• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    Evaluation system (10) for a flow rate of a fluid from a reservoir (20, 21), comprising measuring means (17, 22, 23) for measuring a fluid level in the reservoir (20) , 21) and characterized in that it comprises means for estimating the flow rate of the fluid using an odorless Kalman filter, said estimation means comprising means for obtaining (16) the raw flow rate of fluid and correction means (18) connected to the obtaining means and the measuring means and configured to correct the raw flow obtained by the obtaining means (16) according to the level measured by the measuring means. Process implemented by such a system.
    公开号:FR3021740A1
    申请号:FR1455024
    申请日:2014-06-03
    公开日:2015-12-04
    发明作者:Gonidec Serge Le；Antoine Romet；Tarek Madani
    申请人:SNECMA SAS；
    IPC主号:
    专利说明:

    [0001] FIELD OF THE INVENTION The present disclosure relates to a method of evaluating a flow of fluid from a reservoir, and more particularly to the reliability of such a process. It also relates to a system for implementing this method. BACKGROUND ART A method of evaluating a flow rate of a fluid from a reservoir, including measuring a level of fluid in the reservoir, is known. For simple cases, for example in the case where the level measurement is permanently available, where the tank has a cylindrical shape and where there is no accumulation of fluid between the outlet of the tank and the point d flow rate, the fluid flow can be assessed with certainty by a simple affine function.
    [0002] However, in practice, the measurement of the fluid level is rarely available continuously, which requires the use of interpolations or estimators during periods when the level measurement is unavailable. However, there may be uncertainties and biases related to the level measurement and / or the estimators used. In addition, the reservoir may have a non-cylindrical shape, which introduces non-linearity into the evaluation function. There is therefore a need for a new type of rate evaluation method. PRESENTATION OF THE INVENTION The present disclosure relates to a method of evaluating a flow rate of a fluid from a reservoir, comprising measuring a fluid level in the reservoir and characterized in that it comprises a step of estimating the flow rate of the fluid using an odorless Kalman filter, said estimation step comprising a step of obtaining the raw fluid flow and, when a level measurement is available, a step of 3021740 2 correction during which the raw flow obtained is corrected according to the level measurement. A Kalman filter is a computationally implementable calculation method for estimating the states of a dynamic system from a series of incomplete or noisy input data. The Kalman filter is modeled by a state equation (1), which represents the evolution of the dynamic system, and a measurement equation (2), which expresses the relationship between the measured observable quantities and the intrinsic state of the system. . These equations are of the type: rk ± i = f (xk, uk, wk) (1) Yk = h (xk, vk) (2) where k is the current moment, k-1-1 the next instant, x the (intrinsic) state of the system, u the input data, y the measured (observed) quantity, f the state function, h the measurement function and v and w are noises (respectively measurement noise and state noise). The variables x, y, u, y, w, can be vectors with one or more components. In order to implement a Kalman filter in a process according to the invention, the state x comprises the fluid flow, which is observed from the measurement y which comprises, for example, the level of fluid in the reservoir. The input data u include, for example, system parameters such as the speed of a pump and / or the fluid pressure at one or more positions of the fluid circuit. The input data u can be estimated from other measurements than y or be system operating data. The Kalman filter comprises a step of obtaining a value of the state (k + 1) according to the current state (k) from the input data uk and a step of correction of the value obtained from of the measure yk. These steps are largely detailed in the literature and will not be explained here again in the general case.
    [0003] 3 0 2 1 7 4 0 3 The odorless Kalman filter (or unscented Kalman filter, hereinafter UKF), described in "A new extension of the Kalman filter to nonlinear systems "(Winger and Uhlmann, in the 11th International symposium on Aerospace / Defense sensing, 5 simulation and controls, Vol multi sensor fusion, tracking and resource management II, Orlando, Florida, 1997), is a variant of the Kalman filter particularly suitable for non-linear systems such as the flow rate of a fluid depending on the level of the reservoir from which the fluid comes. In addition to the steps of the Kalman filter, the odorless Kalman filter includes a step called "odorless transformation" of approximating the current state xk with a random Gaussian variable represented by a set of cleverly chosen points named sigma-points. This set of points faithfully reproduces the mean and the covariance of the Gaussian random variable. In the obtaining step, the state equation (1) is applied to each of the sigma-points in order to obtain the next state. The result is a set of sigma-points at time k + 1, which reproduces the mean and covariance of the next state X -k + 1 with second-order precision (in terms of Taylor series decomposition). ). In contrast, the extended Kalman filter, which is another variant of the Kalman filter applying to non-linear systems, achieves only first order accuracy. Moreover, unlike the extended Kalman filter, the UKF does not require an explicit heavy linearization calculation (Hessian or Jacobian matrix calculations).
    [0004] The obtaining step consists of an initial, possibly provisional, determination of the flow rate that is to be evaluated. It can be performed by measurement, by calculation (in particular thanks to the state equation (1) of the UKF) or by a combination of the two. In other words, the state x of the UKF includes the bit rate that is to be evaluated. The raw rate obtained by the obtaining step is then subjected to the correction step.
    [0005] As indicated above, the correction step uses the results of the measuring step. The measurement step does not assume that the measurement of level is always available: it consists in checking if a measurement is available and, if it is the case, to raise it. If the measurement 5 is not available at a time, the measurement step does not return any data to the correction step and, therefore, the rate obtained at this time is not corrected. The evaluation method can be used to evaluate an instantaneous flow rate, a cumulative flow rate or any quantity calculated from the flow rate.
    [0006] The evaluation method therefore provides a fluid flow rate evaluation which is accurate, consistent with the level measurements in the tank and which intrinsically corrects, through the use of a UKF, the biases and errors of the stage. obtaining. In addition, this method is compatible with any form of tank, including non-cylindrical tanks.
