In this repository are available codes for implementation of our study.

The version of python should preferably be greater than 3.7 our environment(for reference only): tensorflow==2.3.0 keras==2.4.0 scikit-learn==1.1.2

1. https://github.com/MingjunZhong/NeuralNetNilm
2. https://github.com/MingjunZhong/transferNILM/
3. C. Zhang, M. Zhong, Z. Wang, N. Goddard, and C. Sutton. Sequence-to-point learning with neural networks for non-intrusive load monitoring. In Proceedings for Thirty-Second AAAI Conference on Artificial Intelligence. AAAI Press, 2018.

Some already well-trained models ('*.h5' files) are in the folder directory '/models' Change the file path (refer to the parameter 'param_file') in the AugLPNNILM_test.py, and you will get the results soon. For example: param_file = args.trained_model_dir '/UK_DALE' '/AugLPN_' args.appliance_name '_pointnet_model'

1. REDD and UK_DALE datasets are available in (http://redd.csail.mit.edu/) and (https://data.ukedc.rl.ac.uk/browse/edc/efficiency/residential/EnergyConsumption/Domestic/UK-DALE-2015/UK-DALE-disaggregated).
2. Put the raw data into the folder directory dataset_preprocess, and named low_freq and UK_DALE respectively.
3. Run redd_processing.py and uk_dale_processing.py to get the prepared dataset for training and test. (note that the preprocessing of UK_DALE dataset needs another step :put the preprocessed data in "dataset_preprocess/created_data/UK_DALE/" ) The structure of folder directory is as follows: dataset_processing/ created_data/ REDD/ UK_DALE/ low_freq/ house_1/ house_2/ ... UK_DALE/ house_2/ redd_processing.py ukdale_processing.py

You can run AugLPNNILM_train.py to verify the results in our paper after you have preprocessed all the dataset. The best results('*.h5' files) will be stored in the file directory '/models'

Change the file path (refer to the parameter 'param_file') in the AugLPNNILM_test.py, and you will get the results soon. for example: param_file = args.trained_model_dir '/AugLPN_' args.appliance_name '_pointnet_model'

Contact e-mail: [email protected]