Simultaneous Temporal-Frequency-Variable Modeling for Power Forecasting

This is the origin Pytorch implementation of SimTFV in the following paper: [Simultaneous Temporal-Frequency-Variable Modeling for Power Forecasting] (Manuscript submitted to IEEE Internet of Things Journal). The data preprocessing, hyperparameter settings, experimental setups (including ablation studies), training duration, hardware specifications, and inference latency can be found in the manuscript.

Model Architecture

Figure 1. The architecture of our proposed SimTFV consists of three main components: (a) RevSTIN operation (right side), which mitigates the impact of nonstationarity on the statistical characteristics of the input sequence and fuse the temporal-frequency features. An example that splits the input sequence into four segments is presented. (b) TVA module, which efficiently extracts temporal-variable features in the encoder. (c) TCA module, which generates the output feature maps in the decoder for producing the prediction results.

Requirements

• python == 3.11.4
• numpy == 1.24.3
• pandas == 1.5.3
• scipy == 1.11.3
• torch == 2.1.0+cu118
• scikit-learn == 1.4.2
• thop

Dependencies can be installed using the following command:

pip install -r requirements.txt

Raw Data

ECL was acquired at: here. Solar dataset was acquired at: Solar. Wind was acquired at: Wind. Hydro was acquired at: Hydro.

Data Preparation

We supply all processed datasets and put them under ./data, the folder tree is shown below:

|-data
| |-ECL
| | |-ECL.csv
| |
| |-Hydro_BXX
| | |-Hydro_BXX.csv
| |
| |-Solar
| | |-solar_AL.csv
| |
| |-Wind
| | |-Wind.csv
| |
| ...

The processing details for the four datasets are as follows. We place ECL in the folder ./electricity of here (the folder tree in the link is shown as below) into folder ./data and rename it from ./electricity to ./ECL. We rename the file of ECL from electricity.csv to ECL.csv and rename its last variable from OT to original MT_321. The processed file can be found at ./data/ECL/ECL.csv

The folder tree in https://drive.google.com/drive/folders/1ZOYpTUa82_jCcxIdTmyr0LXQfvaM9vIy?usp=sharing:
|-autoformer
| |-electricity
| | |-electricity.csv

To standardize the data format, we convert the data file of Solar from 'solar_AL.txt' to 'solar_AL.csv'. We place the processed file into the folder ./data/Solar. For convenience, we processed the Wind and Hydro datasets and you can obtain the processed files at ./data/Wind/Wind.csv and ./data/Hydro_BXX/Hydro_BXX.csv, respectively.

Usage

Commands for training and testing SimTFV of all datasets are in ./scripts/Main.sh.

More parameter information please refer to main.py.

We provide a complete command for training and testing SimTFV:

python -u main.py --data <data> --long_input_len <long_input_len>  --short_input_len <short_input_len> --pred_len <pred_len> --encoder_layers <encoder_layers> --decoder_layers <decoder_layers> --patch_size <patch_size> --d_model <d_model> --decoder_IN --learning_rate <learning_rate> --dropout <dropout> --batch_size <batch_size> --train_epochs <train_epochs> --itr <itr>  --train --patience <patience> --decay <decay>

Here we provide a more detailed and complete command description for training and testing the model:

Parameter name	Description of parameter
data	The dataset name
root_path	The root path of the data file
data_path	The data file name
checkpoints	Location of model checkpoints
long_input_len	Input length
short_input_len	Input length
pred_len	Prediction length
enc_in	Input variable number
dec_out	Output variable number
d_model	Hidden dims of model
encoder_layers	The num of layers in each encoder stage
decoder_layers	The num of layers in each decoder stage
patch_size	The initial patch size in patch-wise attention
Not_use_CV	Whether not to adopt the cross-variable attention in TVA
decoder_IN	Whether to use decoder_IN
dropout	Dropout
num_workers	Data loader num workers
itr	Experiments times
train_epochs	Train epochs of the second stage
batch_size	The batch size of training input data
decay	Decay rate of learning rate per epoch
patience	Early stopping patience
learning_rate	Optimizer learning rate

Results

The experiment parameters of each dataset are formated in the Main.sh files in the directory ./scripts/. You can refer to these parameters for experiments, and you can also adjust the parameters to obtain better mse results or draw better prediction figures. We present the multivariate forecasting results of the four datasets in Figure 2.

Figure 2. Multivariate forecasting results.

Contact

If you have any questions, feel free to contact Li Shen through Email (shenli@buaa.edu.cn) or Github issues. Pull requests are highly welcomed!