论文部分内容阅读
鉴于常规煤与瓦斯突出BP预测模型的不足,将改进DE算法用于BP网络模型参数的优化及训练,提出结合两者优点的改进差分进化神经网络(IDEBP)煤与瓦斯突出预测模型.模型通过对变异模式、变异交叉因子自适应确定等改进,有效提高了标准DE的性能.实现了DE全局优化搜索与BP自适应、自学习的有机结合,稳健性得到加强,更能充分辨识煤与瓦斯突出样本的复杂非线性知识.以36组工程实例数据,进行了IDEBP和DEBP模型与BP模型仿真对比实验.结果表明:该模型能有效避免常规BP的不足,在收敛迅速、结果辨识和预测精度等方面均大为提高,为瓦斯智能预测提供了新的解决方案.
In view of the shortage of conventional coal and gas outburst BP prediction model, the improved DE algorithm is used to optimize and train the parameters of BP network model, and an improved coalge outburst prediction model based on the improved differential evolution neural network (IDEBP) is proposed. The improvement of the standard DE is achieved by the improvement of the mutation pattern, the adaptive crossover factor of mutation, etc. The combination of DE global optimization search with BP self-learning and self-learning is realized, the robustness is enhanced, and the coal and gas Highlighting the complex nonlinear knowledge of samples.Comparison experiments between IDEBP, DEBP model and BP model are carried out by using 36 sets of engineering example data.The results show that this model can effectively avoid the shortcomings of conventional BP, with the advantages of fast convergence, accuracy of results identification and prediction And other aspects are greatly improved, providing a new solution for gas intelligent prediction.