• 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:
    The invention relates to a method for predicting a malfunction of the equipment of an aircraft, the equipment parameters being monitored and recorded during flights by measurements also called signals, the aircraft further comprising a detection means equipment failures, said failures being also recorded in the memory of the aircraft. During the maintenance phases between flights, the measurements and failures are retrieved in a programmable computer to form a database and (a) during a maintenance phase between flights, a program is executed at least once. data analysis system for determining a set of parameter pairs whose signals of the pair evolve in a positively correlated manner over time and in the absence of failure, and (b) during a maintenance phase between once the set of parameter pairs has been determined, a first detection program is executed to calculate the correlations for the pairs of the determined set of pairs and the value of the computed correlation of a pair for each pair of parameters. a given flight falls below a determined positive detection threshold, a first type of anomaly is reported.
    公开号:FR3016710A1
    申请号:FR1450434
    申请日:2014-01-20
    公开日:2015-07-24
    发明作者:Jean-Hugues Pettre
    申请人:Jean-Hugues Pettre；
    IPC主号:
    专利说明:

    [0001] The present invention relates to a method for predicting a malfunction of the equipment of an aircraft of an aircraft fleet. It has applications in the field of aircraft maintenance. During the operation of an aircraft, the on-board electronic systems that monitor the aircraft equipment and record measurements of status parameters or equipment operation may further generate fault messages that are also recorded. These measurements and messages are stored in one or more memories during the operation of the aircraft, and are then extracted and retrieved by the mechanic during the maintenance phases between the flights. After the flight, for maintenance, the ground engineer retrieves the messages and, for each message, the mechanic will perform a confirmation search process or non-confirmation of malfunction. At the end of this process, the dysfunction is either confirmed or unconfirmed. This process is time consuming because many messages of the same type are produced at each flight, which forces the mechanic to intervene on the aircraft even for repetitive messages that do not cause, in fact, real dysfunction. To reduce the maintenance time, it is proposed a method of analyzing the recorded measurements to warn the mechanic of an anomaly that did not necessarily degenerate into a fault message during the flight but which may be a sign of a future breakdown. The invention thus makes it possible to act preventively with respect to failures. In addition, the maintenance may concern a fleet of aircraft that includes several aircraft of the same type, that is to say, aircraft that are built in the same way and therefore have the same equipment. Also, it is interesting to use information on a set of aircraft of the same type to improve maintenance, which the proposed method allows. In principle, the invention is based on the search for a loss of correlation between recorded measurements or parameter signals monitored during operation of the aircraft during flights. First of all, we look for strong correlations between signals for flights during which there was no failure. We keep the corresponding parameters and we search for other flights, in practice the last flights, if there has been no loss of correlation between these same parameters and therefore the corresponding signals. As a simplification, we are interested in correlations between two signals. More particularly, therefore, correlations are considered for pairs of signals. A loss of correlation for the pair is then significant of an anomaly. In variants one can consider correlations for three (triplets) signals or more signals. Thus, the invention relates to a method for predicting a malfunction of the equipment of an aircraft of a fleet of aircraft of the same type, each of said equipment of each aircraft having at least one state or operating parameter, said parameters being monitored during flights, in time, by measurements of equipment parameters recorded in a memory of the aircraft, said recorded measurements, also called signals, being detectable in time and for the corresponding parameter and for corresponding equipment and for the corresponding aircraft, the aircraft further comprising means for detecting equipment failures, said failures being also recorded in the memory of the aircraft, said recorded failures being also detectable in time and for at least the corresponding equipment and for the corresponding aircraft, the recorded measurements and the recorded failures being in o can be found for each corresponding flight. According to the invention, during the maintenance phases between flights, the recorded measurements, or signals, and possible failure (s) are copied into a programmable computer to form a database of aircraft aircraft signals. the fleet, and: a) during a maintenance phase between flights, the computer performs, at least a first time, a data analysis program on the aircraft signal data base, and with said Data analysis program: - a set of parameter pairs is determined whose signals of the pair evolve in a positively correlated manner over time and in the absence of failure recorded for at least one given flight of at least one of the aircraft of the fleet, the correlation corresponding to a result of calculating the correlation value between signals of the pair greater than a determined positive correlation threshold, b) during a maintenance phase between flights, once the set of peer parameters were determined, whether during the maintenance phase during which said pair determination was made or during a maintenance phase subsequent to the maintenance phase during which said pair determination was performed. performed, the computer performs a first detection program on the basis of aircraft signal data, and with said first detection program: - correlations are calculated for the pairs of the set of pairs determined for one or flights subsequent to those used for said pairing determination, and when the value of the computed correlation of a pair for the given flight falls below a determined positive detection threshold, a first type of anomaly is reported for the equipment corresponding to the signals of the pair for said flight in its entirety. The terms "recorded measurements" and "signals" are equivalent in the context of the invention.
