TITLE OF THE INVENTION: System and method for regulating a physical parameter of a real system of a turbomachine from a setpoint physical parameter

TECHNICAL FIELD OF THE INVENTION

The present invention relates to a system for regulating a physical parameter of a real system of a turbomachine from a setpoint physical parameter. For example, the physical parameter may correspond to a displacement speed of a valve of a turbojet engine, a fuel flow rate, an angle of orientation of a blade, etc.

In known manner, a regulation system REG comprises a corrector comprising a correction function C1 (p) and a parameter gain K. The performance of the regulation system is evaluated in particular according to its response time and its stability. In practice, a control system is configured to ensure a compromise between response time and stability.

[0003] In practice, a real system of a turbomachine includes characteristics and variables which are likely to change and deteriorate over time (wear, drift, etc.). Also, a control system whose operation is optimal when it is put into service can be non-optimal a few months later, in particular, in terms of stability. To eliminate this potential drawback, during commissioning, the regulation system is configured in such a way as to have a large stability margin, which affects the response time.

The present invention aims to eliminate at least some of these drawbacks by proposing a regulation system making it possible to adapt dynamically in order to have good performance in terms of stability and response time over time.

[0005] Incidentally, in the prior art patent application US2004/0123600 teaches a control system to optimize the definition of the real system over time in order to integrate the faults and malfunctions appearing during time. The optimization of such a real system is complex (definition of new transfer functions, etc.) and does not allow reactive regulation. The computational cost is very important.

Also known from patent US5537310 is an adaptive correction model during which the parameter gain of a model of the real system is adapted during a transient phase. This document deals only with stability for the transients and does not make it possible to treat the defects of speed.

PRESENTATION OF THE INVENTION

The invention relates to a system for regulating a physical parameter of a real system of a turbomachine from a setpoint physical parameter, the regulation system having a response time and comprising:

a corrector comprising a correction function and a parameter gain K,

a theoretical inverse transfer function of the real system, and

a system for optimizing the parameter gain K during regulation, the optimization system comprising:

a stability correction module configured to determine a first gain constant K1, of positive value, upon detection of instability of the regulation system during regulation of the physical parameter,

a regulation system response time correction module configured to determine a second gain constant K2, of negative value, upon detection of a delay during regulation of the physical parameter, the stability correction module being configured to inhibit the response time correction module upon detection of an instability during the regulation of the physical parameter, and

a determination module configured to determine the parameter gain K as a function of the first gain component K1 and the second gain component K2 previously determined.

The invention is remarkable in that the regulation system makes it possible to dynamically correct a stability defect by increasing the parameter gain and a delay by decreasing the parameter gain, so that the regulation system has performances optimal. The control system is thus self-adaptive. It is advantageously no longer necessary to sacrifice the response time with regard to stability as in the prior art. The regulation is thus more reactive. Advantageously, the improvement in speed is inhibited in the event of instability. In other words, the stability is corrected as a priority, the response time being improved only when the regulation system is stable. Thanks to the regulation system according to the invention,

[0009] The regulation system thus makes it possible to correct itself according to its own response.

[0010] Preferably, a difference is defined between the physical parameter and the setpoint physical parameter. The stability correction module comprises a stability detection module configured to compare the deviation with a high deviation threshold and with a low deviation threshold. The stability detection module is configured to detect an instability if the deviation is successively greater than the high deviation threshold then lower than the low deviation threshold. In other words, the stability detection module makes it possible to measure a deviation from a predetermined range. Such detection is fast and robust.

[0011] Preferably, the oscillating deviation during an instability, the stability detection module is configured to count the oscillations following a detection of an instability and to determine a TopCS stability correction parameter according to the number of counted oscillations NB-osc, the first gain component K1 depending on the stability correction parameter TopCS. Thus, by counting the number of oscillations, the degree of instability is determined and the appropriate degree of correction can be deduced therefrom.

Preferably, the stability correction module is configured to reset the number of counted oscillations NB-osc in the event of detection of a transient phase generated by a significant variation in the physical setpoint parameter.

[0013] In other words, the stability detection module is dedicated to ensure stability during a stationary phase, stability during a transient phase being ensured by dedicated means. This improves the correction by allowing the calculation of optimal correction values ​​according to the type of instability.

