine flowmeters, resulting in coupled errors during pressure–flow product calculation and reducing the reliability of test-bench measurements. To address the limitation of conventional single-channel temperature compensation in suppressing the coupled effects of dual-sensor thermal drift, an analytical inverse compensation method for pressure–flow dual channels based on a second-order thermal-drift parameterized model is proposed. First, a mathematical model describing the parameter drift of pressure and flow channels under temperature variations and its error propagation to hydraulic power is established. Then, temperature-dependent offsets and sensitivities of the pressure and flow channels are identified using full-temperature-range static calibration data, and the actual pressure and flow rate are analytically reconstructed based on the sensor measurement equations to achieve online hydraulic power compensation. Dynamic loading experiments with continuous temperature increase were conducted within a temperature range of 34.87~80.72 °C. Using the first test (lowest initial oil temperature 34.87 °C) as the reference baseline, the proposed method reduced the maximum relative error of hydraulic power from 12.3% to below 1.1%, with the average relative error controlled within 0.82% and error reductions exceeding 87% under all tested conditions. Compared with the BP neural network compensation method, the proposed approach achieved comparable compensation accuracy while requiring only 85 μs for a single computation. The results demonstrate that the proposed method provides a lightweight solution for thermal-drift correction and reliable test-bench measurement of tractor hydraulic output power by balancing compensation accuracy, physical interpretability, and real-time computational efficiency.