Boeing is working on a system called “component fault prediction” to determine if their planes need repairs. This system uses a prediction model that analyzes sensor data collected during flight, such as speed, temperature, and altitude. By filtering out correlated data, the model focuses on non-correlated information to make accurate predictions with less computational power.
When the system determines that a component needs to be replaced, it can send the prediction to a maintenance system, which then instructs autonomous robots or systems to perform the replacement work. In the event of a mid-flight prediction, the system may redirect the aircraft to the nearest airport that has the necessary component.
While the patent focuses on component replacement, the prediction model itself could be applied in various scenarios. It reduces the amount of data needed to arrive at a solution, making it useful for handling large, complex datasets.
Although Boeing’s application is narrow, other researchers have explored similar methods. However, Boeing’s focused approach may increase their chances of securing the patent.
Implementing this technology could save Boeing time and resources. The patent claims that it can make predictions 10 times faster than other sorting algorithms, enabling proactive actions to prevent breakdowns. Moreover, this technology does not require specialized knowledge to interpret the data.
Like any AI model, there are risks. False positives and false negatives may occur if the data leads to incorrect conclusions. It is important for Boeing, as a major provider of commercial jets, to ensure the accuracy of their system and have appropriate fail-safes in place.