Method for determining an efficiency fault of a module of a turbine shaft engine of an aircraft

Abstract

The invention relates to a method for determining an efficiency fault (R11-R15) of at least one module (11-15) of a turboshaft engine (T) of an aircraft (A), the determining method comprising: a step of determining an estimated actual map (CARE); a step of determining an actual indicator (IRE) from the estimated actual map (CARE); a step of determining (E3) a plurality of simulated maps for different efficiency configurations from a simulation of a theoretical model of the turboshaft engine (T); a step of determining (E4) a simulated indicator (ISx) for each simulated map (CARSx); a step of training (E5) a mathematical model (CLASS) by associating the simulated indicators (ISx) with the efficiency configurations (CR); and a step of applying (E6) said mathematical model (CLASS) to the actual indicator (IRE) in order to derive therefrom the actual efficiency configuration (CR).

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis in aircraft turbine shaft engines.

[0002] In existing technologies, it is known to record flight data (temperature, pressure, speed, etc.) of turboshaft engines and compare it with a database to identify potential faults. In a known manner, the database includes flight data related to the faults. This database is constructed from fault detection experienced in actual flight or from theoretical data obtained by simulating theoretical faults.

[0003] Determining a turboshaft engine failure depends not only on the quality of flight data but also on the relevance of the database. In practice, such comparisons are complex, and different diagnoses are not very accurate, potentially corresponding to similar data. In particular, for turboshaft engines composed of multiple modules (low-pressure compressor module, high-pressure compressor module, combustion chamber, high-pressure turbine module, and low-pressure turbine module), this method cannot accurately identify the defective module and the cause of the failure.

[0004] One solution to achieve accuracy is to integrate sensors into all modules of the turboshaft engine, allowing each module to be characterized independently. Such a solution could be considered for bench testing, but it's impractical at the industrial level. First, installing these sensors is time-consuming and complex. Most importantly, these sensors are placed in critical environments (high temperature and high pressure), making them either too costly or too fragile.

[0005] One of the objectives of this patent application is to enable accurate determination of the state of each module of a turboshaft engine without the need to add additional sensors to the modules of the turboshaft engine. Summary of the Invention

[0006] This invention relates to a method for determining an efficiency failure of at least one module in an aircraft turboshaft engine, the turboshaft engine including at least one compressor module, at least one combustion chamber module, and at least one turbine module, each module having its own specific efficiency. The determination method includes:

[0007] The step of determining the performance cartographie of the turboshaft engine, hereinafter referred to as "estimated actual cartographie", defines multiple mathematical relationships between the parameters of the turboshaft engine, each of which is determined based on the measured values ​​of the turboshaft engine parameters obtained during a specific flight of the aircraft.

[0008] The step of determining actual indices from actual mappings, which correspond to predetermined operating singularities of turboshaft engines.

[0009] The steps involved in determining multiple performance mappings for different efficiency configurations, hereinafter referred to as "simulation mappings", from simulations of a theoretical model of a turboshaft engine, which determine the efficiency of each module;

[0010] The step of determining simulation indices for each simulation mapping, the simulation indices corresponding to a predetermined operating singularity of the turboshaft engine,

[0011] The steps of training a mathematical model by linking simulation metrics to efficiency configurations, and

[0012] The steps involve applying the mathematical model to actual indicators in order to derive the actual efficiency configuration.

[0013] Because of this invention, the estimated actual mapping of the turboshaft engine can be defined from a few parameter measurements during a specific flight period, which advantageously makes it possible to calculate the relevant actual indicators and is easy to calculate.

[0014] Using the theoretical model of a turboshaft engine, a related mathematical model can be developed. This mathematical model can be directly applied to practical indicators to determine the efficiency configuration of the turboshaft engine, i.e., the local efficiency variations of each module. Advantageously, the mathematical model can be determined in a biased manner by processing large amounts of data. Efficiency failures of one or more modules can be determined shortly after flight by analyzing measurements obtained during the flight. Highly advantageously, there is no need to increase the number of sensors in the turboshaft engine.

