METHOD FOR CONTROLLING A GAS TURBINE POWER PLANT AND GAS TURBINE POWER PLANT

Description

TECHNICAL FIELD

[0001] The present invention relates to a method of controlling a gas turbine power plant and to a gas turbine power plant.BACKGROUND

[0002] As is known, gas turbine electric power plants normally comprise a gas turbine engine as a drive unit, a generator and a main breaker to an electric power grid. The gas turbine engine in turn includes a compressor with orientable inlet guide vanes, a combustor assembly and a turbine as an expansion section. The compressor, the turbine and the generator are all mechanically connected to one and the same shaft. Gas turbine power plants are also equipped with control systems configured to perform several functions intended to ensure correct operation, and which also ensure compliance with increasingly strict regulations defining system requirements in terms of safety, stability, and the ability to respond to variations in grid power demand.

[0003] More specifically, a control system is configured at least to: adjust gas turbine engine speed at no-load conditions, i.e. when the generator is disconnected from the grid (for example, during start up transients); check and adjust the electrical load supplied to the network, and participate in grid frequency control, if required; check and adjust turbine exhaust temperature, acting on inlet guide vanes of the compressor to modify the inlet airflow and maintain a desired temperature level; carry out all the auxiliary machine protection functions.

[0004] Protection functions may include, for example: limiting high exhaust temperatures by reducing the load: in fact, the "fuel controller" reduces the load, if the temperature increases in order to exceed the limit value; limiting the load in accordance with the compression ratio, by reducing fuel supply if the compression ratio exceeds a threshold value; limiting maximum load, which may be necessary for machines that must never exceed certain delivered power.

[0005] The above functions of power plant control systems are mostly achieved by controlling the fuel supply through fuel control valves, except for exhaust temperature control based on intake airflow and adjustment of the compressor inlet guide vanes.

[0006] In known power plants, functions of control systems are often performed by respective PID (Proportional-Integral-Derivative) controllers. Moreover, since several controllers concur on the same actuator, i.e. the fuel feed valve, selection devices are provided to avoid conflict based on programmed rules. A switch selectively firstly enables electric load control and no-load speed adjustment, depending on whether the generator is connected or not to the grid (i.e. depending on the main breaker setting), and then a minimum-selection port assigns fuel feed valve access to the controller involving the least control effort (i.e. supplying the control signal with the lowest module).

[0007] Known control systems, however, suffer from some limitations. A first problem lies in calibrating the PID controllers, which, given the numerous parameters involved and the complex nature of gas turbine engines and power plants, on the one side is invariably difficult and hardly repeatable, even when dealing with similar systems, and on the other side is sensitive to parameter variations and ageing. As a result, start-up of gas turbine engines is a long, painstaking job, and the sensitivity to variations in machine parameters makes repeated adjustments and recalibrations necessary, with a corresponding increase in operating costs. Other drawbacks have to do with performance, and are caused, in particular, by the integral action effect of the minimum-selected PI controllers. That is, since the non-selected controllers are unable to act on the respective controlled variables to reduce errors, the contribution of the integral term tends to rapidly saturate the control signal. Such a situation is obviously to be avoided, as it would cause unacceptable delays and would prevent timely intervention of the excluded controller when requried. Anti-wind-up techniques have therefore been proposed, which basically provide for artificially forcing feedback to keep the out-of-control signals at a slightly higher value than the competing signal at the minimum-selection port (so-called tracking mode). In this case, however, the non-selected controller tends to intervene when not strictly necessary, when the competing controller attempts to track a rapid variation in the respective controlled variable, thus also resulting in a significant impairment in performance. In fact, conventional systems fail to react promptly, especially in the case of sudden variations in load, and often exhibit undesired oscillations incompatible with the increasingly strict stability requirements of recent regulations. More specifically, systems are required to assist in grid control, and to withstand sudden switching from nominal conditions, in which the grid is virtually able to absorb infinite loads, to so-called "load island" conditions, in which a section of the grid is isolated, e.g. due to a fault, and only absorbs a limited load.