[0014] Preferably, the stability detection module is configured to reset the number of oscillations counted following the determination of the parameter gain K from the first gain component K1. In other words, a new correction is inhibited as long as a previous correction has not yet produced its effects.

According to one aspect of the invention, the stability correction module comprises an overflow detection module configured to determine a TopOS overflow parameter, the first gain component K1 depending on the

TopOS overflow. According to the invention, the stability during a transient phase, here a transient acceleration phase, which corresponds to an increase in the setpoint physical parameter, is monitored by dedicated means, which ensures optimum correction.

[0016] Preferably, a difference being defined between the physical parameter and the setpoint physical parameter, the overflow detection module is configured to start a period of monitoring of the difference following an increasing significant variation of the setpoint parameter . The overflow detection module is configured to compare the deviation to at least one overflow threshold and the overflow detection module is configured to detect an overflow if the deviation is greater than the overflow threshold during of the monitoring period. In other words, only overruns during a previously verified transient phase are taken into account. Overshoots during a stationary phase are advantageously ignored in the overshoot detection module. A separate processing of an overshoot advantageously makes it possible to carry out a reactive correction making it possible to eliminate any future overshoot as soon as a first overshoot appears. Since the monitoring window is narrow, the correction made can be more relevant and reactive.

[0017] Preferably, the overflow detection module is configured to inhibit detection of an overflow in the event of detection of a decreasing variation of the setpoint parameter. In other words, the monitoring period is stopped if the increasing transient conditions are no longer respected. Any erroneous correction is thus avoided.

According to one aspect of the invention, the stability correction module comprises an undershoot detection module configured to determine an undershoot parameter TopUS, the first gain component K1 depending on the undershoot parameter TopUS. Advantageously, the acceleration and deceleration transients are processed separately in order to achieve tailor-made regulation for each type of instability.

[0019] De préférence, le module de correction de stabilité comporte un module de détection de transitoire configuré pour mesurer une variation de l’écart e par rapport à la réponse en boucle fermée yBF du paramètre physique. Autrement dit, pour détecter un transitoire, on calcule de manière préalable la réponse théorique en boucle fermée afin de former un étalon de comparaison. Un tel étalon de comparaison dynamique est avantageux pour déterminer la période de surveillance et réaliser une correction réactive.

Preferably, the response time correction module is configured to determine a tolerance range around the setpoint physical parameter and to determine a second gain constant K2 if the physical parameter does not belong to the tolerance range. . Thus, if a delay or an excessive advance is detected, a correction is made dynamically.

The invention also relates to a method for regulating a physical parameter by implementing an REG regulation system as presented previously, the regulation method comprising:

a stability monitoring step during the regulation of the physical parameter,

a step of determining a first gain constant K1, of positive value, upon detection of instability during regulation of the physical parameter,

a step for monitoring the response time of the regulation system REG during the regulation of the physical parameter in the event of absence of instability,

a step of determining a second gain constant K2, of negative value, upon detection of a delay during regulation of the physical parameter,

a step of determining the setting gain K of the corrector C(p) from the first gain constant K1 and the second gain constant K2 so as to guarantee the stability of the regulation while optimizing the response time.

The invention also relates to a computer program comprising instructions for the execution of the steps of the regulation method as presented above, said program being executed by a computer.

The invention further relates to an electronic control unit for a turbomachine comprising a memory including instructions from a computer program as presented previously.

The invention also relates to a turbomachine comprising an electronic unit as presented previously.

PRESENTATION OF FIGURES

The invention will be better understood on reading the following description, given solely by way of example, and referring to the accompanying drawings given by way of non-limiting examples, in which identical references are given to similar objects and on which:

Figure 1 is a schematic representation of a correction system of a real system by an inverse model according to the prior art;

Figure 2 is a schematic representation of a control system according to one embodiment of the invention for a correction system of a real system by an inverse model;

Figure 3 is a schematic representation of one embodiment of a stability correction module;

Figure 4 is a schematic representation of one embodiment of an instability detection module;

Figure 5 is a curve of evolution of the difference e over time for the detection of instabilities;

