FR2727809A1 - SPEED CONTROL DEVICE FOR TURBO-ALTERNATOR GROUP - Google Patents

Abstract

Speed regulation device for a turbo-alternator group, characterized in that it comprises a first regulator (I) for small movements intended to ensure an operational power control, this first regulator (I) being optimized to maintain the stability of the alternator around its steady state whatever the network configuration considered, a second regulator (II) for large movements intended to effectively oppose the most violent disturbances likely to occur, and a detector device (IV) variations in the electrical power signal ensuring the switching of the first and second regulators (I, II) with a view to controlling by the first regulator or by the second, the degree of opening of the high pressure and low pressure valves of the turbine turbo-alternator drive.

Description

The present invention relates to the provisions

  speed control devices for turbo-alternating groups

  nators. A primary frequency / power regulator for a turbo generator group aims on the one hand to

  participate in maintaining the production / consumption balance

  summation and on the other hand to ensure a good transient operation of the alternator. In this case,

  to help avoid loss of synchronism during incidents

  teeth occurring on the network (in the case of short circuits) and limiting overspeed when traveling in a separate network

or islands.

  To achieve these two objectives, the

  conventional ingredients have two characteristics

additional.

  On the one hand, the objective of participation in

  the balance between production and consumption is generally ensured

  ral, by a proportional regulator whose input is the signal Po-k.Af, where P. is a power setpoint and Af

  is the difference between the measured frequency and the frequency

nominal network frequency (50Hz).

  On the other hand, this regulator having perfor-

  mances very limited, it is necessary most of the time, in large transients, to supplement it by a particular device allowing the fast closing of the valves of the turbine of drive of the alternator following an islanding, a passage in separate network or a

  significant short circuit near the power plant. This arrangement

  tif, which can be called a regulator in "large movements" acts on exceeding a speed, acceleration or imbalance threshold between mechanical and electrical power. It is indeed imperative to permanently maintain the speed of the alternator within the admissible operating range. This being very narrow, it

  it is necessary to react energetically when a disturbance

  important bation appears on the network.

  In particular, an all-device is known

  or nothing, called threshold accelerometer, used to ensure transient operation. The role of the threshold accelerometer consists in rapidly closing the steam intake valves and consequently in rapidly reducing the engine torque when the acceleration of the group is greater than a threshold fixed in advance, which varies according to the type of group. For example, for the groups in the CP1 stage, this threshold is set at a value equal to 35% of the nominal acceleration, while for the 1300 MW power groups the threshold is 50% of the nominal acceleration. This procedure presents several

  disadvantages, both with regard to behavior

  much in small movements than that in large movements-

ment.

  a) Behavior in small movements.

  The electric power delivered by the alter-

  nator is generally not measured. Conventional regulators, which are called opening controls first of all generate a Po-k.Af signal, which they then apply directly to the input of the valve control device by means of a module providing a reversal of the static flow-opening characteristic of the high pressure valves. The low pressure valve does not

  little or no participation in behavior in small movements.

  This device does not allow very precise regulation of the permanent regime obtained. Indeed, it assumes that the static characteristic of the valves is very well known, and that it does not change over time. However, the wear caused by the movement of the valves modifies this characteristic, which decreases the precision.

  of the setting obtained by this device.

  In addition, this regulator has only one degree of freedom (adjustment parameter) which is the parameter k (the reverse of which is commonly called droop). It is therefore not possible to independently regulate the steady state and the transient behavior of the regulation. In particular, with a current droop (k = 25), we

  obtains a transitional regime which solicits fairly violently

  the valves, and therefore causes more wear

than that strictly tolerable.

  The Applicant also uses a device called power control which uses a measurement

  of the electrical power P. supplied by the group.

  In this device a Po-P signal, - K.Af is

  applied at the input of a PI regulator (Proportional -

  Integral). This device avoids the first drawback of the opening control. On the other hand, its dynamic characteristics are very bad: during a short circuit for example, it generates at the first instant a signal which tends to open the valves, then

  that they should be closed as soon as possible.

