Method for monitoring a start-up sequence of a turbine engine and monitoring system implementing said method

Abstract

One aspect of the application relates to a method for monitoring a start-up sequence of a turbomachine comprising a compressor equipped with a rotor, a starter able to rotate the rotor and a combustion chamber, the method comprising the operations of: determining (100) a time zone range (E) in which the ignition instant occurs, said range being defined on the one hand by a lower limit (El) corresponding to an event that necessarily occurs before the ignition instant and by an upper limit (E2) corresponding to an event that necessarily occurs after the ignition instant; and between said lower limit (El) and upper limit (E2), determining (130-190) a break (A) in the variation over time of the measurement signal, said break corresponding to the ignition instant of the air-fuel mixture in the combustion chamber.

Description

Technical Field

[0001] This invention relates to a method for monitoring the start-up sequence of a turbine engine to detect any signs or trends indicating engine degradation that could affect the proper deployment of the turbine engine's start-up sequence. The invention also relates to a monitoring system for implementing this method.

[0002] This invention can be applied to the field of health monitoring and predictive maintenance of turbine engines, particularly aircraft turbojet engines and turboprop engines. Background Technology

[0003] In the field of turbine engines, such as those used in aircraft turbine engines, it is necessary to detect engine ignition within the framework of engine conditioning and control. In fact, it is essential to know whether engine ignition has occurred correctly. For this purpose, the engine start-up sequence needs to be monitored. The term "start-up sequence" refers to a set of steps performed in the following order:

[0004] -1) Starter opens.

[0005] -2) The starter begins to rotate the engine, which in turn causes the high-pressure and low-pressure components to rotate.

[0006] -3) Engine speed increases

[0007] -4) Fuel is introduced into the combustion chamber.

[0008] -5) The ignition device (usually a spark plug) is opened to ignite the air-fuel mixture in the combustion chamber. -6) When the air-fuel mixture is ignited, the engine begins to rotate on its own. -7) When the engine reaches a stable engine speed, the starter disengages from the engine.

[0009] For example, an incorrect sequence of starting operations in a turbocharged engine can lead to poor ignition of the air-fuel mixture in the engine. There are several reasons for no ignition, including insufficient or no fuel due to deterioration of the fuel pump, metering device, or injectors, or insufficient or no energy due to deterioration of the spark plugs or the spark-generating system.

[0010] The degradation of systems that play a role in the start-up sequence of a turbocharged engine can be monitored by the ignition duration of the air-fuel mixture, which is defined as the time between the moment fuel is injected into the engine combustion chamber and the moment the mixture is ignited.

[0011] Many methods are known to monitor the ignition sequence during the ignition duration. However, while the moment of fuel ignition is relatively easy to detect because it requires a control system that regulates the engine (the moment of ignition is known), the exact moment of ignition is difficult to determine because it can vary depending on the conditions under which the engine starts.

[0012] For monitoring ignition sequence within the framework of engine regulation and control, the precision of the ignition moment is not important; what matters is knowing whether ignition has occurred. Within the framework of monitoring health status (known as health monitoring) and predictive maintenance of turbine engines, monitoring ignition sequence involves comparing ignition sequences with each other during aircraft flight to infer trends and derivatives, thereby predicting failures and taking preventative measures to prevent non-starting events.

[0013] Therefore, within the framework of monitoring health status and predictive maintenance, measurements must be accurate so that comparisons between these continuous measurements are reliable. In fact, due to imprecise methods, any dispersion introduced into measurements and / or detections will produce inconsistent results, leading to unreliable comparisons.

[0014] Methods for determining ignition duration have been proposed for monitoring ignition sequence within a health monitoring and predictive maintenance framework, as described in patent application FR2998003A1. This method proposes determining the ignition time—considered the time interval between the start-up instant and the ignition instant—in each engine run and studying its derivative. However, this method cannot easily distinguish between variations in duration corresponding to deterioration in engine health and variations in duration related to measurement inaccuracies.