    [0007] In addition, the obtaining step makes it possible, thanks to the UKF state equation, to evaluate the flow of the fluid precisely even when the level measurements are not available. Finally, this method adapts perfectly to the flow variations and to the operating point variations of the system on which the flow rate is measured.
    [0008] In some embodiments, the gross flow rate value is measured by at least one sensor, including a flow meter. In these embodiments, the obtaining step is essentially carried out by the flowmeter and the correction step makes it possible to correct the flowmeter from the measured level values. When a flowmeter is present, the evaluation method thus makes it possible to make it reliable, to reset it and / or to recalibrate it. In some embodiments, the gross rate value is calculated by a mathematical estimate with iteratively evaluated coefficients, in particular by an artificial neural network. By mathematical estimation with iteratively evaluated coefficients, physical measurements and mathematical expressions whose coefficients are constant and known a priori are excluded. Iteratively evaluated coefficients can be evaluated in particular by successive adaptations of a model to a set of data, as is done by an artificial neural network, or by convergence of a solution of equation solved numerically. An artificial neural network (ANN) is a computational model comprising one or more neurons, each neuron being provided with a transfer function. The RNA thus has a global transfer function for calculating at least one output as a function of at least one input. The transfer functions of each neuron or the relative weights of neurons in the network can be weighted by coefficients called synaptic weights (or simply weight) and bias. The weights can be modulated according to an RNA learning.
    [0009] Learning involves providing the RNA with a set of situations in which inputs and outputs are known. During learning, the RNA adapts its synaptic weights and biases to fit the situations learned, possibly with some tolerance. RNAs are therefore mathematical expressions whose coefficients are adapted iteratively as a function of learning. The weights and biases, once determined at the end of the learning phase, can then remain constant or not during the exploitation phase. The RNA is thus able to intelligently model a system by learning from the actual behavior of said system, without having to know the theoretical laws that govern it. The RNA also has a generalization capability (also called "inference"), that is, it is able to determine the output values of a situation for which it is given the input data even if This situation was not learned during the learning phase, provided that the entries are contained in a validity domain (in other words, it is ensured that the values of the entries from the situation to the calculate are between or near the extreme input values that were used for learning). This is particularly useful for systems having a wide range of operating points and for which it is industrially unthinkable to explicitly parameterize all possible situations. For better accuracy, we must verify that the learning situations constitute a sufficiently refined mesh of the field of operation. Finally, RNA is an easily scalable system. These advantages are also valid in the case of other types of mathematical expressions with coefficients evaluated iteratively. Moreover, in such embodiments, the obtaining step is therefore performed by a mathematical estimation, which makes it possible to move from the flowmeter and to realize a gain in costs, in mass and in size. In contrast to some otherwise known systems, for which an odorless Kalman filter is used to determine the synaptic weights of an RNA, the present method uses a mathematical expression with iteratively evaluated coefficients (including RNA) in the same way. integrating at the stage of obtaining the UKF. The operation and purpose are therefore quite different. In some embodiments, the fluid mass in the reservoir is a non-linear function of the fluid level in the reservoir. In some embodiments, the reservoir is non-cylindrical. In some other embodiments, the reservoir is a cylinder whose generatrices are not perpendicular to the average level of the fluid in the reservoir. It will be recalled that a cylinder is a surface defined by the passage of a line of fixed direction, called a generator, along a closed planar curve. A cylinder trunk is also referred to as a cylinder, that is to say the solid delimited by a cylinder and two parallel planes, the two planes being non-parallel to the generatrices of the cylinder. The aforementioned characteristics reflect the shape of most tanks, including tanks with a spherical bottom. When the tank is non-cylindrical, the emptying rate of the tank (i.e., the derivative of the level as a function of time) is not an affine function of the outlet flow of the tank. As mentioned above, this non-linearity is particularly well taken into account by the UKF.
    [0010] In some embodiments, a state of the odorless Kalman filter includes a bias of the raw rate value. In some embodiments, it is assumed that the rate is equal to the sum of the gross rate obtained at the obtaining step and a bias (positive or negative). More generally, in other embodiments, it is assumed that the rate is an affine function of the gross rate obtained, the bias then being a vector constituted by the coefficients of the affine function. The fact that the state of the Kalman filter includes the bias itself allows the value of the bias to be extracted and used subsequently, for example to detect an anomaly when the value of the bias exceeds a certain threshold. In some embodiments, a state of the odorless Kalman filter includes the raw rate value as well as a bias of that value. In some embodiments, the evaluation method further includes a step of filtering fluid level fluctuations. Fluid level fluctuations refer to transient changes in fluid level at constant fluid volume in the reservoir, as opposed to changes in level due to fluid flow. Fluid level fluctuations may be due to slugging of the fluid in the tank or to the broadband noise of the signal (white noise essentially related to reservoir interactions with its mechanical and hydraulic environment). This filtering step can be carried out for example with a low-pass filter or by providing, in the UKF, a rejection of the proper mode of sloshing. For example, it is possible to write the fluid level in a form comprising a fluctuation function and to determine the coefficients of this fluctuation function (for example the amplitudes, frequencies and phases of the main terms of its series decomposition. Fourier) by including these coefficients in the state vector of the UKF.
