ECMDN is a TOOL for 3D Lithology Probability Modeling and Grade Estimation.
Please download the .zip file or run
git clone https://github.com/sherryfive/ECMDN.git
in your command line.
Required:
- Python: >= 3.8.1
- Torch: 2.7.1
- python packages: scipy, numpy, sys, matplotlib, pandas, sklearn
There are two main folders in this repository: CODE1-rock and CODE2-grade.
This folder contains the code for 3D Lithology Probability Modeling.
- rock_FunEval.py, which is to evaluate the accuracy of results.
- FunModel.py, which is to construct a Deep Neural Network model. Based on this model, both direct data (boreholes) and expert-derived data (exploration profiles, geological maps) are utilized.
- rock_nn_Predict.py, which is to predict model.
- rock_nn_Train.py, which is to train model.
This folder contains the code for Grade Estimation.
- FunEval.py, which is to evaluate the accuracy of results.
- FunModel_self3.py, which is ti simultaneously characterizes local grade variability and the constraining effects of lithology for grade estimation.
- main.py, which is to train and predict.
Jiangmei Wang, [email protected]
- Lithological results
- Scatter plots of measured and estimated values
