论文部分内容阅读
煤与瓦斯突出影响因素多,难以为其建立合适的多指标非线性预测模型,为提高突出预测的准确性和增强预测预报方法的实用性,采用改进的BP算法建立煤与瓦斯突出预测数学模型。通过研究不同算法的突出预测效果,对已建模型的泛化能力进行检验,利用Matlab GUI和神经网络工具箱设计开发煤与瓦斯突出预测系统,通过向系统输入已知的突出样本数据,经过学习、训练,实现对未知参数的预测。仿真结果表明:网络在训练300次后,误差训练曲线的均方差(MSE)可以达到10-15,实际预测误差也小于0.1,系统得到的5组数据预测结果与实际情况相符。
There are many influencing factors of coal and gas outburst, so it is difficult to establish suitable multi-index nonlinear forecasting model. In order to improve the accuracy of outburst forecasting and enhance the practicability of forecasting and forecasting methods, an improved BP algorithm is used to establish mathematical model of coal and gas outburst prediction . By studying the prominent prediction effect of different algorithms, the generalization ability of the established model is tested, the prediction system of coal and gas outburst is designed and developed by using Matlab GUI and neural network toolbox, and the known prominent sample data are input to the system. After learning , Training, to achieve the prediction of unknown parameters. The simulation results show that the mean square error (MSE) of the error training curve can reach 10-15 after the network is trained for 300 times, and the actual prediction error is less than 0.1. The predicted results of the five sets of data obtained by the system are in accordance with the actual situation.