    [0011] Such a filtering step makes it possible to attenuate the noise of the level measurements and to make the evaluation method even more reliable. The present disclosure also relates to a method for evaluating two flow rates of fluids coming respectively from a first reservoir and a second reservoir, in which the flow rates of the fluids are evaluated using separately: a first evaluation method comprising a step of estimating the flow rates of the fluids (4) using an odorless Kalman filter, no level measurement is taken into account in a correction step of the odorless Kalman filter; a second method of the type previously proposed, in which only a level measurement of the first reservoir is taken into account; a third method of the type previously proposed, in which only a level measurement of the second reservoir is taken into account; a fourth method of the type proposed above, in which the two level measurements are taken into account; and in which the flow rates evaluated by that of the four processes are returned which takes into account exactly the available measurements and only these. In the present method, it is sought to evaluate two flow rates whose values are corrected on the basis of two level measurements. Both level measurements may be simultaneously available, simultaneously unavailable, or only one or both of them may be available. There are therefore four situations of availability of the level measurements. The idea of such a method is to evaluate the two flow rates according to four variants of the previously described method and then to select the most appropriate variant 5 for the level measurements available. The four methods differ in that they partially, totally, or in no way take into account the level measurements returned by the measurement step. Each of the four processes can therefore be optimized according to the measures it is able to take into account.
    [0012] Among the four methods used, one retains the one that takes exactly into account the available measurements, that is to say the one which, at the moment considered, takes into account the measurements returned by the measuring step and does not take not count the measurements not returned by the measurement step. This selection therefore consists in choosing the method which is the most adapted to the situation under consideration. The selection among the four methods can be carried out before, after or during the application of the methods. In addition, this method is generalizable to a larger number of flow rates and / or cases of availability of measurements. The present disclosure also relates to a system 20 for evaluating a flow rate of a fluid coming from a reservoir, comprising measuring means able to measure a level of fluid in the reservoir and characterized in that it comprises means estimating the flow rate of the fluid using an odorless Kalman filter, said estimating means comprising means for obtaining the raw fluid flow rate and correction means connected to the means of obtaining and measuring means and configured to correct the raw rate obtained by the obtaining means as a function of the level measured by the measuring means. Such a system is particularly suitable for implementing the method described above.
    [0013] The present disclosure also relates to a propulsion system, in particular for a space launcher, comprising two tanks each containing a propellant, a combustion chamber into which the two propellants are injected, and a system for evaluating the flow rate of at least one of the propellants. 5 propellants as previously described. For a propulsion system, a correct estimate of propellant flows is particularly important. In addition, mass relief and reduced congestion due to the removal of flow meters are also desired gains. If such a propulsion system comprises a system for evaluating the two flows, or a system for evaluating each of the flow rates, it is possible to calculate the propellant mixture ratio, which is the ratio of the propellant flow rates to the entrance to the combustion chamber. The mixing ratio of two propellants is an important parameter for controlling the propulsion system.
    [0014] The present disclosure also relates to a program comprising instructions for performing the steps of the evaluation method according to any one of the previously described embodiments when said program is executed by a computer. In a particular embodiment, the various steps of the evaluation method are determined by computer program instructions. This program can use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any another desirable form. The present disclosure also relates to a computer-readable recording medium on which a computer program is recorded including instructions for performing the steps of the evaluation method according to any one of the previously described embodiments. . The information carrier may be any entity or device capable of storing the program. For example, the medium may comprise storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or a magnetic recording means, for example a floppy disk or a Hard disk. On the other hand, the information medium may be a transmissible medium such as an electrical or optical signal, which may be conveyed via an electrical or optical cable, by radio or by other means. The program according to the invention can be downloaded in particular on an Internet type network. BRIEF DESCRIPTION OF THE DRAWINGS The invention and its advantages will be better understood on reading the following detailed description of embodiments of the invention given as non-limiting examples. This description refers to the accompanying drawings, in which: FIG. 1 represents a propulsion system equipped with a flow rate evaluation system according to a first embodiment; FIG. 2 represents the mass of propellant contained in the tanks of the propulsion system of FIG. 1 as a function of the level; FIGS. 3A and 3B show the availability of reservoir level probe measurements as a function of time; FIGS. 4A to 4C illustrate the application of the flow rate evaluation system of FIG. 1; FIG. 5 represents a propulsion system equipped with a flow rate evaluation system according to a second embodiment; FIG. 6 represents a fluid circuit equipped with a system 30 for evaluating flow rate according to a third embodiment.
    [0015] DETAILED DESCRIPTION OF THE INVENTION FIG. 1 shows a propulsion system 50, in particular for a space launcher, comprising two reservoirs 20, 21 each containing a propellant (for example liquid hydrogen and liquid oxygen respectively), a combustion chamber 30 in which the two propellants are injected, and a propellant flow rate evaluation system 10 at the inlet of the combustion chamber 30. The combustion gases develop a thrust during their ejection into a nozzle 32 downstream of the combustion chamber 30.
    [0016] In the present case, the propulsion system 50 is an integrated flow system in which heated propellant (for example hydrogen) drives turbopumps 24, 25 before being reinjected into the combustion chamber 30. Control valves V1, V2 serve to modulate the flow rate of heated propellant entering the turbines of the turbopumps 24, 25, in order to control the liquid propellant flow rate pumped by the turbopumps 24, 25. As can be seen in FIG. 1, the reservoirs 20, 21 are non-cylindrical reservoirs. More particularly, in the example under consideration, they are tanks with a spherical bottom. It follows that the relationship between the level of propellant in each reservoir and the mass of propellant in said reservoir is nonlinear. In the present case, an example of such a relationship is given in FIG. 2. Curve G20 (curve G21, respectively) represents the mass of propellant m in reservoir 20 (or reservoir 21) as a function of the level of ergol n in this same tank. Each of the curves G20, G21 consists of a substantially linear central portion 25 (corresponding to a substantially cylindrical central portion of the reservoir). For the highest levels or the lowest levels, the curves are flattened, which means that the tanks are narrowed relative to their central portion. This curve shape is characteristic of a hemispherical shape or the like at the ends of the tanks.
