基于FOA优化SOM-RBF的压力传感器温度补偿研究

Temperature Compensation Research of Pressure Sensor Based on FOA Improved SOM-RBF

硅压阻式差压传感器受硅片和封装介质温度特性的影响,温变时,传感器输出呈现明显的非线性。文中提出一种全新温度补偿方法,采用基本RBF网络确定网络隐含层节点数,建立自适应SOM网络得到RBF网络中心值,采用FOA对RBF网络扩展参数进行寻优得到最终优化RBF网络,最后将测试数据输入网络得到补偿输出。结果表明:补偿精度随着RBF网络的优化逐步提升,补偿后平均最大相对误差精度为9.642 7×10-4,相对均方误差精度为2.476 2×10-7。模型输出结果验证了算法用于硅压力传感器温度补偿的有效性,抑制了温度对传感器输出的影响。

In order to alleviate the temperature effect on high precision silicon pressure sensor,a temperature compensationmodel based on the self-organizing feature maps（SOM）modified radial basis function neural network（RBF）and the Fruit Fly Op-timization Algorithm（FOA）was presented in this research.By incorporating the hierarchical clustering capability of SOM and thefast converging feature and the global optimizing ability of FOA into the modeling process,the proposed method was able to avoidthe randomness in the selection of the RBF centers and spread values,however,it made the pressure measurement more accuracy.A pressure calibration experiment was given and from which the data set with respect to key variables was obtained.The compen-sation results show that the maximum relative error is 9.6427×10-4,mean square error can reach 2.4762×10-7,which demon-strates that the proposed method is highly effective and applicable,effect of temperature on sensor output is suppressed.