    [0002] In various embodiments of the invention, the following means can be used alone or in any technically possible combination, are used: when determining the set of pairs of parameters whose signals of the pair are changing in a positively correlated manner over time and in the absence of a failure recorded for at least one given flight of at least one of the aircraft in the fleet, said parameters of the pair must be parameters monitored in all aircraft of the fleet. the fleet, - in an equivalent manner, during the analysis program, a set of parameter pairs is determined whose signals of the pair evolve in a positively correlated manner over time and in the absence of recorded failure. for at least one given flight of at least one of the aircraft of the fleet, the correlation corresponding to a correlation value calculation result between signals of the pair greater than or equal to a positive determined correlation threshold, in an equivalent manner, during the first detection program, the correlations for the pairs of the set of pairs determined for one or more subsequent flights to those used for said determination of pairs are calculated , and when the value of the computed correlation of a pair for the given flight falls below or becomes equal to a positive determined detection threshold, a first type of anomaly is reported for the equipment corresponding to the the pair for said flight in its entirety, - the pairs are determined within recorded measurements, or signals, for parameters coming from the same equipment, the signals of a pair always corresponding to the same given equipment, pairs different from different equipment, - each pair of parameters corresponds to two signals of the same equipment, 15 - the pairs are determined by in recorded measurements, or signals, for parameters which may come from different equipments, the signals of a pair being able to correspond to two different equipments, - pairs of parameters may correspond to signals from two different equipments, preferably 20 the pairs are determined on a set of flights of the past for which no failure has been recorded, - the pairs are determined on a set of successive past flights for which no failure has been recorded up to, not included , a first flight of said past for which a failure has been recorded, the pairs are determined on a set of past flights in which the flight signals for which a failure has been recorded for the flight are excluded; equipment corresponding to the signal excluded, - in addition to or in replacement of step b): c) during a maintenance phase between flights, once the set of 30 parameter pairs has been determined, whether during the maintenance phase during which said pairing determination was carried out or during a maintenance phase subsequent to the maintenance phase during which said pairing determination was carried out, it is performed in the computer a second detection program 35 on the basis of aircraft signal data, and with said second detection program: the correlations for the pairs of pairs are calculated over time periods of each flight determined for one or more subsequent flights to those used for said pair determination, each period for defining a block of data in the signals of the computer database, and when the value of the computed correlation of a pair in the considered period of the given flight or flight falls below a positive determined detection threshold, and eventually becomes equal to that uil, a first type of anomaly is reported for the equipment corresponding to the signals of the pair for said considered period of said flight, - the time periods have identical durations, a given flight being cut in equal time periods, - the periods of time have durations which may be variable and correspond to a specific phase of flight, the flight phases being in particular rolling, take-off, cruising flight, landing, - in addition, for each period, the variation in the correlation between two successive periods of each flight for which the correlations per periods for the pairs of the set of pairs determined were calculated, and when the value of the calculated variation of a pair between the two successive periods considered or a given flight exceeds a determined threshold of variation and, preferably, signaling a decrease in the correlation of the pair from one period to the next, a second type of anomaly is reported for the equipment corresponding to the signals of the pair for said two considered periods of said flight, - for a given flight, the variation of the correlation for the first flight period is calculated between the first period and the second period and the variation of the correlation for the second or subsequent period of the flight is calculated between said period and the previous fair period, the correlation variation between two periods is calculated by difference between the values of the coefficients. correlations of the corresponding periods, - the correlation variation between two periods is calculated by the absolute value of the difference between the correlation coefficient values of the corresponding periods, - the correlations are correlations of Pearson or Spearman, - for the execution of the data analysis program, a correlation threshold is used equal to 0.7, the parameter pairs having to have a signal correlation value calculation result greater than 0.7; the sets of pairs are determined and the detection program executed individually for each given aircraft of the fleet, the signals of the different aircraft of the fleet are not used together in common for the data analysis program and the detection program; - the results of correlation calculations of the detection program (s) are represented graphically and to represent the The correlation information by flight or by period uses the difference of 1 of the correlation calculation result so that a strong remaining correlation is represented close to zero and the correlation calculation result between [0-1] is bound to that correlations becoming negative be displayed at 1, - more generally and for all the calculations we use the difference at 1 of the result of e correlation calculation so that a strong correlation is close to zero, - it limits the calculations of the detection program (s) to a subset of the set of determined pairs, said subset corresponding to a statistical selection of the pairs of the set, the selection corresponding in particular to one or more of the following criteria: a maximum number of pairs, the pairs having the least variation between flights in the case where several flights are used for the determination of the set pairs, 25 - the pairs having the least variation between the aircraft in the case where several aircraft are used for the determination of the set of pairs, - recorded measurements and / or recorded failures are furthermore identified for the flight corresponding, 30 - Recorded measurements and / or recorded failures are locatable for each corresponding flight by comparison with temporal information of start and end of each flight, said time information may not be recorded in the memory of the aircraft or can be.