    [0017] In the present case, the non-linearity of the relationship between the propellant flow rate at the inlet of the combustion chamber 30 and the propellant level in the tanks 20, 21 stems not only from the shape of the reservoirs (geometric non-linearity), as explained above, but also a second non-linearity related to the expression of the flow rate as a function of the operating parameters of the propulsion system (mechanical and thermodynamic non-linearities). As shown in FIG. 1, the tanks 20, 21 are equipped with respective level probes 22, 23, intended to measure the level in the tank. The size and shape of the tanks as well as the configuration of the level probes 22, 23 may vary from one tank to another. The level probes 22, 23 consist of a set of sensors. These sensors do not return a signal that is taken into account only when they are denuded, that is to say when they are at least partially above the free surface of the propellant contained in the tank. For example, the availability curves of the level probes 22, 23 as a function of time t are shown respectively in FIGS. 3A and 3B during the operation of the propulsion system 50. The tanks 20, 21 emptying at different speeds, the curves the availability of level 22, 23 probes are generally different. Availability is 1 when the level probe is being dewatered and capable of returning a measurement; availability is 0 otherwise. As indicated in FIG. 3A, the level probe 22, composed of several sections, is being dewatered between t1 = 10s and t2 = 40s on the one hand and between t5 = 120s and t7 = 150s on the other hand. During the rest of the time, its availability is zero, ie the level probe 22 returns no measurement: the level of propellant is between two consecutive sensors. Similarly, as shown in FIG. 3B, the level probe 23 is being dewatered between t3 = 60s and t4 = 90s on the one hand and between t6 = 130s and t8 = 160s on the other hand. For the rest of the time, the level 23 probe is not available. It is apparent from Figures 3A and 3B that the level probes 22, 23 provide intermittent measurements. In addition, the availability periods of the level 22, 23 probes do not necessarily coincide. Indeed, the first dewatering period of the level probe 22, between t1 and t2, is completely disjunct from the first dewatering period of the level probe 23, between t3 and t4, while the second dewatering period of the Level probe 22, between t5 and t7, overlaps the second dewatering period of the level probe 23, between t6 and t8, for ten seconds from t6 to t7. The measurements provided by the level probes 22, 23 may therefore not be simultaneously available. The level probes 22, 23 are connected to a rate evaluation system 10, as can be seen in FIG. 1. As previously indicated, the evaluation system 10 comprises measuring means 17 able to measure the fluid level in the tank. The measuring means 17 comprise the level probes 22, 23 and an acquisition card 19 to which they are connected.
    [0018] The evaluation system 10 comprises means for estimating the flow rate of each propellant using an odorless Kalman filter. These means are calculation means comprising in particular a computer 11 of which the acquisition card 19 is part. In the present case, the estimation means comprise an initialization device 12, a transformation device 14, a device 16 and a correction device 18. The evaluation system 10 uses, as input data, in addition to the level measurements collected by the measuring means 17, data on the operation of the propulsion system 50. For example, has shown an acquisition device 15 connected to the turbopumps 24, 25 and 30 receiving as data the speed of rotation of the turbopumps and the propellant pressure at the output of the turbopumps. Other data could also be collected. Furthermore, the acquisition device 15 could be connected to other components or locations of the propulsion system 50 to acquire other data. For example, it could also be connected to the combustion chamber 30 to measure the pressure of the gases in the combustion chamber 30. The odorless Kalman filter is an iterative filter, each new estimate being calculated from the previous state and current data. The values returned by the correction device 18 serve both as input values to the transformation device 14, which ensures the iterative operation of the evaluation system 10, but also overall output values. These output values can be used for different purposes, for example for monitoring the operation of the motor or for regulating the motor, in particular for adjusting the opening of bypass valves V1, V2 as a function of the flow rates evaluated. In FIG. 1, where k is the current time (or the current iteration), it is noted: - xk the current state, Jek an estimate of its mean value and Pk its covariance matrix; - Qk and Rk the respective covariance matrices of the state noise wk and the measurement noise vk of the Kalman filter (see equations (1) and (2) previously exposed), assumed to be known a priori, knowing that the noise measurement can be measured; - Sik the ith sigma-point at instant k; - uk the input data; - 2 - k + iik the raw value of the state x obtained at time k + 1 knowing its same value at time k; 3 0 2 1 7 4 0 16 - P - the raw covariance matrix of the state x at the instant k + 1 knowing the covariance matrix Pk; - yk the measured values from the measuring means 17; - 52k + 1 the value of the state x at the instant k + 1 corrected as a function of the level measurements measured by the measuring means 17 and Pk + 1 its covariance matrix. The state xk of the odorless Kalman filter chosen in the present embodiment comprises the level of ergol nk, the rate qk estimated by the RNA and the bias Ek of this rate. So we have xk = [nk qk Ek] T. The state of the UKF can include as many levels, rates, and biases as the number of flows that the UKF seeks to evaluate. In the example of FIG. 1, the quantities n, q, E are vectors each comprising two components, one of which refers to the propellant of the tank 20 and the other to the propellant of the tank 21. For For reasons of clarity, unless otherwise stated, the description which follows will deal only with a rate, but the elements presented are also valid for several rates. The initialization device 12 is configured to provide the transformation device 14 with an initial state 20, P0, a function of the characteristics of the propulsion system 50.
