Power generation trends are switching from non-renewable energy to renewable energy sources in recent years due to their merits of clean sources of energy, able to be replenished quickly, sustainable, and eco-friendly. Thanks to the intermittent nature of renewable energy source, they are integrated with the diesel system to meet the continuous load demand. This combination of conventional and renewable sources without grid is called isolated hybrid power system (IHPS) [ 1– 3]. The use of IHPS with synchronous generators (SG) and induction generators (IG) for diesel and wind generators respectively are reported in many papers. The configuration may exploit the advantages of both the machines, i.e. improved power factor from the synchronous generator and low power generation cost from the induction generator [ 1– 4]. An induction generator requires reactive power for its excitation and a synchronous generator generates or absorbs reactive power depending on the excitation to some extent. A separate reactive power compensator is used to provide reactive power compensation for improving the voltage stability of the IHPS [ 5]. A static synchronous compensator (STATCOM) is used for improving the voltage response and reactive power compensation with the load and input disturbances using a proportional integral (PI) controller. The integral square error (ISE) criterion is the popular tool for controlling the outputs of compensators using PI control techniques [ 6]. It is also documented that the STATCOM can also enhance the transient stability of the induction and synchronous generators when a network disturbance occurs in the power system [ 7]. To reduce the rating of the STATCOM, the IG may simultaneously be excited by a capacitor and a STATCOM. Moreover, the use of the capacitor bank with a STATCOM provides reactive power demand for exciting the IG under steady-state and transients with reduced rating of the fixed capacitor (FC) and STATCOM. The calculation of the rating of the capacitance and STATOM is also presented [ 8]. Systems are available for voltage variation due to load disturbances for open loop consideration at load side. The importance of the load modeling is well documented [ 9]. In IHPS, systems are studied for voltage variation for open loop considered at load side normally. The interaction of loop model, i.e., the sensitivity of load in terms of the link between IHPS and the dynamics of the response of loads, is presented [ 10]. The interaction of load model, i.e., the sensitivity of load in terms of the interrelation between IHPS and the dynamics of the response of loads, is also attempted in this paper.

Recently advanced control techniques have been proposed for regulating the system voltage in IHPS. These techniques include artificial neural networks (ANN), fuzzy control, adaptive neuro-fuzzy interface system (ANFIS), vector control and so on. The key point of direct power schemes is a correct and fast estimation of the active and reactive power as well as fast PI controllers [ 11]. Due to the nonlinearities of power system, system parameters may be linearized around an operating point using a PI or a proportional integral derivative (PID) controller. The ANFIS has proven to be an excellent function approximation tool.

In this paper, an IHPS is being considered which consists of a variable speed wind generator operated by IG and a diesel generator operated by SG. The load interaction is considered by a closed loop of load reactive power demand. The system is studied by comparing the system performances based on two aspects. First, the STATCOM alone provides static and dynamic compensation. Second, the FC is used for static compensation along with the STATCOM for proving dynamic compensation. Moreover, an ANFIS based approach is presented to tune the STATCOM controller parameters for transient response voltage at load and reactive power of compensators. A performance comparison of the IHPS is presented between the PI controller based and ANFIS tuned STATCOM. The model is tested for sudden changes at two instants in reactive power load demand and for increase in power input at variable wind speed model.

The isolated wind-diesel hybrid power system is considered for the study of voltage stability considering the STATCOM for providing static as well as dynamic compensation, as shown in Fig. 1.

Fig.1

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.1 Isolated wind-diesel hybrid system with STATCOM only

The mathematical model has been developed for the system and the equations are derived for the IHPS with the STATCOM as the compensator [ 12– 14]. The STATCOM can also enhance the transient stability of induction and synchronous generators when a network disturbance occurs in the power system [ 7, 15]. In this paper, the IG excitation is provided by the fixed capacitor bank while the STATCOM is needed to regulate the load voltage for dynamic conditions. The proposed IHPS with the FC as well as the STATCOM is presented in Fig. 2.

