基于猎人猎物优化算法的粉尘浓度BP神经网络预测模型
BP neural network prediction model of dust concentration based on Hunter-Prey optimization algorithm
徐景果 1张宇轩 2王飞 1史默 1李永永1
作者信息
- 1. 陕西彬长文家坡矿业有限公司,陕西咸阳 713500
- 2. 渭南陕煤启辰科技有限公司,陕西西安 710100
- 折叠
摘要
为了更好地预测煤矿井下泡沫降尘技术处理后的粉尘浓度,以文家坡煤矿4-2盘区2#灾害治理巷为工程背景,采用基于猎算法优化的BP神经网络模型预测粉尘浓度.通过接触角试验,结合煤样接触角表征的抑尘效率与经济性,确定对降尘效率影响较大的发泡剂浓度为0.5％,并对掘进巷道内不同参数条件下粉尘浓度进行实测;以水压、风压、初始粉尘浓度3个参数为输入,以不同条件下巷道内的粉尘浓度为输出,对比分析各算法的预测精度及泛化能力.通过比较4种神经网络预测模型的拟合度,"3-9-1"结构的HPO-BP神经网络模型预测拟合度最大,更适用于煤矿井下掘进面粉尘浓度的预测,能够为泡沫降尘技术参数的后续调整提供依据.
Abstract
In order to better predict the dust concentration after the treatment of foam dust reduction technology in coal mine,taking the 2# hazard treating roadway of 4-2 panel area of Wenjiapo Coal Mine as the engineering background,the dust concentration was predicted based on hunting algorithm-optimized BP neural network.Firstly,using the contact angle test,the concentration of foam foaming agent with better influence of dust reduction was determined as 0.5％,and the dust concentration of driving roadway at different conditions of parameters was measured.Taking the water pressure,wind pres-sure and initial dust concentration as three inputs,and the dust concentration in the roadway under different conditions as the output,we analyzed and compared the prediction accuracy and generalization capability of each algorithm.By comparison of the fitting degree of four kinds of neural networks'prediction models,the 3-9-1 structured HPO-BP neural network pre-diction model with the optimal fitting degree is more suitable for dust concentration prediction in the driving roadway.The study provides basis for adjusting the parameters of foam dust reduction in the future.
关键词
粉尘浓度/接触角/BP神经网络/泡沫降尘/预测模型Key words
dust concentration/contact angle/BP neural network/foam dust reduction/prediction model引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据