Maximum Output Power Tracking of Wind Turbine Using Intelligent Control

The Output power of wind turbine is determined by wind speed. The Output power can be adjusted by controlling the generator speed and pitch angle of wind turbine. When the wind speed below the wind turbine rated, the output power of generator can be maximized by controlling the generator speed at point of maximum power coefficient. When the wind speed above the wind turbine rated, output power of wind turbine will exceed the power generators rated. In this condition, the output power of wind turbine needs to be regulated to conform to the generator power rate. Output power of wind turbine can be regulated by adjusting the pitch angle of wind turbine. In this paper is developed the control strategies based intelligent control for controlling the generator speed and pitch angle of wind turbine, so the maximum output power tracking (MOPT) of wind turbine can be obtained at any wind speed variations. Generator speed is controlled using PI Fuzzy Logic Controller (PI-FLC) based Direct Field Oriented Control (DFOC). Pitch angle of wind turbine is controlled using Elman Recurrent Neural Network (RENN). The simulation results with Matlab Simulink shows that the both controller was successfully regulates the output power when the wind speed above the wind turbine rated and the output power can be maximum when the wind speed below the wind turbine rated.


Introduction
Wind turbine operation can be divided in two different operation regions.The first region describes wind turbine operation during below the wind speed rated.In this region, the wind power is lower than the rated power output of wind turbine generator.The second operation region describes wind turbine operation during above the wind speed rated winds.In this region wind power is greater than the rated power output of wind turbine generator.In the first region, the main task of control system is to maximize the power output of wind turbine by maximizing wind power capture.Variable speed wind turbine has been developed to maximizing wind power capture.There have been many control strategies are applied to variable speed wind turbine.Variable speed wind generation system based on vector control has been developed in [1][2][3][4][5][6][7][8][9].Vector control based on intelligent control has been developed in [10][11][12][13][14].In [1][2][3][4][5][6][7][8][9][10][11][12][13][14], the wind turbine is operated with pitch angle fixed.It causes the wind turbine can not be operated on a second region.For wind turbine operation in the second region, has developed a variable speed wind turbine with pitch angle-regulated [15][16][17].Several control strategies have been applied, among others, robust digital control [15], inteligent control based on RBFNN [16], and improved Elma Neural Network based Modified Particle Swarm Optimization [17].
In this paper, a MOPT strategy of WECS is developed to both operation regions, so the wind turbine can be operated for above the wind speed rated.The Maximum output power is obtained by adjusting the generator speed and pitch angle of wind turbines.Generator speed is controlled using PI-FLC based on DFOC method.Pitch angle of wind turbine is controlled using RENN.Control strategy is shown in Figure 1.In zone 1, the wind speed is lower than the wind speed rated (v 1 ≤ v ≤ v 2 ), while in zone 2 the wind speed above the wind speed rated of wind turbine (v 2 ≤ v ≤ v 3 ).For zone 1, the generator speed ω m is controlled at maximum power coefficient C p_max with pitch angle β constan, such that the maximum output power can be obtained at each wind speed variations.For zone 2, the generator speed ω m was maintained constant and pitch angle β is controlled to lower the power coefficient, so that the output power of wind turbine in accordance with the rating of maximum power the generator.

219
DC supply is used for initial operation and supplying the generator when the WECS in the cut-in condition.For generator excitation, the capacitor mounted on the DC side of the converter.Model details are described in the following sections.

Squirrel Cage Induction Generator (SCIG)
In DFOC method, SCIG is modeled in dq axis.General equation of SCIG voltages in dq axis are described as [19][20] where v dqs and v dqr are stator and rotor voltages in dq axis, respectively.i dqs and i dqr are stator and rotor currents in dq axis, respectively.R s and R r are stator and rotor resitance, L s and L r are stator and rotor leakage inductance, respectively.Torque and power of SCIG are given by [19] : ( ) where P e , P m and T m are output power of SCIG, Mechanical power of wind turbine and mechanical torque of wind turbine, respectively.J, D and T e are moment inertia, damping of wind turbine and electromagnetic torque of SCIG, respectively.

Wind Turbine
Wind turbine model is developed based on mechanical power equation of wind turbine.
Mechanical power of wind turbine P m can be written as [18]: where ρ, λ and R are air density, Tip speed ratio (TSR) and blade radius of wind turbine, respectively.Power coefficient is determined by TSR and the pitch angle of wind turbine.The relationship of power coefficient, TSR and pitch angle of wind turbine C p (λ,β) can be expressed as [18] : where c 1 -c 6 are constants value of wind turbine.TSR is ratio of the angular velocity of wind turbine with wind speed, which is formulated by [10]: From equations ( 5)-( 8) can be concluded that the mechanical power can be regulated by controlling the generator speed and the pitch angle of wind turbine.Mechanical power of wind turbine varies on each change of wind speed and has a maximum point at the particular generator speed [18].Maximum mechanical power P m_max is obtained at the point of maximum power coefficient C p_max , as shown in Figure 3.

Generator Speed Control
The generator speed is controlled by DFOC method.In this method, the generator speed is controlled by adjusting the torque component and field component.In DFOC method, rotor angle θ e is calculated from the stator current feedback that has been converted into the dq axis, that defined as [20]: ( ) where ω sl and r ψ are slip frequency and rotor flux, respectively.Output controller DFOC are stator current references in dq axis (i ds * and i qs * ), which can be written as: where L m , p and are magnetizing inductance and number of pole, respectively.In this paper, i qs * is controlled using PI-FLC, whereas the i ds * is made constant, as illustrated in Figure 2. The generator speed control scheme is shown in Figure 4.This controller consists of two FLC, namely FLC1 and FLC2.FLC1 is used to tracking the generator speed reference for maximum output power.For zone 1, the speed reference of generator gives a chance varies according to wind speed variations.For zone 2, speed reference of generator is made constant at each wind speed variations, as illustrated in Figure 1.FLC2 is used to control the generator speed to fit the reference speed obtained from FLC1, so that the generator working at maximum output power point.FLC2 made in the PI-FLC scheme.In PI-FLC, the gain factor K i is used for gain the error input and K p for gain the change of error [20][21].Figure 5 shows the input and output membership function of FLC1 and FLC2.( ) where m and θ are magnitude and phase angle of the voltage.SVPWM will modulate the gate of converter switch based on m and θ.