[0008] As an alternative solution, use of sliding mode control techniques has been proposed. In fact, sliding mode control allows simpler calibration procedures and is less sensitive to variations in plant parameters. Moreover, due to different performance, also concurrent controllers on the same actuator through a minimum port are less critical. Also standard sliding mode control techniques are not free from limitations, however. In particular, sliding mode control is based on a control law that includes a gain factor and a switching function. Determining the gain factor is a critical step in sliding mode control design, because too low gain values may result in not effective control action, whereas too high values may cause chattering, i.e. high frequency oscillations that often affect lifetime of mechanical components.

[0009] EP 1 679 563 A1 discloses a device, for controlling an electric power generating system having a gas turbine, a compressor, a combustion chamber, and an electric power generator connected mechanically to the gas turbine, includes regulators supplying a first control signal for controlling fuel supply to the combustion chamber, and a second control signal for controlling air intake by the compressor. The regulators are configured to generate at least one of the first and second control signals in sliding-mode control manner.

[0010] Other examples of known plants using sliding mode control are disclosed in EP 2 067 932 A2 and in DAKHILI KHATERE ET AL, "Improving the Performance of a Small-Scale CHP Plant using Fuzzy Sliding Mode Controller", 2019 7TH INTERNATIONAL CONFERENCE ON ROBOTICS AND MECHATRONICS (ICROM), IEEE, (20191120), doi:10.1109 / ICROM48714.2019.9071912, pages 577 - 582.SUMMARY OF THE INVENTION

[0011] It is an aim of the present invention to provide a method of controlling a gas turbine power plant and a gas turbine power plant that allow to overcome or at least attenuate the above described limitations.

[0012] According to the present invention there is provided a method according to claim 1 of controlling a gas turbine power plant including a gas turbine engine, and an electric power generator mechanically connected to the gas turbine engine, the gas turbine engine comprising a compressor, a combustor assembly and a turbine; the method comprising determining at least one of a fuel supply and an air supply to the combustor assembly to control at least one of a load and an exhaust temperature of the gas turbine engine in accordance with a sliding-mode control technique; wherein controlling comprises determining a sliding quantity, as a function of a control error between a current value and a reference value of a controlled variable, the controlled variable including at least one of a load and an exhaust temperature of the gas turbine engine, and determining a control law, having a gain factor and a sliding factor and configured to bring the sliding quantity on a sliding manifold defined by states that cause the sliding quantity to be zero; wherein the sliding factor is a first function of the sliding quantity and a time derivative of the gain factor is a second function of the sliding quantity; and wherein the time derivative of the gain factor is negative in a first region around the sliding manifold, is zero in a second region that surrounds the first region and is positive in a third region that surrounds the second region.

[0013] Making the time derivative of the gain factor dependent on the sliding quantity allows to dynamically adjust the gain factor based on the distance of the sliding quantity from the sliding manifold, thus allowing to balance requirements for responsiveness and the need to avoid or reduce chattering, to preserve actuators of the fuel valve and inlet guide vanes. Flexibility of the control action may be thus increased without affecting reliability.

[0014] The gain factor increases as long as the sliding quantity is in the third region, i.e. away from the sliding manifold, and the strength of the control action determined by the control law grows accordingly. The control action is thus effective and tends to rapidly bring the sliding quantity toward the sliding manifold, without substantial risk of causing chattering. The gain factor is constant when the sliding quantity is in the second region and decreases when the sliding quantity remains in the first region. Hence, the strength of the control action is reduced to a minimum when the sliding quantity is close to the sliding manifold and chattering is consequently attenuated. Moreover, the adjustment of the gain factor in sliding mode control is not sensitive to changes in plant parameters, which may occur e.g. because of changes in environmental conditions or ageing. Thus, maintenance operations for recalibration are not necessary or at least may be carried out much less frequently.

[0015] According to an aspect of the invention, the time derivative (k̇(S,t)) of the gain factor is linearly dependent on a module of the sliding quantity in the first region and in the third region.

[0016] According to an aspect of the invention, the time derivative (k̇(S,t)) of the gain factor is defined by k · = − A 1 TH 1 − S TH 1 S < TH 1 k · = 0 TH 1 < S < TH 2 k · = A 2 S − TH 2 TH 2 S > TH 2 k being the time derivative of the gain factor, S being the sliding quantity, A 1 being a first constant coefficient, A 2 being a second constant coefficient, TH 1 being a first threshold and TH 2 being a second threshold, greater than the first threshold, and wherein the first threshold delimits the first region and the second threshold delimits the second region.