    [0003] Finally, the invention relates to a programmable computer means comprising software and hardware means for carrying out the method of the invention. It is also considered a computer data medium comprising a program for a computer and allowing it to perform the method of the invention. More specifically, a computer program that comprises code means capable of executing, when executed on a computer, the method of the invention. Finally, consider a memory support, readable by a computer, storing a program which is the computer program of the invention.
    [0004] The present invention, without being limited thereby, will now be exemplified with the following description of embodiments and implementations in relation to: FIG. 1 which represents a visualization of results of correlation calculations of a pair of signals on two successive flights of the same aircraft B, Figure 2 which represents a visualization of the signals of the pair of one of the two flights of the aircraft B of Figure 1, Figure 3 which represents a visualization of results of correlation calculations for the signal pair of FIG. 1 on successive flights of another aircraft H, FIG. 4 which represents a visualization of the signals of the pair of one of the flights of the aircraft H of FIG. Figure 3 for which there has been loss of the correlation, Figure 5 which represents a visualization of the signals of the pair of Figure 3, correlation calculations by periods and superposition of calculation results of variation of corr lation to said pair of signals for two flights of the aircraft H, and Figure 6 which shows for a particular operation the correlation between variations in time or data blocks of a particular pair of signals. In the context of the presentation of the invention, an aircraft equipment is, in general, both a driving member, e.g. a pump (one of its measurable parameters can then be the intensity absorbed), a controlled valve (one of its measurable parameters can then be its open or closed position) or a mechanical member, e.g. a flap (one of its measurable parameters can then be the angular position of said flap) or, again, a system or part of a system, e.g. a fluidic circuit (one of its measurable parameters can then be the pressure of the fluid or its flow) or an electrical circuit (one of its measurable parameters can then be the voltage or the intensity that flows). Indeed, in order to facilitate the management of the equipment, it is possible to group the equipment by system, for example the fuel system equipment, the electrical system equipment, etc. It is understood that these groupings can be done according to several levels of precision. This notion of measurable parameter of the equipment can be extended to systems or even to the aircraft globally, for example the parameter "flight" (the measurements that can produce two states: in flight / non-in-flight) or "rolling" ( measurements that can be two states: in rolling / non-rolling. Thus, the parameters may also correspond to measurements concerning the operation of the aircraft in general. This is particularly the case of speed, altitude, outside temperature, atmospheric pressure, etc. which are parameters that can also be measured and recorded. Indeed, there could be correlations between such parameters and those of equipment of the aircraft. A lot number for an item of equipment can also be considered as a measurable parameter: for example, a defective lot can be detected. The signals may therefore come directly from the aircraft, which is the general case, or may be added later, for example a flight identifier or equipment item lot number or any other signal that may be useful. The parameters of the aircraft are thus monitored during the flights, and even possibly outside the flights themselves, and they are measured and recorded in the form of recorded measurements, also called signals. These recorded measurements or signals are retrieved from a computer database for analysis during maintenance between flights. The calculator is typically a microcomputer. Conventionally, these signals are identifiable according to various criteria and, in particular the time, the parameter and / or the corresponding equipment, the flight and the aircraft concerned. This means that for a given recorded measurement, one is able to know the moment of the measurement, to which it corresponds: in practice to what parameter it corresponds, to which flight it belongs and to which aircraft it belongs. This is obtained directly and / or indirectly according to the case: for example, for the time, a temporal variable is associated with each measurement, or, then, one knows the time time of beginning of the measurements as well as the periodicity of the measurements and one can to deduce the moment of a particular measure.