Figure 6 is a schematic representation of one embodiment of an overflow detection module;

FIG. 7 is a curve of change in the physical parameter following an increasing variation in the setpoint parameter, that is to say an increasing transient phase;

[0033] FIG. 8 is a set of curves representing the evolution, during a transient phase, of the physical parameter and of the deviation to determine an overshoot correction value;

Figure 9 is a schematic representation of one embodiment of a response time correction module;

[0035] FIG. 10 is a curve showing the evolution of the physical parameter y with respect to the setpoint physical parameter yc in stationary phase;

[0036] FIG. 11 is a set of curves representing the evolution of the physical parameter following a periodic delay affecting the real system for a regulation system according to the prior art;

FIG. 12 is a set of curves representing the evolution of the physical parameter following a periodic delay affecting the real system for a regulation system according to the invention.

It should be noted that the figures expose the invention in detail to implement the invention, said figures can of course be used to better define the invention if necessary.

DETAILED DESCRIPTION OF THE INVENTION

[0039] En référence à la figure 1 , il est représenté un système de correction SC configuré pour déterminer un paramètre de commande y en fonction d’un paramètre de consigne yc. Dans cet exemple, le système de correction SC met en œuvre un correcteur par modèle inverse. Autrement dit, le système de correction SC comporte successivement un correcteur C(p), un modèle inverse F 1(p) et un système réel F(p). Le système réel F(p) correspond au système réel de la turbomachine qui agit sur le paramètre de commande y. En pratique, le système réel F(p) met en œuvre plusieurs fonctions de transfert. Par nature, le système réel F(p) évolue au cours du temps (dérive, usure, etc.) et ne réagit pas de manière identique à un paramètre de consigne yc.

[0040] Dans un système de correction SC mettant en œuvre un correcteur par modèle inverse, il est supposé que le système réel F(p) soit inversable sur le plan mathématique de manière à définir le modèle inverse F 1(p). Selon cette hypothèse, le système de correction SC dépend alors essentiellement du correcteur C(p) étant donné que F 1(p)*F(p)=1. Autrement dit, le temps de réponse et la stabilité du système de correction peuvent alors être directement déterminés par le correcteur C(p).

[0041] Le correcteur C(p) comporte de manière connue une fonction de transfert Ci(p) et un gain de réglage K connu de l’homme du métier sous la désignation de « gain de tuning » de manière à obtenir la formule suivante : C(p)= Ci(p)*1/K.

[0042] Selon la présente invention, le gain de réglage K est modifié au cours du temps de manière à ajuster le temps de réponse et la stabilité du correcteur C(p). Ainsi, si un retard pur apparaît dans le système réel F(p) ou si le gain statistique du système réel F(p) est modifié, le gain de réglage K peut être modifié pour conserver des performances optimales.

[0043] Par la suite, l’ensemble comprenant le correcteur C(p) et le modèle inverse F 1(p) est désigné régulateur REG et fournit un paramètre physique préliminaire ya au système réel F(p). En référence à la figure 2, il est représenté un régulateur REG selon une forme de réalisation de l’invention. Le régulateur REG présente un temps de réponse.

With reference to FIG. 2, the corrector C(p) comprises a transfer function Ci(p) and an optimization system OPTK of the parameter gain K.

The OPTK optimization system comprises a stability correction module 2 configured to determine a first gain component K1, a response time correction module 3 configured to determine a second gain component K2 and a determination module 4 configured to determine the parameter gain K as a function of the gain components K1, K2 previously determined. Subsequently, a deviation e is defined which corresponds to the difference between the setpoint physical parameter yc and the physical parameter y (e=yc-y).

The different modules will now be presented in detail.

With reference to FIG. 2, the stability correction module 2 is configured to determine the first gain component K1 as well as a TopIS instability as a function of the command parameter y and the setpoint parameter yc.

[0048] Stability correction module 2 (Figure 3)

The stability correction module 2 is represented schematically in FIG. 3. It comprises an instability detection module 21 configured to detect a TopIS instability and to determine a TopCS stability correction parameter from the deviation e of the setpoint parameter yc. In practice, as will be presented below, the instability detection module 21 makes it possible to detect an instability around the setpoint parameter yc.

