基于RBF神经网络的湿度传感器建模方法

针对高分子湿度传感器的测量精度易受温度影响的问题，提出了湿度硬件测量电路，建立了高分子湿度传感器的数学模型。此硬件测量电路的湿度传感器选用IH3605，温度传感器选用AD590。此数学模型采用RBF神经网络，利用在线测量的温湿度信号值，应用带遗忘因子的梯度下降算法进行网络参数的调整，同时给出了建模步骤。实际结果表明，该模型的误差小于0．5％RH，具有较高的精度。

Modeling Method of Humidity Sensor Based on RBF Neural Network

According to the problem of the high molecule humidity sensor measure precision easy interfered with temperature, established the hard measure circuit and the mathematical model of high molecule humidity sensor, humidity sensor IH3605 and temperature sensor AD590 were choosed in the hard measure circuit. Utilizing the humidity and temperature value measured on-line, RBF neural network andthe steps of modeling were introduced in detail, and the gradient descend algorithem with a momentum factor was used to adjust the parameters of RBF neural network, Experiment results show that its measure error is less than O. 5 % RH, has high urecision.