  In a separate network, its behavior is even worse since it can lead to a regime of

  very oscillating, even unstable operation.

  These drawbacks mean that this regulator is only used in steady state, and when in service, it is associated with a device which switches on

  the opening control as soon as a fault is detected.

  b) Behavior in large movements.

  The almost universally used device

  in the world is similar to the threshold accelerometer.

  Variants exist according to the manufacturers, but the basic principle is always the same: a device estimates the acceleration of the group; as soon as it is found to be abnormal, a rapid closure of the

  high pressure and low pressure turbine valves.

  This makes it possible to avoid as much as possible the loss of synchronism in overspeed of the group, by significantly increasing the

  time limits for eliminating short circuits.

  The major drawbacks of this system are its rusticity, and in particular the fact that it is an all-or-nothing device, which makes it extremely

  difficult to choose the switching threshold. If the acceleration

  tion of the group is below the threshold, the threshold accelerometer does not act. For an acceleration slightly above the threshold, the valves are closed for a certain time. This represents an extremely brutal action, while a partial closing of the valves

  would have been enough to maintain the stability of the group.

  The obvious annoyances are on the one hand that one causes a very important wear of the valves, and on the other hand that one decreases for a certain time the

  power produced by the group, which affects the balance

  bre overall production-consumption. It is then necessary that other groups on the network compensate for this imbalance, and the disturbance thus caused increases the fatigue of groups which were not a priori concerned by the first incident. Another drawback stems from the fact that the disturbance caused by the rapid closing of the valves causes a jerk which causes oscillations.

  electromechanical at the alternator. These oscillated

  the frequency of rotation of the group in many cases cause other untimely stresses

  the threshold accelerometer. This process increases

  still lies the deficit of power produced, and can lead in certain configurations of the network to a regime of maintained oscillations, even to losses of

synchronism under speed.

  It should be noted that the estimation of the acceleration

  tion of the group is generally carried out by deriving the speed measurement. This procedure has the disadvantage of amplifying the noise of the speed measurement which is often very important. It is then necessary to very strongly attenuate this noise by filtering and this is often the cause of a non-negligible delay in the detection of faults. We also know a regulator called p-rec

  (see D. Deloues "Advantages of the Microrec System" GEC.

  Alsthom Technical Review, 1990 n 3) significantly better than the threshold accelerometer. It is the combination of a so-called dynamic correction element and

  of a high pressure dynamic interaction element -

  low pressure. The dynamic correction element is a

  regulator of large movements acting in a conti-

  naked, not all or nothing. It allows the regulator gain to be temporarily increased when a fault is detected, and therefore to react with the action exactly appropriate to the extent of the fault. The dynamic interaction then makes it possible to coordinate the action of the high pressure and low pressure valves by forcing the low pressure valve to follow the movements of the high valve

  pressure during the first moments of the fault.

  However, the methodology from which this regulator was designed is unclear. Optimizing its setting therefore seems very delicate. On the other hand, it does not seem that this regulator has satisfactory resynchronization capacities, nor that any measure has

taken for this purpose.

  Another important characteristic of a regulator is its ability to restore synchronism to a machine that lost it during a fault.

  too important, because a machine or a region out of sync

  nisée no longer supplies active power to the network and must be disconnected after a few seconds. If she has the option to resynchronize during this time

  time, this saves other machines from having to compensate

  be the power deficit. It is therefore important to put back into service as soon as possible a machine that has lost

synchronism.

  A good cruise control must therefore ensure correct behavior in both small and large movements. These two objectives are too antagonistic to be achieved by means of a single functional module. The invention aims to remedy the drawbacks of known regulators by creating, from clear methodological principles, a speed regulation device which has the performance required for each

  situation and at every moment of its operation.