The stability correction module 2 further comprises an overshoot detection module 22, or "overshoot" detection, configured to detect an overshoot T opOS from epsilon e, of the control parameter y and the setpoint parameter yc. In other words, during a rapidly increasing variation of the setpoint parameter yc, the physical parameter y can exceed the setpoint parameter yc and create instability linked to the transient phase. In what follows, we will also use the language shortcut “a transient” to designate a transient phase.

Similarly, the stability correction module 2 further comprises an undershoot detection module 23, or "undershoot" detection, configured to

detecting a TopUS undershoot from epsilon e, command parameter y and setpoint parameter yc.

The stability correction module 2 finally comprises a stability correction module 24 configured to determine the first gain component K1 as a function of the overshoot detection parameter TopOS, the undershoot detection parameter TopUS and the parameter stability correction TopCS obtained by the other modules 21, 22, 23 of the stability correction module 2.

[0053] Stability detection module 21

The stability detection module 21 is shown schematically in Figure 4. The stability detection module 21 comprises a first module 21 1 to detect TopIS instability if the value of epsilon e is greater than a threshold high SH-CS or below a low threshold SB-CS. A TopIS instability is detected when a high threshold SH-CS and a low threshold SB-CS are successively exceeded.

For example, the first module 21 1 is configured to compare, on the one hand, the difference e to the high threshold SH-CS and, on the other hand, the difference e to the low threshold SB- CS. If the deviation e is greater than the high threshold SH-CS, the overrun is recorded in a memory. Similarly, if the difference e is less than the high threshold SB-CS, the overshoot is recorded in a memory. When two overshoots of different natures are successively detected, a TopIS instability is detected as illustrated in FIG. 5. In the absence of crossing of the thresholds SB-CS, SH-CS, no TopIS instability is detected.

As illustrated in Figure 4, the stability detection module 21 further comprises a second module 212 to count the NB-osc oscillations of the physical parameter y when a TopIS instability is detected. The number of NB-osc oscillations can be obtained in a practical way by counting the successive overshoots and storing them in a memory during a stationary phase.

The second module 212 is also configured to receive a command to reset the number of NB-osc oscillations to zero. To this end, as illustrated in FIG. 4, the stability detection module 21 comprises a third module 213 and a fourth module 214 to determine a reset to zero RAZ.

As illustrated in FIG. 4, the stability detection module 21 includes a fifth module 215 configured to determine the TopCS correction parameter from a certain number of NB-osc oscillations. In this embodiment, if the number of NB-osc oscillations is greater than 3, a correction is performed and a TopCS correction parameter is transmitted by the stability detection module 21. As illustrated in FIG. 4, during a correction, reset information is also sent to avoid repeating the same correction.

[0059] Toujours en référence à la figure 4, le module de détection de stabilité 21 comporte un troisième module 213 pour détecter un transitoire, c’est-à-dire, une variation significative du paramètre de consigne yc, et pour mettre à zéro le nombre d’oscillations NB-osc suite à une détection de transitoire. A cet effet, le troisième module 213 surveille si l’écart e est maintenu entre un deuxième seuil haut SH2 et un deuxième seuil bas SB2. De manière préférée, le deuxième seuil haut SH2 est plus grand que le seuil haut SH-CS et le deuxième seuil bas SB2 est plus petit que le seuil bas SB-CS utilisés précédemment de manière à conserver une marge par rapport à la détection d’instabilité TopIS. Si les deuxièmes seuils SB2, SH2 sont dépassés, il est détecté la présence d’un transitoire, ce qui remet à zéro le nombre d’oscillations NB-osc et stoppe la correction de stabilité. En effet, les corrections liées aux transitoires sont traitées par le module de détection de dépassement supérieur 22 et le module de détection de dépassement inférieur 23.

[0060] Comme illustré à la figure 4, le module de détection de stabilité 21 comporte un quatrième module de temporisation 214 configuré pour détecter si une correction a été réalisée et commander une remise à zéro RAZ en cas de détection. Autrement dit, le quatrième module 4 permet d’inhiber une nouvelle correction lorsqu’une correction précédente n’a pas encore produit ses effets. Dans cet exemple, le quatrième module 214 se présente sous la forme d’une temporisation qui est fonction du temps de réponse du système réel F(p).