[0015] Another method described in patent application WO2017 / 098124A1 proposes to detect the moment of ignition by determining the moment of ignition as the intersection of a nonlinear regression of the engine speed change before ignition and a linear regression of the engine speed change after ignition. Figure 1 The curve illustrates an example of the power increase of an aircraft turbine engine rotor as the engine speed changes over time. The curve shows a first portion C1 between starter activation and ignition point A at time 0, and a second portion C2 between ignition point A and the stabilization starting point R. It can be seen that, according to the first portion C1, the engine speed change to ignition point A follows a semi-parabolic shape, while according to the second portion C2, the engine speed increases linearly from point A to point R, after which the engine speed stabilizes. This stabilized engine speed is referred to as "idle speed."

[0016] It is understandable that although the ignition moment does indeed correspond to the intersection between the first and second curve segments (C1 and C2, respectively), the determination of this intersection point depends directly on the applied nonlinear and linear regressions, and therefore on the sufficiency of these regressions at curves C1 and C2. For the method in document WO2017 / 098124A1 to work, the regression model, particularly the order of the polynomial, must be specifically suited to the curves showing the change in engine speed in curve segments C1 and C2. However, the form of engine speed change can vary from engine to engine, and even from start-up: the overall shape of the curve remains constant, but the slopes of curve segments C1 and C2 may be more or less noticeably flattened. Furthermore, the linearity of the second curve segment C2 is not necessarily perfect. The applied regressions are then incorrect and may deviate significantly from the actual curve segments, especially around ignition point A. The intersection point then does not correspond to the ignition moment and may even have physical anomalies (pre-injection ignition…). Figure 2 Two examples of engine speed variations are shown, applying regressions C'1 and C'2 to curve segments C1 and C2, respectively, and to the intersection point I, which separates from ignition point A. In the example of curve 1, regression C'1 for curve segment C1 is a quadratic polynomial, and regression C'2 for curve segment C2 is a linear polynomial; their intersection point is point I, differing from ignition point A by a value Δ. In the example of curve 2, regression C'1 for curve segment C1 is a quadratic polynomial, and regression C'2 for curve segment C2 is a linear polynomial; however, due to the presence of a plateau on curve segment C1, their intersection point I is determined to be before injection, which is physically impossible.

[0017] Therefore, not only does the determination of the intersection point depend on the regression of the application and thus may be erroneous, but since the intersection point is determined for each start-up of the turbine engine, errors may accumulate and introduce discreteness, making fault prediction unreliable.

[0018] Therefore, there is a real need for a sufficiently reliable and accurate engine start sequence monitoring method for monitoring the health of turbine engines and for predictive maintenance. Summary of the Invention

[0019] To address the aforementioned issues regarding the accuracy of ignition duration, the applicant proposes a method for monitoring engine start-up sequence, which includes defining the time zone at the moment of ignition before determining that moment of ignition.

[0020] According to a first aspect, the present invention relates to a method for monitoring the start-up sequence of a turbine engine, the turbine engine comprising a compressor equipped with a rotor, a starter capable of rotating the rotor, and a combustion chamber wherein a mixture of air and fuel is ignited by an ignition device, the method comprising acquiring a measurement signal of the engine speed of the rotor during the start-up sequence, characterized by comprising the following operations:

[0021] - Determine the time zone range (bracketing) of the moment of ignition, which is defined by a lower limit corresponding to events that must occur before the moment of ignition and an upper limit corresponding to events that must occur after the moment of ignition;

[0022] as well as

[0023] - Between the lower and upper limits, determine the breakpoints in the time-varying measurement signal, the breakpoints corresponding to the instant of ignition of the air-fuel mixture in the combustion chamber.

[0024] This monitoring method can determine the ignition moment with good accuracy and repeatability, so the derivatives and trends of the start-up sequence monitored in repeated flights are reliable.

[0025] In addition to the features mentioned above, the monitoring method according to one aspect of the present invention may have one or more of the following additional features, individually or according to any technically permissible combination:

[0026] -The lower limit is defined as the instant when injection begins when fuel begins to enter the combustion chamber;

[0027] - The upper limit is defined as the end of the rotor start-up sequence.

