论文部分内容阅读
为提高煤矿瓦斯涌出量预测的准确度,引入证据理论组合预测方法。根据瓦斯涌出量及其主要影响因素间的实验数据,采用3个不同的粒子群神经网络模型对涌出量进行初步预测。并由BP、RBF网络对预测误差及预测点的影响因素进行分析建模,以获取每个模型的可信度。再利用证据理论对其进行合成,确定组合模型的权值,最终实现对瓦斯涌出量的组合预测。实例结果表明,该组合预测方法的平均绝对误差、均方误差分别为18.5%、5.8%,均小于神经网络组合法及等权平均法的相应预测误差,适用于煤矿瓦斯涌出量预测。
In order to improve the accuracy of coal mine gas emission prediction, the combination theory of evidence theory is introduced. According to the experimental data of gas emission and its main influencing factors, three different particle swarm optimization neural network models were used to predict the emission. BP and RBF networks are used to analyze and forecast the prediction error and the influencing factors of the prediction points to obtain the credibility of each model. Reusing the theory of evidence to synthesize it, determine the weight of the combined model, and finally achieve the combined forecast of gas emission. The results show that the average absolute error and mean square error of the combined forecasting method are 18.5% and 5.8%, respectively, which are smaller than the corresponding prediction errors of the neural network combination method and the weighted average method, and are suitable for the prediction of gas emission.