Installed between the motor’s shaft and the hoist power input one, an integrate sensor could output not only the total torque signal related to the change for hoist’s load but also 60 pulses per revolution for motor’s shaft. The hoist’s opening could be calculated by associating the number of those pulses with some parameters of the hoist. In the integrate sensor, those rotary circuits could be powered from the rotary secondary of transformer whose primary was fixed with the sensor’s metal shell, which was used as a part of magnetic core of the transformer. By using the photoelectric coupling method, non-contact transmission for the sensor’s signals about the total torque and the opening was realized. Besides the torque caused by constant load, the total one also included some other torques shown as fluctuation and unsteadiness because of those factors, such as friction existing in the hoist’s mechanical structure. In order to get the actual load from the total torque signal, two phases were selected for hoist’s load detection. Firstly, the analog-to-digital converter (ADC) output value corresponding to the total torque was corrected though gray correlation analysis, and the corrected ADC value whose fluctuation and unsteadiness had been greatly reduced was selected to be as the result of the first phase. In the second phase, by using the result of the first phase as input variable, the hoist’s load predicted model was founded according to the least square support vector regression (LS—SVR) or partial least square regression (PLSR). Taking the first predicted value as the expected data sequence and implementing the correction based on the gray correlation analysis again, the hoist actual load was finally attained. Experiment result showed that the error of the result of load detection was less than ±0.6% for LS—SVR and below ±1% for PLSR model cooperated with the correction based on the gray correlation analysis. Meanwhile, since the integrate sensor was directly installed between the motor shaft and the hoist one, the number of pulses would be greatly increased during the hoist’s moving up and down which could well improve the opening detection resolution. The resolution of the hoist opening detection was far below 1mm.