  Its object is therefore a regulatory system

  Speed lation for turbo-generator group, charac-

  terized in that it includes a first regulator for small movements intended to ensure an operational power control, this first regulator being optimized to maintain the stability of the alternator

  around its steady state whatever the configuration

  network ration considered, a second regulator for large movements intended to effectively oppose the most violent disturbances likely to occur, and a device detecting variations in the electric power signal ensuring the switching of the first and second regulators in view of the order by

  the first regulator or the second of the degree of openness

  high pressure and low pressure valves

  turbo-generator drive turbine.

  The invention will be better understood on reading

  the description which follows, given only as

  example and made with reference to the accompanying drawings, in which: - Fig.1 is a block diagram representing

  a linearization device for the low pressure valve

  sion of a regulation device according to the invention; - Fig.2 is a general block diagram of the regulation device according to the invention;

  - Fig.3 is a block diagram of the regulation

  teur of small movements which is part of the regulating device according to the invention;

  - Fig.4 is a block diagram of the regulation

  large movements involved in the construction of the regulation device according to the invention; and

  - Fig.5 is a block diagram of an arrangement

  switching factor used in the construction of the

  regulating device according to the invention.

  For the reasons given above, neither access

  Threshold lerometer nor the improvements made to arrive at the regulator with combined dynamic correction element and dynamic interaction element (p-rec) do not fully meet the needs expressed. These two regulators have satisfactory performance for correct operation of the system.

  electric, and their use gives relatively satisfactory

  faction for the usual operating regimes.

  The device described below in fact makes it possible to obtain superior performance, which is sensitive

  both steady and disturbed.

  The speed regulation of turbo groups

  alternators is a subject that is at the origin of a

  extremely abundant literature, especially abroad.

  Most recent articles are unfortunately academic research whose practical application seems difficult apart from academic examples.

  presented. In particular, the performance of regulators

  The presented authors are highlighted on an always very limited number of examples. While it is easy to achieve each of the objectives described above, none of the work

  presented does not seem to satisfy all of the constraints

  your requested. The most significant achievements which have been the subject of an industrial application are quite limited in number. The study by Y. Morioka, H. Tanaka, T. Kojima, T. Suzuki, S. Hanada. "Application of Multivariable Optimal

  Controller to Real Power Systems ". IEEE PES Winter Mee-

  ting, February 1994 presents results which are integrated

  in small movements. Nothing is specified on the other hand as for the behavior in large movements of the device described, and in particular on the way in which are reached the contradictory objectives of obtaining an effective action in large movements without requesting too much

the valves in small movements.

  NEI-Parsons for its part in P.A.L.

  Ham Digital Turbine Governor and AVR Systems. 10th Int. Conf. on Power Stations, Liège, Belgium, September 1989,

  a regulator associated with ultra-movement valves

  quick. The performances obtained are however not better than for the regulator with dynamic correction element and dynamic interaction element combined, whereas, from a theoretical point of view, the increase in

  valve speed should lead to performance-

  these increased. We can therefore deduce that the control logic of these valves is not optimal for rejecting

  at best the disturbances present on the network.

  The regulation device which will be described with reference to FIGS. 1 to 5 makes it possible to satisfy the

objectives stated above.

  This is in fact a device which calculates the position of the high pressure and low pressure valves of the alternator drive turbine at

from available measurements.

  The equations for the alternator drive turbine are as follows: Tai, = - x1 + u1 Tsi2 = x1 - x2x3 T1X3 = - x3 + U2 TBx4 = - X4 + x2x3 o xl is the flow or high pressure HP , x2 is the superheater pressure, x3 is the opening of the low pressure valve BP and x4, the low pressure BP. We denote by z the outflow from the turbine superheater (z = x2x3), but this is not an additional state variable. ul and u2 are the controls for the HP valve and the BP valve respectively. TN denotes the equivalent time constant of the body HP with the valve control chain, Ts is the time constant of the reheater, T1 the time constant of the valve control chain BP and TB represents the time constant of the body BP . This represents a non-linear model whose behavior in small movements varies according to the operating point of the alternator. These variations make the design of cruise control more difficult: a cruise control optimized for a given operating point will provide poorer performance for others

operating points.