[0015] The term "efficiency" refers to a parameter that affects performance, whether it is effective efficiency, flow rate, or something else. The term "efficiency configuration" refers to the value of the efficiency, and / or the value of the deviation from a reference efficiency.

[0016] Preferably, in the determination step, the actual metric is determined by comparing the actual mapping with the reference mapping. This allows the actual metric to highlight any deviations from previous conditions, such as changes in efficiency.

[0017] Preferably, the mathematical model is a classification or regression model.

[0018] Preferably, the reference mapping is obtained from simulations of a theoretical model of the turboshaft engine, for which the efficiency of each module is perfect. In this way, any module that deviates from perfect efficiency will stand out in the actual performance metrics.

[0019] Preferably, in the step of determining the actual mapping, a correction sub-step and a filtering sub-step are performed on the parameter measurements of the turbine shaft engine. Therefore, the relevant mathematical relationships are formed without bias and without considering marginal measurements.

[0020] Preferably, each mathematical relationship is obtained through computer training based on parameter measurements of the turboshaft engine obtained during a specific flight period of the aircraft. Computer training performs better than statistical training.

[0021] Preferably, each mathematical relation is obtained by training a computer from several models to determine the mathematical relation most relevant to the parameters; the optimal mathematical relation is obtained by regression or classification scores. Determining the optimal relation by analyzing the scores makes it possible to quickly and repeatedly determine the best model for each relation.

[0022] Preferably, each mathematical relation is obtained by computer training based on a subset of parameter measurements of the turbine shaft engine obtained during a specific flight period of the aircraft, while another subset of parameter measurements is used to validate the obtained mathematical relation. Separating the measurements for determining and validating the model in this way makes it possible to avoid doubling the amount of data required.

[0023] The present invention also relates to a computer program comprising instructions, when executed by a computer, for performing the steps of the previously described method for determining efficiency faults. The present invention also relates to a recording medium for the computer program. The recording medium mentioned above can be any entity or device capable of storing a program. For example, the medium can include storage media (such as ROMs (e.g., CD-ROMs or microelectronic circuit ROMs)) or magnetic recording media (e.g., hard disks). On the other hand, the recording medium can correspond to a transmissible medium (such as electrical or optical signals) that can be transmitted via electrical or optical cables, radio, or other means. The program according to the present invention can in particular be downloaded to an Internet-type network. Additionally, the recording medium can correspond to an integrated circuit containing a program, the circuit being adapted to execute or be used to execute the relevant method. Attached Figure Description

[0024] A better understanding of the invention can be achieved by reading the following description, which is by way of example only, and by referring to the accompanying drawings, which are by way of non-limiting example, wherein the same reference values ​​are assigned to similar objects, and wherein:

[0025] Figure 1 This is a schematic diagram of an aircraft, showing a turboshaft engine mounted on it.

[0026] Figure 2 This is a schematic diagram of a turboshaft engine module;

[0027] Figure 3 It is a schematic diagram of the steps to determine the estimated actual mapping and the steps to determine the actual index;

[0028] Figure 4 This is a schematic diagram of parameter measurements obtained during a specific flight of the aircraft.

[0029] Figure 5 This is a schematic diagram of the sub-steps used to determine the actual mapping;

[0030] Figure 6 It is a schematic diagram of several mathematical relationships that form the actual mapping;

[0031] Figure 7 It is a schematic diagram of the reference mapping of a theoretical model of a turboshaft engine with perfect efficiency;

[0032] Figure 8 It is a schematic diagram of determining and using mathematical models.

[0033] It must be noted that the figures illustrate the invention in detail for the purpose of implementing the invention; of course, the figures can be used to better define the invention where applicable. Detailed Implementation

[0034] This invention relates to a method for determining the efficiency variation of a module in an aircraft turboshaft engine. As an example, Figure 1 The image shows aircraft A (helicopter) including a turboshaft engine T. (Example) Figure 2 As shown, the turbine engine T comprises several modules, specifically a low-pressure compressor module 11, a high-pressure compressor module 12, a combustion chamber module 13, a high-pressure turbine module 14, and a low-pressure turbine module 15. It goes without saying that the number and type of modules 11-15 may vary depending on the turbine engine T in question.

