基于LM神经网络瓦斯灾害预测模型的应用研究

Applied Research on the Prediction Model of Coalmine Gas Disaster Based on the LM Neural Network

摘要: 矿山瓦斯突出与爆炸事故的预测预报是当前我国煤矿安全生产中急待解决的问题之一。本文引入BP神经网络的LM优化算法，在保留空间实体相关和多种分布并存的前提下，讨论了建立LMBP神经网络瓦斯灾害预测预报模型的数学模型设计、网络结构设计和程序设计三个部分，并以兖矿集团济宁二号井为实例进行了测试。实验结果表明：该模型稳定、快速、预测精度高，能够较好地模拟矿山瓦斯突出与爆炸事故特征，对瓦斯灾害作出较准确的预测。

Abstract: Currently the forecasting of gas outburst and explosion is one of the issues to be solved in China’s coal mine safety production. Based on the LM optimization algorithm of BP neural network, this paper discussed the mathe- matical model, network architecture and programming design of establishing the neural network prediction model on gas disaster with keeping the relationship among the spatial entities and their distributions, and tested an instance of Jining No. 2 coal mine. The result shows that this model is stable, fast and high prediction accurate and it can simulate the mine gas disaster to get higher accurate predictions.