    [0005] In practice, the aircraft incorporates a flight recorder. This flight recorder is responsible for recording the various signals of the aircraft throughout the flight. This recording is one of the raw materials of the process. In the example given here, we will focus more specifically on the fuel system. The displayed parameters or signals represent the fuel system of the end tanks and the center of the wings of the aircraft and are identified by their names with, for example: CURM_20QN2_3_1, CURM 20QN2_2_1, CURM 20QN2_1_1, CURM 20QN1 3 1, CURM 20QN1_2_1 , CURM 20QN1_1_1, CURM 11QL2_3_1, CURM 11QL2_2_1, CURM 11QL2_1_1, CURM 11QL1_3_1, CURM 11QL1_2_1, CURM 11QL1_1_1, CURM 10QL2_3_2, CURM 10QL2_2_2, CURM 10QL2_1_2, CURM 10QL1_3_2, CURM 10QL1_2_2, CURM 10QL1_1_2, CURM_9QL2_3_2, CURM_9QL2_2_2, CURM_9QL2_1_2, CURM_9QL1_3_2, CURM_9QL1_2_2, CURM_9QL1_1_2, CURM_6QA4_3_2, zo CURM_6QA4_2_2, CURM_6QA4_1_2, CURM_6QA1_3_1, CURM_6QA1_2_1, CURM_6QA1_1_1, CURM_5QA4_3_1, CURM_5QA4_2_1, CURM_5QA4_1_1, CURM_5QA1_3_2, CURM_5QA1_2_2, CURM_5QA1_1_2, CURM 20QA1_3_2, CURM 13QN2_3_1, CURM 13QN2_2_1, CURM 13QN2 1 January 25 CURM 13QN1_3_1, CURM 13QN1_2_1, CURM 13QN1_1_1 , FVLVBROTSH_1_1, FVLVBROTOP_1_1, FVLVBRMTSH_1_1, FVLVBRMTOP_1_1, FVLVBLOTSH_1_1, FVLVBLOTOP_1_1, FVLVBLMTSH_1_1, FVLVBLMTOP_1_1 , FVLVARMTSH_1_1, FVLVARMTOP_1_1, FVLVALMTSH_1_1, FVLVALMTOP_1_1, 30 FQVVT_R0_1, FQVVT_RM_1, FQVVT_L0_1, FQWT_LM_1, FPMPBROTPS_1_1, FPMPBROTEN_1_1, FPMPBRMTPS_1_1, FPMPBRMTEN_1_1, FPMPBLOTPS_1_1, FPMPBLOTEN_1_1, FPMPBLMTPS_1_1, FPMPBLMTEN_1_1, FPMPARMTPS_1_1, FPMPARMTEN_1_1, FPMPALMTPS_1_1, FPMPALMTEN January 1.
    [0006] The CURM (Current Measure) type signals represent the current expressed in ampere flowing in the SSPCs (Solid State Power Control) driving the fuel pumps. The following 4 characters identify the Subsystem. The next 3 digits identify the wing (1 = left wing, 2 = right wing), the phase number (3-phase) and the element number (1 = normal element, 2 = spare element ). Fuel Quantity Weight (FQWT) signals represent the amount of fuel expressed in kilograms remaining in a tank. The following 2 characters allow the identification of the concerned tank (RO = right outer, LO = left outer, MO = m iddle outer, LO = left outer). FVLV (Valve) signals represent the open or closed status of fuel valves. The next character represents the primary or backup system (A, B). The next 3 characters represent the identification of the tank concerned (RMT = right middle tank, LMT = left middle tank, LOT = left outer tank, ROT = right outer tank). The last 2 characters represent the nature of the status (OP = Open, SH = Shut). FPMP (Pump) type signals represent the non-powered or pressurized status. The last 2 characters represent the nature of the status 20 (EN = Energized, PS = Pressure switch). In addition, the aircraft generates reports triggered by a previously configured event such as landing, takeoff ... This report is a photograph of value of certain signals at the time of an event. Finally, the aircraft records all the fault messages. A failure message contains an occurrence time, a system reference (ATA Air Transport Association code), a message class: warning (WN = warning) or failure (FR = failure) and a text message. Firstly, it is necessary to determine pairs of parameters, also called pairs of signals, whose two signals of the pair have variations in time which follow one another, that is to say, which evolve from the same way, as part of the normal operation of the aircraft. For this purpose, a correlation calculation is used between pairs of recorded parameters or signals in search of the pairs having the highest correlation. As is known, the result of a correlation calculation can range from -1 (anti-correlation) to +1 (correlation) through 0 (no correlation). In the example shown, we consider the positive correlations, that is to say close to 1. We understand that if there are negative correlations (anti-correlation) it means that the signals are also linked together in operation normal and that in alternative embodiments of the invention can use these negative correlations and seek the fact that they change for the detection of anomaly. In another variant, it is possible, for negative correlation signals, to perform an inversion calculation on one of the signals to obtain a correlation calculation result that is positive instead of negative.
