Re: Least squares curve fitting


In article <43a376f2$1@xxxxxxxxxxxxxxxxxxxxxx>,
Adrian Jansen <adrian@xxxxxxxxx> writes:
>Peter Spellucci wrote:
>
>>
>> I understood your problem as computing (and possible predicting?)
>> the flight curve from some measured data. but what please is a wind vector?
>> do you want do compute wind direction and speed from the deviation of the
>> flight curve from a predicted one? but this would require some more information
>> e.g. concerning acting forces.
>> hth
>> peter
>
>Yes, we have the data for the flight curve, referenced to the ground.
>What we want to find is the wind vector, ie the speed and direction of
>the airmass in which the aircaft flies, also referenced to ground. We
>assume ( or can pick ) areas of the flight where the acting forces are
>not large - ie areas where we dont make significant pilot control
>movements, or change engine power settings. For instance one of the very
>common conditions is to fly at a constant bank angle and speed
>referenced to the local airmass. So we describe a ( close approximation
>to ) circle in the air, and that translates to a cycloidal pattern on
>the ground, which we can measure from the ground speeds and coordinates
>in real time. From the ground speed data alone, we want to be able to
>get a good estimate of the wind.
>

you could compute the difference of
expected (projected) flight curve with no wind from the measured one and then
use least squares estimation of the wind vector, using the relevant physics
(of which i know nothing)
hth
peter
.

Relevant Pages

  • Re: Least squares curve fitting
    ... the flight curve from some measured data. ... What we want to find is the wind vector, ie the speed and direction of the airmass in which the aircaft flies, also referenced to ground. ... So we describe a circle in the air, and that translates to a cycloidal pattern on the ground, which we can measure from the ground speeds and coordinates in real time. ...
    (sci.math.num-analysis)
  • Re: Least squares curve fitting
    ... >any data set which includes at least some part of a curve, ... I understood your problem as computing (and possible predicting?) ... the flight curve from some measured data. ... but what please is a wind vector? ...
    (sci.math.num-analysis)
Quantcast