Certain things are fairly easy to detect like wheel imbalance vibration or a bad muffler sounds.
but there’s so many “vibes plus experience” things that I don’t think software will catch. The human brain is exceptionally good at picking signal out of noise, and “feeling” a bad set of tires or an old timer being able to “hear” how healthy your motor is, aren’t really things you can teach an algorithm.
I’m sure somebody will try to predict failures, but it might not go well. Surely it will be used to gouge consumers, and of course the owners of self-driving cars won’t know any better.
Could we stop saying that computers “could never do” things? It always gets proven wrong. Anything we can detect as humans has some physical reason that we can detect it. Sensors can detect it more effectively. To suggest that you can’t program a computer to know what those sensors are supposed to be reading is just absurd.
Airliner engines are getting to ludicrous reliability numbers (the latest generation appears to be closing in on 10M hours between inflight shutdowns) largely through predictive maintenance performed far in advance. We’re well past ‘most pilots never see an engine failure’ and approaching ‘most airlines don’t see an engine failure’.
And there are few locations more abusive to sensors than the hot section of a turbine engine.
Certain things are fairly easy to detect like wheel imbalance vibration or a bad muffler sounds. but there’s so many “vibes plus experience” things that I don’t think software will catch. The human brain is exceptionally good at picking signal out of noise, and “feeling” a bad set of tires or an old timer being able to “hear” how healthy your motor is, aren’t really things you can teach an algorithm.
I’m sure somebody will try to predict failures, but it might not go well. Surely it will be used to gouge consumers, and of course the owners of self-driving cars won’t know any better.
Could we stop saying that computers “could never do” things? It always gets proven wrong. Anything we can detect as humans has some physical reason that we can detect it. Sensors can detect it more effectively. To suggest that you can’t program a computer to know what those sensors are supposed to be reading is just absurd.
Airliner engines are getting to ludicrous reliability numbers (the latest generation appears to be closing in on 10M hours between inflight shutdowns) largely through predictive maintenance performed far in advance. We’re well past ‘most pilots never see an engine failure’ and approaching ‘most airlines don’t see an engine failure’.
And there are few locations more abusive to sensors than the hot section of a turbine engine.