UAV Predictive failure/maintaince possible?

[mr_westie]

Didn't get much love at rcg so I though the bright sparks here might be able to comment
http://www.rcgroups.com/forums/showthread.php?t=1750660

Applicable to both fixed wing and multirotors.

Given enough info can we predict the failure of a component/system?

Computer hard drives have SMART.
You car has the Engine light.

Not sure what the higher consumer grade UAV have.
MicroDrones etc.

Prop RPM
Motor Vibration
Motor RPM
Motor Temp
Motor Voltage/Current
ESC Temp
ESC Voltage/Current
Ambient Temp/Humidity
Battery Temp
Frame Vibration

Given a base line we should be able to see a component go bad.


Motor bearings,
.increased vibration
.more power for a given rpm
.increased heat for a given watts input

Prop slipping, motor/prop rpm not matching.

ESC,
.more heat for a given watts input
.less rpm for a given watts input
Bad battery,
.more heat for a given watts output
.more voltage drop for a given watts output
Frame,
.combined motor vibration not matching frame vibration
Others?

How much would such a system cost?
The software would be the hard bit, I like open source

Cost of the sensors?
$20 per motor
$15 for the frame

It could be built into the ESC.
This would be a good candidate
http://autoquad.org/esc32/


Possible? Worth it?

 

[jes1111]

Some of the high-end systems do have some of these kinds of things. Micropilot even has a triple-redundancy system for the flight controller itself. I think it's fairly obvious that it all costs money, adds weight, needs a two-way radio link and, importantly, assumes components about which you have failure prediction data. So at the hobby/AP end of the market, it could only really be implemented effectively in a complete RTF system from a single manufacturer. Tricky.