基于PSO优化RBF-NN的磁浮车间隙传感器温度补偿
Temperature Compensation of Maglev Vehicle Gap Sensor Based on RBF-NN Optimized by PSO

针对高速磁浮列车悬浮间隙传感器的温度漂移现象，建立了基于RBF（radial basis function）神经网络的间隙传感器温度补偿模型.通过对全局最优粒子执行梯度下降寻优，将粒子群优化算法与梯度下降算法结合得到一种寻优能力更强的混合算法，并将该方法用于RBF温度补偿模型参数优化，提高了间隙传感器的补偿精度，最后，使用现场可编程门阵列FPGA（field-programmable gate array）实现了该补偿模型并进行了实验.实验结果表明：该方法能够较好地对间隙传感器进行温度补偿，补偿后的传感器输出不受环境温度影响，全量程范围内最大误差为0.45 mm，8~12 mm工作间隙范围内误差为0.16 mm.
：In order to solve the temperature drift problem of a maglev vehicle gap sensor, a temperature compensator based on RBF-NN (radial basis function neural network) was designed to compensate the temperature drift error. A hybrid algorithm was proposed to combine PSO (particle swarm optimization) algorithm with gradient descent algorithm. In the proposed algorithm, the global optimal particle of the PSO was optimized by the gradient descent method. The hybrid algorithm has stronger optimization ability. The compensation model was optimized by the hybrid algorithm and the accuracy of the compensation model was considerably improved. Finally, the compensation model was implemented in FPGA (field-programmable gate array). Experimental results show that temperature drift error of the gap sensor can be compensated effectively. The compensated output of the gap sensor was independent of the temperature. The gap sensor provides correct gap data with a maximum error of 0.45 mm for full scale and a maximum error of 0.16 mm for a working gap from 8 mm to 12 mm