[0035] Each module 11-15 has its own unique efficiency R11-R15. The efficiency R11-R15 of module 11-15 corresponds to the ratio between its actual performance and 100% of its theoretical performance. A change in efficiency R11-R15 is an indication of a failure in said module 11-15. According to one aspect of the invention, the change in efficiency corresponds to a change in flow rate. Therefore, when the performance of the turbine shaft engine T deteriorates, it is important to determine which module 11-15 can be attributed to this performance degradation.

[0036] Reference Figure 3 First, the steps for determining the performance mapping of the E1 turboshaft engine T will be introduced, hereinafter referred to as the "Estimated Actual Mapping" (CARE). The Estimated Actual Mapping (CARE) is obtained through computer training based on measurements of parameters MiPj obtained during a specific flight of aircraft A. As will be described later, the Estimated Actual Mapping (CARE) allows the parameters Pj of the turboshaft engine T to be compared with at least one other parameter Pk of the turboshaft engine T, which are associated with a model determined through computer training.

[0037] In practice, refer to Figure 4For the specific flight volume (VOL), the parameters MiPj of the turboshaft engine T are measured continuously or periodically. Measurements are performed using sensors located within and near the turboshaft engine T. Preferably, few sensors are used, and there is no need to add additional sensors compared to conventional turboshaft engines.

[0038] Reference Figure 4 The measured value MiPj may include, in particular, temperature measurements, pressure measurements, speed measurements, torque measurements, or others. Preferably, environmental parameters of the turbine shaft engine T (especially external temperature and external pressure) are also measured. (Refer to...) Figure 4 The measured values ​​of three parameters P1, P2, and P3 are obtained.

[0039] Preferably, such as Figure 5 As shown, in the ET11 correction sub-step, the measured value MiPj is corrected to bring it back to comparable environmental conditions (ISA ground reference conditions). For example, taking into account the external temperature, the measured value MiPj is converted to a corrected measured value MiPj'. Preferably, similarly, in the ET12 filtering sub-step, the corrected measured value MiPj' is then filtered by comparing it with certain minimum or maximum measured values ​​or operating conditions (measurements in flight). The ET11 correction sub-step and the ET12 filtering sub-step are optional sub-steps, making it possible to emphasize the relevance of the measured value to establishing the actual mapping CARE for the estimate.

[0040] Since all the measurement results MiPj, MiPj', and MiPj'', at least one mathematical relationship Fg between different parameters Pj and Pk is trained in the training ET13 sub-step. Preferably, the relationship Fg between the parameters Pj and Pk of the turbine engine T is trained by computer learning from the time-related measurements MiPj, MiPj', and MiPj''. Preferably, the mathematical relationship Fg is trained by computer from several models to determine the mathematical relationship most relevant to the parameters Pj and Pk. Preferably, the best mathematical relationship is obtained from regression or classification scores. Preferably, only a portion of the flight measurements are used to determine the mathematical relationship Fg associated with the parameters Pj and Pk, while another portion of the flight measurements are used to verify the mathematical relationship Fg.

[0041] The mathematical relationship Fg between parameters Pj and Pk refers to, for example, internal temperature as a function of rotational speed, torque as a function of temperature, or any other thermodynamic relationship. The redundancy in these associations allows for the powerful use of the results. In this example, refer to... Figure 6The mathematical relationships F12, F13, and F23 between parameters P1, P2, and P3 are shown. The trained mathematical relationships Fg make it possible to form a practical mapping CARE for the estimated turboshaft engine T in a specific flight volume of aircraft A. This estimated practical mapping CARE is used to describe the behavior of the turboshaft engine T in a specific flight volume. Advantageously, due to continuous training, the turboshaft engine T is mathematically modeled based on measurements obtained through existing sensors.

