FR2970526A1 - DIFFERENTIAL REGULATOR FOR WINDMILL - Google Patents

Abstract

Differential controller for wind turbine. The invention relates to a method for regulating (20) an energy collecting device (1), of the wind or tidal type, which comprises at least one rotary drive member (2) intended to be driven by a current of fluid for transmitting energy to a load (3) applied thereto, said method being characterized in that it comprises an observation step during which the evolution of the speed of rotation is observed ( Q) of the drive member in response to a change in its operating conditions and the second derivative (Q ") of this rotational speed evolution (Q) is evaluated, followed by a step of adjusting the speed of rotation (Q). load in which the load (3) is adapted according to said evaluation of the second derivative, and methods for regulating the operation of a wind turbine.

Description

DIFFERENTIAL REGULATOR FOR WINDMILL

The present invention relates to the general field of devices for capturing the energy of a fluid stream, such as turbines, and their regulation, and more particularly the field of wind turbines, and in particular vertical axis wind turbines.

In particular, the present invention relates to a method for regulating an energy sensing device comprising at least one rotary drive member intended to be driven by a stream of fluid in order to transmit energy to a load which is applied to him. The present invention also relates to a control module, a computer program, and a power generation facility relating to such a method. It is known to use rotating machines, in particular of the turbine or wind turbine type, in order to convert the kinetic energy of a moving fluid, such as water, air or steam, into mechanical and / or electrical energy. . In particular, it is known to use wind turbines as a means of driving alternators to exploit renewable energy deposits. If they have undeniable advantages, including ecological, such known devices may however suffer from certain disadvantages. Indeed, the flow of natural fluids currents, including wind, being subject to relatively random fluctuations, sometimes fast or even violent, the exploitation of wind or tidal deposits is generally intermittent, and often delicate. In this respect, it has notably been imagined to regulate the operation of the wind turbines as a function of the wind force applied to them, this regulation being able, for example, to modify the wing of said wind turbine or to vary the wind speed. mechanical or electrical load that is associated with the latter. However, a mechanical modification of the configuration of the wing or the load of the wind turbine generally requires the establishment of additional relatively heavy, bulky mechanisms, likely to degrade the reliability as well as the performance of the wind turbine. In addition, the known control methods can be relatively inaccurate, slow, or even unstable, especially when the wind turbine is exposed to frequent changes of wind, which can sometimes lead, depending on the case, to a braking or even a stop untimely wind turbine, either on the contrary to a runaway of the latter, and in any case to a degradation of its performance. For all these reasons, it remains particularly difficult to size and operate wind turbines effectively, particularly in the "small wind" sector where it is desired to use low power and small wind generators by exploiting particularly disturbed winds. and fluctuating, especially in urban or peri-urban areas. The objects assigned to the present invention therefore aim at proposing a new method of regulating an energy sensing device, of the wind or tidal type, which makes it possible to efficiently and reliably optimize the efficiency of said device whatever the flow conditions of the flowing fluid stream and the fluctuations of said flow. Another object assigned to the invention is to propose a new control method that is particularly simple and economical to implement and operate. Another object assigned to the invention is to propose a new control method that allows a fast, accurate and stable optimization of the operating speed of an energy sensing device. B11612 / FR Another object assigned to the invention is to propose a new particularly versatile control method that allows the regulated device to benefit from a particularly wide operating range. Another object assigned to the invention is to propose a new particularly robust control method, and in particular not very sensitive to the inertia of the sensing device that it controls. Another object assigned to the invention is to propose a new control method particularly effective and well suited to the implementation of the "small wind". Another object assigned to the invention is to propose a new energy production facility, preferably from a renewable energy source, which has an optimized efficiency over a wide range of operation. Finally, another object assigned to the invention is to provide an energy production facility that is autonomous, and inexpensive to manufacture and operate. The objects assigned to the invention are achieved by means of a method of regulating a wind or tidal turbine energy sensing device, which comprises at least one rotary drive member intended to be driven by a fluid stream for transmitting energy to a load applied thereto, said method being characterized in that it comprises an observation step during which the evolution of the rotation speed of the driving member and the second derivative of this evolution of the rotational speed is evaluated, then a load adjustment step during which the load is adapted according to said evaluation of the second derivative. The objects assigned to the invention are also achieved by means of a regulation module intended to be connected to an energy sensing device and containing a regulation circuit making it possible to implement such a method, as well as using a computer program comprising a computer program code means adapted to perform the steps of such a method, and finally by means of a power generation installation comprising an energy capture device

B11612 / FR feeding a load, and a regulator adapted to regulate the operation of the capture device according to said method. Other objects, features and advantages of the invention will appear in more detail on reading the description which follows, and with the aid of the accompanying drawings, provided for purely illustrative and non-limiting purposes, among which: Figure 1 illustrates, in a schematic representation, an installation according to the invention.

FIG. 2 illustrates the output curves of a wind turbine device that can be regulated by the method according to the invention, said curves representing the power produced by said device as a function of the speed of rotation of said device, and for different wind speeds. FIG. 3 illustrates the evolution of the speed of rotation of a wind-sensing device, for a given wind speed, as a function of the load applied to it. FIG. 4 illustrates, in the form of a flow chart, an example of an algorithm corresponding to a method according to the invention. The present invention generally relates to a device 1 for energy capture intended to extract, and more particularly to convert, the kinetic energy of a fluid stream F, as well as a regulation method 20 intended to regulate, and more particularly to enslave, the operation of such a device.

The device 1 comprises at least one rotary drive member 2 intended to be driven by said fluid stream F in order to transmit energy to a load 3 which is applied to said drive member 2. Of course, the device 1 and the process may be suitable for fluids of various kinds, such as air (wind), liquid water (stream, marine current) or any gas such as water vapor. B11612 / FR In addition, they can be adapted to fluid currents both natural and artificially generated by any means. Particularly preferably, the device 1 and the corresponding control method will be particularly suitable for fluctuating fluid streams, preferably natural, and whose intensity, and in particular the flow rate, vary in time, randomly and / or frequent, alternating phases of acceleration and deceleration, and possibly in relatively large proportions within the operating range of the device, for example of the order of 3: 1, 4: 1, even greater than or equal to 5: 1.

Preferably, as shown in FIG. 1, the drive member 2 comprises a rotor 4 provided with at least one blade 5 which allows the fluid stream F to drive said rotor around a rotating rotor. axis of rotation (ZZ), and may for convenience be likened to such a rotor in the following Advantageously, the rotor 4 is supported and guided in rotation by a stator 6, itself fixed or integrated to any support 7, such as a mast, a building, particularly a public, industrial, commercial or residential building, or even a mobile platform or a vehicle, The control method 20 which is the subject of the present invention may of course be used to regulate the operation of any type of sensing device, both horizontal axis, that is to say intended to capture the driving fluid in a direction substantially collinear with its axis of rotation (ZZ ", that vertical axis, that is, that is, intended to capture said fluid in a direction transversely to said axis, however, the axis of rotation (ZZ) will preferably be arranged substantially transversely to the direction of the fluid flow F incident, that is to say such that said flow of fluid approaches the device 1 laterally, in a direction which is not collinear with the axis of rotation (ZZ "and which is instead, preferably, substantially perpendicular to said axis of rotation.