    [0007] In the example described more precisely, we will therefore use the parameter pairs whose correlation calculation results are close to +1 for each flight without report of failure considered, the limit being fixed at 0.7: we keep a pair of signals if its correlation calculation result is greater than 0.7. This gives a set of pairs of parameters or significant signals that will be used in subsequent anomalies searches. The correlation coefficient considered is the Pearson or Spearman coefficient between two signals for a flight. A correlation greater than 0.7 is considered to be a strong correlation. This allows the observation of the degradation of the correlation from one flight to another. It is understood that this determination of the pairs is made on a flight history, in practice a database in which the signals are transferred after each flight. This history can thus be completed as and when the flights. It is therefore possible to reiterate the determination of the pairs over time, following new flights. However, it is ensured that this determination is made on flights where failures have not been reported and, if possible, of which we are almost certain that they are flights well representative of a normal operation of the aircraft. the aircraft. Indeed, in the case of slow and progressive loss of correlation one could lose one / the pairs without there having been a detection of anomaly. It is understood that it is possible to implement other criteria for the selection of flights and / or signals used for the determination of pairs. For the displays during the display of the results by the computer, it is preferred to invert the correlation calculation results, that is to say that in the case where the signals are strongly correlated with each other (result of the calculation of correlation close to +1), it displays a value close to 0 and therefore, in the case of a graphic representation, a small surface is displayed. For this purpose, the difference of 1 is made to the correlation calculation result and an inverse correlation value is obtained with respect to the direct or gross correlation value. We understand that this is a convention of the program, especially for viewing the results, and that it is possible to use another. In the following we will call "CIL C" the result deducted from 1 of the correlation calculation on a pair of signals. In FIG. 1, the CIL Cs of two successive flights of an aircraft B have been represented. These CIL Cs have low representation values, 0.001 for the 0 flight and 0.002 for the 1 flight, which means that the signals of the pair are highly correlated since the representation results from the difference at 1 15 of the correlation computation which has therefore given respectively 0.999 and 0.998. In this case the signals of the pair are: CURM_10QL2_1_2 and CURM 10QL1 _ 1 _2. A correlation calculation was made during the pairing determination and for this pair it did give results above the threshold of 0.7 necessary to keep the pair and, therefore, there is a significant pair there. Note in Figure 1, the existence of a cursor on the flight 1 which provides a detailed display of the characteristics of this flight. More generally, on the displays, "Flight number" means flight number, "flight date" date of flight, "duration" duration, "departure time" time of departure, "arrival time" arrival time, "FUEL" fuel, "Apply" apply, "reset" reset, "value" value, "aircraft" aircraft. In order to better understand this notion of correlation, FIG. 2 shows the two signals of the pair of FIG. 1 for the flight 1 of this aircraft B. It is clearly seen, visually, that the two signals of this pair significant have similar behaviors, explaining the result of the correlation calculation of this pair. Once the significant pairs have been determined, the corresponding parameters will be examined for flights, normally subsequent to those used for the determination, and their correlation will be calculated in search of a loss of this correlation. For example, in FIG. 3, the correlation calculations were performed for the same pair CURM_10QL2_1_2 and CURM_10QL1_1_2 on flights of another aircraft, the aircraft H which is of the same type as the aircraft B, and the CIL C are represented for a series of 7 flights. It can be seen that the CIL C representation of flights 0, 1, 2, 3 and 6 is small indicating the persistence of a strong correlation of these signals of the pair. On the other hand, for flights 4 and 5, the CIL Cs are practically at 1 visualizing a total decorrelation of the signals of the pair, a failure having probably even been recorded during these flights. For the flight 7, the CIL C is represented with a value of 0.0789 which deviates from the CIL C flights 0, 1, 2, 3 and 6. We can already visually detect for the flight 7 that it something abnormal happened. It is understood that by a simple calculation of exceeding the threshold, this anomaly detection can be automated. This threshold may be a fixed value or result from a calculation, e.g. average, on CIL C having certain characteristics, e.g. themselves lower than another threshold. In fact, the existence of this CIL C at 0.0789 can be explained with FIG. 4 which represents the signals of the pair for the corresponding flight. There is an abnormal behavior of one of the signals of the pair, one of the electrical phases falling to zero while the other retains a high value of amperage. This abnormality detection which is done globally by flight, can be refined by a detection by flight periods, a flight period corresponding to a "block" of data signals. In the example given, the