[0042] Furthermore, the mathematical relationship Fg that actually maps CARE can be determined using statistical methods. For example, by conditionally analogizing the measured values ​​MiPj, their distribution can be determined, and the dominant pattern of that distribution can be selected (which is consistent with the mathematical relationship sought).

[0043] As an example, during the operation of turbine engine T, the outlet temperature parameter T45 of the high-pressure turbine and the rotational speed parameter N1 of the gas generator are continuously measured. In this example, environmental parameters (especially external temperature T0 and / or external pressure) are also measured during the operation of turbine engine T.

[0044] Based on environmental parameters, especially the external temperature T0, the measured values ​​of parameters T45 and N1 are thermodynamically corrected to obtain the corrected measured values ​​T45' and N1' determined according to the following functions:

[0045] N1'=N1 / sqrt(T0 / T0ref);

[0046] T45' = T45 / T0 * T0ref, where T0ref corresponds to the reference temperature (ISA ground).

[0047] The corrected measurements T45' and N1' are then filtered, retaining only the most relevant ones. The filtered measurements T45'' and N1'' are shown below. In practice, the corrected measurements T45' and N1'' are compared to predetermined thresholds (maximum and / or minimum thresholds) to ensure that they correspond to a sufficiently high power demand for the turboshaft engine. In other words, marginal measurements are discarded, and the most relevant measurements are selected. The relationship between the filtered measurements T45'' and N1'' is then trained using computer training. Several mathematical relationships Fg are trained to form the estimated practical mapping CARE.

[0048] The estimated actual mapping CARE is an actual image of the turboshaft engine T estimated from multiple measurements MiPj obtained at different times during the flight of the aircraft, within the operating range of the turboshaft engine T. Advantageously, the estimated actual mapping CARE of the turboshaft engine T is updated for each flight volume (VOL) of the aircraft A.

[0049] Refer again Figure 3 The method includes the step of determining the actual E2 index IRE from the estimated actual mapping CARE and the reference mapping CAREF.

[0050] The actual index IRE is designed to highlight changes in the module efficiency of turbine engine T by comparing it to the estimated actual mapping CARE and the reference mapping CAREF. For example, the actual index IRE can be represented by the maximum value of the difference between the mathematical relationship Fg of the two mappings CARE and CAREF, a general statistic of these differences, the center point of the curve slope, etc. The actual index IRE may have different properties and be identified as characteristic singularities reflecting efficiency decline. For example, some actual index IREs are configured to highlight failures in compressor modules 11, 12 or turbine modules 14, 15. The actual index IRE corresponds to a set of mathematically defined singularities.

[0051] The reference mapping CAREF can be obtained in various ways. In particular, the reference mapping CAREF can be an average mapping of the turboshaft engine T, a mapping constructed from previous flights, or a simulated mapping.

[0052] Under the current circumstances, refer to Figure 7 The CAREF reference mapping is obtained from the theoretical model MOTH of the turboshaft engine T, defined from multiple mathematical and / or thermodynamic equations. Such a theoretical model MOTH is known to those skilled in the art. The reference mapping CAREF is not based on physical measurements performed on an actual turboshaft engine T. Preferably, the reference mapping CAREF is based on the theoretical model MOTH of the turboshaft engine, where the efficiencies R11-R15 of modules 11-15 are perfect (100%). The reference mapping CAREF is obtained through theoretical simulation, by changing the inputs of the theoretical model MOTH to obtain different outputs. Similar to before, the mathematical relationship between the input / output of parameters Pj and Pk can be obtained, and the reference mapping CAREF can be derived.

[0053] Preferably, the estimated actual mapping CARE and the reference mapping CAREF have the same mathematical relationship Fg, so they can be easily compared to form an actual index IRE that highlights local efficiency faults.

