基于深度学习的便携式浊度计示值误差校正方法  

Correction method for measured value error of portable turbidimeter based on deep learning

作　　者：束平 曾爱娣 张娟 SHU Ping;ZENG Aidi;ZHANG Juan(Yancheng Environmental Monitoring Center,Yancheng 224000,China)

机构地区：[1]江苏省盐城环境监测中心,江苏盐城224000

出　　处：《黑龙江环境通报》2023年第4期143-146,共4页Heilongjiang Environmental Journal

摘　　要：常规的浊度计示值误差校正方法多采用比值校正法进行设计,应用过程中,无法深入挖掘浊度计示值的各项误差,存在校正精度较低的问题,导致浊度计测量的浊度示值与待测液的实际浊度值之间偏差较大。针对这一问题,笔者引入深度学习原理,以便携式浊度计为例,提出一种全新的示值误差校正方法。先采集便携式浊度计示值,获取浊度计示值的动态变化,然后利用深度学习算法原理,构建神经网络模型结构,通过神经网络模型的示值误差激励、校正量化、迭代训练、最优保存等步骤,得出最优化的示值误差校正结果。实验分析可知,新的方法应用后,测得标准浊度液的浊度示值与其实际浊度非常接近,示值误差较小,校正效果优势显著。Conventional correction methods for measured value errors of turbidity meters often use ratio correction,which cannot effectively identify and correct various sources of errors in turbidity measurements.As a result,the accuracy of turbidity measurement is compromised,leading to significant discrepancies between the measured turbidity values and the actual turbidity of the tested liquid.To address this issue,this paper introduces the principles of deep learning and proposes a novel correction method for measured value errors in portable turbidity meters.The method involves collecting the measured values of a portable turbidity meter and capturing their dynamic changes.A neural network model is then constructed using deep learning algorithms.The model undergoes steps such as error excitation,quantization,iterative training,and optimal preservation to obtain optimized correction results for the measured value errors.Experimental analysis demonstrates that the proposed method achieves close agreement between the measured turbidity values and the actual turbidity of standard turbidity liquids,with significantly reduced value errors and improved correction effectiveness.

关 键 词：深度学习 校正 便携式 浊度计 

分 类 号：TH83[机械工程—仪器科学与技术]