基于小波神经网络的电涡流传感器非线性补偿

Nonlinear compensation of eddy current sensor based on wavelet neural network

为消除电涡流传感器的非线性误差,提高其测量精度,提出了一种基于小波神经网络和遗传算法的电涡流传感器非线性补偿方法。该方法利用小波神经网络的非线性映射能力,使得传感器的输入与输出线性化,并使用遗传算法搜寻网络的最优初始值,加强网络的非线性逼近能力和收敛能力,显著提高电涡流传感器的非线性补偿效果。实验结果表明,经过补偿后,极大提高了传感器的精度,传感器输出电压最大绝对误差为15.55mV,最大相对误差为1.36%,非线性误差为0.34%。

To eliminate the nonlinear error and improve the measurement accuracy of eddy current sensor,a nonlinear compensation method of eddy current sensor based on wavelet neural network and genetic algorithm was proposed.The method uses nonlinear mapping ability of wavelet neural network to make input and output of the sensor linearization,and uses genetic algorithm to search the optimal initial value of the wavelet neural network to strengthen network nonlinear approximation ability and convergence capability,which can significantly improve nonlinear compensation effect of eddy current sensor.The experimental results show that accuraty of the sensor is improved greatly after the compensation,the maximum absolute error of output voltage of the sensor is 15.55 mV,the maximum relative error is1.36%,and nonlinear error is 0.34%.