Advantageously, the stability detection module 21 makes it possible to determine a TopCS correction parameter as a function of the number of NB-osc oscillations measured following the detection of a TopIS instability. Advantageously, any correction is inhibited in the event of a transient or of a correction not yet taken into account by the real system F(p). The correction parameter TopCS thus calculated makes it possible to improve the stability of the regulation as will be presented subsequently.

[0062] Overflow detection module 22

The overflow detection module 22 is represented schematically in FIG. 3. It makes it possible to determine an upper correction value TopOS if the physical parameter y to be regulated performs an overflow at the end of the transient.

In other words, the overflow detection module 22 makes it possible to carry out an immediate correction following an increasing variation of the setpoint parameter yc.

In practice, with reference to FIG. 6, the overflow detection module 22 comprises a module 221 for determining the response of the real system F(p) in closed loop Yb following the reception of a physical parameter setpoint incl. This makes it possible to determine a theoretical ideal response of the real system F(p). In this example, the determination module 221 implements a “low-pass” type filter, in particular, of order 1 or 2.

The overflow detection module 22 further comprises a transient detection module 222, that is to say, a variation of the deviation e with respect to the closed loop response yBF, in particular, of its derivative. It is thus determined whether the regulation is indeed in an increasing transient phase, that is to say, an increasing variation of the control setpoint yc. If the deviation e deviates from the closed loop response yBF, an acceleration is detected TopAccel. An example of a transient with an instability at the output of the transient is schematically illustrated in FIG. 7 (damping fault).

Still with reference to FIG. 6, the overflow detection module 22 further comprises a storage module 223 configured to start a monitoring period when a TopAccel acceleration is detected. Thus, the overflow detection module 22 is focused on detecting instability during a transient acceleration phase. Indeed, the overflow detection module 22 does not aim to detect an overflow of the physical parameter y when the setpoint parameter yc is stationary, which corresponds to an instability of the regulation.

The overflow detection module 22 further comprises a module 224 for monitoring the deviation e from overflow thresholds SD1, SD2. In this embodiment, the monitoring module 222 comprises two overshoot thresholds SD1, SD2, which are, in this example, thresholds of the hysteresis type.

As illustrated in FIG. 6, if an overflow of an overflow threshold SD1, SD2 is detected during the monitoring period, a TopOS overflow correction value is sent by the overflow detection module superior 22.

Advantageously, the overflow detection module 22 comprises modules for stopping the monitoring period of the storage module 224 in the event of detection of stabilization (module 225) or in the event of a deceleration setpoint (module 226). In fact, it is necessary to prevent an overshoot from being detected by a

deceleration of physical reference parameter yc. This makes it possible to avoid making untimely corrections which are a source of instability.

An example of implementation of an overflow detection is illustrated in Figure 8. In this example, following a transient acceleration phase, the difference e exceeds the first overflow threshold SD1, which activates a TopOS overflow correction, then the second overflow threshold SD2, which again activates a TopOS overflow correction. In other words, the correction is reactive and makes it possible to correct the regulation as soon as an overshoot appears at the end of the transient. Such a correction is possible because it is carried out only in the presence of transient.

[0071] Underflow detection module 23

The underflow detection module 23 is shown schematically in Figure 3. This will not be presented in detail later since it is similar to the overflow detection module 22 but aims detecting a "downward" overshoot during a decreasing variation of a setpoint parameter yc, that is to say, a deceleration of the setpoint physical parameter yc.

Analogously to the overflow detection module 22, if a transient phase is detected and if an overflow of a lower threshold is detected, an underflow correction value TopUS is transmitted by the overflow detection module lower 23.

[0074] Calculation module 24

The calculation module 24 is represented schematically in FIG. 3 and has the function of determining the first correction component K1 from the correction values ​​TopCS, TopOS and TopUS.

[0076] Response time correction module 3 (Figure 9)

The response time correction module 3 is represented schematically in FIG. 9. It comprises a stability detection module 31 configured to detect that the setpoint parameter yc is indeed stable. In practice, the stability detection module 31 verifies that the setpoint parameter yc varies slightly, that is to say, without any transient. The stability detection module 31 produces a stability confirmation signal ConfS intended for a calculation module 33.