[0017] The first, second and third region are conveniently defined as bands around the sliding manifold. The way the time derivative of the gain factor is defined ensures that the gain factor increases when the sliding quantity is outside a safety band and decreases when the sliding quantity approaches the sliding manifold. Moreover, calibration is as simple as in conventional sliding mode control techniques and much more straightforward than in PID controllers.

[0018] According to an aspect of the invention, the control law comprises a load component and controlling comprises using the load component.

[0019] According to an aspect of the invention, the gas turbine engine comprises a fuel valve configured to set fuel supply for the combustor assembly and controlling comprises applying the load component to the fuel valve.

[0020] According to an aspect of the invention, the control law comprises an exhaust temperature component and controlling comprises using the exhaust temperature component.

[0021] According to an aspect of the invention, the gas turbine engine comprises an inlet guide vane stage, configured to set an air supply for the combustor assembly and controlling comprises applying the exhaust temperature control law to the inlet guide vane stage.

[0022] The sliding control in accordance with the invention may be applied to either one of the fuel valve and inlet guide vane stage or to both, in accordance with design preferences. When applied to the fuel valve, the control law may be sent in a minimum port with other control signals without affecting performance of the control system.

[0023] According to the present invention, there is also provided gas turbine power plant according to claim 9 comprising: a gas turbine engine, having a compressor, a combustor assembly and a turbine; an electric power generator mechanically connected to the gas turbine engine; a sensor assembly configured to detect quantities associated with operation of the gas turbine engine; and a control system configured to determine at least one of a fuel supply and an air supply to the combustor assembly to control at least one of a load and an exhaust temperature in accordance with a sliding-mode control technique; wherein the control system is configured to determine a sliding quantity as a function of a control error between a current value and a reference value of a controlled variable, the controlled variable including at least one of a load and an exhaust temperature of the gas turbine engine, and to determine a control law, having a gain factor and a sliding factor and configured to bring the sliding quantity on a sliding manifold defined by states that cause the sliding quantity to be zero; wherein the sliding factor is a first function of the sliding quantity and a time derivative (k̇(S,t)) of the gain factor is a second function of the sliding quantity.

[0024] According to an aspect of the invention, the control system comprises: a sliding quantity generator configured to supply current values of the sliding quantity; a first computation stage, configured to receive the current values of the sliding quantity from the sliding quantity generator and to determine values of the time derivative of the load gain factor based on the received current values of the sliding quantity as k · = − A 1 TH 1 − S TH 1 ; a second computation stage, configured to determine values of the time derivative as k · = 0 ; a third computation stage, configured to receive the current values of the sliding quantity from the sliding quantity generator (30) and to determine values of the time derivative of the load gain factor based on the received current values of the sliding quantity as k · = A 2 S − TH 2 TH 2 an integrator; a selector, configured to pass an output of a selected one of the first computation stage, second computation stage and third computation stage to the integrator; and a decider stage, configured to control the selector to select the output: of the first computation stage, when |S|<TH 1 ; of the second computation stage, when TH 1 <|S|<TH 2 ; and of the third computation stage, when |S|>TH 2 .

[0025] According to an aspect of the invention, the control system comprises: the sensor assembly is configured to detect an active power or the control system is configured to determine an estimate of the active power based on the quantities detected by the sensor assembly; the gas turbine engine comprises a fuel valve, which is set by the control system to adjust fuel supply to the combustor assembly; the sliding quantity is defined based on a control error between a load set point as a reference value and the active power; and the control law comprises a load component and wherein the load component is defined as U L S L , t = k L S L , t SF L S L , t . U L being the load component and S L , k L , SF L being the sliding quantity, the gain factor and the sliding factor for the load component, respectively.

