基于AFSA-BP神经网络的湿度传感器温度补偿  

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

作　　者：郭敏[1,2] 行鸿彦[1,2] 张冬冬[1,2] 张兰[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,江苏南京210044 [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,江苏南京210044

出　　处：《仪表技术与传感器》2017年第8期6-10,共5页Instrument Technique and Sensor

基　　金：国家自然科学基金(61671248);江苏省产学研联合创新资金计划(BY2013007-02);江苏省高校自然科学研究重大项目(15KJA460008);江苏省"六大人才高峰"计划;江苏省"信息与通信工程"优势学科资助

摘　　要：针对HMP45D型温湿一体化传感器在使用中受温度影响的问题,提出了基于人工鱼群算法(AFSA)的BP神经网络温度补偿方法,用AFSA算法优化BP网络的初始权阈值并用该值训练BP网络。根据湿度传感器在多温度下所测数据,建立AFSA-BP神经网络模型实现温度补偿,与传统BP神经网络方法对比。结果表明,与传统BP神经网络相比,AFSA-BP神经网络的误差绝对值之和降低了6.14%,改善了BP神经网络易陷入局部极小的缺陷,提高了补偿精度。According to the temperature and humidity sensors of the type HMP45D influenced by temperature in the actual application, this paper put forward the temperature compensation method of the BP network based on artificial fish-swarm algo- rithm（AFSA）. AFSA was used to optimize initial weight of BP neural network,and this value was used to BP network for train- ing.According to the data measured by the humidity sensor in multi-temperature, AFSA-BP neural network model was established to realize the temperature compensation, and compared with the traditional BP rieural network method. The results show that the sum of the absolute error of AFSA-BP neural network is reduced by 6.14% compared with the traditional BP neural network ,AF- SA-BP neural network improves the defect that the BP neural network is easy to fall into local minimum and improves the com- pensation precision.

关 键 词：人工鱼群算法 BP神经网络 湿度传感器 温度补偿 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]