    [0019] The transformation device 14 is configured to carry out the odorless transformation of the UKF at each instant k, i.e. to create a set of Sik sigma-points from the value 2k and the covariance matrix. Pk approximating the current state xk. The obtaining device 16 is configured to perform the step of obtaining the evaluation method. In the present embodiment, the obtaining step is implemented by an artificial neural network. This RNA provides the raw propellant flow rates at the inlet of the corresponding turbopumps 24, 25 taking as inputs the pressure of said propellant at the outlet of the pump, the pressure of the gases in the combustion chamber 30 and / or the rotational speed of the pump. said turbopump. All things being equal 3 0 2 1 7 4 0 17 Moreover, the more input data (which are all criteria for discriminating situations), the more accurate the estimate provided by the ANN. In particular, the RNA can be of the multi-layer perceptron type, in particular with a single hidden layer for the different sets of inputs, established from a database corresponding to a cartography resulting from experimental data or from physical models. . Empirically, it is observed that the output flow output of the RNA is noisy. In this embodiment, a low-pass filter is applied here, of the first order. By noting qk the bit rate obtained by the obtaining device 16 at time k, RNAk the function applied by the RNA to the input data uk, At the time step between k and k + 1, and T the constant time of the filter (here of the order of the second), an example of such a filter is an expression of the type: qk + i = eA; x qk + (1- e2-1) x RNAk (uk) (3). It is further assumed that the rate d is the sum of the raw rate q returned by the RNA and an unknown bias E, positive or negative, representing the effects of the rate modeling errors. In other words, for each propellant, we have the relation dk = qk + Ek (4). In other embodiments, equation (4) can be generalized as dk = Ekqk + Bk, wherein E is vector bias and applies to the rate as an affine function. In particular, the coefficient Ek allows the taking into account of a multiplicative bias, while the coefficient Ek allows the consideration of an additive bias. In the first example of equation (4) above, we have Ek = 1 and a = Ek.
    [0020] It is assumed, for the obtaining device 16, that the bias E is constant; this does not prevent the bias E from being corrected later by the correction device 18. This hypothesis results in the relation Ek + 1 = Ek (5). Other types of evolution relations of the bias Ek are possible. 30 for the obtaining device 16, for example a bias proportional to the thrust of the propulsion system 50, or an increasing bias as a function of the square distance with respect to the nominal operating point for which it can be assumed that the bias is the smallest (since the system is best known by definition). Thus, the 5 equations (3), (4) and (5) define flow rate equations based on the RNA implemented in the obtaining device 16. Moreover, the obtaining device 16 also includes a modeling of the evolution of the levels of propellants nk in the reservoirs. By writing that the mass of ergol mk decreases at each time step At of a quantity At x dk and that the mass is connected to the level by the functions G20, G21 (see FIG. functions generically by the letter G), we obtain for each ergol the relation: nk + 1 = G (G-1 (nk) - At X (qk + Ek)) (6) in which G-1 is the function reciprocal of the G function.
    [0021] The operation of the evaluation device 10 will now be detailed. The initialization device 12 initializes the initial state x0 by taking for example the initial filling level of the tanks, a zero flow and a zero bias. The initialization device 12 supplies the transformation device 14 with the values 20, P0 corresponding to the initial state. For this computation, the state x is assumed to be a Gaussian random variable of mean 2 and covariance P. At each iteration k, the transformation device 14 computes an expanded state vector Xk = [Xk Vk Wk] T. Its covariance matrix Pxk is the diagonal block matrix formed by the respective covariance matrices P 0 Rk. The covariance matrices Qk, Rk are obtained -1c,, k, -k- empirically and stored in a memory 13. Alternatively or in addition, the measurement covariance Rk can be calculated automatically by the estimates of the variances of the measurement noises. vk on a sliding time window.
    [0022] The transformation device 14 generates a set of sigma-points corresponding to the Gaussian random variable -k, -P Xk approximating the current state Xk. The Sik sigma-points are generated according to the conventional technique of the odorless Kalman filter, which will not be detailed here. Each sigma-point Sik, which has the same structure as the vector Xk, is supplied to the obtaining device 16. The role of the obtaining device 16 is to obtain an initial estimate (raw value) of the state vector to the time k + 1 by knowing an estimate of this vector at time k and data of the propulsion system 50 (in this case the input data uk). As indicated above, this obtaining is carried out using equations (3), (4), (5), (6) using an RNA. Since the RNA is a predictive model, the obtaining step is called here more particularly a prediction step. The evolutions of the vk and wk noises are determined according to the standard methods employed when using an odorless Kalman filter. For state noise, in the particular case where a model that is constructed (like a RNA) is used, the results of this model can be compared with known situations to obtain a map of the precision of the RNA in its area of validity.
    [0023] The values of the sigma-points Sik + i at time k + 1, provided by the RNA, are then used to determine, at the output of the obtaining device 16, a value obtained (predicted). k + iik of the state x at the moment - k + 1 knowing the value £ k at the instant k, and the covariance matrix Pk + iik of the state x at the moment k + 1 knowing the covariance matrix Pk.
    [0024] As shown in FIG. 1, these values are then supplied to the correction device 18. The correction device 18 then corrects the values obtained by the obtaining device 16 when measurements are available. For each propellant, if the level measurement is available (i.e., if the corresponding level probe is denuded), the value of the state xk + i is corrected according to the difference between the measured level yk and the predicted level (obtained level) ÿk. The value of the state is corrected in particular by a value proportional to said deviation, the proportionality factor being called "Kalman gain". The Kalman gain is calculated conventionally from the sigma-points and the level measurement. Otherwise, if the level measurement is not available, the value of the state xk + 1 determined using equation (6) is kept by the obtaining device 16, for example by setting the gain from Kalman to zero. According to the fact that no one or two level measurements are available (four possible cases), the correction device 18 consequently corrects the bit rates obtained by the obtaining device 16. The correction device 18 therefore comprises a selection function integrated with the available measures. A variant is shown in FIG. 5, in which the distinction between the cases of availability of the measurements is implemented by arranging in parallel, in the evaluation system 110, four calculation functions 11a, 11b, 11c, 11d, integrated in a computer 111, each set to cover one of the four aforementioned cases of availability of measurements. In FIG. 5, the unchanged parts of the propulsion system 50 (in particular the propulsion chamber 30 and the inputs 20 to the computers) have not been represented. In addition, each calculation function 11a, 11b, 11c, 11d has an architecture similar to that integrated in the computer 11. The calculation function 11a evaluates the flow rates of the fluids coming from the tanks 20, 21 without taking into account a measurement of level. The correction device 18 provided in the computer 11a is therefore not used. The calculation function 11b evaluates the flow rates of the fluids from the tanks 20, 21, taking into account only the measurements returned by the level probe 22, when they are available. Similarly, the calculation function 11d evaluates the flow rates of the fluids from the tanks 20, 21 taking into account only the measurements returned by the level probe 23, when they are available. Finally, the calculation function 11c evaluates the flow rates of the fluids from the tanks 20, 21, taking into account the measurements returned by the two level probes 22, 23, when they are available.