Pitch Angle Control Using Recurrent Elman Neural Network (RENN)
Pitch angle of wind turbine is controlled to remain constant at the maximum power coefficient for zone 1 and varies for zone 2. The variation of pitch angle in zone 2 serves to reduce the power coefficient, so the mechanical power of wind turbine remain under power rating of generator at high wind speed.The pitch angle of wind turbine is controlled using where (k) represents the k th iteration, i i is input values of the input layer for k th iteration.Output the context layer O r can be written as [23]: with α is gain feedback in the context layer.The output of hidden layer O j is activated by tansig transfer function, which is formulated by : ( ) 1, 2,..., 5 r = and 1, 2,..., 5 j = where W ij and W rj are the weight of input neuron to hidden neouron and weight of context neuron to hidden neuron, respectively.The output layer RENN is activated by linear transfer function, so that the β can be written as: with W jo is the weight of hidden neuron to output neuron.RENN trained with supervised learning method.In learning process, the W ij , W rj and W jo is updated steepest descent algorithm [22], can be written as: ( 1) ( ) , , , ( ) ( ) where δ o is error term to be propagated, which is given by: ( ) ( ) where E, P m and P rate_max are error in learning process, mechanical power of wind turbine and maximum power rate of generator, respectively.The pitch angle of wind turbine β is trained with wind speed 5 m/s -20 m/s. Figure 7 shows the performance of training.RENN training reach the error goal in epochs 337 with training error 2.967e-6.

Results and Analysis
The WECS model has been designed as shown in Figure 2 is simulated using Matlab.Design parameter of WECS is shown in Table 1.The first simulation is training the pitch angle β using RENN.The Figure 8 shows the pitch angle β from RENN training result.The power coefficient graph in Figure 8 is obtained from the calculation using the equation ( 6)-( 8).The highest power coefficient found on the pitch angle 0 with C p_max 0.48.For Figure 8, the wind turbine is operated with a reference speed at the C p_max point for zone 1 and a constant 1000 rpm for zone 2. For zone 2, the pitch angle grows large when the wind speed increases.From the Figure 8 can be concluded that the increase in pitch angle of wind turbine will reduce the power coefficient.
The second simulation, the generator is operated in zone 1 with varying wind speed 6 m/s, 8 m/s and 7 m/s, respectively.Figure 9 shows the results of simulation.The generator speed graph shows validility of the PI-FLC.It can be seen from the actual speed can follow the reference speed.For wind speed 6 m/s, PI-FLC just gives error 1.8 rpm at transient condition and ± 0.2 rpm at steady condition, as shown in Figure 11.The power coefficient graph shows that the values can be constant at maximum point 0.48 in each wind speed variations.Change the power coefficient value occurs only during transient changes in wind speed.The RENN controller gives the pitch angle remains constant 0 with error ± 0.00004º at every variation of wind speed.
Figure 10 shows the simulation results of zone 2 with winds speed 15 m/s, 17 m/s and 19 m/s, respectively.The generator speed graph shows the speed remains constant 1000 rpm on every variation of wind speed.Error of generator speed is shown in Figure 11.The output power graph shows the output power constant at rated generator on every variation of wind speed.It shows the validity of RENN to control the pitch angle of wind turbine.When the wind speed increases, the pitch angle also increases and the power coefficient decrease.It is gives the output power remain constant at maximum power of generator.
Figure 11 shows the error of generator speed for zone 1 and zone 2. For wind speed 6 m/s, the error is 1.8 rpm at transient condition and ± 0.2 rpm at steady state condition.For wind speed 15 m/s, the error is 1.3 rpm at transient condition and ± 0.1 rpm at steady state condition.It is shows that PI-FLC is valid for both operation regions.
This result is better compared with the results in [15].In [15], the ripple of generator speed ± 0.7 rpm with the ripple of pitch angle ± 0.4 at wind speed 15 m/s.

Conclusion
Control strategy of MOPT for WECS is developed in this paper.The maximum output power is obtained by controlling the generator speed and pitch angle of wind turbine.The generator speed is controlled using PI-FLC and pitch angle is controlled using RENN.The both controller is valid for two operation regions of wind turbine.It was shown by the simulation results.For zone 1, the output power of generator can be maximized at maximum power coefficient point in each wind speed variations.Speed control using PI-FLC based DFOC method gives the actual speed of generator can follow the reference speed with error 1.8 rpm at transient condition and 0.2 rpm at steady state condition.Pitch angle control of wind turbine using RENN just gives error 0.00004º in each wind speed variations.The same result was also shown by the simulation result for zone 2.

Figure 3 .
Figure 3. Power characteristics of wind turbines

TELKOMNIKA
ISSN: 1693-6930 Maximum Output Power Tracking of Wind Turbine Using Intelligent Control (Muldi Yuhendri) 223 where ∆W x is change of the weight, can be formulated by:

Figure 7 .
Figure 7. Performance of RNN training

Figure 11 .
Figure 11.Error of generator speed

Table 1 .
Design Parameter of WECS