摘要
针对神经网络模型在建模过程中受到各种噪声影响这一问题,提出利用Kalman滤波降低噪声,建立组合模型。经过工程实例验证,证明基于滤波算法的BP神经网络模型在一定程度上可以提高预测值的精度和预测模型的稳定性,更稳定地反映了监测目标的变化趋势,在形变监测中优势明显。
For the problem of the neural network model is affected by noises during the process of modeling,the method of making use of Kalman filter to lower the noise is put forward to establish segment pattern.Verified by engineering project,it has been verified that the BP neural network model which is based on filtering algorithm is able to improve the precision of predicted value and the stability of prediction model in a certain degree,and the variation trend of the monitored object can be more stably reflected,so that it has more advantages in deformation monitoring.
作者
陈盟
姜刚
Chen Meng;Jiang Gang(Institute of Geological Engineering and Surveying,Chang an University,Xi'an 710064,China)
出处
《甘肃科学学报》
2019年第1期17-21,共5页
Journal of Gansu Sciences
基金
国家自然科学基金项目(41501499)
中央高校基本科研业务费专项基金项目(300102268206)