[0028] - The end of the start-up sequence corresponds to the ignition device stopping, the engine speed reaching a threshold, or the engine speed stabilizing within a predetermined range.

[0029] - Determining the breakpoints in the time-varying measurement signal involves the following operations:

[0030] -a) Define the sliding window;

[0031] -b) Divide the window into first and second half-windows, each half-window containing a portion of the measurement signal corresponding to the curve portion;

[0032] -c) By approximating each curve portion with a single polynomial and determining the dominant coefficient of each of these curve portions;

[0033] -d) Compare the dominance coefficient of the curve portion of the first half-window with the dominance coefficient of the curve portion of the second half-window, and

[0034] -e) Identify the window where the difference between the two dominant coefficients is greatest, and this window contains the breakpoints in the change of the measured signal over time.

[0035] - The measurement signal portion of each window in the first and second half-windows is approximated by a straight line, the dominant coefficient of which is the slope.

[0036] - Operation d) for comparing slopes includes the step of determining the differences between these slopes, and operation e) for identifying windows includes the step of comparing the differences between slopes within a bracketed time zone, with the largest difference corresponding to the breakpoint in the measurement signal's change over time.

[0037] - The operation of comparing slopes (d) includes a step of statistical testing, which is suitable for rejecting breakpoints that are inconsistent with the moment of ignition.

[0038] - Determining the breakpoints in the time-varying measurement signal involves the following operations:

[0039] - Determine the derivative of the measured signal with respect to time;

[0040] - Determine the step of the derivative, which corresponds to the break point in the change of the measured signal over time.

[0041] - Determining the breakpoints in the time-varying measurement signal involves the following operations:

[0042] - Determine the second derivative of the measured signal as a function of time;

[0043] - Compare the second derivative with the threshold; and

[0044] - Identify the moment when the second derivative is greater than a threshold, which corresponds to the breakpoint in the change of the measured signal over time.

[0045] A second aspect of the invention relates to a system for monitoring the start-up sequence of a turbine engine, the turbine engine including a compressor equipped with a rotor, a starter capable of rotating the rotor, and a combustion chamber in which a mixture of air and fuel is ignited by an ignition device, the monitoring system being characterized in that it includes a calculator configured to perform the operation of the method described above.

[0046] A third aspect of the invention relates to a computer program product comprising program code instructions for performing operations as defined above. Attached Figure Description

[0047] Other advantages and features of the invention will become apparent when reading the following description, which is illustrated with reference to the accompanying drawings, wherein:

[0048] already described Figure 1 This shows an example of how engine speed changes over time during the start-up phase of a turbocharged engine;

[0049] already described Figure 2 This shows an example of incorrectly determining the moment of ignition using existing techniques;

[0050] Figure 3An example is shown of determining the moment of ignition using a method according to one embodiment of the present invention; and

[0051] Figure 4 One embodiment of the method according to the present invention is shown in the form of a functional diagram. Detailed Implementation

[0052] The following detailed description, with reference to the accompanying drawings, describes an embodiment of a method for monitoring the start-up sequence of a turbine engine, a method adapted to allow reliable detection of the ignition moment. This example illustrates the features and advantages of the invention. However, it should be noted that the invention is not limited to this example.

[0053] In the figure, identical components are labeled with the same reference numerals. For the sake of readability, the size ratios between the represented components are not considered.

[0054] Figure 3 In the form of a block diagram and Figure 4 An example of a method for monitoring the start-up sequence of a turbine engine according to the present invention is shown in the form of a functional diagram. Method 100 first proposes to determine (steps 110, 120) the time zone range in which the ignition instant occurs. Figure 3 An example of the engine speed change over time during the start-up phase of a turbocharged engine is shown (curve C). The change in engine speed over time is a signal, such as a signal measured by a tachometer or sensor, used to control the turbocharged engine. In the method of the present invention, this signal is also used to detect the moment of ignition.