[0026] According to an aspect of the invention, the sensor assembly is configured to detect an exhaust temperature of the gas turbine engine; the gas turbine engine comprises an inlet guide vanes stage, which is set by the control system to adjust an air intake of the compressor; the sliding quantity is defined based on a control error between an exhaust temperature set point as a reference value and the exhaust temperature; and the control law comprises a exhaust temperature component and wherein the load component is defined as U TE S TE , t = k TE S TE , t SF TE S TE , t U TE being the exhaust temperature component and S TE , k TE , SF TE being the sliding quantity, the gain factor and the sliding factor for the exhaust temperature component, respectively. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention will now be described with reference to the accompanying drawings, which show a number of non-limitative embodiments thereof, in which: figure 1 is a simplified block diagram of a gas turbine power plant carrying out a method in accordance with an embodiment of the present invention; figure 2 is a graph showing quantities relating to the power plant of figure 1; figure 3 is a more detailed block diagram of a first portion of the plant of figure 1; figure 4 is a more detailed block diagram of a second portion of the plant of figure 1; and figure 5 is a more detailed block diagram of a third portion of the plant of figure 1. DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS

[0028] Figure 1 shows a gas turbine power plant 1. The power plant 1 is selectively connectable to a distribution grid 2 by a main breaker 3, and comprises a gas turbine engine 5, a generator 6, a sensor assembly 7 and a control system 8.

[0029] The gas turbine engine 5 comprises a compressor 10, a combustor assembly 11, and a gas turbine engine 12. The compressor 10 is provided with a stage of orientable inlet guide vanes or IGV stage 13, which is set by the control system 8 to adjust the air intake of the compressor 12 and air supply to the combustor assembly 11. The gas turbine engine 5 also comprises a fuel valve 15 which is set by the control system 8 to adjust fuel supply to the combustor assembly 11. The generator 6 is connected mechanically to the same shaft as the turbine 12 and the compressor 10, and is rotated at the same angular speed ω. The generator 6 converts the mechanical power produced by gas turbine engine 5 to active electric power PE available for the grid 2.

[0030] The sensor assembly 7 comprises a plurality of sensors configured to detect respective quantities, such as, by way of non-limiting example, an angular speed ω of the gas turbine engine 5 and of the generator 6, an active (electric) power P E delivered by the generator 6, and an exhaust temperature T E at exhaust of turbine 12. The sensor assembly 7 may also detect an ambient temperature T A , intake pressure P I and output pressure P O of compressor 10, a current vane position IGVPOS of the IGV stage 13 of the compressor 10, and a MAIN position signal indicating the position of main breaker 3.

[0031] The detected quantities are supplied to the control system 8, which generates a control law having two components, namely a load control law U L , for an actuator of the fuel valve 15 and an exhaust temperature control law U TE for an actuator of the IGV stage 13.

[0032] The control system 8 is configured to determine at least one of the load control law U L and the exhaust temperature control law U TE in accordance with a sliding mode control technique. Specifically, for each controlled variable, the control system 8 determines a sliding quantity S and a control law, hereinafter indicated by U(S, t), which is a function of the sliding quantity S and of time. The sliding quantity S is determined by the control system 8 as a function of a control error (and possibly one or more time derivatives thereof) between a current value and a reference value of a controlled variable. The control error is in turn a function of a state vector of the gas turbine engine system, which may include quantities detected by the sensor assembly. A sliding manifold S' is defined by states of the system that cause the sliding quantity S to be zero, i.e. S=0.

[0033] The control law U(S, t) has a gain factor k(S, t) and a sliding factor SF(S), which are both dependent on the sliding quantity S. The gain factor k(S, t) also depends on time. The control law U(S, t) is therefore defined as U S , t = k S , t SF S , t

[0034] Specifically, the sliding factor SF(S) is an odd function of the sliding quantity S and in one embodiment is defined by SF S = sat S TH 1 where TH 1 is a first threshold that defines a first region as a first band around a sliding manifold S' (S = 0, see figure 2).

[0035] The gain factor k(S, t) is defined on conditions that depend on the sliding variable S. The control system 8 sets a time derivative k̇(S,t) of the gain factor k(S, t), see also figure 2: to a negative value in a first region B 1 defined by a first threshold TH 1 around a sliding manifold S=0; to zero outside the first region B 1 and in a second region B 2 that surrounds the first region, wherein the second region B 2 is defined by the first threshold TH 1 and a second threshold TH 2 greater than the first threshold TH 1 , such that in the second region B 2 TH 1 <|S|<TH 2 ; and to a positive value outside the second region B 2 and in a third region B 3 that surrounds the second region B 2 , such that in the third region B 3 TH 2 <|S|.

