灰分是衡量煤炭质量优劣的关键指标,是衡量煤矿和选煤厂煤炭产品质量的主要指标之一。针对传统煤灰分含量识别效率低、煤样本质量不高的问题,本文基于粒子群优化算法(PSO)和BP神经网络,提出了基于粒子群神经网络的煤炭灰分预测模型。目的是快速识别出煤炭产品中灰分的含量,为煤炭开采提供技术支撑。研究选取了180个标准煤粉样品,1～140号样本数据用于训练集,141～180号样本数据作为测试集。应用PSO BP模型对煤炭灰分特性进行了研究,仿真结果表明:优化后的6维BP神经网络模型,决定系数R2为088501越接近1,表明建立的PSO BP模型具有较好的预测性能,灰分预测值与灰分真值无限逼近。进而表明所构建的灰分预测模型具有较高的预测精度,提升了模型的泛化能力和预测精度,为后续的LIBS术应用于煤炭检测提供一定的理论依据。
Ash content is a key index to measure the quality of coal and is one of the main indicators of coal product quality in coal mines and coal preparation plants.Aiming at the problems of low identification efficiency of traditional coal ash content and low quality of coal samples,a coal prediction model based on particle swarm optimization (PSO)and BP neural network is built and the purpose is to quickly identify the ash content in coal products and provide technical support for coal mining.180 standard pulverized coal samples are selected for the study,and the sample data of No.1～140 are used for the training set,and the sample data of No.141～180 are used as the test set.The PSO BP model is applied to study the coal ash characteristics,and the simulation results show that the optimized 6 dimensional BP neural network model,with the coefficient of determination R2 of 0.88501,is closer to 1,indicating that the established PSO BP model has better prediction performance,and the predicted value of ash is infinitely close to the true value of ash.In turn,it shows that the constructed gray prediction model has high prediction accuracy,which improves the generalization ability and prediction accuracy of the model,and provides some theoretical basis for the subsequent application of LIBS technique to coal detection.
LI Yun-hong, YU Tian-jiao, ZHOU Xiao-ji, GUAN Jin-ge, ZHENG Yong-qiu, ZHANG Cheng-fe, CHENG Bo. Research on coal ash content characteristics basedon laser induced breakdown spectroscopy[J]. LASER & INFRARED,2023,53(11):1657~1664