Originally Posted by
Smythe
Rather that software algorithms, has AB gone the route of additional sensors? Dynamic Pitot, sideslip, 4 AoA...
Thanks for re-posting the image (I saw it somewhere way back in a thread), which makes a lot more sense in the light of recent discussions. Clearly the A350 sensors are very different from the B737, but they probably require even more software algorithms to benefit from all the information!
A simple (2 out of 3) voting system can eliminate a single point failure, but detecting dual sensor failure (airspeed on AF447) would require computed values from the ADIRU (based on all the other types of sensors). A complex cross check against pitch from internal gyro sensors, would allow the complete state of the aircraft to be determined, and the faulty AOA data to be isolated.
Some "simplistic" calls on social media asked why the B737 AOA data can't just be compared against internal pitch information? The answer is that you need a whole A350 suite of sensors, plus a suite of redundant flight data computers, in order to make a valid comparison at all stages of the flight envelope.