[0054] In order to detect efficiency faults in the module of the turboshaft engine T, refer to Figure 8 The method includes a step of estimating the E3 efficiency adjustment from the mathematical model CLASS applied to the actual index IRE. Using the E3 estimation step, the efficiency variation R11-R15 of each module 11-15 of the turboshaft engine T can be determined, thereby identifying which module 11-15 is defective.

[0055] To obtain accurate and relevant estimates of the efficiency R11-R15 for each module 11-15, the relevance of the mathematical model CLASS is important. To obtain the mathematical model CLASS, multiple simulation performance mappings CARSx are first determined, such as... Figure 8 As shown.

[0056] Each simulated map CARSx is derived from the theoretical model MOTH of the turboshaft engine T, defined from multiple mathematical and / or thermodynamic equations, with different efficiency configurations CR for each MOTH model. For example, the first efficiency configuration CR1 might indicate that the efficiency of the low-pressure compressor module 11 is 50%, while the efficiency of the other modules 12-15 is 100%. Similarly, another efficiency configuration CR2 might indicate that the efficiency of all modules 11-15 is 80%. Therefore, the simulated map CARSx covers a large number of efficiency configurations CR. Thus, the simulated map CARSx is a graph of the degree of efficiency degradation of the individual modules 11-15 of the turboshaft engine T.

[0057] The method then includes the step of determining the E4 simulation index ISx based on the simulation map CARSx and the reference map CAREF. Preferably, the simulation index ISx is obtained in a manner similar to the actual index IRE, with each simulation map CARSx used instead of the estimated actual map CARE. Thus, the simulation index ISx is a fundamental characterization of each efficiency configuration CR of the theoretical model MTH of the turboshaft engine T. It is preferable to use the same reference map CAREF, but it is self-evident that it can be different.

[0058] In the training E5 step, the mathematical model CLASS is trained by associating the simulated index ISx with the efficiency configuration CR. Preferably, the mathematical model CLASS makes it possible to assign an efficiency configuration CR to each simulated index ISx, thereby assigning an efficiency loss (classification), or an efficiency loss value (regression), or a deviation from a reference efficiency (regression) to each module 11-15. Thus, the sign (loss or gain) of the efficiency change can be determined, preferably the numerical value of the change. Therefore, by analyzing the simulated index ISx, the individual change in efficiency for each module 11-15 of the turbine engine T can be determined. In this example, each efficiency configuration CR corresponds to a percentage of efficiency, but needless to say, it can correspond to a change (positive or negative) or an efficiency difference compared to a reference efficiency (perfect or previous).

[0059] The optimal mathematical model CLASS is obtained through methods such as regression operators, classification operators, random forests, support vectors, and nearest neighbor methods (plus proches voisins). Advantageously, the mathematical model CLASS can be obtained through efficient computational methods to achieve the accuracy of various efficiency configurations (CRs), i.e., the fault characteristics. The same mathematical model CLASS can be advantageously used for multiple flights of the same type of turboshaft engine.

[0060] Reference Figure 8 After obtaining the mathematical model CLASS, the method includes the step of applying the mathematical model CLASS to the actual performance index IRE to derive the actual efficiency configuration. Using the mathematical model CLASS requires a suitable amount of computational power, which makes it possible to reactively identify efficiency failures after each specific flight volume (VOL).

[0061] For example, in the case of classification operators, the mathematical model CLASS training relation ,in It's an indicator. It is a vector, if the module The efficiency increased relative to the reference value, where the components corresponding to the module... value If reduced, then In the case of the regression operator, the training relation is... , among which is index, It is a vector, where the components correspond to the modules. yes ,in It is a module The efficiency, whether in the reference state ( ) or in the current state ( ).

[0062] At each new flight volume (VOL), when new flight data becomes available, a new estimated actual mapping CARE is performed using the new actual index IRE. By applying the mathematical model CLASS to the new actual index IRE, the efficiency variations of each module 11-15 of the turboshaft engine T are advantageously derived. Therefore, the operating status of each module 11-15 is known, making it possible to perform appropriate, and especially predictive, maintenance.