[0036] Thus, the gain factor k(S, t) increases as long as the sliding quantity S is in the third region B 3 , i.e. away from the sliding manifold S' (S=0), and the strength of the control action determined by the control law U(S, t) grows accordingly. The control action is thus effective and tends to rapidly bring the sliding quantity S toward the sliding manifold S', without substantial risk of causing chattering. The gain factor k(S, t) is constant when the sliding quantity S is in the second region B 2 and decreases when the sliding quantity S remains in the first region B 1 . Hence, the strength of the control action is reduced to a minimum when the sliding quantity S is close to the sliding manifold S'.

[0037] In one embodiment, the time derivative k̇(S,t) of the gain factor k(S, t) is linearly dependent on a module of the sliding quantity S at least in the first region B 1 and in the third region B 3 . For example, the time derivative k̇(S,t) of the gain factor k(S, t) is defined as k ˙ = − A 1 TH 1 − S TH 1 S < TH 1 k ˙ = 0 TH 1 < S < TH 2 k ˙ = A 2 S − TH 2 TH 2 S > TH 2 where A 1 is a first constant coefficient and A 2 is a second constant coefficient. Accordingly, on the sliding manifold S', which in sliding mode control can be reached in a finite time, the time derivative k̇(S,t) of the gain factor k(S, t) is equal to -A 1 , which may be selected to meet design preferences.

[0038] Equations (2)-(4) may be used by the control system 8 to determine one or both of a load component U L and an exhaust temperature component U TE of an overall control law (see figure 1).

[0039] Specifically, the load component U L is applied to the fuel valve 15 and is defined as U L S L , t = k L S L , t SF L S L , t

[0040] Here, the sliding quantity S L is defined based on a control error between a load set point as the reference value and the active power P E detected by the sensor assembly 7. As an alternative, the control system 8 may determine an estimate of the active power P E based on the quantities detected by the sensor assembly 7.

[0041] A load sliding factor SF L (S L , t) is determined by the control system 8 as SF L S L = B L sat S L TH 1 L where B L is a constant coefficient and TH 1L is the first threshold defining the first region B 1L for load control.

[0042] A load gain factor k L (S L , t) is determined by time integration from k ˙ L = − A 1 L TH 1 L − S L TH 1 L S L < TH 1 L k ˙ L = 0 TH 1 L < S L < TH 2 L k ˙ L = A 2 L S L − TH 2 L TH 2 L S L > TH 2 L where TH 2L is the second threshold defining the second region-B 2L and the third region B 3L for load control.

[0043] In one embodiment, a load processor module 20 of the control system 8 that supplies the load gain factor k L (S L , t) may have the structure shown in figure 3. The load processor module 20 comprises a first computation stage 21, a second computation stage 22, a third computation stage 23, a selector 24, a decider stage 25 and an integrator 27. The load processor module 20 is coupled to a sliding quantity generator 30 of the control system 8 that supplies current values of the load sliding quantity S L .

[0044] The first computation stage 21, the second computation stage 22 and the third computation stage 23 receive the current values of the load sliding quantity S L from the sliding quantity generator 30 and supply unbound values of the time derivative k̇(S L ,t) of the load gain factor k L (S L , t) in accordance with equations (8), (9) and (10), respectively, except in that constraints on the module of the load sliding quantity S L are ignored.

[0045] The selector 24 is configured to pass the output of a selected one of the first computation stage 21, second computation stage 22 and third computation stage 23 to the integrator 27.

[0046] The decider stage 25 controls the selector 24 in accordance with the current values of the load sliding quantity S L received from the sliding quantity generator 30. Specifically, the decider stage 25 sets the selector 24 based on the module of the load sliding quantity S L so that the integrator 27 receives the appropriate output from one of the first computation stage 21, second computation stage 22 and the third computation stage 23. In the embodiment herein described, the decider stage 25 controls the selector 24 to select the output: of the first computation stage 21, when |S L |<TH 1L ; of the second computation stage 22, when TH 1L <|S L |<TH 2L ; and of the third computation stage 23, when |S L |>TH 2L .

[0047] Likewise, the exhaust temperature component U TE , which is applied to the fuel valve 15, in one embodiment is defined as U TE S TE , t = k TE S TE , t SF TE S TE , t

[0048] The sliding quantity S TE is defined based on a control error between an exhaust temperature set point as the reference value and the exhaust temperature T E detected by the sensor assembly 7.

