Environmental Engineering Reference
In-Depth Information
Wind
Wind Turbine
Measured
wind speed
+
+
+
-
T-S PMIO based
T-S controller
Fig. 7.11
Wind turbine PMIO based sensor FTC scheme
unavoidable spikes that specifically affect the drive train torsion of low inertia
wind turbines. Also the performance of the proposed FTC strategy is highly
affected by the robustness and the computation time of the residual evaluation unit.
Moreover, the T-S model in this reference is derived based on measured wind
speed which in turn causes clear modelling uncertainty since the wind varies
stochastically and faster than wind turbine dynamic and hence it cannot schedule
the controller commands appropriately. Furthermore, there is a significant prob-
ability of simultaneous occurrence of generator and rotor speed sensor faults.
Within the framework of the proposed strategies, the use of wind speed and
rotor rotational speed as scheduling variables will ensure that the T-S fuzzy model
can represents a wide range of operation scenario. Specifically, the model can
cover cases in which the system operates away from the ideal power/wind speed
characteristic shown in Fig.
7.4
. In fact, large inertial wind turbines frequently
operate away from their ideal power/wind characteristics and hence the use of two
scheduling variables is the best approach to handle this challenge.
The main contributions involved in the proposed strategy are: (1) the use of
the PMIO to hide or implicitly compensate the effect of drive train sensor faults.
This obviates the need for residual evaluation and observer switching (see [
21
] for
example). (2) The PMIO simultaneously estimates the states and the sensor fault
signals. Hence, information about the fault severity can also be provided through
the fault estimation signals. (3) The fuzzy PMIO scheme is shown to give good
simultaneous state and abrupt sensor fault estimate. Figure
7.11
schematically
illustrates the proposed strategy.
In this strategy, the controller forces the generator rotational speed to follow the
optimal generator speed. Additionally, this strategy makes use of the measured
wind speed as an approximation of the EWS.
As derived in Sect.
1.2
, the T-S fuzzy model of the non-linear wind turbine
system given in Eq. (
7.20
) with additive sensor can be expressed as:
