[editorial note: This article is based on materials supplied by Boeing]

Description

Airplane condition monitoring is a method to track and monitor the overall health of an aircraft, its components, and systems by analyzing data obtained from the aircraft itself, according to materials supplied by Boeing.

Airplane condition monitoring can have various approaches. The reactive approach is maintained once something happens (a fault or component failure, for example) and the data notifies the operator that something has happened. The diagnostic approach is maintained once something happens, and data is used to determine the best way to fix the problem. The prognostic approach is maintained when analyzing the data to detect patterns that predict what may be going to happen. “Data analysis would feed predictive maintenance, wherein a component would be proactively replaced before it failed, turning unscheduled maintenance into scheduled maintenance”, reports Boeing.

Continuous Parameter Logging (CPL)

In this context of aircraft data collection and analysis, continuous parameter logging (CPL) has emerged as a method for recording pre-determined parameters of data at a pre-determined rate. The principle is similar to the gathering of quick access recorder (QAR) data from the flight data recorder. As an example, CPL is in use on the B787 and B777X aircraft. “CPL can record a few thousand parameters at about 1Hz - once per second, even if this is an average as some are sampled faster and some slower. This data is recorded during flight and then off loaded once the aircraft lands for processing and analysis”, reports Boeing.

It is important to note the differences between CPL and aircraft condition monitoring system (ACMS), which works using data that is processed onboard, during flight, and then is sent off as individual values (i.e., a few data points per flight) that are then further processed and analyzed. “This approach is effective for engineering or logic-driven alerting and follows basic rules or heuristics, such as monitoring the quantity of a fluid where it is already known what data to record and how to process it”, reports Boeing.

Predictive Analysis of Data

The capabilities that big data has developed has led to the ability to store and process vast amounts of data that in turn make a greater use of data science techniques possible. “In the case of an aircraft on a 10-hour flight, for example, if a single parameter is recorded at 1 hz for the full flight this means 60 values per minute X 60 minutes per hour X 10 hours, this would yield 36,000 data points for a single parameter”, reports Boeing. “Scaling that up and assuming 3,000 parameters are being recorded by the airplane condition monitoring system, this would yield 108,000,000 data points for one 10-hour flight for one aircraft”.

With this significant data capture, data scientists have created tools to search through and analyze that data to detect and then watch for patterns or signatures that detect component or system degradation, leading to predictive alerts. “Passengers experience less disruption when predictive maintenance actions replace components before a failure occurs. Operators can reduce maintenance costs and avoid flight delays and cancellations”, reports Boeing.