[0049] An exhaust temperature sliding factor SF TE (S TE , t) is determined by the control system 8 as SF TE S TE = B TE sat S TE TH 1 TE where B TE is a constant coefficient and TH 1TE is the first threshold defining the first region B 1TE for exhaust temperature control.

[0050] An exhaust temperature gain factor k TE (S TE , t) is determined by time integration from k ˙ TE = − A 1 TE TH 1 TE − S TE TH 1 TE S TE < TH 1 TE k ˙ TE = 0 TH 1 TE < S TE < TH 2 TE k ˙ TE = A 2 TE S TE − TH 2 TE TH 2 TE S TE > TH 2 TE where TH 2TE is the second threshold defining the second region B 2TE and the third region B 3TE for exhaust temperature control.

[0051] In one embodiment, an exhaust temperature processor module 40 of the control system 8 that supplies the exhaust temperature gain factor k TE (S TE , t) may have the structure shown in figure 4. The exhaust temperature processor module 20 comprises a first computation stage 41, a second computation stage 42, a third computation stage 43, a selector 44, a decider stage 45 and an integrator 47.

[0052] The first computation stage 41, the second computation stage 42 and the third computation stage 43 receive the current values of the exhaust temperature sliding quantity S TE from the sliding quantity generator 30 and supply unbound values of the time derivative k̇(S TE ,t) of the exhaust temperature gain factor k TE (S TE , t) in accordance with equations (13), (14) and (15), respectively, except in that constraints on the module of the exhaust temperature sliding quantity S TE are ignored.

[0053] The selector 44 is configured to pass the output of a selected one of the first computation stage 41, second computation stage 42 and third computation stage 43 to the integrator 47.

[0054] The decider stage 45 sets the selector 44 based on the module of the exhaust temperature sliding quantity S TE so that the integrator 47 receives the appropriate output from one of the first computation stage 41, second computation stage 42 and the third computation stage 43. In the embodiment herein described, the decider stage 45 controls the selector 44 to select the output: of the first computation stage 41, when |S TE |<TH 1TE ; of the second computation stage 42, when TH 1TE <|S TE |<TH 2TE ; and of the third computation stage 43, when |S TE |>TH 2TE .

[0055] In one embodiment (figure 5), the load component U L of the control law concurs in a minimum port 50 with other control signals SC 1 , ..., SC N , that may be used by the control system 8 to carry out additional control and / or protection functions. Protection function may include e.g. limiting high exhaust temperatures by reducing the load, limiting the load in accordance with the compression ratio by reducing fuel supply if the compression ratio exceeds a threshold value and limiting maximum load. In this case, the minimum port 50 sends a load control law U L ' requiring the minimum control effort to the actuator of the fuel valve 15.

[0056] Finally, it is clear that modifications and variants can be made to the method and to the gas turbine power plant described herein without departing from the scope of the present invention, as defined in the appended claims.

Claims

1. A method of controlling a gas turbine power plant including a gas turbine engine (5), and an electric power generator (6) mechanically connected to the gas turbine engine (5), the gas turbine engine (5) comprising a compressor (10), a combustor assembly (11) and a turbine (12); the method comprising determining at least one of a fuel supply and an air supply to the combustor assembly (11) to control at least one of a load (PE) and an exhaust temperature (TE) of the gas turbine engine (5) in accordance with a sliding-mode control technique; the method being further characterized in that controlling comprises determining a sliding quantity (S; SL; STE), as a function of a control error between a current value and a reference value of a controlled variable, the controlled variable including at least one of a load (PE) and an exhaust temperature (TE) of the gas turbine engine (5), and determining a control law (U(S, t); UL(SL, t); UTE (STE, t)), having a gain factor (k(S, t); kL(SL, t) ; kTE(STE, t)) and a sliding factor (SF(S, t); SFL(SL, t); SFTE(STE, t)) and configured to bring the sliding quantity (S; SL; STE) on a sliding manifold (S') defined by states that cause the sliding quantity (S; SL; STE) to be zero; the sliding factor (SF(S, t); SFL(SL, t); SFTE (STE, t)) is a first function of the sliding quantity (S; SL; STE) and a time derivative (k̇(S,t); k̇(SL,t); k̇(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) is a second function of the sliding quantity (S; SL; STE; and the time derivative (k̇(S,t); k̇(SL,t); k̇(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) is negative in a first region (B1) around the sliding manifold (S'), is zero in a second region (B2) that surrounds the first region (B1) and is positive in a third region (B3) that surrounds the second region (B2).

