改进GWO-BP算法的概率积分法预计参数求取

针对BP神经网络模型在求取概率积分法预计参数时的缺陷，提出了一种基于改进灰狼优化算法(GWO)的BP神经网络参数预测模型。主要通过对灰狼算法的收敛因子a进行非线性收敛的改进，再利用粒子群算法(PSO)的速度更新公式更新搜索灰狼搜索位置。用改进的灰狼优化算法对BP神经网络的初始权值和阈值进行优化，然后利用最优的初始权值和阈值对模型进行训练和预测，从而得到概率积分法参数的预测结果。结果显示经过改进的灰狼算法优化BP神经网络的参数预测结果明显优于单一的BP神经网络模型和不改进的灰狼算法优化BP神经网络模型的预测结果，可以在矿区开采沉陷预计方面得到应用。

Probabilistic integral method based on improved GWO-BP algorithm

In view of the shortcomings of the BP neural network model in obtaining the parameters of the probability integral method, we proposed a BP neural network parameter prediction model of the improved gray wolf optimization algorithm. Mainly through the improvement of the nonlinear convergence of the gray wolf algorithm convergence factor a, and then use the particle swarm algorithm Particle Swarm Optimization, speed update formula to update the search gray wolf search position. We used the improved grey wolf optimization algorithm to optimize the initial weights and thresholds of the BP neural network, and used the optimal initial weights and thresholds to train and predict the model, so as to obtain the predicted results of the probability integral method parameters, The results show that the improved gray wolf algorithm optimized BP neural network parameter prediction results are significantly better than the single BP neural network model and the unimproved gray wolf algorithm optimized BP neural network model prediction results, which can be applied to the prediction of mining subsidence in the mining area.