  It is easy to note that the only

  linearity of the system comes from the low pressure valve

  if we. The idea is therefore to insert a linear device

  between the cruise control and the LP valve. Instead of simply applying the output of the regulator to the control of the LP valve, it is applied at the input of the linearizing device. This then calculates the control of the BP valve, so that the non-linearity due

  at the BP valve is not visible from the outside.

  In other words, the association turbine-linear regulator-

  health exhibits linear behavior (which therefore does not vary depending on the operating point). This can

  be summarized by the diagram in Fig.l.

  Fig.l is a block diagram showing a turbine 1 for driving an alternator not shown. The outlet of turbine 1 is connected by a

  part at the input of a linear-quadrati- type regulator

  that, LQ, and secondly to a command input of a

  non-linear looping circuit 3.

  The output of circuit 3 is connected to an input 5 for controlling the low pressure valve of the

turbine 1.

  The regulator 2 has a first output connected to an input 4 for controlling the high valve

  HP pressure from turbine 1 and a second connected outlet

  ted to another input of the nonlinear loop circuit 3.

  In Fig. 2, the synop-

  general tick of the regulation device according to the invention

  tion. This device includes a regulator for small movements I and a regulator for large movements Il whose inputs are connected via

  conventional acquisition cards with sensors not shown

  felt: - high pressure body pressure, - superheater pressure, - low pressure valve position, - low pressure body pressure, - Af, and

- electrical power.

  The assembly constituted by regulators I and II, and switch III controlled by the detector

  IV defects is implemented for example on a micro-

  PC type computer and controls the actuators of the

  high pressure HP and low pressure BP valves.

  We will subsequently designate v2 as the input of the

linearizing device.

  This device ensures operation that is both linear and as close as possible to the natural operation of the turbine. It is easy to verify that it can be achieved by the following function: X20- V2 = (u2-1) x2 + Tl / T, (xl-z) (z / x2-1) (1) where x20 denotes the design point of the regulator, which can for example be the nominal operating point

of the turbine: x20 = 1.

  This regulator ensures that the turbine behaves

  following linear ment: TiX1 = - x1 + u1 TsX2 = X1 - Z TlZ = T1 / T. (xl-z) + X2 - z + v2 Tai4 = - X4 + z We will now describe the regulators for small and large movements which calculate ul and v2 to

  from the measurements available on the turbine.

  The regulator for small movements is shown

  felt as a block diagram in Fig. 3.

  This regulator includes a first integrator 7 to develop the quantity: E1 = Jot (Po-P. - kAf) dt (2)

  which guarantees the absence of static error in perma-

  nent. The integrator 7 receives at its inputs, the output signal Pe from a non-electric power sensor

represented and a signal Af.

  To obtain a correct transient behavior materialized by a sufficiently fast response time and no weakly damped oscillations, it is necessary to use for the regulation the important variables of

alternator operation.

  In addition, the aforementioned signals Pe and Af, it is necessary

  measure the vapor pressure values in the different

  rents body designated hereinafter by Pi, i representing the body of the high pressure valve, the superheater, the

low pressure valve body.

  This leads to a regulator which controls the position of the high pressure HP and low pressure BP valves (ul and u2) according to the following law: [ulv2] T = f (X) with X = [El E2 Af P. Pi ] T with E2 = / '(u2 - 1) dt (3) which ensures that the low pressure valve remains

open to 1 in steady state.

  Equation (3) is shown in Fig. 3 by

a second integrator 8.

  The outputs E1 and E2 of the integrators 7 and 8 are applied to a circuit 9 generating signals ul and v2 which receives on four other of its inputs respectively an output signal from a pressure sensor connected to the high pressure body HP, d ' a superheater pressure sensor, a low pressure valve position sensor and a low body pressure sensor

pressure (not shown).

