论文部分内容阅读
为准确预测瓦斯涌出量,选取某煤矿的开采煤层、临近煤层、采空区3个瓦斯涌出源作为实例研究,将BP神经网络、粒子群算法(PSO)、Ada Boost迭代提升算法和瓦斯涌出分源预测法相结合,建立基于PSOBP-Ada Boost算法的瓦斯涌出量分源预测模型,并将其与BP神经网络算法进行比较分析。结果表明,PSOBP-Ada Boost算法预测的3个瓦斯涌出源平均相对误差分别为3.24%,2.11%,3.21%;BP神经网络的平均相对误差分别为6.73%,3.19%,4.27%,基于PSOBP-Ada Boost模型的预测精度明显优于BP神经网络模型。
In order to accurately forecast the amount of gas emission, three gas emission sources in the coal seam of a certain coal mine, adjacent coal seam and goaf are selected as an example to study. The BP neural network, Particle Swarm Optimization (PSO), Ada Boost iterative lifting algorithm and gas Gushing out the source prediction method to establish the source prediction model of gas emission based on PSOBP-Ada Boost algorithm, and compared with BP neural network algorithm. The results show that the average relative errors of the three gas emission sources predicted by PSOBP-Ada Boost algorithm are 3.24%, 2.11% and 3.21%, respectively. The average relative errors of BP neural network are 6.73%, 3.19% and 4.27% -Ada Boost model prediction accuracy is significantly better than the BP neural network model.