Professional Documents
Culture Documents
386 Õ Vol. 125, NOVEMBER 2003 Copyright © 2003 by ASME Transactions of the ASME
冤 冥冤 冥 冤 冥冤 冥
1 0 0 0 0 0 0 0 1 0 0 0 0 0
ẋ1 x1
0 M11 M14 0 0 0 M17 ⫺K11 ⫺C11 ⫺C14 0 0 ⫺K17 ⫺C17
ẋ2 x2
0 2M14 Irot 0 0 0 0 ẋ3 0 ⫺2C41 ␥ ⫺Cd ⫺1 Cd ⫺K47 ⫺C47 x3
0 0 0 1 0 0 0 ẋ4 ⫽ 0 0 Kd 0 ⫺Kd 0 0 x4
0 0 0 0 Igen 0 0 ẋ5 0 0 Cd 1 ⫺Cd 0 0 x5
ẋ6 x6
0 0 0 0 0 1 0 0 0 0 0 0 0 1
ẋ7 x7
0 2M71 0 0 0 0 M77 ⫺2K71 ⫺2C71 ⫺C74 0 0 ⫺K77 ⫺C77
冤冥 冤冥
0 0
b ␣b
␣
⫹ 0 ␦⫹ 0 ␦w
0 0
0 0
t ␣t
冤冥
x1
x2
冋册冋 册
0 0 0 0 1 0 0 x3
y1
⫽ x4 . (2)
y2 0 0 0 0 0 1 0 x5
x6
x7
where, tive pitch. Because rotor collective pitch is the only control input
x1 ⫽ perturbed rotor symmetric flap mode displacement, in this study, the vector uគ has dimension 1⫻1, represented by
x2 ⫽ perturbed rotor symmetric flap mode velocity, perturbations in rotor collective pitch ␦.
x3 ⫽ perturbed rotor speed, The dimension of u D is 1⫻1 and is represented by perturba-
x4 ⫽ perturbed drive train torsional spring force, tions in windspeed ␦w. This disturbance is considered to contain
only one component of wind speed, the component normal to and
x5 ⫽ perturbed generator speed,
uniform across the rotor disk. The only nonzero elements in ⌫ are
x6 ⫽ perturbed tower first fore-aft mode displacement, and the second, third, and seventh rows 共for the seven-state model兲.
x7 ⫽ perturbed tower first fore-aft mode velocity. Those elements are the partial derivatives of the blade flap normal
Please note that the mass matrix in Eq. 共2兲 is not symmetric due force, rotor aerodynamic torque, and rotor thrust force with re-
to the selection of the states in this system. For example, x2 , spect to wind speed. These values reflect the influence of uniform
which is associated with rotor first symmetric flap 共which is a wind-speed fluctuations on the rotor symmetric flap mode, rotor
aerodynamic torque, and rotor thrust.
linear combination of blades one and two flap兲 共row two兲, couples
Models with fewer states can be obtained from this equation by
to the tower fore-aft motion (x7 ) through M17 while x7 共row deleting appropriate rows and columns. For the models
seven兲 couples to x2 through 2 M71 . without tower motion, only generator speed was measured, so y
Elements in A consist of various combinations of damping and was 1⫻1.
stiffness terms for the turbine. The elements of B show how the
control input enters the system. For the seven-state model shown Control Design
above, the only nonzero elements in B are the second, third, and
DAC allowed us to regulate turbine rotor speed in the presence
seventh rows. These quantities represent the partial derivatives of of wind-speed disturbances, while placing plant poles through
the blade flap normal force, rotor aerodynamic torque, and rotor full-state feedback 关14兴. It also allowed us to use state estimation
thrust force with respect to pitch angle. These values reflect the to provide the controller with values for those states that were not
capability to control the rotor symmetric flap mode, the rotor aero- measured. This is important when limited turbine measurements
dynamic torque, and tower first fore-aft mode using rotor collec- are available. A basic idea of DAC is the augmentation of the state
Fig. 2 Plot of rotor aerodynamic torque versus blade pitch for Fig. 3 Plot of rotor aerodynamic torque versus wind speed for
various wind speeds various pitch angles
Fig. 10 Plot of generator speed and blade pitch during control Fig. 11 Plot of actual and estimated wind speeds using DAC
using DAC designed from five-state model designed from five-state model
Fig. 13 Plot of tower-top fore-aft deflection during control us- Fig. 15 Plot of estimated wind speed for case with turbulent
ing DAC designed from five- and seven-State models inflow using DAC designed from five-state model at 18 mÕs