  Circuit 9 is a circuit allowing the implementation of a linear function, which leads to the following regulator: [ulV2] T = K (X) (4) o K is a matrix which comprises two lines and as many

  columns than measured variables.

  K can be calculated for example by a method of linear quadratic type (LQ). This method is particularly appropriate in the present case, because the different variables brought into play in the system are weighted: the weighting on El makes it possible to adjust the speed with which the setpoint is reached; that on X makes it possible to adjust the transient behavior of the group, and those on ul and u2 define the respective action of the two valves in the achievement of these objectives. We weight

  very strongly u2 so as to limit its action as much as possible.

  Only the HP valve is actually really useful in

small movements.

  The output of circuit 9 delivering the signal v2 is connected to a circuit 10 for linearizing the control of the low pressure valve implementing equation (1). The large movement regulator is going to be

  described with reference to Fig. 4. This regulator is simi-

  laire to that described with reference to Fig.3 except that it includes a low-pass filter 11 connected to the input of the integrator 7 connected to the electric power sensor. This regulator which enters into service after a fault must have an energetic action in order to avoid

  maximum risk of loss of synchronism. Functional-

  Its structure is very close to that of the regulatory system.

  of small movements, but two fundamental differences

  tales explain his different behavior.

  On the one hand, the gain matrix K of equation (4) has significantly larger values than in small movements. This makes it possible to obtain greater valve movements for a similar variation of the

  measured quantities (frequency of rotation for example).

  This gives the regulator greater bandwidth and enables the balances that had been broken following the fault to be restored as quickly as possible. This gain matrix can be calculated as before by an LQ method

  by increasing the weights corresponding to and to El.

  The weighting matrix on the controls also makes it possible to choose the respective participation of the two valves HP and BP in maintaining the balance. In this case, we wish a parallel movement of the two valves

  to make the most of their efficiency.

  On the other hand, and this is one of the points which constitute

  kills a fundamental difference from the regulatory

  Known factors, the signal E1 is no longer calculated in accordance with equation (2). It checks the equation: E1 d.o (Po-P.f - kAf) dt

  o P, is the value P. filtered by the low-pass filter 11.

  We will choose for example an order Butterworth filter

2, 1 Hz cut-off pulse.

  This procedure improves the behavior of the regulator for the following reason. The behavior at low frequencies is not modified, and E1 always makes it possible to reach the same steady state. On the other hand, during a fault, when P. varies rapidly, P does not move in the first moments. Therefore, E1 represents the integral of the variations of the frequency, which amounts to saying that one takes into account the angular variations of the group considered compared to the rest of the network. In practice, we obtain the following effect: when the overspeed is controlled, we check if the group is ahead of the rest of the network. If yes, keep the valves closed for a while to slow down

  the group and compensate for this advance.

  It should be noted that this prolonged closure of the valves is favorable to resynchronization. When the group has lost synchronism, its speed is higher than the reference speed. The relative angle therefore increases rapidly, which results in a variation of E1. The corresponding gains in the matrix K then make it possible to keep the valves closed longer, which of course has a favorable effect for

  resynchronization. Simulations confirmed the effectiveness

effectiveness of this device.

  The switching device between the regulators

  tor for small movements and the regulator for large movements will now be described with reference to Fig. 5.

  As soon as a disturbance occurs, it is essential

  thinkable to react quickly. To do this, it is necessary to detect the disturbance as quickly as possible, in order to

  take appropriate action.