[0063] Preferably, the method includes the step of monitoring the efficiency R11-R15 of each module 11-15, and determining a fault if at least one efficiency is below a threshold or its slope is below a threshold. Changes in efficiency are monitored individually and / or holistically.

Claims

1. A method for determining a failure of the efficiency (R11-R15) of at least one module (11-15) of a turboshaft engine (T) of an aircraft (A), the turboshaft engine (T) comprising at least one compressor module (11, 12), at least one combustion chamber module (13) and at least one turbine module (14, 15), each module having its own specific efficiency (R11-R15), the method comprising: The steps for determining the performance mapping (E1) of the turboshaft engine (T), hereinafter referred to as the "estimated actual mapping" (CARE), define multiple mathematical relationships (Fg) between parameters (Pj) of the turboshaft engine (T). Each mathematical relationship (Fg) is determined based on the measured values ​​of the parameters (Pj) of the turboshaft engine (T) obtained during a specific flight (VOL) of the aircraft (A). The step of determining the actual index (IRE) by comparing the estimated actual mapping (CARE) with a reference mapping (CAREF), wherein the reference mapping (CAREF) is an average mapping of the turboshaft engine (T), a mapping constructed from previous flights, or a simulated mapping, and the actual index (IRE) corresponds to a predetermined operating singularity of the turboshaft engine (T), wherein the actual index (IRE) is represented by the maximum value of the difference between the mathematical relationship between the estimated actual mapping (CARE) and the reference mapping (CAREF), the general statistics of the difference, or the center point of the slope of the curve. The steps of determining (E3) multiple performance mappings for different efficiency configurations (CR), hereinafter referred to as "simulation mappings" (CARSx), from the simulation of the theoretical model (MOTH) of the turboshaft engine (T), wherein the efficiency configuration (CR) determines the efficiency (R11-R15) of each module (11-15). The step of determining a (E4) simulation index (ISx) by comparing each simulation map (CARSx) with the reference map (CAREF), the simulation index (ISx) corresponding to a predetermined operating singularity of the turboshaft engine (T), wherein the simulation index (ISx) is represented by the maximum value of the difference between the mathematical relationship between the simulation map (CARSx) and the reference map (CAREF), the general statistic of the difference, or the center point of the slope of the curve. The steps for training a mathematical model (CLASS) by correlating simulation indices (ISx) with efficient configurations (CR), and The steps for applying the mathematical model (CLASS) (E6) to the actual index (IRE) to derive the actual efficiency configuration (CR) are as follows.

2. The method according to claim 1, wherein the reference mapping (CAREF) is obtained from a simulation of a theoretical model (MOTH) of a turboshaft engine (T), for which the efficiency (R11-R15) of each module (11-15) is perfect.

3. The method according to claim 1, wherein in the determination (E1) step of actual mapping (CARE), a correction (ET11) sub-step and a filtering (ET12) sub-step are performed on the measured values ​​of parameters (Pj) of the turbine shaft engine (T).

4. The method of claim 1, wherein each mathematical relation (Fg) is obtained by computer training based on measurements of parameters (Pj) of the turboshaft engine (T) obtained during a specific flight (VOL) of the aircraft (A).

5. The method of claim 4, wherein each mathematical relation (Fg) is obtained from several models through computer training to determine the mathematical relation most relevant to the parameter (Pj), and the optimal mathematical relation (Fg) is obtained from regression or classification scores.

6. The method of claim 4, wherein each mathematical relation (Fg) is obtained by computer training based on measurements of a subset of parameters (Pj) of the turboshaft engine (T) obtained during a specific flight of the aircraft (A), and measurements of another subset of parameters (Pj) are used to verify the obtained mathematical relation (Fg).

7. A computer program comprising, when executed by a computer, instructions for performing the steps of the method for determining an efficiency fault according to any one of claims 1 to 6.

8. A computer medium comprising the computer program according to claim 7.

Images