The purpose of furling is to protect the mill. You're protecting it from three things:
- Burnout from overcurrent. (Resistive heating of the coils is the main concern, and it's proportional to current squared times resistance.)
- Tear-apart of the turbine from overspeed if the load can't keep its speed down in high wind.
- Tear-apart of the turbine and/or supporting structure from excessive wind loading in high wind.
Furling to reduce current will also reduce torque, letting the load keep the speed in check. And it also reduces the wind load, protecting the tower and turbine from excessive forces.
So it seems to me that, rather than trying to infer the speed (which is complicated), you should be furling to regulate (put an upper limit on) the current (whose measurement is straightforward). That directly attacks the problem and doesn't involve complexity to tune out distortions.
However, it doesn't deal with the case where the load has come disconnected, allowing the mill to run away. (Of course the offset-turbine/pivot-tail mechanical furling systems don't handle this well, either: The torque load on the shaft from the current produces most of the wind load on the turbine that resists the tail until it rises.)
With ground-controlled furling you can also deal with this by ALSO furling to limit the voltage. Then the mill will normally furl to limit the current but will also furl to limit speed if the load becomes disconnected.
Furling to limit both current and voltage measured at the ground gives you another advantage: You can set the voltage limit to prevent overcharging - eliminating the need for a dump load.
If you want to get fancy you can control the mill's furling on a function of mill current (for protection), charging current, and rectifier (battery) voltage, to optimize the battery charging and perform equalization cycles when there is enough more wind than load. B-)