Unscheduled aircraft maintenance accounts for 88% of the $33 billion annual bill the sector faces, according to solutions provider Veryon.
Kris Volrath, chief product officer, told the first PAM MENA conference being held in Dubai this week, said addressing this requires a prescriptive approach to maintenance.
He said this allows carrier to identify repeat problems based on global data and apply a fix that works first time through automated workflows.
“We have a lot of data,” he said. “We are seeing around 100 million maintenance events where we can track what was actually recorded and how did we solve that problem.
“One of the benefits of working across the global fleet is that we can see problems manifest themselves in different places every day. Often people are working from different data sets.”
Volrath said many airlines are returning parts to OEM’s believing them to be faulty but the supplier finds no fault only for the part to be reinstalled and the problem to recur.
“Our focus is on how to diagnose the problem and then prescribe the right solution. Our focus is on reactive, prescriptive analytics. We are sitting in 10 to 15 years of analytics.”
Veryon’s AIRE ‘back bone’ technology uses natural language processing and machine learning to identify “clustering of defects” in the data,” added Volrath.
“We are looking for patterns to cluster similar events. Linking these events together reveals that in fact they are the same problem happening repeatedly and not being fixed the first time.”
Once the issue has been identified Veryon works to “close the loop” by suggesting the right next step to resolve it at the first time of asking.
Volrath said that enables every technician to learn from what’s happening across the global fleet of aircraft and what is the most effective first-time resolution.
“We can help small providers compete with large providers. Software providers can help to level the playing field. AI is levelling the playing field. What we can level the playing field in it data – organised, trusted, cleaned data.”
Artificial Intelligence will be used by today’s workforce, said Volrath, but if hallucinations can’t be spotted it is worthless.
“What we really care about is that you have this problem with the aircraft and what’s the best way to solve it.”