RESUMO
The torque is a significant indicator reflecting the comprehensive operational characteristics of a power system. Thus, accurate torque measurement plays a pivotal role in ensuring the safety and stability of the system. However, conventional torque measurement systems predominantly rely on strain gauges adhered to the shaft, often leading to reduced accuracy, poor repeatability, and non-traceability due to the influence of strain gauge adhesion. To tackle the challenge, this paper introduces a photoelectric torque measurement system. Quadrants of photoelectric sensors are employed to capture minute deformations induced by torque on the rotational axis, converting them into measurable voltage. Subsequently, the system employs the radial basis function neural network optimized by simulated annealing combined with particle swarm algorithm (SAPSO-RBF) to establish a correlation between measured torque values and standard references, thereby calibrating the measured values. Experimental results affirm the system's capability to accurately determine torque measurements and execute calibration, minimizing measurement errors to 0.92%.
