GeoPDNN 1.0: a semi-supervised deep learning neural network using pseudo-labels for three-dimensional shallow strata modelling and uncertainty analysis in urban areas from borehole data

A program to build 3D geological model from borehole data.

PDNN depends on several standard Python packages that should be shipped with any standard distribution (and are easy to install): numpy torch scipy.

Before running this program, you need to change the file path in the code. The borehole data is stored in the mdb file. In the borehole data, the following parameters used for processing are saved in the txt file: ZKX, ZKY, TCHD, TCCDSD, ZKBG, TCZCBH, ZKID, GCSY, TCYCBH, TCCYCBH. These represent borehole coordinates x, borehole coordinates y, soil layer thickness, soil layer bottom depth, borehole elevation, main layer label, borehole id, engineering index, sublayer label, sublayer label of sublayer.

Finally, after training with GeoPDNN, you get the pt file, the training accuracy, and the test accuracy.

You can predict the results based on the modeling range of your data, and the gird data will be saved in txt file by using Pridict.py. HUll3D.py is from https://github.com/rgmyr/pyConvexHull3D, it is used by Predict.py. If you have any questions, please contact us.