摘要
采用模拟滑坡实验装置,模拟坡角为30°、降雨为后锋降雨的滑坡,用超清摄像头收集不同时刻的滑坡变化平面图.在滑坡平面图中画出滑坡断裂曲线,进行图像二值化处理.由于对处理后的数据直接进行预测误差较大,因此先进行经验模态分解,然后对其进行人工神经网络预测,计算预测值和真实值之间的均方根误差RMSE和r2.结果表明:经验模态分解与人工神经网络相结合的方法进行滑坡趋势预测,有助于提高预测精度,进而对不同的滑坡进行趋势预测.
A simulated landslide experiment device is used to simulate a landslide experiment with a slope angle of 30 degrees and rainfall as the rearward rainfall,and then an ultra-clear camera is used to collect the landslide change map at different times.The landslide fracture curve is drawn in the landslide change diagram to perform image binarization processing.Since the direct predic⁃tion error of the processed data is large,the empirical mode decomposition is performed first,and the artificial neural network pre⁃diction is finally performed to calculate the root mean square error RMSE and r 2 between predicted and true value.The results show that the method of combining empirical mode decomposition and artificial neural network for landslide trend prediction is helpful to improve the prediction accuracy,and then to predict the trend of different landslides.
作者
倪晓兰
NI Xiaolan(School of Mathematics and Physics,Lanzhou Jiaotong University,Lanzhou,Gansu 730070,China)
出处
《宜宾学院学报》
2022年第6期73-78,共6页
Journal of Yibin University
关键词
滑坡
经验模态分解
人工神经网络
landslide
empirical mode decomposition
artificial neural network