This solution is based on a real world predictive maintenance system for a major airline. The data comes from a variety of sources including IoT data streams from aircraft engines, flight plans, weather information and logs. The solution leverages Azure HDInsight for feature engineering, Azure Machine Learning to detect operational anomalies and Azure SQL DW to enable high performance querying on the resulting petabytes of data. This report also shows the use of custom visualizations designed for the airline industry.
Operations Engineers at major airlines, responsible for a fleet of aircraft, must constantly weigh the cost and disruption of ad hoc maintenance against the risk and even higher cost of technical failures.
The report in this solution provides an overview of the fleet’s status as well as a summary of predictions for near term changes to the fleet’s technical health. The predictions are based on multiple machine learning models and use the aircrafts' Quick Access Recorder (similar to the Black Box) along with other data sources. The report shows a detailed output from one of the contributing machine learning models which predicts the remaining useful life for critical engine components.
The report is visually rich and provides an overview of the flight plans and locations to help decide where an aircraft should be serviced and which other aircraft is best positioned to replace it. The report also includes a custom Sankey chart to rationalize the fleets’ different KPI weightings based on the types of airframes, and a striking 3D heat map in the shape of a jet engine. These visuals provide insights in a manner that is intuitive to those working in the airline industry.
Visit our Partner Solution Showcase to see how they're solving problems and improving efficiency with Power BI.