基于小波神经网络的煤矿瓦斯涌出量预测

MINE GAS GUSHING QUANTITY FORECASTING BASED ON WAVELET NEURAL NETWORK

针对影响瓦斯涌出量的因素复杂多样化以及各因素之间的非线性问题,综合利用神经网络的自学习能力和小波变换的局部化性质,采用了一种基于黄金分割原理获得隐含层节点数的寻优算法,结合MATLAB强大的运算功能,建立了基于小波神经网络的煤矿瓦斯涌出量预测模型。仿真结果表明整个系统具有较强的逼近和容错能力,以及较快的收敛速度和良好的预报效果。

Comprehensive utilizing the self-learning ability of neural network and localization character of wavelet transform, an optimization algorithm for getting the number of hidden layer nodes is introduced based on the principle of golden section, and the powerful calculation functions of MATLAB is combined to set up a mine gas gushing quantity forecasting model based on wavelet neural network for every diversified factors which influence mine gas gushing quantity, as well as the nonlinear problem among them. The simulation results shown that the system has stronger approximation and fauh-tolerance ability, better convergence and forecast effect.