  The fault detection techniques used in France are all based on the same principle. The starting point is that it is necessary to avoid that the group reaches too high a speed (or reminds that there is a protection triggering the group as soon as the rotation speed exceeds a fixed threshold, 53 Hz for example). To anticipate, the speed measurement is electronically derived in order to estimate the acceleration, and we compare this acceleration to a threshold. As soon as the threshold is crossed, we switch to the regulator for large movements. The disadvantages of this method come from the fact that the speed measurement is derived. As this measurement is very loaded with noise (especially by

  torsional oscillations of the tree line), the estimate

  tion of this derivative is unreliable and often causes

  nuisance trips. To limit these triggers

  we have to filter the derivative, which intro

  due to delay which is detrimental to entry into action

  fast movement regulator.

  To overcome these drawbacks, the Applicant

  thought of going back to the causes of the speed variation.

  According to the equation of rotating masses, the acceleration of the group is due to the difference between the engine torque supplied by the turbine and the mechanical torque sent to the network. Instead of deriving the speed measurement, it is

  therefore natural to use the measurement of the electric power

  that and look at its variations. Since we don't have

  performed bypass, this signal is not very loaded with noise.

  It is therefore useless to filter it too vigorously, and then one avoids introducing harmful delay to the good performances of the whole.

  As soon as a fault is detected, we then switch to the regulator for large movements (Fig. 4). In order to avoid a sudden jerk on the valves, we add a so-called anti-jerk device which ensures that the regulator output which is not in service follows that

  from the other regulator. When switching, the toggle

  between the two regulators is then done without discrepancy

continuity.

  Switching in the opposite direction is made by observing the alternator speed and power measurements. As soon as one has reached the equilibrium state (this generally lasts from a few seconds to a few tens of seconds), one can again switch to the regulator of small movements in order to save the valves. The switching device shown in Fig. 5 includes a circuit 12 for measuring the electric power delivered by the group and the output of which is connected to a threshold circuit 13 which controls the passage of the regulator for small movements (Fig. 3) to the regulator for large movements (Fig. 4) or vice versa depending on the

process described above.

  The device which has just been described makes it possible to obtain performances superior to all that is currently known, both in steady state and in

disturbed regime.

* It performs power control correctly, and therefore ensures perfect setpoint monitoring

  in steady state, even in the presence of uncertainty

  turbine valve modeling.

  This power control functions correctly whatever the network topology considered, that is to say both on an intercon-

connected only on an isolated network.

  The transition from the regulator for small movements to the regulator for large movements is rapid and is

  insured with little risk of nuisance tripping.

  In addition, unnecessary trips are much less annoying than with the threshold accelerometer since the action of the regulator for large movements is not all-or-nothing. There is therefore no major disturbance introduced into the network, nor any deficit

temporary production.

  The regulator for large movements has an energetic but perfectly balanced action. It guarantees shorter short circuit elimination time limits than those usually observed, while damping oscillations faster than the threshold accelerometer.

  electromechanical following the fault.

  The global approach to synthesizing the regulator

  allows great flexibility in the realization of

  ters for different models of turbines. It is easy to optimize the behavior of regulators for small and large movements using the different weighting matrices of the linear quadratic device LQ. One can for example adjust the bandwidth of the regulators to find the best compromise between the performance of the regulation and the wear of the valves, and also decide the respective amplitude of the movements of the HP and LP valves. In small movements, for example, we will ask that the BP valve move very little, while in large movements, we will prefer a parallel movement of the two

valves.

  The good resynchronization capabilities of the regulator make it possible to protect against the most serious incidents, which ensures good safety at

  network even in the most disturbed configurations.

  Finally, the linearization of the BP valve ensures optimal behavior of the regulation for all

operating regimes.