[0055] exist Figure 3 In the example of engine speed changing over time, an example of range E is shown, which defines a time zone around the ignition instant a. This range E includes two limits, called the lower limit E1 and the upper limit E2, defining the first limit before the ignition instant and the second limit after the ignition instant, respectively. The two limits E1 and E2 correspond to two moments or events that must be temporarily located before and after the ignition instant, changing with time C.

[0056] For example, the lower limit E1 can be defined by a speed threshold or any other data from the engine. A definite and easily identifiable event for the lower limit E1 could be, for example, the instant when fuel injection begins in the combustion chamber. This instant of injection must occur before engine ignition and is known because it corresponds to engine control.

[0057] The upper limit E2 can be defined by a speed threshold or any other data from the engine. For the upper limit E2, a definite and easily identifiable event could be, for example, the cessation of spark plug arcing, a moment that inevitably occurs after engine ignition, corresponding to engine data (engine controls spark plug arcing and the cessation of arcing). Alternatively, the upper limit E2 could be the end of the start-up. The end of the start-up corresponds to the engine speed exceeding a known threshold, or to a stable engine speed within an interval, which corresponds to the engine speed that should be reached when the engine has completed the start-up process.

[0058] Therefore, the operations of determining the lower limit 110 and the upper limit E2 can define a time zone range that includes the moment of ignition. This time zone is preferably chosen to be as small as possible to improve accuracy and reduce the processing time of the operations defined below.

[0059] Once the range is determined, the method according to the invention proposes to identify a breakpoint A within range E where the engine speed changes over time. This breakpoint A corresponds to the instant of engine ignition. In fact, breakpoint A is the inflection point of curve C, i.e., the moment of fuel ignition and the moment the engine begins to rotate on its own without starter assistance. As described above, curve portion C1 corresponds to the engine speed when the engine is driven by the starter, and curve portion C2 corresponds to the engine speed when the engine is rotating on its own. The instant of ignition occurs at point A, i.e., the point where the engine begins to rotate on its own. At point A, the engine speed is interrupted, and this interruption corresponds to the inflection point between C1 and C2. Regardless of the type of engine, the instant of ignition is always at inflection point A.

[0060] Therefore, the method of the present invention proposes to find the inflection point A on curve C. It particularly proposes to find the inflection point A between the lower and upper limits, so as to restrict the process of finding point A to a finite time zone. Several implementations can be considered for this purpose.

[0061] exist Figure 3 and 4 In the illustrated embodiment, the ignition instant A is determined by finding the inflection point between the curve portions C1 and C2, which are displacements between the lower limit E1 and the upper limit E2, within a sliding window f. To this end, the method includes operation 130 of determining or extracting the sliding window f. This window f is divided into two adjacent half-windows f1 and f2 (steps 140, 145). The curve portion C within each half-window f1, f2 is approximated by a polynomial. Therefore, in each half-window f1, f2, there is an adjustment polynomial independent of the other half-window. These polynomials can be of the same order or different orders. Regardless of the order, the two polynomials are separate. The method then proposes to determine the dominant coefficients of each of the two polynomials and compare these coefficients, wherein the window with the two dominant coefficients furthest apart from each other is identified as containing the inflection point A or the breakpoint.

[0062] In some implementations, the size of the window, particularly its width, can be configured according to various parameters (e.g., variations in engine type or engine speed). When a sufficiently small window is selected, such as a few points, the curved portion within each half-window is relatively short so that it can be approximated by a first-order polynomial, i.e., a straight line. The dominant coefficients of the polynomial are the slopes of the straight line. The method then includes determining the slopes a1 and a2 of the curved portions of the first and second half-windows f1 and f2, respectively, using mathematical methods well known to those skilled in the art. Figure 4 Steps 150 and 155 of the function diagram.

[0063] Then, according to Figure 4 The method of implementation includes comparing the slopes a1 and a2 of the straight lines of half-windows f1 and f2 in step 160. Steps 140 to 160 are repeated until the window slides over the entire interval length between limits E1 and E2 (step 170). In other words, the calculation and comparison of slopes are repeated for the entire length of curve C contained in the range E.

