摘要
针对选煤厂对浮选尾矿灰分识别的工程需求,提出了一种基于增量型极限学习机(I-ELM)的尾矿灰分识别方法。首先通过实验得到浮选尾矿图像的灰度直方图与尾矿灰分之间的关系,将由尾矿图像灰度值分布和尾矿的入射光强及反射光强组成的向量作为输入,尾矿灰分作为输出,然后利用I-ELM建立预测模型,对尾矿灰分进行识别,并与用BP神经网络和固定型极限学习机(ELM)建立的模型进行了对比。结果显示,I-ELM具有较高的预测精度,同时具有较快的学习速度,是一种比较有效的浮选尾矿灰分识别方法。
Aimed at the engineering requirement of coal separating plant in recognition of flotation tailing ash contents, a new recognition ruethod for flotation tailing ash contents based on the incremental extreme learning machine is proposed. Firstly, the relation between the gray level histogram of flotation tailings and ash content of tailings is obtained through experiments and the vector formed of the gray value distribution of railings image and the incident light and reflected light intensity of tailing is regarded as the input and the ash content of tailings as the output.Then, the prediction model is built by using I-ELM, which is used to recognize the ash contents of tailings. Then the model is compared with the model using BP nerve net and the model using fixed extreme learning machine. The results show that the I-ELM model has higher prediction accuracy and faster learning speed, which is one more effective ash identification method for flotation tailing.
出处
《机械设计与制造》
北大核心
2017年第8期17-19,共3页
Machinery Design & Manufacture
基金
山西省科技攻关项目(20120321004-03)
