基于HHO-KELM的FBG流量温度复合传感解耦

Decoupling of FBG flow and temperature composite sensing based on HHO-KELM

针对光纤布拉格光栅(Fiber Bragg Grating,FBG)流量温度复合传感耦合干扰严重的问题,以小型探针式FBG流量温度复合传感器为研究对象,提出了基于哈里斯鹰算法优化核极限学习机(Harris Hawks Optimizer algorithm Optimized Kernel Extreme Learning Machine,HHO-KELM)的解耦算法。首先,设计了以空心圆柱悬臂梁为受力载体的小型探针式FBG流量温度复合传感器,揭示了该传感器波长漂移量与流量温度的映射关系;然后,构建实验系统进行了流量温度复合传感实验,分析了流量温度耦合特征;最后,利用哈里斯鹰算法优化核极限学习机,获取核极限学习机的最优正则化系数和核函数参数组合,建立了HHO-KELM算法流量温度解耦模型,解耦后流量在2~30 m^(3)/h内,流量平均误差为0.038 m^(3)/h,均方误差为1.91×10^(-3)m^(3)/h,温度平均误差为0.027℃,均方误差为1.03×10^(-3)℃。为验证解耦效果,将HHO-KELM算法与BP算法、ELM算法的解耦结果进行对比。实验结果表明:HHO-KELM算法具有较好的解耦精度和解耦效率,能够有效降低流量温度耦合干扰,提高了传感器的测量精度和稳定性,可实现流量温度的实时动态解耦。

To address the issue of serious coupling interference in fiber Bragg grating(FBG)flow and temperature composite sensing,this paper proposes a decoupling algorithm based on the Harris hawks optimizer algorithm optimized kernel extreme learning machine(HHO-KELM) by taking the small probe FBG flow and temperature composite sensor as the research object. First,a small probe FBG flow and temperature composite sensor with a hollow cylindrical cantilever beam as the force carrier is designed,and the mapping relationship between the wavelength drift,flow,and temperature is revealed. Then,an experimental system is constructed to carry out the flow and temperature composite sensor experiment,and the coupling characteristics of flow and temperature are analyzed. Finally,the HHO is used to optimize the KELM,and the KELM optimal regularization coefficient and kernel function parameter combination are obtained. Further,the flow and temperature decoupling model of HHO-KELM algorithm is established. After decoupling,in the flow rate range of from 2 m^(3)/h to 30 m^(3)/h,the average flow error is 0. 038 m^(3)/h,the mean square error is 1. 91×10^(-3)m^(3)/h,the mean square error of temperature is 0. 027 ℃,and the mean square error is 1. 03×10^(-3)℃. Meanwhile,to verify the decoupling effect,the decoupling results of the HHO-KELM,BP,and ELM algorithms are compared. The experimental results show that the HHO-KELM algorithm has good decoupling accuracy and decoupling efficiency,and it can effectively reduce the coupling interference of flow and temperature. It can also improve the measurement accuracy and stability of the sensor and can realize the real-time dynamic decoupling of flow and temperature.