煤与瓦斯突出自适应小波基神经网络辨识和预测模型

SELF-ADAPTIVE WAVELET NEURAL NETWORK FOR RECOGNIZING AND PREDICATING OUTBURST OF COAL AND GAS

由于煤与瓦斯突出是一典型的复杂非线性动力系统,影响因素很多,如应力、煤层特征、构造等,且各影响因素相互关联,因此,采用非线性人工神经网络进行煤与瓦斯突出的模式辨识与预测是十分必要的。针对具体的不同煤层条件,建立自适应小波基神经网络激活函数模型,用于煤与瓦斯突出系统的辨识和预测,实现由网络本身自动确定神经单元的数目,避免人为因素的影响,提高辨识和预测的可靠性和智能化程度。实例分析结果表明,所建立的自适应小波基神经网络激活函数模型,辨识和预测精度高,具有重要的推广应用价值。

As matter of fact,the evolution system of coal and gas outbursts is a typical nonlinear dynamical one;there are many factors such as stress,thickness variation of seam,geological fault,etc.influencing it,and these factors are correlative.It is necessary to build a nonlinear artificial neural network（ANN） to recognize the pattern of coal and gas outbursts and to predicate coal and gas outbursts intensity.A self-adaptive wavelet neural network for recognizing pattern of coal and gas outbursts and for predicating coal and gas outbursts intensity has been built by considering different coal seam and gas conditions,which can generate the neural element numbers automatically and can avoid the jamming for determining the element number in BP network artificially.This ensures the reliability and intelligence of recognition and predication.It is verified by some examples that the model has a high accuracy for recognition and predication;and it is valuable for generalizations and applications.