B11612 / FR More particularly, the device 1 can thus constitute a wind turbine or a tidal turbine, preferably with a vertical axis, and particularly preferably a Darrieus type wind turbine, propelled essentially by the lift phenomenon resulting from the circulation of the fluid from and other blades 5.

In what follows, for convenience of description and without this constituting a limitation of the invention, said device 1 will be likened to a vertical axis wind turbine, and the fluid flow F to the incident wind. The load 3 applied to the drive member 2 may be of any kind, the device 1 may equally be used to produce a direct mechanical energy, for example for the drive of any mechanism for example for the drawing of water. water, or pneumatic energy (by driving a compressor), thermal energy (for example by induction by moving magnets vis-à-vis a metal heating body), or, preferably , electrical energy.

In a particularly preferred manner, as illustrated in the figures, said load 3 will comprise a generator 8 intended to produce electric current when it is driven by the drive member 2. Advantageously, said generator 8 may be integrated in the stator 6, between the rotor 4 and the latter, to gain compactness.

If necessary, it can be provided between the drive member 2 and the generator 8 a coupling either direct or through a gearbox, gear reducer type whose transmission ratio may optionally be adjustable. According to the invention, the load 3 is modifiable so that it is possible to modify the latter over time, depending on the drive conditions of the device, and therefore to adapt the resistive torque CR it opposed to the motor torque CM which is generated by the action of the wind on the drive member 2. B11612 / FR More particularly, the invention aims in particular to provide a control method that optimizes the efficiency of the device 1 continuously, dynamically, adapting at each instant said load 3 to the maximum power that can produce the device 1 at the instant considered, depending on the regime of the fluid current F applied to it at this time . The invention also relates of course to a power generation installation 10 comprising a power sensing device 1 supplying a load 3 as described above, as well as a regulator 11 able to regulate the operation of said sensing device 1 preferably in a process according to the invention.

Moreover, the energy produced by the charge, that is to say extracted from the fluid stream and made available, in an appropriate form, at the output of said device 1, may optionally be stored, such a buffer storage allowing in particular a deferred and progressive return and use of the energy produced, even though the mode of production is itself fluctuating.

For this purpose, the installation 10 will preferably include an accumulator 12 capable of storing the energy produced. Depending on the nature of the energy produced by the load 3, this accumulator 12 may be for example pneumatic (compressed air tank), electrochemical (battery), electric (capacitor large capacity) or thermal.

In the latter case, a relatively dense heating body, liquid or solid, may accumulate heat, itself produced for example either by direct induction or by Joule effect in an electric charge circuit 13 having one or more resistors 14 In a particularly preferred manner, as illustrated in FIG. 1, the accumulator 12 will be formed by a hot water tank, intended for example for the production of hot water or domestic heating, the electricity produced by the generator 8 for feeding one or more resistor (s) 14 heating the water contained in said reservoir.

B11612 / FR Advantageously, the energy produced, and especially hot water, can be stored and used locally, at the production site, at the user 15 where the installation 10 is located. Moreover, it is remarkable that the device 1 and the control method can be particularly adapted to the realization of the "small wind", in the form of an individual installation whose dimensions and power are for example compatible with a location on or in the immediate vicinity of buildings of any nature, particularly in inhabited, industrial or commercial areas, and more particularly in an urban or peri-urban environment, and / or which are adapted to the autonomous exploitation of disturbed wind regimes, in particular at a lower or equal height of implantation 20 meters from the ground or roof of the building concerned. In particular, the method may be adapted to devices whose nominal power is substantially between 0.5 kW and 100 kW, and in particular to wind turbines, preferably vertical axis, the scanned surface is of the order of 1 m2 to 30m2. Particularly preferably, the control method will be designed to control vertical axis wind turbines whose nominal power, considered for a wind speed of 12 m / s (about 43 km / h) and including, where appropriate, the efficiency of the generator 8, will be substantially equal to or between 1 kW and 2 kW, 3.5 kW or 5 kW. As an indication, the powers less than or equal to 3.5 kW may correspond to swept surfaces less than or equal to 10 m2. More particularly, said scanned surfaces may be substantially equal to 2.6 m2, 5.9 m2 and 9.7 m2, for respective rotor diameters of 2 m, 2.8 m and 3.3 m, and for respective nominal powers of 1 kW, 2 kW and 3.5 kW respectively. The operating range of said wind turbines may advantageously correspond to incident wind speeds V substantially between 4 m / s (starting threshold) and 16 m / s. If necessary, the installation 10 will allow the safety

B11612 / FR beyond a safety threshold speed, which will be determined in particular according to the power range and resistance of the constituent materials of the wind turbine, and for example of the order of 16 m / s to 25 m / s. This safety can be done for example by braking the rotor with an electromagnetic brake and a ballast resistor. According to the invention, the control method 20 comprises an observation step 21 during which, among other things, the evolution of the rotational speed Q of the drive member is observed and the derivative is evaluated. second Q "of this evolution of the speed of rotation 0, then a load adjustment step 22 during which the load 3 is adapted according to said evaluation of the second derivative 0". Advantageously, such a method makes it possible to observe the evolution over time, preferably over a period corresponding to an iteration of the regulation cycle, of the dynamic behavior of the device 1, and more particularly of its speed of rotation, in response to a disturbance (an initial, punctual or lasting change) of its operating conditions, and in particular in response to a variation in the speed of the fluid flow F, which affects the motor torque CM, and / or in response to a controlled change in the load 3, which affects the resistive torque CR, in order to deduce characteristics of this evolution, by an analysis of the evolution function thus identified, the changes to be made to the load to optimize the efficiency of the device. As will be described in greater detail in the following, the inventors have discovered that such information relating to the angular velocity and acceleration of the driving member 2, and more particularly to the trend of evolution of the acceleration, directly and faithfully reflect the dynamic behavior of said drive member, and therefore the state of the device at any time, and that they can be advantageously exploited not only as a reliable indicator of wind changes detection, which makes it possible to evolve in a timely and "intelligent" way the nature of the regulation to adapt it to the wind regime, but also like a base of analysis for a search algorithm B11612 / FR of the optimum operating point which is able to react and converge very quickly. In a particularly advantageous manner, the method according to the invention makes it possible to analyze in almost real time the dynamic behavior of the drive member 2, and more particularly of the rotor 4, and consequently to detect very rapidly the evolution of the its behavior during changes in the flow regime of the fluid stream F, and in particular during the wind acceleration or deceleration phases, when the latter is blowing gusty or when it undergoes any marked disturbance, a transition or alternating transitions from one established regime to another. More particularly, the method aims to monitor the evolution trends of the angular acceleration of the rotor 4, being recalled that said angular acceleration corresponds to the first derivative of the rotational speed 4 = d ()) and is representative of the torque which results from the difference between the motor torque CM and the resistive torque CR. Advantageously, the inventors have indeed found that it is possible to regulate particularly efficiently, reliably and reactively, with a very low response time, the operation of the device 1 by modulating its charge 3 on the basis of evolution trends motor torque CM obtained by the evaluation of the second derivative Q "and its evolutions over time, this information giving particular information on the variations of said engine torque under the effect of wind fluctuations As illustrated in FIG. 4, the evaluation of the second derivative Q "is preferably carried out on the basis of three discrete successive measurements of the speed of rotation 0.