    [0025] The flow rate values evaluated by the calculation functions 11a, 11b, 11c, 11d are transmitted to a selection device 118. The selection device 118 is also connected to the level probes 22, 23 in order to know their availability. The selection device 118 selects among the four sets of received rates as a function of the availability of the measurements returned by the level probes 22, 23. The selection device 118 returns, as output, the set of flow rates. provided by the calculator which takes into account exactly the available measures. For example, if only the measurement returned by the level probe 22 is available, the selection device 118 will return the flow rate set received from the computer 11b. Alternatively, the selection device 118 may be upstream or at the same level as the calculation functions 11a, 11b, 11c, 11d. In all cases, the evaluation system 10, 110 thus permits a resetting of the value obtained by the obtaining device 16 on the basis of discontinuous and asynchronous measurements. The method implemented by the evaluation system 10 allows instantaneous evaluation of flow rates. As can be seen in FIG. 1, in order to evaluate successive bit rates, a recurrence loop makes it possible to go on to the next instant and to repeat the previously described steps. At each step, the propellant flow rate evaluated, i.e., obtained and corrected, is a -k =, k + - kf, in which the estimated gross flow rate qk is essentially determined by the RNA and esk estimated bias is determined primarily by the UKF. An application example will now be described with reference to FIGS. 3A-3B and 4A-4C. Figures 4A-4C show the operation of the propulsion system 50 for two phases (two operating points) of one hundred seconds each. FIG. 4A shows the evolution of the propellant flow from the tank 20 and entering the combustion chamber 30, estimated by a single RNA (curve Q20a) and by the evaluation device 10 (curve Q20b). The actual value of the flow is plotted for comparison (curve Q20c). As can be seen in FIG. 4A, during the first seconds, the estimated flow rate by a single RNA (curve Q20a) is closer to the actual flow (curve Q20c) than the flow rate estimated by the evaluation device (curve Q20b). . This is due to the arbitrary values that can be provided by the initializer 12. After a few seconds, this gap is resorbed. At time t = t1 (10s), as shown in FIG. 3A, the level probe 22 is denuded and the measuring means 17 acquire level measurements. The correction device 18 can therefore, on the basis of these measurements, correct the values provided by the obtaining device 16. It follows a transient regime of the curve Q20b, then a stabilization of the flow rate evaluated at a close level. the actual flow (curve Q20c). The level probe 22 remains denatured until t = t3 (40s) and is again unavailable. Between t3 and t5, the evaluated flow represented by the curve Q20b is evaluated only with the aid of the obtaining device 16 since the correction device 18 does not perform any action. It should be noted that from the first registration based on measurements, and in particular on both sides of the operating point change (t = 100s), the flow rate evaluated by the evaluation device 10 follows the actual flow rate. (curve Q20c) with great accuracy while the flow rate evaluated by a single RNA (curve Q20a) always keeps a bias or offset approximately constant compared to the actual flow. FIG. 4B, similar to FIG. 4A, relates to the evolution of the propellant flow from the tank 21 and entering the combustion chamber 30.
    [0026] The curve Q21a represents the estimated flow rate by a single RNA, the curve Q21b represents the flow rate evaluated by the evaluation device 10 and the curve Q21c represents the actual flow rate. It can be seen that the curve Q21b follows the curve Q21a as long as the correction device is not effective (see FIG. 3B: until t = t3, 60s, the level probe 23 returns no measurement). Next, the curve Q21b approaches substantially the curve Q21c. The right magnifying glass of FIG. 4B makes it possible to observe a similar effect during the second dewatering of the level probe 23, that is to say between t6 (130s) and t8 (160s). The use of the odorless Kalman filter therefore substantially improves the results that an estimator alone could provide, including an RNA-type estimator. Figure 4C shows the evolution of the mixing ratio as a function of time. The mixing ratio is defined as the ratio of the two propellant flows. Each MRz curve is obtained by performing the ratio Q21z / Q20z, where z is a, b or c. The mixing ratio is a quantity often used to regulate the operating point of a propulsion system such as the propulsion system 50. The evolution of the MRb curve representing the mixing ratio calculated from the flow rates evaluated by the device Evaluation 10 shows the respective readings of each flow at times t1 and t3, then 20 t5 and t6. The evaluation device 10 thus provides an accurate estimate of the mixing ratio by successive readjustments allowed by the odorless Kalman filter. FIG. 6 shows another embodiment of an ergol flow rate evaluation device 210. In this figure, the elements 25 corresponding to or identical to those of the first embodiment will receive the same reference sign, to the number of hundreds, and will not be described again. The evaluation device 210 of FIG. 6 is integrated within a fluid circuit 250 comprising a reservoir 220 from which a fluid flows.
    [0027] Downstream of the tank, a flowmeter 215a measures the flow of the fluid.
