Abstract: In order to solve the major technical problems such as high labor intensity, insufficient detection frequency and high testing cost in the manual monitoring of dynamic liquid level in ground-driven screw pump wells, an innovative detection method is proposed based on the fusion of three parameters including the axial force, torque and rotary speed of the polished rod. Firstly, the pump differential pressure mechanism model is constructed by analyzing measured data of the polished rod axial load. And then, the cable-free computation model for the working fluid level is established by combining the dynamic compensation algorithm for pressure loss under variable speed conditions, the dynamic production prediction model is established according to the torque-displacement characteristics simultaneously. Finally, the pump pressure differential model is integrated to form a collaborative solution framework of liquid level-output. The empirical results of 10 typical wells in Daqing Oilfield show that the average relative error of working fluid level calculated by this method is 12.32%, which is more than 3 times more accurate than the traditional methods, with test frequency increased by over 4,000 times. The prediction error of liquid production is 9.39%, representing a 74.8% reduction compared to the electric parameter method. The monitoring cost of a single well is reduced by 82.7%. This technical solution successfully enables real-time monitoring of dynamic liquid level and production parameters, promotes the digitalization of oilfield production management, reduces operating costs, and improves operational efficiency and economic benefits.