Thus, during said observation step 21, one can successively record, at three distinct moments tn, 1, tn, 2 and tn, 3 distributed according to known or timed time intervals, the speed of rotation, or at any other time. the least an image of said speed of

B11612 / FR rotation, the three corresponding values being denoted respectively Qn, 1, Qn, 2 and Qn, 3. Unless otherwise indicated, the index "n" indicates the nth iteration of the method 20, and more particularly of the current step (here the observation step).

The discrete second derivative of the rotational speed, here expressed with respect to the time t, can then be evaluated by the following formula 1 (by analogy with the properties of the limited developments):

0,, n (t) _ N _ SZn, 3 + SZn, I - 2 x nn, 2 D (tn, 3 - tn, 2) x (tn, 2 - tn, 1)

Thus, the observation step 21 preferably comprises an acquisition sub-step 21A during which an image of the rotation speed On, 1, Qn, 2, Qn, 3 is recorded at at least three successive instants. distinct tn, 1, tn, 2, tn, 3 and a sub-step 21B of calculation where, on the basis of these readings, the discrete second derivative whose numerator corresponds to N = Qn, 3 + On is evaluated, 1 - 2 x Qn, 2. Of course, it is also conceivable to access the dynamic information and to evaluate the second derivative by means other than speed readings, and for example by using an accelerometer mounted on the rotor and provided with an elastically suspended weight. whose displacement makes it possible to quantify the angular acceleration of the rotor. By taking the acceleration at two different instants, one can calculate, by linear approximation, the first derivative of said acceleration, and consequently the second derivative of the velocity. However, preference will preferably be given to a calculation based on angular velocity measurements, since these can be carried out in a particularly simple, fast and reliable manner. As such, particularly preferably, the evaluation of the rotational speed Q will be done by measuring the frequency of the alternating current produced by the generator 8 coupled to the rotor 4, corrected if necessary the corresponding reduction ratio. B11612 / FR (formula 1) In addition, it will be noted that, while the observation of the evolution of the dynamic behavior of the rotor supposes to carry out several measurements (preferably of speed) at distinct intervals of time, in a certain period of time following the disturbance, it is possible to correlate this evolution with variables other than time. In particular, the speed of rotation depending in particular on the load 3 applied, said load being able to be dynamically adjusted during the cycles, it is conceivable to relate the numerator N to a denominator D corresponding to the product of the load variations ( and no longer time), that is to say to define the second derivative Q "with respect to another variable than time, in this case with respect to the load 3 applied, in order to examine the properties of the function of evolution with respect to said variable, however, preference will be given to the study of the second time derivative of the speed of rotation, denoted Q "n (t) or, for convenience, 0", for various reasons It is furthermore remarkable that, according to a characteristic which can constitute an entire invention, independently of the very nature of the steps of the method 20, the unwinding, and more particularly the repetition of The steps of said control method 20 are preferably indexed to a relative frequency reference which depends on the rotational speed Q of the drive member 2. In other words, the acquisition, the measurement, the steps of analysis, calculation, iteration test, increment, etc. are advantageously clocked and synchronized according to a frequency base based on the number of revolutions made by the drive member 2, and not indexed on an absolute time frame.

Thus, the faster the wind turbine rotates, the more often the process 20, and more particularly the algorithm that governs it, is repeated, the duration assigned to each regulation operation over a given period corresponding to the time required for the rotor B11612 / FR for perform a predetermined number of revolutions at the speed that animates it during this period. Such a characteristic advantageously makes it possible to achieve a regulation that is both fine and economical in terms of means, insofar as one refines the temporal resolution of the regulation when the wind turbine accelerates, whereas it is reduced when the speed of said wind turbine decreases. This also saves the resources and energy mobilized by the regulator 11 for acquisition, calculation and load adjustment operations, by deploying these resources only to the extent necessary for proper regulation, sufficiently safe and fast, but not unnecessarily heavy or redundant. In addition, such a feature makes it possible to simplify the various calculations (formulas) of the algorithm while avoiding the considerations related to the incidences of the inertia of the rotor 4 on the dynamic behavior of the device. For example, an iteration cycle n may correspond to 40 turns of the drive member 2 on itself, each iteration being subdivided into as many phases as necessary. Thus the records of speeds Qn, 1, Qn, 2, Qn, 3 will intervene at 20 turns of interval, the first at 0 turns (tn ,,), the second at 20 turns (tn, 2), and the third at 40 turns (tn, 3) - A clock or stopwatch will be used to date these readings and / or to determine the associated time intervals, which will correspond to the absolute duration that the rotor 4 sets to perform twenty laps between the moment when one performs a first measurement and the moment when the next measurement is made, this time itself depending on the speed of rotation Q of said rotor on said period. As has been said above, the invention aims to adapt the load 3 to the instantaneous production capacity of the device 1, and more particularly to oppose to the engine torque CM a resistive torque CR which is high enough to substantially balance said motor torque, so as to recover the maximum useful power that can provide the device 1 at a given moment, but also sufficiently

B11612 / FR moderated so as not to brake the drive member 2 at the risk of causing it to slow down, and consequently a yield drop, or even a complete stop. Of course, depending on the nature of the charge and the energy produced, it is conceivable to modify said charge in different ways, and in particular mechanically by selecting for example the transmission ratio between the driving member 2 and the load 3, and more particularly the generator 8, or by electrical switching by modifying the electric charging circuit 13 for example by adding or removing resistors 14, etc. However, according to a preferred embodiment which may constitute an invention as such, the generator 8 being connected to an electric charging circuit 13, the method 20 comprises a hashing step 25 during which the charge circuit is connected. 13 to the generator 8 intermittently, according to a selected cyclic ratio, as illustrated in Figure 1. The load adjustment step 22 then comprises a substep 26 of modifying said duty cycle year in the course of which one increases or on the contrary is reduced said cyclic ratio year of a determined adjustment step Aan, in order to increase, or on the contrary to decrease, the average power consumed by the load 3, and more particularly by the resistive circuit of 13, 14, as shown in Figure 4, said charging circuit 13, and more particularly its electrical resistance, preferably being substantially invariant. The intermittent connection may correspond to a substantially rectangular signal, the cyclical ratio an being equal to the quotient of the duration of the connection to the generator by the period of said signal, that is to say to the fraction of period during which the circuit charge 13 is connected to the generator 8, which therefore delivers through the latter (while said generator runs empty over the rest of the period). Preferably, it will thus be possible to regulate the average electrical power consumed by the load, which will correspond substantially to:

