Introduction

Welcome to the repository for the study "Predicting the Urban Stormwater Drainage System State using the Graph-WaveNet", published in Sustainable Cities and Society. This repository contains the code used in our research to train a Graph-WaveNet model utilizing historical network state and rainfall data from SWMM simulations.

For the original PyTorch implementation of Graph-WaveNet, please refer to the paper Graph WaveNet for Deep Spatial-Temporal Graph Modeling, IJCAI 2019. (https://arxiv.org/abs/1906.00121)

Requirements

• Python 3.6
• See requirements.txt
• Pytorch installation：
  1. download nvidia cuda12.0, or latest version (13 March 2023 update)
  2. check driver version in CMD using 'nvidia-smi', high version will match cuda better
  3. download nvidia cudnn
  4. install torch(the following version works for my cuda), be careful, copy from https://pytorch.org/get-started/previous-versions/ pip install torch==1.10.1+cu102 torchvision==0.11.2+cu102 torchaudio==0.10.1 -f https://download.pytorch.org/whl/cu102/torch_stable.html

Data Preparation

Generating training data from SWMM simulations

1. Prepare event-based SWMM output files
2. Run 'SWMM2TrainingData.ipynb'

Generating .pkl file with adjacency matrix

1. Extract node IDs and their distances from network data
2. Run 'gen_adj_mx.py'

Training

python train.py