基于改进GA-BP网络算法的边坡力学参数反演分析

针对BP神经网络收敛速度慢和易陷入局部极小值等不足,通过改进遗传算法,显著提升遗传算法的全局寻优能力,进而优化BP神经网络初始权值和阈值。结合工程算例,采用正交法设计参数样本,利用边坡工程的有限元正分析模型计算出反演分析所需的样本,建立基于改进的GA-BP网络算法反分析模型,经过网络训练,得到符合实测效应量值的反演参数值,对比GA-BP网络算法和改进GA-BP网络算法的反分析模型结果可知,改进GA-BP网络算法反分析模型在解的稳定性和求解精度上均得到了较大提高。研究成果可供类似工程参考。

Back Analysis of Slope Mechanics Parameters Based on Improved GA-BP Network Algorithm

Aiming at the shortcomings of BP neural network such as slow rate of convergence and tendency to trap into local minimum easily, genetic algorithm was improved to significantly promote its global optimization capacity. And then the initial weight value and threshold of BP neural network were optimized. Combined with engineering examples, the orthogonal method was adopted to design parameter samples. The finite element analysis model of slope engineering was used to find out the samples required for inverse analysis. An inverse analysis model based the improved GA-BP network algorithm was established. The values of inverse parameters reflecting the measured effect were obtained through network training. Compared with the model of GA-BP network algorithm and improved inverse analysis model of GA-BP network algorithm, it can be seen that the inversion analysis model of the improved GA-BP network algorithm has been greatly improved in the solving stability and accuracy. The study can be used for similar engineering reference.