[0064] Once the slopes a1 and a2 of the entire curve portion C between limits E1 and E2 are compared, the method suggests identifying (step 180) a window containing the two slopes furthest from each other, which contains the moment of ignition. In fact, the window with the largest difference between the two slopes contains the inflection point A, which corresponds to the moment of ignition. Therefore, the moment of ignition is determined directly from detecting the inflection point A (step 190).

[0065] According to the implementation, the comparison of slopes a1 and a2 is performed by calculating the difference (a2-a1) for each window. The window with the largest difference is then determined to contain inflection point A. Determining the maximum difference (a2-a1) is considered relative (not absolute) because the slope after the breakpoint is expected to be larger than the slope before the breakpoint during nominal engine start-up. Relatively speaking, considering the difference (a2-a1) eliminates any inflection points that do not correspond to ignition (cases where a2 is less than a1). Figure 3 This displays an example window with two distinct locations, more simply referred to as window f1 and window f2, each magnified. The magnified view in window f1 shows the inflection point B between the lines with slopes a1 and a2, but this inflection point is not identified as the ignition moment because (a2 – a1) < 0. The magnified view in window f2 shows the inflection point A between the lines with slopes a1 and a2, which, since (a2 – a1) > 0, will be identified as the ignition moment.

[0066] According to the alternative approach, comparing slopes a1 and a2 involves a statistical test step, which is applied to reject any inflection points that do not correspond to the moment of ignition. Based on these statistical tests, the two slopes a1 and a2 are considered to follow the normal law defined below: Where a and σ are known, and ∑ is the empirical correlation matrix between the data. These laws are approximated by Gaussian and thus by the following relationship: (a2-a1) / σ, where σ is the local standard deviation calculated over two windows f1 and f2, located to the left and right of instant ii, respectively, and follows Student's law: T(2 fen_temp-1), which is easily calibrated with the expected rejection rate. The rejection rate is the probability that an interruption occurs despite no interruption. For example, the rejection rate could be 5%.

[0067] This alternative approach restores the detection method to the instant when the derivative of engine speed over time undergoes a step change, or the instant when the second derivative of this change with engine speed over time exceeds a predefined threshold. Therefore, this alternative approach can help calibrate and determine the threshold where an interruption occurs. Thus, its advantage is that it makes the detection at the moment of ignition more robust.

[0068] According to other implementations, the breakpoint A or inflection point can be determined by the derivative of engine speed over time (i.e., curve C) and by detecting a step on that derivative. Indeed, estimations of the derivative show that it is almost zero at point A and quickly becomes positive again. In other words, the derivative of curve C is calculated only in the interval between the lower limit of E1 and the upper limit of E2, and a step is sought within this same interval.

[0069] According to other embodiments, the breakpoint A can be determined by the second derivative of the engine speed over time (i.e., curve C) and by detecting the instant when the second derivative exceeds a predefined threshold. In fact, at point a, a change in curvature is observed, and the second derivative should pass through 0. As in the above embodiment, the second derivative of curve C is calculated only within the interval between the lower limit E1 and the upper limit E2, and the instant when the second derivative exceeds the threshold is sought only within this same interval.

[0070] The aforementioned operation, which allows for the measurement of the moment of ignition, can be implemented in various monitoring methods for the start-up sequence of turbine engines, particularly within the framework of monitoring the health status of turbine engines and predictive maintenance.

[0071] Regardless of the implementation method, the method of the present invention can easily isolate the breakpoint, thereby accurately determining the ignition moment. Due to this precision, the method can be implemented in the field of monitoring the mechanical health of turbine engines, and the measurement of the ignition moment can be continuously repeated in each flight without the risk of dispersion due to the method, thus allowing the determination of trends and derivatives.

[0072] The method according to the invention also has the following advantages: it is not only robust to changes in engine speed over time from one flight to another, but it can also operate even when the slope interruption is not significant.

[0073] The method of the present invention also offers the advantage of requiring only one measurement: engine speed, high-pressure volume, or low-pressure volume measurement. This measurement is already common in engines because it is necessary for other applications and typically features high sampling frequency, accuracy, and resolution, making it particularly suitable for implementing the method according to the invention.

