Manufacturing
Manufacturing
Anomaly detection during flight tests
The company is a leading international manufacturer of helicopters. With them, we realized a data quality and anomaly detection project, aimed at identifying and removing anomalous data (spikes) recorded by the on-board sensors during prototypes' flight tests.
- Description and benefits
During the flight tests of the new prototypes, a specialized department of the company collects and monitors the data stored by the sensors and instrumentation on the aircraft. Once analyzed, this data is provided to the design engineers who thoroughly verify the helicopter's behavior in flight. During flight testing, adverse situations, like a momentary loss of data connection or other minor sensor malfunctions can occur. These inconveniences lead to the return of anomalous spikes that are very difficult to identify in the huge amount of data collected for each flight.
To distinguish and eliminate these anomalies, we have developed automatic anomalous spike identification algorithms that can handle the huge stream of data generated. With the help of a dashboard, the company's specialized operators can check the anomalous detections reported by the algorithm and decide whether to eliminate them because they are the result of a sensor error or to forward them to the designers for further investigation of the aircraft.
Big-Data analysis
Massive control and analysis of more than 30.000 signals, collected during a single flight, instead of the traditional manual control that could be performed on small samples of a hundred data at a time.
Reduced timelines
Less time is spent checking and cleaning data
Improved data quality
The remediation of anomalies due to signal spikes takes place upstream, before the flight data is passed to the design engineers
Deepening case study
A long history solved thanks to anomaly detection
Anomaly detection of spikes is a problem that our customer had been trying to solve for a long time. In order to develop the new algorithms, we also studied a multitude of documents stored in their archives. It was very exciting for us to find old documents, still written on typewriters and with hand-drawn graphs. Studying these historical sources, we realized how crucial innovation can be today to solve age-old problems whose origins are sometimes lost in time.
C.F. e P.IVA 07015910966 | Via Schiaffino 11, 20158 Milano