B11612 / FR U2 PMoy = anxUx1 = an U (formula 2), R14 where U corresponds to the peak of the crenellated voltage. More generally, the load adjustment step 22 will comprise a substep of analysis during which the behavior of the device 1, and more particularly of its drive member 2, is evaluated, where this behavior is examined. according to a predefined criterion, then a setting step during which a load adjustment step is defined, and more particularly a duty cycle adjustment step Aan, in order to redefine the load 3 by incrementing, or on the contrary by decrementing, preferably in progressive stages, said load 3 according to the fact that the power output and usefully exploited in said load is less than or greater than the power that the device 1 can produce at the instant in question. As has been pointed out above, the evaluation of the second derivative Q "of the rotational speed Q of the driving member 2 has at least two uses, in that it makes it possible not only to converge towards the optimal efficiency with a good reactivity and a good accuracy, but also, independently or in combination, to define a switching criterion between different modes of regulation, in this respect, according to a preferential characteristic which can constitute a complete invention, irrespective of the nature of the steps of the control method 20, the latter may include, especially when it applies to a wind turbine, at least a first control mode MD and a second distinct MP control mode, which are respectively governed by a first law and a second law distinct from the first, said regulation method then comprising a step 40 of selecting the regulation mode during the which the flow regime of the fluid stream F to which the drive member 2 is subjected is analyzed, and then one of the said control modes MD, MP is selected according to the said flow regime (and more particularly, it retains the mode of regulation best suited to said flow regime). The analysis of the flow regime, aimed at applying the switching criterion, and more particularly the determination of the established and permanent character or, conversely, variable and transient, of said flow, may be carried out by a direct subsidiary measure of the speed of the incident fluid stream, for example by means of an anemometer disposed near or at the top of the wind turbine, or indirectly by analyzing the dynamic behavior of the driving member 2, and more particularly acceleration variations of the latter in response to the application of said fluid flow. Moreover, the method according to the invention, whatever it is, preferably comprises at least a first iterative MD regulation mode, called "derivative regulation mode", or "derived mode", or "mode" differential ", according to which the load 3 is changed until substantially reaching a critical operating point where the second derivative Q" of the rotational speed vanishes, because it should be remembered that the efficiency of the rotating machines varies. In particular, as shown in FIG. 2, the power produced by a Pprod wind turbine depends not only on the wind speed V (the speed value increasing by convention with the wind turbine). index i = 1, 2 ...) which drives the rotor, but also, for a given wind speed V;, the speed of rotation of the rotor as it is conditioned by the load 3 which is applied to said rotor. So, for ch At a given velocity V, the yield curve P = f (0) has a substantially bell-shaped shape whose apex corresponds to the optimum operating point (Popt, Qopt) at which the maximum power produced is obtained. and therefore the best performance possible. However, the inventors have discovered that for a wind speed V; given substantially constant, when one varies the load 3, and more particularly the duty cycle a, gradually over time, and more particularly over the iterations, the speed response of rotation Q of the wind turbine follows a curve , illustrated in solid lines in Figure 3, which has a particular point A = (aA, QA), and more particularly a critical point corresponding to a point of inflection.

B11612 / FR They have also discovered that the operating point corresponding to the point of inflection A, that is to say the power produced when it is placed at the load value aA corresponding to the point of inflection A and therefore at the speed of rotation QA, is close to the optimal operating point (Popt, Qopt). Empirically, the difference found is generally less than 10%. In other words, by adjusting the load to aA a rotation speed QA is obtained which corresponds, within 100/0, to the speed of rotation Qopt of the optimum efficiency point (Popt, Qopt). Therefore, it is possible to quickly converge towards an operating point close to the optimum by moving towards said inflection point, substantially to the latter. In a particularly advantageous way, the inventors have also found that this point of inflection, which corresponds in principle to the point in which the second derivative of the speed of rotation with respect to the load vanishes, coincided with the cancellation of the derivative. discrete time second 0 ", during the iterative process in which the load is progressively varied in time, so that the point A can be easily detected and identified when, under the effect of an nth modification of charge, said second derivative 0 ", as calculated for each iteration according to formula 1, passes through zero.

As such, considering that this cancellation of the second derivative (and therefore the critical point) corresponds to an extremum, in this case a maximum, of the curve representative of said second derivative (shown in phantom in the figure 3), which itself has a bell shape, it is possible to detect the crossing of said extremum, and thus the zero crossing, by studying the sign of the slope (tangent) to said curve and detecting the change of the here, which can be realized here very simply by comparing between them, by subtraction, two successive values of the second derivative. B11612 / EN Indeed, as long as the value of second derivative increases over the iterations, while the load evolves monotonously, either by increase or by decrease, it means that one approaches the top of the curve representative of the second derivative, therefore of the point of inflection. On the other hand, if one continues to advance the load in the same direction after having crossed this maximum, the second derivative decreases, which changes the sign of the subtraction. That is why the load adjustment step 22 may comprise for this purpose a second derivative comparison analysis sub-step during which the current value Q "n of the second derivative evaluated for the iteration n in progress to the previous value Q "n_1 of this same second derivative evaluated during the preceding iteration n-1, then a sub-step of setting, or of" redefinition of the load "26, 31 during of which a load adjustment step is defined to increase the load 3 if the current second derivative value Q "n is greater than the previous value Q" n_1, or conversely to reduce said load 3, and more in particular, the duty cycle an, if the current value of the second derivative Q "n is less than or equal to the said previous value Q" n-1 More particularly, the subtraction between the two successive values of the second derivative can advantageously be compared andthe result of the operation to zero, as indicated in box 30 of the flowchart of Figure 4.

As long as the second time derivative Q "increases, ie its value at the iteration n is greater than its value at the iteration n-1, it means that one approaches the optimal operating point. Conversely, when said second time derivative Q "decreases as iterations proceed, this means that one moves away from said optimum.

By way of example, if we consider that we are to the left of the point of inflection A on the curves of FIG. 3, the motor torque CM being greater than the resistive torque CR, that is to say say that the system is working underload, the second derivative will increase when the load will increase, an increasing evolution of the second derivative

B11612 / FR thus indicating that the load can be increased, and therefore the duty cycle a. Conversely, in the part to the right of the point of inflection, the system works in overload, the resistive torque CR being greater than the engine torque CM (it tends to brake the wind turbine), so that it is necessary to reduce the load, and thus the duty cycle a, to increase the second derivative and get closer to the optimum. By applying such a criterion, it can advantageously converge quickly and substantially monotonically towards the point of inflection A, until the moment when an oscillating speed will appear, the charge setpoint alternately passing the device on either side the desired inflection point (any increase in load resulting in such a decrease in the second derivative that it causes a decrease in load and therefore a second derivative increase which causes the return to a load increase, etc.). such a second derivative control mode has the great advantage of making it possible to move the operating point of the device 1 towards the point corresponding to the optimum efficiency in a fast convergence, and in particular much faster than with the control methods previously known. . It is remarkable that insofar as, for a given wind speed, at each increment of iteration corresponds - in fact - an increment of time, a variation of load, and an induced variation of angular velocity, the point of cancellation (or change of sign) of the second derivative of the rotational speed calculated by the discrete method remains a priori identical whatever the variable (time, iteration or load) with respect to which said second derivative is defined, since cancellation of the numerator N, as defined in formula 1, (or more particularly the change of sign of the difference between two successive numerators) exclusively depends, in practice, on changes in the angular velocity between the three successive survey points. B11612 / EN However, the use of a second time derivative will advantageously be preferred insofar as the knowledge of such information correlated with time also makes it possible to recognize almost instantaneously the wind regime that applies to the wind turbine. and select appropriate control mode.

