基于卷积神经网络算法判定稀土总量滴定终点技术研究

由于稀土产业发展讯速,市场需求越来越大,应用范围越来越广,需建立一个长期有效且适用于测定各类稀土的方法。针对目前传统稀土总量检测方法效率低、准确度低、滴定终点差异大等问题,难以满足在线检测的需要。本文提出一种基于卷积神经网络的稀土总量浓度在线分析仪。卷积神经网络算法通过摄像头记录样品在滴定过程中的溶液颜色变化,对溶液进行图像特征提取和学习,从而有效、准确地实现化学反应过程中溶液颜色的自动化识别,配合步进电机和注射泵等部件实现自动滴定过程。图像识别本质上是对图像信息进行特征提取,而卷积神经网络(CNN)有着传统识别方法不具备的优点,比如能够自行训练、识别速度更快、所需特征更少等。本设备将自动滴定与神经网络相结合,实现了滴定流程的自动化和样品前处理、滴定、终点判定等过程的一体化,且设备内可同时进行五个样品滴定试验,提高了滴定效率。

Research on Titration Endpoint Determination of Total Rare Earth Based on Convolutional Neural Network Algorithm

Due to the rapid development of the rare earth industry, increasing market demand, and wider application range, it is necessary to establish a long-term effective and suitable method for determining various types of rare earths. In response to the problems of low efficiency, low accuracy, and large differences in titration endpoint of traditional rare earth total detection methods, it is difficult to meet the needs of online detection. In this paper, an online analyzer of total rare earth concentration based on Convolutional neural network is proposed. The Convolutional neural network algorithm records the color change of the solution during the titration process of the sample through the camera, extracts and learns the image features of the solution, so as to effectively and accurately realize the automatic recognition of the color of the solution during the chemical reaction process, and realizes the automatic titration process with the components such as the stepping motor and the injection pump. In essence, image recognition is to extract features from image information, while Convolutional neural network (CNN) has advantages that traditional recognition methods do not have, such as self training, faster recognition speed, fewer features required, etc. This device combines automatic titration with neural networks to achieve automation of the titration process and integration of sample pretreatment, titration, endpoint determination, and other processes. Additionally, the device can simultaneously perform five sample titration tests, improving titration efficiency.