Fig.2

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.2 Isolated wind-diesel hybrid system with STATCOM and FC

Under steady-state conditions, the reactive power balance equation for Fig. 2 is given in Eq. (1). If this hybrid system consists of a STATCOM only, as indicated in Fig. 1, the role of reactive power through FC is obsolete in Eq. (1).

Due to the disturbances, the reactive power demand will change in the IHPS, and the transients will appear in the load voltage of the system. To compensate this voltage, an incremental change in reactive power of other components will activate and try to stabilize this voltage. The net reactive power released to support Δ V voltage disturbance is given in Eq. (2).

The reactive power balanced loop for Fig. 1 which is developed without load interaction and FC [ 13, 14], including the factor of FC transfer function is given in Eq. (3).

Equations showing the deviations in the reactive power of the individual component used in the IHPS have been derived in Ref. [ 8] except the load interaction in the system and FC contribution, and their corresponding transfer functions are presented in Eqs. (4)–(6). The transfer function given in Eq. (3) includes open loop load characteristics while the load interaction in the IHPS through closed loop will be discussed separately in another paper.

The FC is supposed to fulfill the reactive power demand of IG at steady-state conditions. The capacitor is assumed to be fully charged in starting and hence it will generate reactive power at its full capacity from t = 0 s. By the relation of FC between reactive power and voltage given in Eq. (7), the changes in the reactive power of the FC will also be affected by the system voltage variation, as given in Eq. (8).

An open loop of the load model is considered in many papers for the study of voltage stability and reactive power compensation of IHPS. The block diagram of the IHPS with open loop of load reactive power demand (load RPD) is given in Fig. 3.

Fig.3

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.3 Block diagram of IHPS with open loop of load RPD

The concept of interaction of the load is defined as when the load RPD changes, the voltage varies and the variation in voltage further changes the reactive power demand of load [ 16, 17]. In general, many factors may contribute to the inaccuracy of predictions. The load interaction with the closed loop shows the improvement in the system dynamics. The controller gains K_P and K_I increase with closed loop load model. The first swing amplitude of voltage goes down while its settling time reduces [ 10]. The block diagram of the IHPS with closed loop of load RPD is represented in Fig. 4.

Fig.4

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.4 Block diagram of IHPS with closed loop of load RPD

The linear model for load feedback gives the relation between reactive power and voltage derived with the help of load model and its characteristics [ 9, 10], as presented by Eq. (9).

Both the actual ΔQLA of load disturbance of the system and the feedback ΔQLF of load disturbance provide the net load reactive power ΔQL, as presented in Eq. (10).

In the IHPS, the use of the STATCOM and the exciter of SG in case of dynamic compensation with high gains contribute to oscillatory instability in the power system. This type of instability is characterized by low frequency oscillations which can persist or even grow in magnitude. The use of the conventional controllers such as the P and PI controllers provides liberalized model of the power system around a nominal operating point. The operating condition does change as a result of load variations and major disturbances, making the dynamic behavior of the power system become different, at different operating points. Thus, if the parameters of the controllers are kept fixed, the STATCOM performance will degrade whenever the operating point changes due to sudden load changes [ 18].

Fuzzy logic controllers (FLCs) and ANN controllers (ANNCs), used as power system stabilizers, have been developed and tested in Ref. [ 14]. Unlike other classical control methods, FLCs and ANNCs are model-free controllers; i.e. they do not require an exact mathematical model of the controlled system. Moreover, the speed and robustness of these controllers are the most significant properties in comparison to the other classical schemes. But these controllers also have some disadvantages. There are no practical systematic procedures for the fuzzy PSS (FPSS) design, so the rules and the membership functions of the controller are tuned subjectively, making the design laborious and a time-consuming task. ANNCs have the capability of learning and adaptation, but they work like a “black-box” and it is difficult to understand the behavior of the network. The ANFIS combines the advantages of FLCs and ANNCs, avoiding their problems on the other hand[ 19, 20].