periods are equal time periods splitting the flight. For example, a flight can be divided into 10 equal periods of time. Thus, in FIG. 5, the signals of the same pair CURM 10QL2 1 2 and CURM 10QL1 1 2 are represented in superimposition for the flights 6 and 7 as well as some of the periods or blocks in greyed out and the values of variation of the correlation. or "slope" between periods. For Flight 6, at the top of Figure 5, the correlation variation between periods is small: 0.005 and 0.001 respectively. The same is true for the first period of Flight 7 at the bottom of Figure 5, where the correlation variation is small at 0.003. On the other hand, for the second period of the flight 7, the variation of correlation is important at 0.251 and corresponds to the abnormal behavior of the signals indicated previously. It is therefore also visually detectable for the flight 7 that something happened abnormal because of the high value of the correlation variation. It is understood that by a simple calculation of exceeding threshold can automate this detection. This threshold may be a fixed value or result from a calculation, e.g. average, on correlation variations with certain characteristics, e.g. they even lower than another threshold. We will now explain in relation to Figure 6, how to calculate the variations of the correlation or "slope" between periods. The flight was segmented into periods corresponding to blocks of recorded data. We are interested here in the pair of signals formed of signal A and signal B. The first block considered is block 1 located at the beginning of the flight, on the left in FIG. 6. The inverse correlation calculation has been implemented. , i.e. the CIL for the signal pair in question. A first CIL was calculated for block 1, a second for block 2, and so on. The variation of CIL or slope is calculated by difference between two CIL of successive blocks. Slope 1 for block 2 has the value 0.089 and is calculated by [CIL (signal A, signal B) of block 2] - [CIL (signal A, signal B) of block 1]. Slope 2 for block 3 has the value 0.064 and is calculated by [CIL (signal A, signal B) of block 3] - [CIL (signal A, signal B) of block 2]. Slope 3 for block 4 has the value 0.004 and is calculated by [CIL (signal A, signal B) of block 4] - [CIL (signal A, signal B) of block 3]. The principle of calculation of the slope is thus to make the difference of the CIL with the block preceding the block considered. If we had calculated the slope for block 1, we would have made the opposite difference, that is to say with respect to block 2 (the following and not the previous one), by convention, since there is no block before the block 1. It is understood that these conventions are illustrative and that the principle of calculation of slope can use as an alternative the difference with the following block. In addition to the predictive use of the means of the invention, it is also possible to implement the invention in a retrospective mode starting, for the analysis of the data, a flight V for which at least one failure has been recorded and focusing on previous flights V-1, V-2 ... in order to determine an evolution of correlations, called "pattern", for at least one pair of parameters. More precisely, the evolution of the correlation or "pattern" of a given pair over a given number of flights, typically 5, is determined, comprising for the last flight the flight V of the recorded failure, ie typically the V-4 flights, V-3, V-2, V-1 and V. For the given pair, the correlations for these flights are calculated and these correlations form the "pattern". These correlations can be the raw / direct correlations or, then, the inverse ones that correspond to CILs. Preferably, the values of inverse correlations or CIL are used. In a variant, this evolution or "pattern" rather than being a series of values, is a linear combination of these values according to a determined function. In a particular case that can be compared to the linear combination, we group the flights in question and calculate the correlation or its inverse on all of these flights, which allows to obtain a single value for evolution or "pattern". Once this evolution or "pattern" determined, we sweep in the history of all flights looking, by comparison, a similar evolution or "pattern" on a sequence of the same number of successive flights, typically 5 flights, like a moving window in the flight history. It is understood that this search is carried out by performing the same types of calculations on these successive historical flights as those made to determine the evolution or "pattern" initial so that the comparison is significant. For the comparison, thresholds of acceptance or non-acceptance are preferably used rather than a search for equality in evolutions. For example, in the case where the evolution or "pattern" corresponds to a single value, we look for this "pattern" in the history by defining a threshold s of tolerance, and we consider that we find the "pattern" if: value calculated on a window of the history - initial value (that calculated with flight V) <s. We can thus determine the number of times that we find the "pattern" in question in the history. It is understood that this a posteriori analysis can also be used for a predictive analysis having firstly determined the "patterns" that are significant of a given failure, that is to say in practice the significant pair (s). and the correlation value (s), and using this "pattern" for comparisons with a new flight, the last (most recent) flight of the comparison window being the new flight, the previous flights already being in the history.
