基于PCA-GA-BP神经网络的瓦斯含量预测分析

为提高煤层瓦斯含量预测的效率和准确率，提出了先采用主成份分析（PCA）方法来降低变量间的相关性，然后将遗传算法（GA）与BP神经网络相结合的煤层瓦斯含量预测的新方法。为了避免BP神经网络收敛速度慢、易陷入局部极小值等问题，算法采用GA对BP神经网络的权值和阈值进行优化，利用Matlab软件进行编程，建立了BP神经网络和GA-BP神经网络瓦斯含量预测模型。选取淮南某矿瓦斯含量及其影响因素作为实验数据对该模型进行了实例分析，将主成份回归和BP网络算法预测结果与该模型进行了对比分析。结果表明:PCA-GA-BP网络预测模型平均相对误差为2.759%，预测效果明显优于主成份回归和BP网络预测模型，可以准确的预测煤层瓦斯含量。

Prediction of gas content based on PCA-GA-BP neural network

In order to improve the efficiency and accuracy of prediction on gas content in coal seam, in this paper, a new method was raised to predict the gas content, which adopted principal component analysis (PCA) to reduce the correlation between variables first, then combined BP neural network with genetic algorithm (GA). Considering the problem of slow convergence and easily trapping into the partial minimum, the algorithm adopted GA to improve the weights and thresholds of BP neural network. Taking Matlab for writing programs, the prediction models of gas content based on BP neural network and GA-BP network were established. The gas content and influence factors in a coal mine of Huainan were taken as experimental data to conduct practical analysis on this model, and the prediction results of principal component regression and BP network algorithm were compared with the result of the model. The results showed that the average relative error of PCA-GA-BP network prediction model was 2.759%, which was better than those of principal component regression and BP network prediction model, and it can accurately predict gas content in the coal seam.