In this paper, the STATCOM is tuned by an ANFIS trained from the input-output voltage data obtained by a conventional PI controller. The voltage is stabilized for PI controllers used in the STATCOM controlling blocks first using ISE criterion. Then this data is trained by the ANFIS. The flowchart showing the algorithm steps followed to obtain the STATCOM constants using ANFIS based training is demonstrated in Fig. 5. For initializing the fuzzy system, a FIS file is trained with 100 epochs for the study. The number of MFs for the input variables is 15. The gbellmf membership function used for input variables and output membership function type is a linear type.

The algorithm for the tuning of the ANFIS based STATCOM can be described as:

1) Tuning of PI controller constants K_P and K_I for the STATCOM through ISE criterion.

2) Each tuning provides a voltage structure with time, which is chosen as an input data for tuning.

3) Store this data for genfis and ANFIS.

4) Initialize the fuzzy system and define number of iterations (epochs), membership function, tolerance (error), and optimum method.

5) Start the learning process by commanding ANFIS.

6) The output is achieved after suggested iteration and the tolerance is achieved.

Fig.5

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.5 Flowchart showing steps for ANFIS based tuning of STATCOM

The results for reactive power compensation of wind-diesel based IHPS with load interaction model and ANFIS tuned STATCOM controller has been obtained by developing SIMULINK model in MATLAB. The STATCOM output is controlled for voltage stability by the training process of ANFIS on MATLAB 7.10 editor window. A program is developed for getting the optimum values of the PI controllers K_P and K_I mathematically first using ISE criterion concept. Then this data is trained using the syntax of ANFIS programming in the main program as depicted in Fig. 5. The Simulink block diagram is designed for the complete IHPS with all the components expressed above in Eqs. (2) to (10). The overview of the Simulink block diagram developed in MATLAB 7.10 is presented in Fig. 6.

Fig.6

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.6 Simulink block diagram of IHPS

The IHPS includes wind-diesel (variable speed/slip model) power generating units. Diesel operated SG uses the IEEE type-I excitation system. The MATLAB Simulink model is designed by combining the transfer function of each component used in the IHPS according to Eq. (3). The transfer function of SG and IG are given in Eqs. (4) and (5) respectively. The steady-state reactive power demands of the wind operated IG is fulfilled by the FC while the PI controlled STATCOM is also supplying reactive power as a dynamic compensation required for the system. Their corresponding transfer functions are presented by Eqs. (6) and (8) respectively. The closed loop load model is also included for the load interaction in the system as discussed in Fig. 4 and in the corresponding transfer function in Eq. (9). The simulation time for analyzing the system response is 0.05 s. This model is tested for a 1% step increase in reactive power load demand at t = 0 s and then a sudden change of 3% from the 1% at t = 0.01 s at a 1% step increase in power input at variable wind speed model.

Figure 6 shows the reactive power load disturbances for the system given at t = 0 s and again at t = 0.01 s. The corresponding reactive power demand of load changes from 0.01 pu to 0.03 pu. The small ripples in Fig. 7 indicate the role of load feedback loop.

Fig.7

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.7 Disturbance in load demand for system study

The load and input responses of the IHPS based on the block diagram (given in Fig. 2) are studied. The IHPS used in this paper presents the comparison of the voltage stability at load and reactive power changes of the IHPS components, i.e. STATCOM, SG and IG.

The data used for designing the IHPS is given in Appendix. The system is modeled in MATLAB 7.10, as shown in Fig. 6. Figures 8 to 11 represent the behavior of the system under transients. Both static and dynamic type compensation are considered with load interaction model for Figs. 8 to 11.

