基于GA-WNN的电涡流传感器的温度补偿
|
摘要
针对电涡流传感器的温度漂移对其测量精度带来较大影响的问题,提出了基于遗传优化小波神经网络(GA-WNN)算法对电涡流传感器进行温度补偿修正模型。通过对电涡流传感器做标定实验,并且利用LM35温度传感器监测其工作温度,建立GA-WNN神经网络模型。该模型用遗传算法对小波神经网络的权、阈值进行全局的优化,改善了小波神经网络训练速度慢的问题,克服了易陷入局部最优的缺陷。研究结果表明,补偿后的灵敏度温度系数由8.69×10~(-3)/℃提升到3.48×10~(-4)/℃;零位温度系数由4. 78×10~(-3)/℃提升到1.85×10~(-4)/℃,均提高了一个数量级,成功实现了温度补偿的目的。
In order to solve the problem of temperature drift of eddy current sensors, a temperature compensation correction model based on genetic optimization wavelet neural network( GA-WNN) algorithm for eddy current sensor is persented. The GA-WNN neural network model was established by calibrating the eddy current sensor and monitoring its working temperature with LM35 temperature sensor. The model uses genetic algorithm to optimize the weight and threshold of the wavelet neural network, which improves the slow training speed of the wavelet neural network and overcomes the defect that is easy to fall into local optimum. The results show that the sensitivity temperature coefficient after compensation is raised from 8. 69 ×10~(-3)/℃ to 3. 48 ×10~(-4)/℃and the zero temperature coefficients is raised from 4. 78 ×10~(-3)/℃ to 1. 85 × 10~(-4)/℃. Both have increased by one order of magnitude, and successfully achieved temperature compensation.
In order to solve the problem of temperature drift of eddy current sensors, a temperature compensation correction model based on genetic optimization wavelet neural network( GA-WNN) algorithm for eddy current sensor is persented. The GA-WNN neural network model was established by calibrating the eddy current sensor and monitoring its working temperature with LM35 temperature sensor. The model uses genetic algorithm to optimize the weight and threshold of the wavelet neural network, which improves the slow training speed of the wavelet neural network and overcomes the defect that is easy to fall into local optimum. The results show that the sensitivity temperature coefficient after compensation is raised from 8. 69 ×10~(-3)/℃ to 3. 48 ×10~(-4)/℃and the zero temperature coefficients is raised from 4. 78 ×10~(-3)/℃ to 1. 85 × 10~(-4)/℃. Both have increased by one order of magnitude, and successfully achieved temperature compensation.
引文
[1]李如发.基于多元回归和支持向量机的电涡流传感器的研究[D].上海:东华大学,2014.
[2]刘焱,王烨.位移传感器的技术发展现状与发展趋势[J].自动化技术与应用,2013,32(6):76-80.
[3]秦晓慧,戴蓉,宋野.工件尺寸非接触在线检测方法与实验研究[J].测控技术,2015,34(4):5-8.
[4]卢文科,朱长纯,方建安.霍尔元件与电子检测应用电路[M].北京:中国电力出版社,2005.
[5]傅国刚.电涡流传感器在汽轮机TSI系统中的应用[J].城市建设理论研究,2014(21):4382-4383.
[6]王海洋.电涡流位移传感器在设备检测诊断中的应用[J].硅谷,2011(4):124-124.
[7]付立华,王刚.基于小波神经网络的电涡流传感器非线性补偿[J].工矿自动化,2015,41(9):74-77.
[8]Tang X J,Liu J H.Research on dynamic characteristics of multi-sensor system in the case of cross-sensitivity[J].Science in China Series Engineering&Materials Science, 2005,48(1):1-22.
[9]王立威.人工神经网络隐层神经元数的确定[D].重庆:重庆大学,2012.
[10]张静端.一种用于图书档案库的小波加权式声表面波温度传感器的关键问题研究[J].电子学报,2016,5:1162-1163.
[11]韩瑞锋.遗传算法原理与应用实[M].北京:兵器工业出版社,2010.
[2]刘焱,王烨.位移传感器的技术发展现状与发展趋势[J].自动化技术与应用,2013,32(6):76-80.
[3]秦晓慧,戴蓉,宋野.工件尺寸非接触在线检测方法与实验研究[J].测控技术,2015,34(4):5-8.
[4]卢文科,朱长纯,方建安.霍尔元件与电子检测应用电路[M].北京:中国电力出版社,2005.
[5]傅国刚.电涡流传感器在汽轮机TSI系统中的应用[J].城市建设理论研究,2014(21):4382-4383.
[6]王海洋.电涡流位移传感器在设备检测诊断中的应用[J].硅谷,2011(4):124-124.
[7]付立华,王刚.基于小波神经网络的电涡流传感器非线性补偿[J].工矿自动化,2015,41(9):74-77.
[8]Tang X J,Liu J H.Research on dynamic characteristics of multi-sensor system in the case of cross-sensitivity[J].Science in China Series Engineering&Materials Science, 2005,48(1):1-22.
[9]王立威.人工神经网络隐层神经元数的确定[D].重庆:重庆大学,2012.
[10]张静端.一种用于图书档案库的小波加权式声表面波温度传感器的关键问题研究[J].电子学报,2016,5:1162-1163.
[11]韩瑞锋.遗传算法原理与应用实[M].北京:兵器工业出版社,2010.