A ship’s engine is a critical and complex engineering system comprising several systems and auxiliary subsystems. Maintenance through conventional methods resulted in longer turnaround time to repair or replace components of engines and restore normal operations.

The client partnered with Infosys to develop an Artificial Intelligence (AI) based condition monitoring system to assess and predict the health of the engines and help reduce the maintenance cost and improve the overall safety of the ship.

Key Challenges

  • Need to analyze and correlate relevant data and events related to the multiple systems and auxiliary sub-systems of ship engines
  • Lack of sufficient historical data from engine sensors impeding the development of a reliable prediction model
  • The existing OEM system supported sensor data extraction in PDF format only
  • Past events were manually logged in ledgers and needed to be digitized

Ready to experience?

TALK TO EXPERTS

The Solution

Infosys developed an AI/ML-based condition monitoring system

  • Automated data extraction from PDF format and conversions to time series data
  • Provision for a system administrator to perform activities such as:
    • Raw data correlation analysis and feature selection
    • Data cleansing (missing value imputation/outlier removal)
    • Exploring and analysis of the data using charts
    • Automatic labeling of data
    • Run multiple algorithms, compare KPI’s and finalize the best model based on accuracy and error metrics
  • Provision for a system operator to perform activities such as:
    • View dashboard to get information on the predicted health of an engine and its subsystems based on the latest data
    • Perform ‘what-if’ analysis of the predicted health by changing any of the engine parameters or values

Modular, scalable and extendable solution built using open source technologies

  • Enables extension to multiple data providers for real-time data
  • Provides a configurable solution to predict the health of any asset with sensor and event data (not limited to ship engine)

Benefits

   

Delivered engine and subsystem level health predictions at least 72 hours in advance

Improved accuracy of prediction models

Enabled what-if analysis in case of any change in engine parameters

Reduced inventory, maintenance and operational costs

Enhanced performance and safety of the ship