2. The method according to claim 1, wherein the time derivative (k(S,t); k̇(SL,t); k̇(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) is linearly dependent on a module of the sliding quantity (S; SL; STE) in the first region (B1) and in the third region (B3).

3. The method according to claim 2, wherein the time derivative (k(S,t); k̇(SL,t); k̇(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) is defined by k ˙ = − A 1 TH 1 − S TH 1 S < TH 1 k ˙ = 0 TH 1 < S < TH 2 k ˙ = A 2 S − TH 2 TH 2 S > TH 2 k being the time derivative of the gain factor, S being the sliding quantity, A1 being a first constant coefficient, A2 being a second constant coefficient, TH1 being a first threshold and TH2 being a second threshold, greater than the first threshold, and wherein the first threshold delimits the first region (B1) and the second threshold delimits the second region (B2).

4. The method according to any one of the preceding claims, wherein the control law (U(S, t); UL(SL, t); UTE(STE, t)) is in the form U S , t = k S , t SF S , t U(S, t) being the control law, k(S, t) being the gain factor and SF(S, t) being the sliding factor, and wherein the sliding factor is an odd function of the sliding quantity (S), for example the sliding factor is defined by SF S = sat S TH 1 5. The method according to any one of the preceding claims, wherein the control law comprises a load component (UL(SL, t)) and controlling comprises using the load component (UL(SL, t)).

6. The method according to claim 5, wherein the gas turbine engine (5) comprises a fuel valve (15) configured to set fuel supply for the combustor assembly (11) and controlling comprises applying the load component (UL(SL, t)) to the fuel valve (15).

7. The method according to any one of the preceding claims, wherein the control law comprises an exhaust temperature component (UTE(STE, t)) and controlling comprises using the exhaust temperature component (UTE(STE, t)).

8. The method according to claim 7, wherein the gas turbine engine (5) comprises an inlet guide vane stage (13), configured to set an air supply for the combustor assembly (11) and controlling comprises applying the exhaust temperature control law (UTE(STE, t)) to the inlet guide vane stage (13).

9. A gas turbine power plant comprising: a gas turbine engine (5), having a compressor (10), a combustor assembly (11) and a turbine (12); an electric power generator (6) mechanically connected to the gas turbine engine (5); a sensor assembly (7) configured to detect quantities (PE, TE, ω, TA, PO, IGVPOS, MAIN) associated with operation of the gas turbine engine (5); and a control system (8) configured to determine at least one of a fuel supply and an air supply to the combustor assembly (11) to control at least one of a load (PE) and an exhaust temperature (TE) in accordance with a sliding-mode control technique; the gas turbine power plant being further characterized in that the control system (8) is configured to determine a sliding quantity (S; SL; STE) as a function of a control error between a current value and a reference value of a controlled variable, the controlled variable including at least one of a load (PE) and an exhaust temperature (TE) of the gas turbine engine (5), and to determine a control law (U(S, t); UL(SL, t); UTE(STE, t)), having a gain factor (k(S, t); kL(SL, t); kTE(STE, t)) and a sliding factor (SF(S, t); SFL(SL, t); SFTE (STE, t)) and configured to bring the sliding quantity (S; SL; STE) on a sliding manifold (S') defined by states that cause the sliding quantity (S; SL; STE) to be zero; the sliding factor (SF(S, t); SFL(SL, t); SFTE(STE, t)) is a first function of the sliding quantity (S; SL; STE) and a time derivative (k̇(S,t); k̇(SL,t); k̇(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) is a second function of the sliding quantity (S; SL; STE); and the time derivative (k(S,t); k̇(SL,t); k̇(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) is negative in a first region (B1) around the sliding manifold (S'), is zero in a second region (B2) that surrounds the first region (B1) and is positive in a third region (B3) that surrounds the second region (B2).

