Server simulation for monitoring remaining useful life of turbofan jet engines

To demonstrate the working of proposed models in a real-time environment of https://github.com/ritu-thombre99/RUL-Prediction, we developed a website to simulate the server that can monitor the health of the engines as follows:

• After the simulation is started, health scores of 20 engines are displayed in a table
• Health of an engine falls under four categories:
  • Extremely Critical (0-0.25)
  • Critical (0.25-0.5)
  • Normal (0.5-0.75)
  • Optimal (0.75-1)
• Statistics of health and range of health scores of engines is displayed as a pie chart and a line graph
• Health scores of engines are refreshed every 1-2 second

1. We used flask for the web development.
2. All the models (regression and LSTM) are saved using pickle and save_model.
3. These models are imported in the backend app.py file.
4. VAR forecasts time-series of the randomly chosen 20 engines for 20 cycles into the future.
5. LSTM with lookback=1 then predicts RUL for these 20 cycles and the corresponding health score is calculated.
6. These health scores are then sent one cycle at a time to monitor.html file.

1. We have 3 html files: home.html, monitor.html and finish.html.
2. home.html will let a user to start the simulation and redirects to monitor.html.
3. Once the simulation ends, monitor.html redirects to finish.html, from where user can restart the simulation.
4. monitor.html file receives the health scores of 20 engines one cycle at a time from the backend.
5. These health scores are displayed in a table as shown in the following figure .
6. Range of health scores and categories are generated using jinja2 and they are displayed in a bar chart and a pie chart respectively, using google visualization tools as shown in the following figure .