Genetic-Algorithm-Guided-Satellite-Anomaly-Detection

File list:

main.py contains the code of our proposed Anomaly detection method.

Background:

Papers:

The source code for the paper titled "Genetic Algorithm Guided Ensemble of Neural Networks for Satellite Anomaly Detection", submitted to IEEE Trans. on Aerospace and Electronic Systems, March 2022.

Citation:

There are one main citations for this work.

By default, consider using the following:

@Article{Malekisadr2022,
  author="Mohammadamin Malekisadr, Yeying Zhu, Peng Hu",
  title="{Genetic Algorithm Guided Ensemble of Neural
Networks for Satellite Anomaly Detection}",
  journal="IEEE Transactions on Aerospace and Electronic Systems ",
  year="2022",
  month="March",
  day="13",
}

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes.

Before Installing

To ensure a smooth process, please ensure you have the following requirements.

Hardware

Nvidia GPU with Compute Capability 3.5 or higher

Software

The following Softwares and Packages are recommended to be used before installation

Python: 3.6.1
Numpy: 1.12.1
Pandas: 0.20.1
Keras: 2.0.6
Scikit-Learn: 0.18.1
Theano: 0.9.0
Tensorflow: 1.2.1
Pydot: 1.0.29
GraphViz: 2.38.0
CUDA: 11.0

Installation

Clone this repository, and then install it and its requirements. It should be something similar to this:

git clone https://github.com/aminmalekisadr/Genetic-Algorithm-Guided-Satellite-Anomaly-Detection.git
pip3 install -e Genetic-Algorithm-Guided-Satellite-Anomaly-Detection/
pip3 install -r Genetic-Algorithm-Guided-Satellite-Anomaly-Detection/requirements.txt

Dataset

We use the satellite telemetry data from NASA. The dataset comes from two spacecrafts: the Soil Moisture Active Passive satellite (SMAP) and the Curiosity Rover on Mars (MSL). There are 82 signals available in the NASA dataset. We found that 54 of the 82 signals to be continuous by inspection, and the remaining signals were discrete. We only consider the time-series sequences from the telemetry signals in our evaluation, where the telemetry values can be discrete or continues in these signals.

The dataset is available here. If the link is broken or something is not working properly, please contact me through email ([email protected]).

Experiments

Configuration

The parameters that used to setup the Genetic Algorithm, Recurrent Neural Networks and Random Forests, MC dropout used in training for the model are in /experiments/Config.yaml folder.

Running an Experiment

Future Direction:

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Acknowledgments:

Yeying Zhu and Peng Hu, my research supervisors;
University of Waterloo, who hosted my research;
National Research Council Canada, who funded my Research.

aminmalekisadr / mvts2023 Goto Github PK

mvts2023's Introduction