    [0008] Note that, preferably, in the case where one carries out calculations by periods or blocks, these computations are made within a given flight, that is to say that the blocks considered belong only to on a given flight.5
    权利要求:
    Claims (10)
    [0001]
    REVENDICATIONS1. A method for predicting a malfunction of the equipment of an aircraft of a fleet of aircraft of the same type, each of said equipment of each aircraft having at least one state or operating parameter, said parameters being monitored during flights, in time, by measurements of equipment parameters stored in a memory of the aircraft, said recorded measurements, also called signals, being identifiable in time and for the corresponding parameter and for the corresponding equipment and for the corresponding aircraft, the aircraft further comprising means for detecting equipment failures, said failures also being recorded in the aircraft memory, said recorded failures also being detectable in time and for at least the corresponding equipment and for the corresponding aircraft, recorded measurements and recorded faults are also identifiable for each v corresponding ol, characterized in that during the maintenance phases between flights, the recorded measurements, or signals, and any possible failure (s) are copied into a programmable computer to form a database of aircraft aircraft signals. the fleet, and in that: a) during a maintenance phase between flights, the computer executes, at least a first time, a data analysis program on the basis of aircraft signal data, and in that with said data analysis program: - a set of parameter pairs is determined whose signals of the pair evolve in a positively correlated manner over time and in the absence of failure recorded for at least a given flight of at least one of the aircraft of the fleet, the correlation corresponding to a result of calculating the correlation value between signals of the pair greater than a determined positive correlation threshold, b) during a maintenance phase between vo 1s, once the set of parameter pairs has been determined, whether during the maintenance phase during which said pair determination was made or during a maintenance phase subsequent to the maintenance phase in which said pairing determination has been performed, the computer performs a first detection program based on aircraft signal data, and in that with said first detection program: - correlations are calculated for pairs of the set of pairs determined for one or more subsequent flights to those used in said pair determination, and in that when the value of the computed correlation of a pair for the given flight falls below a determined positive detection threshold is reported a first type of anomaly for the equipment corresponding to the signals of the pair for said flight in its entirety.
    [0002]
    2. Prediction method according to claim 1, characterized in that in complement or replacement of step b), c) during a maintenance phase between flights, once the set of pairs of parameters has been determined, whether during the maintenance phase during which said pairing determination was carried out or during a maintenance phase subsequent to the maintenance phase during which said pairing determination was carried out, it is performed in the computer a second detection program on the basis of aircraft signal data, and in that with said second detection program: - on determined time periods of each flight the correlations for the pairs of the set of pairs determined for one or more subsequent flights to those used for said pair determination, each period for defining a block of data in the signals of the database of the calculator, and in that when the value of the calculated correlation of a pair for the considered period of the given flight or falls falls below a positive determined detection threshold, and possibly becomes equal to this threshold, a first type of anomaly is reported for the equipment corresponding to the signals of the pair for said considered period of said flight.
    [0003]
    3. Prediction method according to claim 2, characterized in that the periods of time have identical durations, a given flight being cut in equal time periods.
    [0004]
    4. A prediction method according to claim 2, characterized in that the periods of time have durations that can be variable and correspond to a specific flight phase, the flight phases being in particular rolling, takeoff, cruising flight, landing.
    [0005]
    5. prediction method according to one of claims 2, 3 or 4, characterized in that calculates further, for each period, the variation in the correlation between two successive periods of each flight for which the correlations per periods for the pairs of the set of pairs determined have been calculated, and in that when the value of the calculated variation of a pair between the two successive periods considered of the given flight or flight exceeds a determined threshold of variation, and Preferably signaling a decrease in the correlation of the pair from one period to the next, a second type of anomaly is reported for the equipment corresponding to the signals of the pair for said two considered periods of said flight.
