改进的WOA-BP算法在概率积分法参数预计中的应用  

Application of improved WOA-BP algorithm in parameter prediction of probability integral method

作　　者：郭立 赵尚民[1] GUO Li;ZHAO Shangmin(College of Mining Engineering,Taiyuan University of Technology,Taiyuan 030024,China)

机构地区：[1]太原理工大学矿业工程学院,太原030024

出　　处：《测绘科学》2023年第6期131-139,共9页Science of Surveying and Mapping

基　　金：国家自然科学基金面上项目(42271432)。

摘　　要：针对BP神经网络在预测概率积分法参数时将引起初始偏置值和权值的随机性的问题,该文旨在解决由此导致预测参数不稳定的问题。首先对鲸鱼算法(WOA)的收敛因素和位置更新方式进行改进,使其寻优能力提升;再利用改进的WOA算法对BP神经网络的初始偏置值和权值进行优化,使用最优的偏置值和权值获得概率积分法预计参数,并在矿区沉降监测中应用。利用实测数据验证表明,与BP神经网络和普通WOA-BP算法相比,改进WOA-BP算法具有较高的一致性指数和决定系数,且均方根误差和相对误差平均值明显减小,迭代次数显著减少。因此,改进的WOA-BP算法可有效提高概率积分预计的质量,在矿区沉降监测中发挥一定作用。Aiming at the problem that back-propagation(BP)neural network will cause the randomness of the initial bias value and the weight value when predicting the parameters of the probability integral method,this paper aims to solve the problem that the prediction parameters are unstable.Firstly,the convergence factor and position update method of whale optimization algorithm(WOA)were improved to improve its optimization ability.Then,the improved WOA algorithm was used to optimize the initial bias value and weight of BP neural network,and the optimal bias value and weight were used to obtain the predicted parameters of probability integral method,which were applied in subsidence monitoring of mining area.Experimental results showed that compared with BP neural network and WOA-BP algorithm,the improved WOA-BP algorithm had higher consistency index and determination coefficient,and the mean value of root mean square error and relative error was significantly reduced,and the number of iterations was significantly reduced.Therefore,the improved WOA-BP algorithm could effectively improve the quality of probability integral prediction and play an role in mining subsidence monitoring.

关 键 词：概率积分法参数预计 改进的WOA-BP算法 BP神经网络 WOA算法 矿区沉降监测 

分 类 号：TD325.2[矿业工程—矿井建设] TP183[自动化与计算机技术—控制理论与控制工程]