    [0028] However, the flow meter 215a may be inaccurate or biased. The function of the device 210 is to correct the flow rate value returned by the flow meter 215a based on fluid level measurements in the reservoir 220. The evaluation device 210 functions as the evaluation device 10, with the exception that it uses the measurements returned by the flowmeter 215a instead of using a predictive mathematical estimator such as a RNA. In other words, in equation (3) above, the term RNAk (uk) is replaced by the current value measured by flowmeter 215a and acquired by acquisition provision 215. The flow rate value obtained by the obtaining device 216 is then supplied to the correction device 18 which corrects it as a function of the measurements yk received from the acquisition card 19. As shown in FIG. 6, the corrected evaluated flow rate is returned to the flowmeter 215a by the evaluation device 210, so as to modify the calibration of the flow meter so as to reduce the bias of the measurements. Although the present invention has been described with reference to specific exemplary embodiments, modifications can be made to these examples without departing from the general scope of the invention as defined by the claims. In particular, individual features of the various embodiments illustrated / mentioned can be combined in additional embodiments. Therefore, the description and drawings should be considered in an illustrative rather than restrictive sense.
    权利要求:
    Claims (12)
    [0001]
    REVENDICATIONS1. A method of evaluating a flow rate of a fluid from a reservoir (20, 21), comprising measuring a fluid level (yk) in the reservoir and characterized in that it comprises a step of estimating the flow rate of the fluid (dk) using an odorless Kalman filter, said estimating step comprising a step of obtaining the raw fluid flow (qk) and, when a level measurement (yk ) is available, a correction step in which the raw bit rate obtained is corrected according to the level measurement.
    [0002]
    2. Evaluation method according to claim 1, wherein the gross flow rate value (qk) is measured by at least one sensor (215a), in particular a flow meter.
    [0003]
    3. Evaluation method according to claim 1, wherein the raw rate value (qk) is calculated by a mathematical estimate with coefficients evaluated iteratively, in particular by an artificial neural network.
    [0004]
    4. Evaluation method according to any one of claims 1 to 3, wherein the mass of fluid (mk) in the reservoir (20, 21) is a non-linear function of the fluid level (nk) in the reservoir .
    [0005]
    The evaluation method according to any one of claims 1 to 4, wherein a state of the odorless Kalman filter comprises a bias (Ek) of the raw rate value (qc). 3021740
    [0006]
    6. Evaluation method according to any one of claims 1 to 5, further comprising, before the correction step, a step of filtering the fluctuations of the level measurement (yk).
    [0007]
    7. A method for evaluating two flow rates of fluids respectively from a first reservoir and a second reservoir, wherein the flow rates of the fluids are evaluated by separately implementing: a first evaluation method comprising a step of estimating fluid flow rates (dk) using an odorless Kalman filter, in which no level measurement is taken into account in a step of correcting the odorless Kalman filter; - a second method according to any one of claims 1 to 6, wherein only a level measurement of the first reservoir is taken into account; A third method according to any one of claims 1 to 6, wherein only a level measurement of the second reservoir is taken into account; a fourth method according to any one of claims 1 to 6, wherein the two level measurements are taken into account; 20 and in which the flow rates evaluated by the four processes are returned which takes into account exactly the available measurements.
    [0008]
    8. Evaluation system (10, 110, 210) of a flow rate of a fluid from a reservoir (20, 21, 220), comprising measuring means (17, 22, 23) able to measure a fluid level (yk) in the reservoir (20, 21, 220) and characterized in that it comprises means for estimating the flow rate of the fluid (dk) using an odorless Kalman filter, said estimation means comprising means (16, 216) for obtaining the raw fluid flow rate and correction means (18) connected to the obtaining means and to the measuring means and configured to correct the gross flow rate. obtained (4k) by the obtaining means (16, 216) as a function of the measured level (yk) by the measuring means.
    [0009]
    9. Propulsion system (50), in particular for a space launcher, comprising two tanks (20, 21) each containing a propellant, a combustion chamber (30) into which the two propellants are injected, and an evaluation system (10). the flow rate of at least one of the propellants according to claim 8.
    [0010]
    10. A flowmeter registration system (215a), comprising an evaluation system (220) according to claim 8 for evaluating the flow rate through the flowmeter.
    [0011]
    11. Program comprising instructions for performing the steps of the evaluation method according to any one of claims 1 to 7 when said program is executed by a computer (11).
    [0012]
    A computer-readable recording medium on which a computer program is recorded including instructions for performing the steps of the evaluation method according to any one of claims 1 to 7.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3152528B1|2020-01-15|Method and system for evaluation of a fluid flow
    EP3126659B1|2019-12-11|Method and device for monitoring a parameter of a rocket engine
    EP2904353B1|2018-11-21|Multi-sensor measuring system method and system
    FR2895537A1|2007-06-29|DETECTION OF DEFECTS IN AERODYNAMIC DATA SYSTEMS BASED ON ARTIFICIAL INTELLIGENCE.
