基于小生境等维BP神经网络的沉降预报

Sedimentation Forecast Based on Niche Genetic Algorithm and Equal Dimensional BP Neural Network

针对传统BP神经网络全局优化能力低、无法学习的缺陷,引入遗传算法中的小生境技术,研究了基于小生境等维BP神经网络模型,同时利用MATLAB进行编程实现。该模型的核心思想是借助小生境遗传算法优化神经网络的连接权和阈值,进而提高了等维BP神经网络模型的全局优化能力,改善了模型的收敛性。结合宁波某大楼沉降监测实例,利用小生境等维BP神经网络、GM(1,1)模型、等维BP神经网络模型分别对沉降数据建模预测,结果表明,小生境等维BP神经网络模型更加符合实际情况、预测效果更佳。

The traditional low global optimization BP neural networks,study of defects introduced niche genetic algo-rithm technology,research niche dimensions is based on BP neural network model using Matlab programming. This model is the core idea of using niche genetic algorithm optimized neural network connection weights and thresholds,thereby in-creasing dimensions,such as global optimization BP neural network model to improve the convergence of the model. With Ningbo settlement monitoring of a building,niche such as BP neural network,GM（1,1） model,dimension data modeling of BP neural network model for settlement prediction,results show that niche such as BP neural networks model consist-ent with the actual situation,forecast better results.