In practice, the method according to the invention is therefore capable of distinguishing the cause of the rotational speed variations of the wind turbine, depending on whether these originate in a variation (in particular sudden) of the wind speed incident or on the contrary, in a load variation (more progressive), and to react accordingly, and in particular to reconfigure automatically to adapt its reaction. As such, it will be noted that although the derivative mode MD advantageously allows to quickly converge towards the optimum operating point (Popt, Qopt), this convergence can sometimes remain relatively approximate. This is why this first mode can advantageously be completed by a second MP regulation mode, which is slower but more precise and more stable than the said derived mode MD, which makes it possible to refine the search for the optimum operating point after the interval research has been reduced, and in some ways "sided" by the derivative mode. For this purpose, the method may advantageously comprise a second iterative regulation mode, referred to as the "power regulation mode" MP, according to which the load adjustment step 22 comprises a substep of analysis which comprises a evaluation phase 32 of the power Pn consumed by the load 3, then a sub-step of redefinition of the load during which a load adjustment step is determined in order to adapt the power consumed to the power available, then the load is modified by applying to it said adjustment step. In other words, while the first derivative mode MD examines the dynamic changes in the behavior of the rotor directly at the latter, the second mode of regulation by the power MP observes the indirect effects that this

B11612 / FR behavior produced on the load, at the output of the installation, and can more particularly measure the power consumption of the charging circuit 13. As such, the measurement 32 of the power consumed Pn may for example be performed, so well known, by measuring the voltage U applied to the charging circuit 13 and the intensity passing through said circuit, or by evaluating the power dissipated by Joule effect using the Ohm's law (see formula 2). Advantageously, there are several methods for operating such a regulation by the power. A first method consists in determining, for example by a test campaign, the optimal operating curve corresponding to all the optimum operating points (Popt, Qopt) established for a plurality of wind speeds V1, V2, V3, V4, V5 distributed over the operating range of the wind turbine. The optimization is then done as follows: at each iteration, the power is measured and this power is horizontally projected on the ideal curve of the optimum points in order to determine a new rotational speed and a corresponding increment of load. If the optimum is not reached, at this new speed corresponds a new operating point (obtained by vertical projection from the ideal curve on the bell curve) and thus leads to a new power reading, which is projected in turn horizontally on the ideal curve to define a second speed, etc. By successively projecting the non-optimized operating points onto the optimum curve, and then from this curve to a corresponding new operating point, an "ascent" optimization as shown in FIG. 2 is obtained between the points A1, A2, A3 and (Popt, Qopt). According to a second method, the characteristic curve of the wing, ie the yield curve of the device, is unknown, and progress is made by fuzzy logic, increasing the load as long as it is found that the power increases, and reducing said charge when it is found that said power decreases. According to this second method, the regulation method 20, and more particularly the substep of analysis, may comprise a phase 33 during which the rate of variation of the power consumed is evaluated with respect to the speed of the evaluation. rotation of the driving member 2 between the current iteration and a previous iteration, and preferably the immediately preceding iteration, according to the formula m = Pn-P (n-1) ~ n - n (n _1) It is then decided to increase the load 3 (in order to reduce the speed) if the rate of variation of power consumed m is negative (which means that one is on the decreasing part of the bell curve of the Figure 2, located to the right of the optimum, and it is necessary to increase the load to reduce the speed and thus return to the optimum), or conversely to reduce the load 3 (to increase the speed) if said rate of change m is positive (which means that one is uve on the ascending part of the yield curve of Figure 2, located to the left of the optimum). In other words, one converges progressively toward the optimum operating point by using the finite increments method, the rate of variation m corresponding to the directing coefficient of the affine function passing through two successive effective operating points. As such, it is remarkable that if the rate of change m changes sign between two successive iterations, it means that the optimum point has been crossed and that one moves away from it, so that it is necessary to turn back, that is to say to reduce the load if it had grown continuously, or on the contrary to increase it if it decreased so far substantially monotonous. In power mode MP, the load adjustment step can be particularly fine, so that, even if it is in practice almost impossible to converge the operation of the device 1 until exactly a single optimum operating point, considering in particular the intrinsic limits of the regulation as well as the variability of the actual operating conditions, the fact remains that the possible residual oscillation of the load setpoint B11612 / FR around the optimal operating point remains despite all contained in a very narrow range (advantageously narrower than the oscillation interval of the derivative mode) bordering the immediate vicinity of said optimum operating point. Preferably, the method 20 is advantageously capable of alternately adopting the first regulation mode MD by the derivative and the second regulation mode MP by the power, which gives it a great versatility, insofar as it can adapt automatically to all wind regimes that apply successively and randomly to the device 1, in order to make the most of each regime.

For this purpose, during step 40 of selecting the regulation mode, the flow regime of the fluid stream F to which the drive member 2 is subjected is preferably analyzed in order to determine whether it corresponds to a transition regime or at a steady state, and then alternatively the control mode by the derivative MD is selected in the case of a transition regime, or the regulation mode by the power MP in the case of an established regime. Thus, in case of rapid variation and / or high amplitude of the wind speed, preference is given to the derivative mode which has the best reactivity (the shortest response time), despite a certain inaccuracy. On the other hand, if the wind speed is sufficiently undisturbed to be considered as established, and in particular if its speed is substantially constant over a sufficient duration, much greater than the response time of the plant 10 in derivative mode, and for example not is not known to vary more than 15% for at least 15 seconds, the accuracy and stability of the servo can be improved by approaching the optimum operating point even further, and possibly slower, by switching to regulation by the power MP.

Advantageously, the device 1 regulated by a method 20 according to the invention can therefore fully exploit both the wind regimes established, and since

B11612 / FR low speeds of the order of 4.3 m / s to high speeds of the order of 16 m / s, but also transient gusts of wind. Preferably, the step 40 of selecting the regulation mode may comprise, in order to determine the conditions in which it is desirable to switch from a regulation mode by the power, which is in principle the default mode, to the mode derivative control system, a sub-step 41 for calculating the progression of the second time derivative during which the difference between the current value of the time second derivative of the speed of rotation Q "n evaluated for the iteration is calculated. in progress n, at the previous value Q "n_1 of this same second derivative temporal evaluated during the previous iteration n-1, then a sub-step 42 of conditional switching in control mode by the derivative where it switches to control mode by the derivative MD if the calculated progression is greater than or equal to a predetermined switching threshold SC. In other words, the monitoring of the time second derivative of the rotational speed and its evolution over time, and consequently iterations, makes it possible to detect rapidly, with a very good sensitivity, a change in the speed of rotation. driving regime of the device 1 and thus adapt without delay the regulation and behavior of the device 1 to its fluctuations. Preferably, the switching threshold, which will depend in particular on the intrinsic characteristics of the wind turbine and the installation in general (dimensions, weight, etc.) will be defined, for example by a test campaign, so as to allow the switching when the wind turbine responds to a wind variation of magnitude close to 10 km / h occurring in less than five seconds, and in particular in one to two or three seconds, or even less than one second.