    WO2016066538A1|2016-05-06|Method of estimating a navigation state constrained in terms of observability
    WO2019239062A1|2019-12-19|Method for calibrating magnetometers fitted in an object
    FR2606872A1|1988-05-20|NAVIGATION SYSTEM IMPLEMENTING A RECURRENCE ESTIMATING CIRCUIT FOR THE INTEGRATION OF NAVIGATION
    FR3069316B1|2019-08-16|METHOD FOR ESTIMATING THE MOVEMENT OF A EVOLVING OBJECT IN A MAGNETIC FIELD
    FR3084170A1|2020-01-24|METHOD FOR DETERMINING A COVARIANCE STATE NOISE MATRIX FOR SETTING A BATTERY CHARGE OBSERVER AND DEVICE THEREOF
    FR3059777B1|2019-08-09|TEST METHOD
    WO2022043629A1|2022-03-03|Method for determining a fault for a set of sensors
    CA2745977C|2017-10-24|Method for estimating a jet temperature in a jet engine
    FR3099590A1|2021-02-05|Method for estimating the depth of a gas-liquid interface for a gas well
    FR3109991A1|2021-11-12|Method of estimating a physical quantity
    EP3785000A1|2021-03-03|Method and system for processing a temperature measurement signal delivered by a sensor
    EP3658853A1|2020-06-03|Method and device for characterising a heading determined from the measurement of the magnetic field
    EP2372912A1|2011-10-05|Calibration method and system, recording medium for said method
    FR3096137A1|2020-11-20|PS3 model registration in a turbomachine
    EP3286468A1|2018-02-28|Valve, and control method
    FR2955190A1|2011-07-15|METHOD FOR DETECTING A TARGET IN AN IMAGE ACQUIRED FROM A VEHICLE OVERLOOKING A CELESTED BODY
    FR3065824A1|2018-11-02|METHOD FOR PROCESSING BY SUPERVISION DEVICE ACQUIRED AND TRANSMITTED DATA BY A SENSOR
    WO2010004098A1|2010-01-14|Estimation of the impulse response of a system on the basis of binary observations
    同族专利:
    公开号 | 公开日
    RU2016151798A|2018-06-27|
    CN106415215A|2017-02-15|
    EP3152528A1|2017-04-12|
    FR3021740B1|2016-06-24|
    RU2016151798A3|2018-12-26|
    RU2690080C2|2019-05-30|
    KR20170012550A|2017-02-02|
    US10234316B2|2019-03-19|
    JP6529993B2|2019-06-12|
    US20170205266A1|2017-07-20|
    WO2015185842A1|2015-12-10|
    JP2017524907A|2017-08-31|
    EP3152528B1|2020-01-15|
    CN106415215B|2020-06-09|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    EP1510795A2|2003-08-22|2005-03-02|Snecma Moteurs|Device for estimating fuel mass flow rate|
    FR2965305A1|2010-09-28|2012-03-30|Snecma|PROPULSIVE SPATIAL LAUNCHER SYSTEM USING A METHOD OF CONTROLLING THE CONSUMPTION OF ERGOLS|
    US20120150517A1|2010-12-10|2012-06-14|The Boeing Company|Calculating Liquid Levels in Arbitrarily Shaped Containment Vessels Using Solid Modeling|
    WO2012156615A2|2011-05-17|2012-11-22|Snecma|Power supply system and method for eliminating the pogo effect|
    FR2996302A1|2012-10-01|2014-04-04|Snecma|METHOD AND SYSTEM FOR MULTI-SENSOR MEASUREMENT|
    US6655201B2|2001-09-13|2003-12-02|General Motors Corporation|Elimination of mass air flow sensor using stochastic estimation techniques|
    NO327870B1|2007-12-19|2009-10-12|Norsk Hydro As|Method and equipment for determining interface between two or more fluid phases|
    WO2012161716A1|2011-05-26|2012-11-29|Hach Company|Fluid quantification instrument and method|US8999139B2|2011-05-26|2015-04-07|Hach Company|Oxidation/reduction measurement|
    CN108229012B|2017-12-29|2020-08-07|武汉理工大学|Channel water level flow relation model solving method|
    CN108470017B|2018-03-29|2021-08-31|淮阴师范学院|Trace fluid jet quality matching method|
    CN110514260B|2019-07-26|2021-02-09|上海空间推进研究所|Measuring equipment and method suitable for rocket engine injector side area flow|
    CN111426353B|2020-04-08|2022-02-11|中国民用航空飞行学院|Accurate flow obtaining device and method|
    CN113297679A|2021-06-19|2021-08-24|中国人民解放军国防科技大学|Propellant mass flow observation method of variable thrust rocket engine|
    法律状态:
    2015-06-16| PLFP| Fee payment|Year of fee payment: 2 |
    2015-12-04| PLSC| Publication of the preliminary search report|Effective date: 20151204 |
    2016-06-10| PLFP| Fee payment|Year of fee payment: 3 |
    2017-04-28| PLFP| Fee payment|Year of fee payment: 4 |
    2018-06-05| PLFP| Fee payment|Year of fee payment: 5 |
    2018-06-29| CD| Change of name or company name|Owner name: SAFRAN AIRCRAFT ENGINES, FR Effective date: 20170719 |
    2020-05-20| PLFP| Fee payment|Year of fee payment: 7 |
    2021-05-19| PLFP| Fee payment|Year of fee payment: 8 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1455024A|FR3021740B1|2014-06-03|2014-06-03|METHOD AND SYSTEM FOR EVALUATING A FLUID FLOW|FR1455024A| FR3021740B1|2014-06-03|2014-06-03|METHOD AND SYSTEM FOR EVALUATING A FLUID FLOW|
    EP15729550.2A| EP3152528B1|2014-06-03|2015-06-02|Method and system for evaluation of a fluid flow|
    PCT/FR2015/051444| WO2015185842A1|2014-06-03|2015-06-02|Method and system for evaluating a flow rate of a fluid|
    RU2016151798A| RU2690080C2|2014-06-03|2015-06-02|Fluid medium flow rate estimation method and system|
    US15/315,554| US10234316B2|2014-06-03|2015-06-02|Method and system for evaluating a flow rate of a fluid|
    CN201580029797.0A| CN106415215B|2014-06-03|2015-06-02|Method and system for estimating fluid flow|
    KR1020177000195A| KR20170012550A|2014-06-03|2015-06-02|Method and system for evaluating a flow rate of a fluid|
    JP2016571152A| JP6529993B2|2014-06-03|2015-06-02|Method and system for evaluating fluid flow rate|
    [返回顶部]