[0078] Toujours en référence à la figure 9, le module de correction de temps de réponse 3 comporte un module de détermination d’une plage de tolérance 32 autour du paramètre physique de consigne yc. Dans cet exemple, on détermine préalablement un gabarit bas GabBTR et un gabarit haut GabHTR et on en déduit une plage de tolérance définie entre yc-GabBTR et yc+GabHTR. La plage de tolérance correspond au trainage qui est considéré comme admissible entre le paramètre physique y et le paramètre physique de consigne yc. De manière préféré, les gabarits GabBTR, GabHTR sont déterminés à partir de la réponse du système réel F(p) en boucle fermée yBF suite à la réception d’un paramètre physique de consigne yc. Ainsi, les gabarits GabBTR, GabHTR sont déterminés de manière idéale pour définir une plage de tolérance de référence.

The response time correction module 3 further comprises a calculation module 33 configured to determine a second gain constant K2 if the physical parameter y does not belong to the tolerance range during a confirmed stationary phase. by the confirmation signal ConfS.

In practice, the response time correction module 3 makes it possible to monitor any delay of the physical parameter y with respect to the setpoint physical parameter yc. Such a delay can for example be linked to a parameter gain K which has been increased too much, in particular, following detection of an instability. A second gain constant K2 of negative value makes it possible to improve the response time. Advantageously, several point corrections are thus carried out.

As illustrated in FIG. 10, the physical parameter y deviates from the monitoring range in the zones P31 and P32, which reflects a delay in regulation. The response time correction module 3 determines the second gain constant K2 according to the number of deviations outside the monitoring range.

[0082] Module 4 for determining the parameter gain K (Figure 2)

As illustrated in Figure 2, the module 4 for determining the parameter gain K is configured to sum the first gain constant K1, determined by the stability correction module 2, and the second gain constant K2, determined by the response time correction module 3. Preferably, a static gain constant is also added to determine the parameterization gain K. In fact, the parameterization gain K is modified in real time.

Since the first gain constant K1 is positive and the second gain constant K2 is negative, the parameter gain K is dynamically modified during regulation in order to adapt to changes and correct any derivative.

Example of implementation of the regulation method with dynamic optimization of the adjustment gain K

According to the regulation method according to the invention, the method comprises a step of monitoring the stability during the regulation of the physical parameter, a step of determining a first gain constant K1, of positive value, during detection of an instability during the regulation of the physical parameter y, a step of monitoring the response time during the regulation of the physical parameter in the event of absence of instability, a step of determining a second gain constant K2, of negative value,upon detection of a delay during regulation of the physical parameter and a step of determining the parameter gain K of the corrector C(p) from the first gain constant K1 and the second gain constant K2 of so as to guarantee the stability of the regulation while optimizing the response time.

By way of example, in order to present the advantages of the invention with regard to the prior art, FIG. 11 shows the evolution of the physical parameter y as a function of the set physical parameter yc (curves higher) when the real system F(p) experiences a periodic delay RET (middle curve) for a static parameter gain (lower curve). In this example, the delay RET varies between 0.2s and 3s.

As illustrated in FIG. 11, when the delay RET becomes significant, significant instabilities of the physical parameter y appear, both in stabilized operation and during transients. In addition, the speed is not optimal. Such regulation is not satisfactory.

With reference to FIG. 12, the evolution of the physical parameter y as a function of the physical parameter yc is shown (upper curves) when the real system F(p) undergoes a periodic delay RET (middle curve) for a dynamic parameter gain (lower curve) as optimized by the regulation system REG according to the invention.

As illustrated in FIG. 12, when the delay RET becomes significant, there appears an instability of the physical parameter y which is corrected reactively by an increase in the parameter gain K. This increase is linked to the stability correction module 2 which increased the first gain constant K1 following the detection of the TopIS instability. When the delay RET becomes smaller, the past increase in the parameter gain K penalizes the response time and introduces lag. This lag is corrected reactively by a drop in the parameter gain K. This drop is linked to the response time correction module 3 which has reduced the second gain constant K2 following the detection of a lag.