10. The gas turbine power plant according to claim 9, wherein the time derivative (k̇(S,t); k̇(SL,t); k̇(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) is linearly dependent on a module of the sliding quantity (S; SL; STE) in the first region (B1) and in the third region (B3).

11. The gas turbine power plant according to claim 10, wherein the time derivative (k̇(S,t); k̇(SL,t); k̇(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) is defined by k ˙ = − A 1 TH 1 − S TH 1 S < TH 1 k ˙ = 0 TH 1 < S < TH 2 k ˙ = A 2 S − TH 2 TH 2 S > TH 2 k being the time derivative of the gain factor, S being the sliding quantity, A1 being first constant coefficient, A2 being a second constant coefficient, TH1 being a first threshold and TH2 being a second threshold, greater than the first threshold, and wherein the first threshold delimits the first region (B1) and the second threshold delimits the second region (B2), wherein the control law (U(S, t); UL(SL, t); UTE(STE, t)) is in the form U S , t = k S , t SF S , t U(S, t) being the control law, k(S, t) being the gain factor (k(S, t) ; kL(SL, t) ; kTE(STE, t)) and SF (S, t) being the sliding factor, and wherein the sliding factor is an odd function of the sliding quantity (S), for example the sliding factor is defined by SF S t = sat S TH 1 12. The gas turbine power plant according to any one of claim 10 or 11, wherein the control system (8) comprises: a sliding quantity generator (30) configured to supply current values of the sliding quantity (S; SL; STE); a first computation stage (21; 41), configured to receive the current values of the sliding quantity (S; SL; STE) from the sliding quantity generator (30) and to determine values of the time derivative (k̇(S,t); k̇L(SL,t); k̇TE(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) based on the received current values of the sliding quantity as k ˙ = − A 1 TH 1 − S TH 1 ; a second computation stage (22; 42), configured to determine values of the time derivative (k̇(S,t); k̇L(SL,t); k̇TE(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) as k ˙ = 0 ; a third computation stage (23; 43), configured to receive the current values of the sliding quantity (S; SL; STE) from the sliding quantity generator (30) and to determine values of the time derivative (k̇(S,t); k̇L(SL,t); k̇TE(STE,t)) of the gain factor (k(S, t); kL(SL, t); kTE(STE, t)) based on the received current values of the sliding quantity(S; SL; STE) as k ˙ = A 2 S − TH 2 TH 2 an integrator; a selector (24; 44), configured to pass an output of a selected one of the first computation stage (21; 41), second computation stage (22; 42) and third computation stage (23; 43) to the integrator; and a decider stage (25; 45), configured to control the selector (24; 44) to select the output: of the first computation stage (21; 41), when |S|<TH1; of the second computation stage (22; 42), when TH1<|S|<TH2; and of the third computation stage (23; 43), when |S|>TH2.

13. The gas turbine power plant according to claim 11, wherein: the sensor assembly (7) is configured to detect an active power (PE) or the control system (8) is configured to determine an estimate of the active power (PE) based on the quantities detected by the sensor assembly (7); the gas turbine engine (5) comprises a fuel valve (15) which is set by the control system (8) to adjust fuel supply to the combustor assembly (11); the sliding quantity (SL) is defined based on a control error between a load set point as a reference value and the active power (PE); and the control law comprises a load component (UL(SL, t)) and wherein the load component (UL(SL, t)) is defined as U L S L , t = k L S L , t SF L S L , t . UL being the load component and SL, kL, SFL being the sliding quantity, the gain factor and the sliding factor for the load component, respectively; and / or wherein: the sensor assembly (7) is configured to detect an exhaust temperature (TE) of the gas turbine engine (5); the gas turbine engine (5) comprises an inlet guide vanes stage (13), which is set by the control system (8) to adjust an air intake of the compressor (10); the sliding quantity (STE) is defined based on a control error between an exhaust temperature set point as a reference value and the exhaust temperature (TE); and the control law comprises a exhaust temperature component (UTE(STE, t)) and wherein the load component (UTE(STE, t)) is defined as U TE S TE , t = k TE S TE , t SF TE S TE , t UTE being the exhaust temperature component and STE, kTE, SFTE being the sliding quantity, the gain factor and the sliding factor for the exhaust temperature component, respectively.