Moreover, in order to be able to return to the regulation mode by the power after the derivative mode has performed its function and oscillates durably in the vicinity of the optimum operating point, it is advantageous to provide that the step 40 of selecting the regulation mode comprises a substep 45 of instability detection of the control mode by the derivative MD during which the history of the successive orientations of the various load adjustments, as they have been defined over a period of several years, is examined. monitoring corresponding to a predetermined number of several successive iterations in regulation mode by the derivative MD, then a substep 46 of conditional switching where it is switched to regulation mode by the power MP if the examination reveals that the number of alternations of said orientations exceeds a predetermined threshold on said monitoring period. In other words, the respective signs of a predetermined number of load adjustment steps, and more particularly of adjustment steps of the duty cycle Aan, are examined in a sliding manner in order to determine whether this mode of regulation is progressing. substantially monotonically to an optimum, which means that the search for optimization is still in progress, or if on the contrary it oscillates between several (and especially two or three) neighboring charge values, which means that it has reached are higher degree of optimization, having allowed the device to reach a point relatively close to the optimum, but that it can not further refine the regulation. In a particularly advantageous manner, this detection of instability can use a convergence counter 47 which determines the number of alternations from which the system is considered to be oscillating. This convergence counter 47 may advantageously be initialized to a predetermined value, for example equal to 3 in the flowchart of FIG. 4, when the derivative mode MD starts, and then either maintained or decremented, depending on whether the sequence of the last two no adjustment is "identical", that is to say that the progression is monotonous, or on the contrary "different", that is to say that one observes an alternation and a crawl in the orientation the charging instruction.

The more alternating the orientations of the charge setpoint, the lower the value of the counter, and finally, when it reaches zero, cause switching to power regulation mode. B11612 / FR Moreover, it is remarkable that, according to a feature which can constitute a complete invention, whatever the control mode considered, the load adjustment step, which more particularly corresponds to the adjustment step Aan cyclic ratio, is preferably variable according to the iterations and chosen according to a predetermined determination law, depending on the historical progression of the adjustment of the load. In other words, the step of increment, respectively of decrement, of the load, that is to say the amplitude of the change of the load reference applicable to each iteration, can advantageously be modified over time, and especially between the different successive iterations. The dual objective is to compensate more quickly, ie in a time and a number of iterations as small as possible, the difference between the current operating point and the optimum operating point, in order to value without delay the energy potential of the device 1, while limiting the overshoot of said optimum and / or the residual error related to oscillations of the load setpoint actually applied around the load value which would correspond theoretically to the optimum. Thus, it is possible for example to choose to differentiate the approach speed (convergence towards the optimum) by choosing a step of higher load increment if one finds that one is at a remote operating point the optimal operating point (Popt, Qopt), or on the contrary a lower step if it is found that one is at an operating point closer to said optimal operating point (Popt, Qopt). For this purpose, a person skilled in the art may in particular provide different laws for determining the adjustment step, and / or tables or charts showing the different steps of adjustment of duty cycle applicable, said laws may be different depending on the mode considered. By way of example, it may be decided to increase the value of the adjustment step, that is to say the amplitude of variation of the load 3 for the iteration considered, as long as the

B11612 / FR progression of the duty cycle is monotonic, that is to say as long as each successive iteration tends to vary said duty cycle (and therefore the load) repeatedly in the same direction (increase or, conversely, reduction monotone), then reduce the value of said adjustment step when the optimum is exceeded and that one must turn back, that is to say when, during an iteration, it detects conditions which lead to decide a new evolution of the inverse charge to that which had been decided at (the) iteration (s) previous, said new evolution then operates at less "speed" than the previous one. In practice, such a law for determining the adjustment step not only makes it possible to accelerate the overall convergence of the load setpoint, but also to dampen the oscillations of said load setpoint in the vicinity of the optimum. Indeed, during the appearance of the oscillating speed, the amplitude of the successive alternations (of adjustment step) is gradually reduced until a residual oscillation of particularly low amplitude, which corresponds to the most end of the control mode concerned, the load setpoint is thus maintained in a narrow "tunnel" closely bordering the load value which corresponds substantially to the optimal operating point targeted. By way of example, it will be possible to define a list of J (for example J = 10) possible values of amplitude of step of adjustment [ai] j =, .. j, ordered in a strictly increasing manner, and to select, when from the current iteration n, the value of the list whose rank is immediately greater than the rank of the value retained during the previous iteration n-1, if the regulation method indicates that it is necessary to continue, during said iteration n, the trend of evolution of the charge already initiated during the previous iteration n-1, that is to say if the instructions Aan and Aan_1 have the same sign (positive or negative), so that the we get: I ~ anl = aj + ,, the sign of Aan depending on him of the evolution that must follow the load. If one is in a phase of monotonous growth of load, one will have thus Aan = + aj + ,. On the other hand, if the process concludes that it is necessary to reverse the trend, the rank value immediately below is selected as step amplitude, so that B11612 / FR is obtained if the instructions Aan and Aan_1 are signed. different, 'Aar,' = a; _ ,. In the previous example, if the load must be reduced after a monotonic growth phase, then we will have Aan = -a; _, In power mode MP, it will be possible to choose the amplitude of modification of the duty cycle directly in depending on the absolute value of the rate of change m, or indirectly, depending on the corresponding history of previous load adjustments. Preferably, especially in power mode MP, the different values of the list will follow a progression law proportional to the cube of their rank, of type: a; = ko + k, x j3.

Such a law of progression notably makes it possible to significantly accelerate the convergence when it is found that it is necessary to adapt the load monotonously over several successive iterations, that is to say that one is in an operating point far from the optimum.

By way of example, the limits of variation of the adjustment step can be fixed substantially as follows: amIN = a, = 0.0015% and amAx = at) = 2.5%, with a2 = 0.019%, a3 = 0.0665%, etc. According to a principle similar to that which is performed in MP power mode, it is possible to modify the value of the increment step Aan during the regulation in MD derivative mode, depending on whether one is more or less distant from the point the list of values [a;] can however differ in its absolute values (which may in particular be higher than those of the power mode in order to accelerate the first coarser approach to the optimum) and / or in its evolution law (which could be, as required, linear, quadratic, cubic, exponential, etc.) In regulation mode by the derivative MD, the law of determination could for example define the value of the adjustment step for the current iteration is based on the sign and magnitude of the difference directly observed between the current value Q n and the value

B11612 / FR previous Q "n_1 of the second derivative, or preferably, depending on the historical progression of said load adjustment step which results itself from the analysis of the successive second derivatives. The invention also relates, as such, to a regulation module 11 intended to be connected to a wind energy-type power-sensing device 1, and containing a regulation circuit making it possible to implement a regulation method 20 in accordance with FIG. invention.

