基于混沌BP算法的数字温度传感器温度误差模糊控制方法  

Fuzzy control method for temperature error of digital temperature sensor based on chaotic BP algorithm

作　　者：郑薇 ZHENG Wei(School of General Education,Xi an Mingde Institute of Technology,Shaanxi Xi an 710124,China)

机构地区：[1]西安明德理工学院通识教育学院,陕西西安710124

出　　处：《工业仪表与自动化装置》2023年第3期122-126,133,共6页Industrial Instrumentation & Automation

基　　金：高等学校大学数学教学研究与发展中心2022年教学改革项目,项目名称:新时代高等数学教材建设的研究与探索(CMC20220508)。

摘　　要：为了保证数字温度传感器对温度的准确读取,需要对传感器温度误差展开有效控制。为了提高数字温度传感器温度误差控制效果,提出基于混沌误差反向传播(Error Back Propagation Training,BP)神经网络算法的数字温度传感器温度误差模糊控制方法。首先利用经验模态分解法对收集到的温度测量数据展开去噪处理,计算不同温度区间内的温度测量超差概率,进而实现误差特征阈值的提取;建立三层神经网络,通过对温度误差的反复训练达到温度补偿目的。引入Logistic映射法与混沌扩频序列法,计算混沌系数间的互协方差函数,采用M-N-L结构的前馈网络对原BP网络的三层建构展开优化,并以此提高温度误差模糊控制的精度。测试结果表明:方法对传感器温度误差的提取值与实际超差值的差距低于0.03,收敛步数小于120步,误差补偿后误差比率低于19.3%,残差平方和低于0.23,对传感器温度误差的控制更加精准,温度补偿的效率更高,提高了误差控制效果。In order to ensure the accurate reading of temperature by digital temperature sensor,it is necessary to effectively control the sensor temperature error.In order to improve the control effect of temperature error of digital temperature sensor,a fuzzy control method of temperature error of digital temperature sensor based on chaos error back propagation training(BP)neural network algorithm is proposed.Firstly,the empirical mode decomposition method is used to de-noise the collected temperature measurement data,calculate the out-of-tolerance probability of temperature measurement in different temperature ranges,and then extract the error characteristic threshold.Three-layer neural network is established to achieve temperature compensation through repeated training of temperature error.Logistic mapping method and chaotic spread spectrum sequence method are introduced to calculate the cross-covariance function between the chaotic coefficients,and the feedforward network of M-N-L structure is used to optimize the three-layer construction of the original BP network,so as to improve the accuracy of temperature error fuzzy control.The test results show that the difference between the extracted value of the sensor temperature error and the actual excess value is less than 0.03,the number of convergence steps is less than 120,the error ratio after error compensation is less than 19.3%,and the sum of squares of the residual error is less than 0.23,which makes the control of the sensor temperature error more accurate,the efficiency of temperature compensation higher,and improves the error control effect.

关 键 词：Logistic映射法 混沌扩频序列法 互协方差函数 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]