基于PSO-BP神经网络的湿度传感器温度补偿  

The Temperature Compensation for Humidity Sensor Based on the PSO-BP Neural Network

作　　者：行鸿彦[1,2] 邹水平[1,2] 徐伟[1,2] 张强[1,2] 

机构地区：[1]南京信息工程大学气象灾害预报预警与评估协同创新中心,南京210044 [2]南京信息工程大学,江苏省气象探测与信息处理重点实验室,南京210044

出　　处：《传感技术学报》2015年第6期864-869,共6页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(61072133);江苏省产学研联合创新资金计划项目(BY2013007-02,BY2011112);江苏省高校科研成果产业化推进项目(JHB2011-15);江苏省“信息与通信工程”优势学科和江苏省“六大人才高峰”计划项目

摘　　要：针对自动气象站采用的HMP45D型温湿一体化传感器在实际应用过程中易受温度影响的问题,提出了基于粒子群优化算法(PSO)的BP神经网络温度补偿模型,利用粒子群优化算法对BP神经网络的初始权值阈值进行全局寻优,将粒子群优化算法优化好的权值阈值赋给BP神经网络,对BP神经网络进行训练。根据不同温度条件下测得的多组湿度传感器数据,通过建立模型,实现温度补偿,与传统BP神经网络补偿结果进行比较。实验表明,与传统BP神经网络模型相比,利用PSO-BP神经网络模型进行温度补偿后所得的误差绝对值之和降低了10.3887%RH,PSO-BP神经网络可以克服传统BP神经网络易陷入局部极值的局限,补偿精度更高,能更加有效地补偿温度对湿度传感器的影响。According to the temperature and humidity sensors of the type HMP45D on the automatic weather stations influenced easily by temperature in the actual application, compensation of humidity sensor model by the Back Propagation（BP） network based on Particle Swarm Optimization（PSO） algorithm has been proposed. The initial weight and threshold of BP network can be searched globally in PSO algorithm, then assigns the optimized weight and threshold to BP network for training. Multiple groups of the humidity sensor datas has been measured under the condition of different temperatures, Using this method to establish a model for temperature compensation, and the results were compared with general BP neural network method. The experimental results show that the sum of absolute value of error by the use of PSO-BP neural network model for temperature compensation is reduced by 10.3887% （RH） compared with that of the traditional BP neural network model. PSO-BP neural network not only can overcome the limitations that the traditional BP neural network is easy to fall into local minima, but also have the higher precision, and it can more effectively compensate the influence of temperature on humidity sensor.

关 键 词：温度补偿 粒子群优化算法 BP神经网络 湿度传感器 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]