Said regulation module may advantageously be contained in a housing, advantageously removable, provided with connection terminals for connecting on the one hand its inputs to the generator, and on the other hand its outputs to the charging circuit 13. Advantageously, such a regulation module may be interchangeable and programmable, and may in particular be installed in a retrofit solution on an already existing wind turbine. In addition, the invention also relates to a computer program comprising computer program code means adapted to perform all or part of the steps of a method according to any one of the variants described above when said program is executed on a computer. By "computer" is meant in particular any programmable electronic circuit allowing control, possibly remote, of the device 1 and the control method 20, that said circuit is integrated in a control unit or that it constitutes an industrial controller or a controller. PC. Of course, the present invention also extends to the implementation of such a computer program on a computer-readable medium, as well as to a support that can be read by a computer and on which is recorded such a program.

The operation of the method 20 will now be described in connection with a preferred embodiment variant of the installation 10. B11612 / FR Initially, the wind turbine 1 is at a standstill, the speed of the fluid stream F being insufficient to make it start, or at least insufficient to keep the wind turbine in its production range. Typically, this situation may correspond to a speed of rotation less than or equal to 60 rpm.

Below the production range, the duty cycle is zero, the electric charging circuit 13 being (physically or virtually) disconnected from the generator 8 so that it runs empty so as not to create inertia and favor the start. Suppose that the wind rises or forces so that its speed increases to stabilize substantially beyond the production threshold of the wind turbine 1. Then the engine torque CM exceeds the resistive torque CR and the drive member 2 accelerates its rotation. During the step of observing the same iteration, we thus measure increasing speeds of rotation Q n, 1, n, 2 and Q n, 3, from which the second derivative Q "n is calculated. beginning of the transient regime in which the rotor adapts its speed to the new wind conditions, the drive member 2 tends to accelerate faster and faster, to the point that the second derivative progression calculated in the sub-step 41 exceeds the switching threshold SC The regulation then goes into derivative mode MD, the convergence counter being initialized to the value 3. The algorithm then performs a comparison of the second derivatives, which is, notably at the beginning of the transient regime, higher It therefore decides to increase the load, and more particularly the duty cycle, by a step value which can in particular be between 0.0015% and 2.5% B11612 / FR the load allows to extract more energy device 1 and has the effect of increasing the resistive torque that opposes the acceleration of the rotor. As long as the "ascending" phase of the acceleration of the driving member 2 continues, that is, as the second derivative 3 grows, the algorithm progresses by iteration in a derivative mode and increases the charge monotonously. Advantageously, the amplitude of the adjustment step also increases from one iteration to another, the need to progress the load repeatedly, here by increasing it, indicating that one is at a point (again ) away from the optimum (or at least that we started from a point relatively far from the optimum). When the acceleration begins to bend, to the point that the second derivative decreases and the test of the block 30 is negative, it means that the device has been loaded too much with respect to its production capacity, and that the we approach a rotor braking cycle at the risk of degrading or even stop production quickly by slowing down or stopping said rotor. The algorithm then decides to reduce the load (right branch of block 30). Because the load decreases while it increased until then, it creates an alternation of sign in the sequence of load instructions, and more particularly the load adjustment steps. This alternation first of all leads to a lowering of the amplitude (in absolute value) of the adjustment step with respect to the last step applied, so that the cusp, and more generally the final phase of approach of the mode of current regulation (derivative mode), is performed at a slower pace in order to refine the convergence. This alternation also results in a lowering of one unit of the convergence counter 47. B11612 / FR The one at 2, it is always greater than zero, so that the algorithm is maintained in derivative mode, the switching condition not being filled. However, lowering the load, and therefore the load torque, while the wind speed and thus the engine torque CM is substantially constant, has the consequence that the rotor accelerates again. Thus, during the next iteration, the difference between the second derivatives, which was previously negative, becomes positive again, which causes the algorithm to increase the load, and therefore the duty cycle. We thus find ourselves in the presence of a new alternation in the sequence of the 10 cyclic ratios, which has the effect of lowering again the adjustment step Aan as well as the convergence counter whose value reaches 1. When the next alternation, the increase in load braking the rotor, it reverses the trend of evolution of second derivatives, and thus alternates again the orientation of the load instructions. This third alternation has the effect of bringing the convergence counter 47 to zero, which causes the return switching in regulation mode by the power MP. Advantageously, this power regulation applies from a "pre-optimized" intermediate operating point, relatively close to the optimum, such that it results from the first rapid approximation performed by the derivative mode. For convenience, it will be possible for example to consider, with reference to FIG. 2, that the wind speed is established at the speed V5, and that the derivative mode has made it possible to reach the intermediate operating point A1.

In the power regulation mode, first a first instantaneous speed Q n, 1 is measured first and then, twenty turns later, a second instantaneous speed measurement Q n, 2 and an evaluation of the instantaneous speed are performed. power consumed by the B11612 / FR charge Pn, then, again twenty turns later, a third speed measurement Q n, 3 in order to calculate the second derivative. As the device is no longer in a sustained acceleration phase, the absolute difference between the last two second derivatives is less than the switching threshold Sc, so that the power mode MP is maintained. The rate of change m is then calculated. If the rate of change m is positive, that is to say that, in response to the last change in load, the device has changed the power consumed in the same direction as the rotational speed, that is to say that is to say that we have seen the power increase while the speed increases or on the contrary the power decreases while the speed decreases, then it means that one is graphically in the ascending portion of the yield curve, situated to the left of the optimum point in FIG. 2, and it is therefore necessary to increase the rotational speed θ, and consequently to reduce the load 3, that is to say to reduce the duty cycle a, for get closer to 15 the optimum. On the other hand, if this rate of variation m is negative, it means that the evolution of the power is contrary to that of the speed, that is to say that the power increases while the speed decreases or at the same time. On the contrary, the speed increases as the power decreases, which indicates that the downward side of the curve 20 of FIG. 2 has been swung to the right of the optimum point, and that it is therefore necessary to reduce the speed. rotation, that is to say to increase the load, to return to the optimum operating point. Advantageously, the amplitude of the adjustment step increases as the load 3 progresses (up or down) in a monotonous manner, and then decreases progressively over the load reference oscillations around the optimum point, and until to reach its minimum amIN. It is thus possible to progress substantially tangentially and particularly precisely along the yield curve, and to approach at most

B11612 / FR near the optimum operating point by a succession of iterations using tests and rules of empirical evolution particularly intuitive and simple. In the case in point, the point Al being to the left of the optimum, the load is reduced, which has the effect of increasing the speed and of transferring the device to the next operating point A2, etc. Convergence in MP power mode will continue until we stabilize substantially around the optimum operating point of the current wind regime (with minimal residual oscillations, corresponding in some way to a static error "incompressible "of the servo system).

Advantageously, the algorithm will continue to regulate in power mode MP as long as the wind will not be subject to a sufficiently large fluctuation to cause a significant acceleration of the wind turbine and consequently a switching in derivative mode. Thus, the device and the adaptive method according to the invention make it possible to optimize the energy available in a wind farm by automatically adapting the actual configuration of the regulation to the instant wind conditions that are exerted on the wind turbine. Moreover, the control method employed uses measurement and acquisition chains, as well as calculation formulas that are particularly simple, save time and energy and do not require a complex architecture or programming. It is therefore possible to realize and operate an installation according to the invention at lower cost.

