Skip to content

Dashboard designed to demonstrate the power of Machine Learning to predict failures (Remaining Useful Life (RUL)) in wind turbines. To predict the date when equipment will completely fail (RUL), XGBoost is used and achieved RMSE error is 0.033964 days, which is highly accurate.

Notifications You must be signed in to change notification settings

iameminmammadov/dash-predictive-maintenance

Repository files navigation

Predictive Maintenance for Wind Turbines Dashboard

Introduction

dash-predictive-maintenance is a dashboard designed to demonstrate the power of Machine Learning to predict failures (Remaining Useful Life (RUL)) in wind turbines. The data covers periods from May, 2014 to January, 2015. To predict the date when equipment will completely fail (RUL), XGBoost is used. The achieved RMSE error is 0.018534 days, which is highly accurate.

Screenshots

initial

initial

Built With

  • Dash - Main server and interactive components.
  • Dash DAQ - Styled technical components for industrial applications.
  • XGBoost - Machine Learning model that was used to predict the RUL. The model was fine-tuned with RandomSearch.

Requirements

Clone this repo and create a clean environment:

git clone https://github.com/iameminmammadov/dash-predictive-maintenance.git
cd dash-predictive-maintenance
python3 -m virtualenv venv

To activate the virtualenv in UNIX:

source venv/bin/activate

To activate the virtualenv in Windows:

venv\Scripts\activate

To install the libraries, needed to run this dashboard:

pip install -r requirements-predeploy.txt

To run this app:

python app.py

The app will be run on http://127.0.0.1:8050/.

The app is currently running @ https://dash-gallery.plotly.host/dash-turbine-maintenance/

About

Dashboard designed to demonstrate the power of Machine Learning to predict failures (Remaining Useful Life (RUL)) in wind turbines. To predict the date when equipment will completely fail (RUL), XGBoost is used and achieved RMSE error is 0.033964 days, which is highly accurate.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published