With the PI controller, the settling time and first peak overshoot of the voltage stability and reactive power variation of each component used in the IHPS has considerably been reduced after t = 0.02 s but the ripples have not vanished yet. Presence of ripples are quite obvious because of the presence of the FC with the STATCOM and closed load model and sudden change in load demand at two instants of time i.e. at t = 0 s and 0.01 s. As the operating point changes frequently, the STATCOM performance is degrading due to its fixed parameters evaluated by the ISE criterion technique. Hence, the use of the conventional PI controllers in such an IHPS is not suitable. The soft computing technique ANFIS presents a better result, as shown in Figs. 8 to 11. It has also been investigated that the characteristics of voltage variation and reactive power variation will remain almost the same in the condition in which only a STATCOM is used. However, the use of the FC with the STATCOM reduces the overall magnitude of the reactive power generated by the STATCOM.

Fig.8

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.8 Δ V deviation with closed loop load model and STATCOM+ FC in IHPS

Fig.9

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.9 Δ QSG deviation with closed loop load model and STATCOM+ FC in IHPS

Fig.10

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.10 Δ QST deviation with closed loop load model and STATCOM+ FC in IHPS

Fig.11

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.11 Δ QIG deviation with closed loop load model and STATCOM+ FC in IHPS

As the IHPS considered in this paper involves a complex hybrid system which includes additional components, i.e. FC block Δ Q_FC, load interaction block Δ Q_LF and a sudden load increase at two instants of time, i.e. at t = 0 s and 0.01s, as shown in Fig. 7. Figure 11 presents that the 1% step change in power input at variable speed model of IG increases the IG reactive power demand which is supplied by the STATCOM. Hence the net reactive power drawn from the STATCOM, as shown in Fig. 10, is the sum of demand increased at load (Fig. 7) and IG (Fig. 11). Figures 8 to 11 depict the improperly damped system transient responses with PI controller and therefore, the ANFIS tuned STATCOM provides better results for the system.

It can be observed from Figs. 12 to 15 that at t = 0.035 s, the values of Δ V, Δ Q_SG, Δ Q_ST and Δ Q_IGare almost stabilized. This illustrates that the load disturbances at t = 0.01 s will change the system dynamics but this hybrid system will regain its original position in 0.25 s at time t = 0.35 s, and now the system will reach its steady-state condition again. Figures 12 to 15 exhibit the comparative performances of the IHPS with the STATCOM only and both the STATCOM and FC, based on ANFIS tuned control parameters. Figures 12, 13 and 15 present the variation of voltage and reactive power of SG and IG respectively. In Fig. 12, the voltage variation with STATCOM+ FC is almost zero and lower compared to with the STATCOM only. In Fig. 13, the SG supplies less power to the IHPS with STATCOM+ FC combination. So a better performance of the SG may be achieved toward active power generated by the SG. In Fig. 15, the IG reactive power demand stabilizes fast with FC+ STATCOM after load variation at t = 0.01 s. Figure 14 shows the behavior of the reactive power generated by the STATCOM which is almost the same in both cases. This means that the use of both compensating devices (STATCOM+ FC) improves the voltage stability, transient response and performances of the IG and SG in the IHPS.

Fig.12

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.12 Comparison of transient response of voltage variation for two options in IHPS

Fig.13

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.13 Comparison of transient response of SG reactive power variation deviation for two options in IHPS

Fig.14

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.14 Comparison of transient response of STATCOM reactive power variation deviation for two options in IHPS

Fig.15

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.15 Comparison of transient response of IG reactive power variation deviation for two options in IHPS

In this paper, an IHPS is considered for the study of reactive power compensation and voltage stability problem. The proposed use of the FC with a STATCOM gives better options of reactive power compensation in hybrid power system. The system is studied considering the load interaction, static compensation by the FC, and dynamic compensation STATCOM simultaneously with wind-diesel IHPS. Transient responses are studied with frequent changes of load demand at two instants of time and input power through variable speed IG. The importance, advantage and necessity of soft computing controller ANFIS over a conventional PI controller are also demonstrated and discussed with the results obtained. In the conventional controllers, the parameters of the controllers are kept constant, so the STATCOM performance degrades whenever the operating point changes due to sudden load changes. But the ANFIS tuned STATCOM provides better results for the IHPS during dynamic reactive power compensation.