Claims (7)

  1. Speed regulation device for a turbo-alternator group, characterized in that it comprises a first regulator (I) for small movements intended to ensure an operational power control, this first regulator (I) being optimized to maintain the stability of the alternator around its steady state whatever the network configuration considered, a second regulator (II) for large movements intended to effectively oppose the most violent disturbances likely to occur, and a device (IV ) detector of the variations in the electrical power signal ensuring the switching of the first and second regulators (I, II) for the control by the first regulator (I) or by the second (II) of the degree of opening of the high valves pressure and   low pressure of the turbo drive turbine alternator.   2. Regulation device according to the claim   cation 1, characterized in that the first regulator (I) for small movements includes a first integrator (7)   receiving at its inputs a signal Pe of electric power   than delivered by the alternator and a signal Af of different   between the measured frequency and the nominal frequency   from the network, a second integrator whose input is connected   tee to a circuit (10) for linearization of the control of the low pressure valve of the drive turbine   the alternator, the outputs of the first and second integration   teurs being connected to a circuit (9) for implementing a linear function [u1v2] T = K (X) in which ul is the control signal of the high pressure valve, v2 is the output signal of the circuit ( 9) controlling the low pressure valve, K is a matrix with two lines and as many columns as there are measured variables, said circuit   (9) also having entries to which are applied   pressure signals from the high pressure body, from the low pressure body, from the turbine superheater and from the position of the low pressure valve, and in that the output of the circuit (9) for implementing said linear function is in further connected to the input of the linearization circuit (10). 3. Regulation device according to one of the   claims 1 and 2, characterized in that the second   regulator (II) for large movements, comprises a first integrator (7) receiving at its inputs a signal Pe of electrical power delivered by the alternator and a signal Af of difference between the measured frequency and the nominal frequency of the network, a second integrator   the input of which is connected to a linear circuit (10)   sation of the control of the low pressure valve of the alternator drive turbine, the outputs of the first and second integrators being connected to a circuit (9) for implementing a linear function [u1v2] T = K (X) in which u1 is the control signal for the high pressure valve, v2 is the output signal from the circuit (9) controlling the low pressure valve, K is a matrix with two lines and as many columns as there are   of measured variables, said circuit (9) also having   their inputs to which pressure signals of the high pressure body, of the low pressure body, of the turbine superheater and of the position of the low pressure valve are applied, and in that the output of the circuit (9) for implementing said linear function is also connected to the input of the linearization circuit (10), said second regulator (II) for large movements further comprising a low-pass filter (11) connected to the input of the first integrator (7) receiving the signal Pe of electric power.   4. Regulation device according to the claim   cation 2, characterized in that for the first regulator (1) for small movements, the first integrator (7) is intended to develop a quantity E1 = t (pp, - kAf) dt in which P, is a power setpoint, P. is the electrical power supplied by the group, Af is the difference between the measured frequency and the nominal frequency of the network, and the second integrator (8) is   intended to develop a quantity E2 satisfying the equa-   tion E2 = Sot (U2 - 1) dt in which u2 is the signal of   control of the low pressure turbine valve.   5. Regulating device according to one of the   claims 2 and 3, characterized in that for the   second regulator (II) for large movements, the   first integrator (7) is intended to develop a large   deur E1 satisfying the equation El = 5t (Po-P.f - kAf) dt in which P, is a power setpoint, Pf is the value of the electric power P. supplied by the group,   filtered by the low-pass filter (11).   6. Regulating device according to one of   Claims 2 to 5, characterized in that the circuit   (10) linearization of the low pressure valve control implements the equation: x20 v2 = (u2-1) x 2 + T1 / T. (xl-z) (z / x2 - 1) in which x20 denotes the point of conception of the   regulator, which can for example be the point of function   nominal turbine operation x20 = 1, v2 is the input signal of the linearization circuit, u2 is the control signal of the low pressure valve, x2 is the pressure of the superheater, Tl is the time constant of the chain   low pressure valve control, T. is the constant   te of superheater time, xl is the flow or the high   pressure, z is the output rate of the superheater.   7. Regulation device according to one of   claims 1 to 6, characterized in that the device   detector of variations in electrical power   includes a circuit (12) for measuring the electric power   plate delivered by the group and the output of which is connected to a threshold circuit (13) which determines the control of high pressure and low pressure valves   the turbine by the first regulator for small movements   elements (I) or by the second regulator for large movements (II).