AI and Predictive Aircraft Maintenance

DC – How well do you like your current airplane’s maintenance schedule? Do you feel like it could be more precise? Today’s airplanes are maintained on a schedule created by the FAA and listed in the Federal Aviation Regulations (FARs). These inspections are designed to help detect and fix maintenance issues before they become in-air emergencies. Let’s explore preventative and predictive maintenance and how AI may be able to improve maintenance and mechanical safety.

Preventative Maintenance

When you, as the pilot, are inspecting your aircraft or taking it for its annual, progressive, or 100-hour inspections, you are participating in preventative maintenance. This means that you are having your aircraft inspected for worn parts and damage before those parts fail or cause an in-flight emergency. All aircraft owners are required to get certain inspections, including preflight inspections, annual inspections, and 100-hour inspections. It’s important to note that 100-hour inspections are only required on aircraft that are used for hire. Additionally, some aircraft owners may decide to implement progressive inspections, instead of annual or 100-hour inspections, to minimize the downtime of the aircraft.

Airworthiness Directives

In addition to having all of the required aircraft inspections performed, pilots need to remain vigilant for airworthiness directives, often abbreviated to AD. The FAA puts out ADs when a component of an aircraft or system is found to be unsafe. Also included in that AD are conditions by which the aircraft must be inspected or repaired and any new limitations for the aircraft.

Predictive Maintenance

Predictive maintenance involves predicting needed repairs before the equipment fails or its operational capacity is negatively affected. This type of maintenance is often facilitated with sensors that collect data and route it into a software program where those metrics can be analyzed.

You might think of this like the sensors in your car. Modern cars have sensors that monitor tire pressure, fuel flow, air, temperature, knocking, fuel levels, and vacuum pressure. As you drive your car, your dash might tell you that you have a tire that’s low on air, meaning the tire’s air pressure fell below the desired threshold. This alerts you, as the driver, to the fact that you need to stop and get air before your tire starts wearing unevenly or goes completely flat.

Now, imagine taking those sensors to the next level with machine learning. In this scenario, your AI-enhanced sensors would be able to tell you that in 400 miles, your tires are usually low on air, so you might want to check them now before they dip below the minimum standards for air pressure. Now, let’s apply this to airplanes.

How AI May Be Able to Help With Predictive Maintenance for Airplanes

Think about your airplane. Right now, you’re having maintenance performed on a specific schedule or when you notice problems during your preflights and while operating your aircraft. What if a sensor with machine learning could alert you to future problems before they happen?

This is where AI enters the picture. AI or artificial Intelligence is simply a very large database of information that’s been tagged in a specific way. After all, AI systems are trained on existing information. Then, the AI is taught to extrapolate that data in a way that makes sense to humans.

AI-Enhanced Aircraft Maintenance

When it comes to aircraft maintenance, an AI system could be programmed with all the data for a specific aircraft. That data can include operating parameters, flight characteristics, expected useful lifespans of the various components of that aircraft, known wear and tear metrics from previous annuals and 100-hour inspections, frequency of parts replacement given various flight scenarios, and even parts failures and crash data. Using all of that data, the sensors and the AI would monitor the aircraft’s components and flight data. Then, using that data, the AI would predict when a part would most likely need to be inspected or replaced.

Sensors, AI, and Aircraft Monitoring

With the implementation of AI, certain scenarios may even be entirely avoidable, like the plug doors blowing out of the Boeing airplanes. In that scenario, AI-enhanced sensors might have been able to detect odd vibrations around the plug door and minute losses in air pressure. The AI could then alert the cockpit crew that the plug door was vibrating outside of tolerance and that it should be examined by a mechanic immediately upon landing.

Another common scenario in pressurized aircraft is metal fatigue. Currently, the metal skins of pressurized aircraft need to be inspected after so many cycles. Structural failures and equipment failures account for about 20 percent of all aircraft accidents. That percentage can range from 21 to 23 percent, depending on the source.

Now, imagine a web of sensors combined with AI to detect imperceptible changes in the metal that might mean it’s experiencing fatigue cracks. In this scenario, the AI may alert the cabin crew or maintenance team to unusual metrics in a certain panel on the aircraft. They could then be brought into the hangar, and the maintenance team could then remove that panel and inspect it for microscopic cracks before the airplane’s flight safety is ever put into jeopardy.

While it’s hard to predict what IA will do for predictive maintenance and maintenance in general for aviation, there are a few ideas, especially when it comes to Advanced Air Mobility (AAM). Advanced Aircraft Mobility refers to various flying machines that have unusual characteristics. For example, the aircraft could take off horizontally or vertically, or it could have a crew or be unmanned. You may even be able to program a flight into an AAM, and it’ll take off, fly, and land on its own.

In that scenario, there’s no crew to observe the flight characteristics of that aircraft. However, there may be operators on the ground viewing the images from the in-flight cameras and observing certain metrics, like altitude and speed. AI, in this scenario, would provide valuable information on how the aircraft handled during the flight, if there were any issues, and if any components of that aircraft need to be inspected for repairs.

All in all, AI could become very useful when it comes to predicting maintenance and improving the mechanical safety for all types of flights.