[0074] The method just described can be implemented in a monitoring system for the start-up sequence of a turbine engine. This system includes a compressor equipped with a rotor (also called an engine) and a starter capable of rotating the rotor prior to the ignition phase. The system also includes a combustion chamber in which an air-fuel mixture is ignited by an ignition device (e.g., a spark plug) to ensure the rotor rotates on its own. The system also includes a ground-based or airborne computer configured to perform operations according to the method of the invention.

[0075] Although described by way of a number of examples, alternatives and implementations, the method for monitoring startup sequence according to the present invention includes various alternatives, modifications and improvements that will be obvious to those skilled in the art, and it should be understood that such alternatives, modifications and improvements are part of the scope of the present invention.

Claims

1. A method for monitoring the start-up sequence of a turbine engine, the turbine engine comprising a compressor equipped with a rotor, a starter capable of rotating the rotor, and a combustion chamber wherein a mixture of air and fuel is ignited by an ignition device, the method for monitoring the start-up sequence of the turbine engine comprising acquiring a measurement signal of the engine speed of the rotor during the start-up sequence, characterized in that... Includes the following operations: - Determine the time range (E) of the moment of ignition (110), the lower limit (E1) of which corresponds to an event that must occur before the moment of ignition in chronological order, the lower limit (E1) being defined as the moment of injection when fuel begins to enter the combustion chamber; the upper limit (E2) of which corresponds to an event that must occur after the moment of ignition in chronological order. as well as Between the lower limit (E1) and the upper limit (E2), determine (130-190) the breakpoint (A) of the measurement signal changing with time, the breakpoint corresponding to the instant of ignition of the air-fuel mixture in the combustion chamber; determining the breakpoint of the measurement signal changing with time includes the following operations: -a) Determine the sliding window (f); -b) Divide the window into first and second half-windows (f1, f2), each half-window containing a portion of the measurement signal corresponding to the curve portion (C); -c) By approximating each curve portion with a single polynomial and determining the dominant coefficients (a1, a2) of each of these curve portions. -d) Compare the dominant coefficient (a1) of the curve portion of the first half-window (f1) with the dominant coefficient (a2) of the curve portion of the second half-window (f2), and e) Determine the window with the largest difference between the two dominant coefficients, the window with the largest difference containing the breakpoint (A) of the measurement signal changing over time.

2. The method for monitoring the start-up sequence of a turbine engine according to claim 1, characterized in that, The upper limit (E2) is defined as the end of the rotor start-up sequence.

3. The method for monitoring the start-up sequence of a turbine engine according to claim 2, characterized in that, The end of the start-up sequence corresponds to the ignition device stopping, the engine speed reaching a threshold, or the engine speed stabilizing within a predetermined range.

4. The method for monitoring the start-up sequence of a turbine engine according to claim 1, characterized in that, The measurement signal portion of each of the first and second half-windows (f1, f2) is approximated by a straight line, with the dominant coefficient being the slope.

5. The method for monitoring the start-up sequence of a turbine engine according to claim 4, characterized in that: - The operation d) comparing the slopes includes the step of determining the differences between these slopes, and - The operation e) of identifying the window includes the step of comparing the differences between slopes within the time zone range (E), where the largest difference corresponds to a break in the measurement signal over time.

6. The method for monitoring the start-up sequence of a turbine engine according to claim 4, characterized in that, The operation d) comparing slopes includes a statistical test step, which is adapted to reject breakpoints that are inconsistent with the moment of ignition.

7. A system for monitoring the start-up sequence of a turbine engine, the turbine engine comprising a compressor equipped with a rotor, a starter capable of rotating the rotor, and a combustion chamber wherein a mixture of air and fuel is ignited by an ignition device, the system for monitoring the start-up sequence of the turbine engine being characterized in that it comprises a calculator configured to perform the operation of the method for monitoring the start-up sequence of a turbine engine according to any one of claims 1 to 6.

8. A computer program product comprising program code instructions for performing operations of the method for monitoring the start-up sequence of a turbine engine as described in any one of claims 1 to 6.

Images