固体废弃物膏体充填料浆质量的神经网络研究  

Research on Quality of Solid Waste Paste for Backfilling Based on Neural Network

作　　者：何荣军[1] 张丽[1] 周华强[2] 武龙飞[2] 

机构地区：[1]重庆工程职业技术学院,重庆400037 [2]中国矿业大学矿业工程学院,江苏徐州221116

出　　处：《采矿与安全工程学报》2008年第3期352-356,共5页Journal of Mining & Safety Engineering

摘　　要：固体废弃物膏体充填在我国煤炭系统是一种新的胶结充填模式.充填料浆质量的研究至关重要.它是一典型的多输入、多输出、非线性的模糊模型.一方面,运用神经网络结合遗传算法构造了膏体充填料浆质量的隐式模型,建立该模型的方法以神经网络为基础,用遗传算法来学习神经网络的权系数,既保留了遗传算法的强全局随机搜索能力,又具有神经网络的鲁棒性和自学习能力.该模型具有较强预测能力,为优化固体废弃物膏体充填料浆质量的影响因素提供了理论依据.另一方面,利用已训练好的膏体充填料浆质量模型获得遗传算法,对充填料浆质量的影响因素进行优化,该法在配比设计时,可在较少的试验次数下获得较好的配比.Solid waste paste filling is a new method of cemented backfilling. The study of the backfill pulp quality is of great importance because it is a typical multi-input, multi-output, and non linear fuzzy model. On one hand, the implicit model of the backfill pulp quality is formed based on GA and BP. The method is based on the neural network, and the weights of neural network are trained by genetic algorithm. So it remains the global stochastically searching ability of genetic algorithm and the robustness and self-learning ability of neural network, making the model have a good prediction capacity, which provides a theoretical support for optimizing affecting factors of the backfill pulp quality. On the other hand, optimizing affecting factors of the backfill pulp quality using genetic algorithm are obtained using the GA-ANN model, making the good mixture proportion obtained with less testing times.

关 键 词：固体废弃物膏体充填 充填料浆质量 神经网络 遗传算法 

分 类 号：TD325[矿业工程—矿井建设]