Thanks to the invention, the stability and the response time of the regulation system REG are corrected dynamically and reactively over time. The performance of the REG regulation system is optimal due to its self-adaptation.

CLAIMS

1. Regulation system (REG) of a physical parameter (y) of a real system F(p) of a turbomachine from a setpoint physical parameter (yc), the regulation system (REG) having a response time and including:

- a corrector comprising a correction function C1 (p) and a parameter gain K,

- a theoretical inverse transfer function F 1 (p) of the real system F(p), and

- an optimization system (OPTK) of the parameter gain K during regulation, the optimization system (OPTK) comprising:

i. a stability correction module (2) configured to determine a first gain constant K1, of positive value, upon detection of instability of the regulation system (REG) during regulation of the physical parameter (y),

ii. a correction module (3) for the response time of the regulation system (REG) configured to determine a second gain constant K2, of negative value, upon detection of a delay during regulation of the physical parameter (y ), the stability correction module (2) being configured to inhibit the response time correction module (3) upon detection of an instability (TopIS) during the regulation of the physical parameter (y), and

iii. a determination module (4) configured to determine the parameter gain K as a function of the first gain component K1 and the second gain component K2 previously determined.

2. Regulation system (REG) according to claim 1, in which, a difference (e) being defined between the physical parameter (y) and the set physical parameter (yc), the stability correction module (2) comprising a stability detection module (21) configured to compare the deviation (e) with a high deviation threshold (SH) and with a low deviation threshold (SB), the stability detection module (21) is configured to detect instability (TopIS) if the deviation (e) is successively greater than the high deviation threshold (SH) then lower than the low deviation threshold (SB).

3. Regulation system (REG) according to claim 2, in which, the deviation (e) oscillating during an instability (TopIS), the stability detection module (21) is configured to count the oscillations following detection of instability (TopIS) and to determine a stability correction parameter (TopCS) as a function of the number of oscillations counted (NB-osc), the first gain component K1 depending on the stability correction parameter (TopCS ).

4. REG regulation system according to claim 3, in which the stability correction module (2) is configured to reset the number of counted oscillations (NB-osc) in the event of detection of a transient phase generated by a significant variation of the setpoint physical parameter (yc).

5. Regulation system (REG) according to one of claims 1 to 4, in which the stability correction module (2) comprises an overflow detection module (22) configured to determine an overflow parameter (TopOS ), the first gain component K1 depending on the overshoot parameter (TopOS).

6. Regulation system (REG) according to claim 5, in which, a deviation (e) being defined between the physical parameter (y) and the setpoint physical parameter (yc), the overflow detection module (22) is configured to start a period of monitoring the difference (e) following a significant increasing variation of the setpoint parameter (yc), the overflow detection module (22) being configured to compare the difference (e) to at least one overshoot threshold (SD1, SD2) during the monitoring period, the overshoot detection module (22) being configured to detect an overshoot (TopOS) if the difference (e) is greater than the upper overshoot threshold (SD1 , SD2).

7. Regulation system (REG) according to one of claims 1 to 6, in which the response time correction module (3) is configured to determine a tolerance range around the setpoint physical parameter (yc) and to determining a second gain constant K2 if the physical parameter (y) does not belong to the tolerance range.

8. Method for regulating a physical parameter (y) by implementing a regulation system (REG) according to one of claims 1 to 7, the regulation method comprising:

- a stability monitoring step during the regulation of the physical parameter (y)

- a step for determining a first gain constant K1 , of positive value, when an instability (TopIS) is detected when regulating the physical parameter (y),

- a step for monitoring the response time of the regulation system REG during the regulation of the physical parameter (y) in the absence of instability (TopIS),

- a step of determining a second gain constant K2, of negative value, upon detection of a delay during the regulation of the physical parameter (y),

- a step of determining the parameter gain K of the corrector C(p) from the first gain constant K1 and the second gain constant

K2 in order to guarantee the stability of the regulation while optimizing the response time.

9. Computer program comprising instructions for the execution of the steps of the control method according to claim 8 when said program is executed by a computer.

10. Electronic control unit for a turbomachine comprising a memory including instructions of a computer program according to claim 9.

1 1. Turbomachine comprising an electronic unit according to claim 10.