基于多元非线性回归和BP神经网络的矿石加工质量预测和控制  

Prediction and Regulation of Ore Processing Quality Based on Multivariate Nonlinear Regression and BP Neural Network

作　　者：唐鹏翔 杨能 张晓美 TANG Pengxiang;YANG Neng;ZHANG Xiaomei(College of Pharmacy,Jilin Medical University,Jilin 132013,China;College of Science,Yunnan Agricultural University,Kunming 650201,China)

机构地区：[1]吉林医药学院,药学院,吉林吉林132013 [2]云南农业大学,理学院,云南昆明650201

出　　处：《微型电脑应用》2025年第1期43-48,共6页Microcomputer Applications

基　　金：吉林省职业教育与成人教育教学改革研究重点课题(2022ZCZ051);吉林省高等教育教学改革研究课题(JLJY202269980687)。

摘　　要：原矿质量和生产温度影响着矿石产品的质量,矿石的原矿质量和产品质量可以分别用原矿参数和质量评价指标来描述,在已知原矿质量下,矿石的加工生产过程主要通过调节生产系统温度Ⅰ、Ⅱ完成。根据某生产车间记录的加工时间、系统温度、原矿参数、质量指标数据,建立产品质量评价指标与原矿参数、系统温度之间的多元非线性回归模型,通过该模型可以计算出不同原矿参数、系统温度下的矿石产品质量;根据客户对矿石产品质量和产品质量评价指标的要求,建立BP神经网络模型,通过训练后的BP神经网络可以快速计算出生产系统温度Ⅰ、Ⅱ,便于实际生产操作。Raw ore quality and production temperature affect the quality of ore product.The quality of raw ore and product can be described by raw ore parameters and quality evaluation index,respectively.Given the quality of raw ore,the processing and production of the ore is mainly achieved by adjusting the temperature of production systems Ⅰ and Ⅱ.Based on processing time,system temperature,raw ore parameters and quality index data recorded in a certain production workshop,a multivariate nonlinear regression model is established between raw ore parameters,system temperature and product quality evaluation index.Through this model,the quality of ore products under different raw ore parameters and system temperature can be calculated.The BP neural network model is established based on customer requirements for ore product quality and product quality evaluation index.The trained BP neural network can quickly calculate the temperature of production systems Ⅰ and Ⅱto facilitate actual production operations.

关 键 词：质量指标 原矿参数 温度调节 多元非线性回归 BP神经网络 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]