• Multiple data sources
  • Python
  • MongoDB
  • Time-series forecasting
  • Classification algorithms
  • IOT
  • Asset monitoring
  • Industry 4.0
  • Asset performance
  • Asset reliability
  • Big Data


  • Which are the indicative parameters for industrial pump failures? How to optimize these parameters?
  • Real-time monitoring of the functioning of pumps for critical alerts.
  • How to predict the likelihood of pump failures?
  • How to optimize the maintenance costs?


  • Exploratory Data Analysis: We analyzed last 3 years data for around 5000 pumps for last 3 years. We measured the correlation between downtime and aberration for different parameters.
  • Real-time monitoring: With time-series analysis and forecasting we built accurate real-time monitoring of pumps to identify: 1) Normal behavior of each operating parameter 2) Anomaly detection in each operating parameter. Using this analysis, we created an alert generating mechanism.
  • Predictive modelling: With all the correlating parameters and anomaly detection data we built an advance classification model to predict the failure of pumps.
  • Recommendations: We also charted the recommendations and prescriptions for all the relevant stakeholders to deal with the alerts and predictive failure data.

Business Benefits

  • Instead of relying on scheduled maintenance, now we could maintain pumps for their uptime.
  • Real time monitoring and predictive failure analysis saved 1500 hours of downtime.
  • Savings of $100K per month in the maintenance costs.
Predictive maintenance helped this Industrial firm to save almost $100K/m for each asset in maintenance costs.