In addition, this simplicity is a guarantee of robustness and longevity. In addition, the installation 10 has excellent ability to adapt to changing training conditions, and more particularly to fluctuations in wind speed, thanks to a differential operation that allows it to alternate, depending on the wind speed, either a derivative regulation mode allowing an approximate and particularly fast approach B11612 / FR optimum operating point, or a power regulation mode allowing, in a second step, to perform a second refined approach, in a progression slower but more precise and more stable than the first, in order to optimize the search for the optimum operating point when the establishment of a substantially constant and sustainable wind speed allows it. Moreover, it is remarkable that the installation 10 may advantageously have a small footprint and cause very little noise, especially noise, through the use of a vertical axis wind turbine.

Finally, such an installation can be advantageously completely autonomous, the energy required for regulation can be provided by the wind turbine itself, a battery can optionally provide a standby function in case of temporary stopping of the rotor.

B11612 / EN

Claims (19)

CLAIMS1 - A method of regulating (20) an energy sensing device (1), of the wind or tidal type, which comprises at least one rotary drive member (2) intended to be driven by a fluid stream so as to transmitting energy to a load (3) applied thereto, said method being characterized in that it comprises an observation step (21) during which the evolution of the speed of rotation is observed ( 0) of the drive member and the second derivative (0 ") of this evolution of the rotational speed (0) is evaluated, followed by a load adjustment step (22) during which one adapts the load (3) according to said evaluation of the second derivative. 2 - Control method according to claim 1 characterized in that the observation step (21) comprises an acquisition sub-step (21A) during which there is an image of the rotation speed (On, 1 , Qn, 2, Qn, 3) at three distinct successive instants (tn ,,, tn, 2, tn, 3) and then a calculation sub-step (21B) where it is evaluated on the basis of these readings, the discrete second derivative whose numerator corresponds to N = Qn, 3 + On, 1 - 2 x Qn, 2. 3 - Control method according to claim 1 or 2 characterized in that, the load (3) comprising a generator (8) to which is connected an electric charging circuit (13), the method comprises a hashing step (25) during which the charging circuit (13) is connected to the generator intermittently at a selected duty ratio (an), and in that the charge adjustment step includes a substep of modifying said duty cycle ( an) during which one increases or on the contrary is reduced said cyclic ratio (year) of a determined adjustment step (Aan) in order to increase or on the contrary to reduce the average power consumed by the load. 4 - Method according to one of the preceding claims characterized in that it comprises at least a first control mode (MD) and a second control mode (MP) distinct, respectively governed by a first law and a second B11612 / FR law distinct from the first, and in that it comprises a step of selection (40) of the control mode during which the flow regime of the fluid stream to which the drive member is subjected is then analyzed. one of said regulation modes is selected according to the flow regime. 5 - Control method according to one of the preceding claims characterized in that it comprises a first iterative regulation mode called "control mode by the derivative" (MD) according to which one changes the load (3) to substantially reach a critical operating point where the second derivative of the rotational speed vanishes. 6 - Process according to claim 5 characterized in that the step (22) for adjusting the load comprises a sub-step (30) analysis by comparison of second derivative during which the current value of the second derivative (Q "n) evaluated for the current iteration with the previous value of this same second derivative (Q" n_1) evaluated during the previous iteration, then a sub-step of setting (26, 31) during of which a load adjustment step is defined to increase the load (3) if the current derivative second value is greater than the previous value, or conversely to reduce said load if the current value is less than the value earlier. 7 - Method according to one of the preceding claims characterized in that it comprises a second iterative regulation mode called "power regulation mode" (MP), wherein the adjustment step (22) of the load comprises a sub-step which comprises a phase of evaluation (32) of the power consumed by the load, then a sub-step of redefinition of the load (26, 31) during which a step of adjustment of the load is determined charge to adapt the power consumed to the available power, then the load (3) is modified by applying to it said adjustment step. 8 - Process according to claim 7, characterized in that the analysis sub-step comprises an evaluation phase (33) during which the rate of variation (m) of the power consumed (Pn) is evaluated with respect to at the speed of rotation B11612 / FR (0) of the drive member between the iteration (n) in progress and the previous iteration (n-1), then the load (3) is increased if the power consumption variation is negative, or conversely, the load is reduced if said rate of variation is positive. 9 - Control method according to claim 4, one of claims 5 or 6 and one of claims 7 or 8, characterized in that, during the step (40) for selecting the control mode, one analyzing the flow regime of the fluid stream (F) to which the driving member (2) is subjected in order to determine whether it corresponds to a transition regime or to an established regime, and then alternatively selected either the derivative control mode (MD) in the case of a transition regime, or the power regulation mode (MP) in the case of a steady state. 10 - Control method according to claim 9 characterized in that the step (40) for selecting the control mode comprises a substep (41) for calculating the progression of the second time derivative during which the difference is calculated between the current value (Q "n (t)) of the temporal second derivative evaluated for the current iteration to the previous value (Q" n_1 (t)) of this same second temporal derivative evaluated during the previous iteration, and then a sub-step (42) of conditional switching in the derivative regulation mode where the derivative is switched to regulation mode if the calculated progression is greater than or equal to a predetermined switching threshold (SC). 11 - Control method according to one of claims 9 or 10 characterized in that the step (40) for selecting the control mode comprises a substep (45) detecting instability of the control mode by the derivative during which the history of the successive orientations of the different load adjustments as defined over a monitoring period corresponding to several successive iterations in the regulation mode by the derivative is examined, then a substep (46) conditional switching system where it is switched to power regulation mode (MP) if the examination reveals that the number B11612 / FR of alternations of said orientations exceeds a predetermined threshold on said monitoring period. 12 - Control method according to one of the preceding claims characterized in that, the adjustment of the load being done by iteration in a load adjustment step (Aan) determined, said load adjustment step is variable according to the iterations and chosen according to a predetermined pitch determination law, as a function of the historical progression of the load adjustment. 13 - Control method according to one of the preceding claims characterized in that the course of its steps is indexed on a relative frequency reference which depends on the speed (0) of the drive member (2). 14 - Control module (11) intended to be connected to a wind-powered energy-sensing device and containing a regulation circuit for implementing a method according to one of claims 1 to 13. A computer program comprising computer program code means adapted to perform the steps of a method according to one of claims 1 to 13 when said program is executed on a computer. 16 - A computer-readable medium on which a program according to claim 15 is recorded. 17 - Installation (10) for producing energy characterized in that it comprises a device (1) for collecting energy supplying a load (3), and a regulator (11) able to regulate the operation of said device capture according to a method according to one of claims 1 to 13. 18 - Installation according to claim 17 characterized in that the energy sensing device (1) is formed by a wind turbine, preferably vertical axis. B11612 / EN 19 - Installation according to claim 17 or 18 characterized in that it comprises an accumulator (12), preferably electrochemical, such as a battery, or thermal, such as a hot water tank, suitable for storing the energy produced. B11612 / EN