    [0006]
    6. Method according to any one of the preceding claims, characterized in that the correlations are correlations of Pearson or Spearman.
    [0007]
    7. Method according to any one of the preceding claims, characterized in that for the execution of the data analysis program, a correlation threshold equal to 0.7 is used, the parameter pairs having to have a result of correlation value calculation between signals greater than 0.7.
    [0008]
    8. Method according to any one of the preceding claims, characterized in that the sets of pairs are determined and the detection program executed individually for each given aircraft of the fleet, the signals of the different aircraft of the fleet not being used together for the data analysis program and the detection program.
    [0009]
    9. Method according to any one of the preceding claims, characterized in that graphically represents the results of correlation calculations of the detection program (s) and in that to represent the correlation information by flight or by period, uses the difference at 1 of the correlation calculation result so that a strong remaining correlation is represented close to zero and the correlation result of the correlation between [0-1] is bound so that correlations that become negative are displayed to 1.
    [0010]
    10. Method according to any one of the preceding claims, characterized in that limits the calculations of the detection program (s) to a subset of the set of determined pairs, said subset corresponding to a statistical selection of pairs of the set, the selection corresponding in particular to one or more of the following criteria: - a maximum number of pairs, - the pairs having the least variation between flights in the case where several flights are used for the determination of the set pairs, - the pairs with the least variation between aircraft in the case where more than one aircraft is used for the determination of the set of 20 pairs.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3097455B1|2019-02-20|Method for predicting an operational malfunction in the equipment of an aircraft or aircraft fleet
    CA2746537C|2017-05-09|Standardization of data used for monitoring an aircraft engine
    EP2376989B1|2019-10-16|Identification of defects in an aircraft engine
    EP3008531B1|2019-07-31|Methods of creating a database and of formulating a map of operating states of aircraft and a method of monitoring the operation of an associated aircraft
    EP2500267B1|2016-03-02|Monitoring of a flight control actuator of an aircraft
    EP3039497B1|2018-08-29|Monitoring of an aircraft engine to anticipate maintenance operations
    FR3052273A1|2017-12-08|PREDICTION OF TROUBLES IN AN AIRCRAFT
    CA2888716A1|2014-05-01|System for monitoring a set of components of a device
    FR2991486A1|2013-12-06|METHOD AND DEVICE FOR ASSISTANCE IN FOLLOWING MISSION OF AN AIRCRAFT
    EP3080670A2|2016-10-19|Forecasting maintenance operations to be applied to an engine
    FR3046268A1|2017-06-30|AIRCRAFT FLIGHT DATA OPERATION SYSTEM
    Dienst et al.2016|Automatic anomaly detection in offshore wind SCADA data
    Letourneau et al.2008|Improving preciseness of time to failure predictions: Application to APU starter
    FR3035232A1|2016-10-21|SYSTEM FOR MONITORING THE HEALTH CONDITION OF AN ENGINE AND ASSOCIATED CONFIGURATION METHOD
    FR3056002B1|2019-11-22|METHOD FOR ANALYZING DATA OF VEHICLE COMPONENTS
    FR3066844A1|2018-11-30|METHOD FOR PREDICTING THE RUNNING OF A FAILURE ON AN APPARATUS IN TIME
    FR3095424A1|2020-10-30|System and method for monitoring an aircraft engine
    EP3503274B1|2020-05-13|Method for determining the health of a fuel cell using measurements during real use
    CA3121061A1|2021-12-08|Method and control system for a damage level of at least one aircraft part, associated aircraft
    JP2021108110A|2021-07-29|System and method for monitoring and predicting vehicle damage
    FR3033661A1|2016-09-16|METHOD OF ANALYZING A FLIGHT OF AN AIRCRAFT TO EVALUATE THE EXPOSURE TO A RISK GIVEN THEFT
    FR3076634A1|2019-07-12|METHOD FOR ANALYZING PLATFORM FAILURE SYMPTOMS, AND SYSTEM THEREOF
    FR3076632A1|2019-07-12|METHOD FOR ANALYZING MAINTENANCE MESSAGES GENERATED BY AT LEAST ONE PLATFORM, AND ASSOCIATED SYSTEM
    FR3018546A1|2015-09-18|METHOD FOR MONITORING THE CONDITION OF AN ENGINE BY MONITORING EXHAUST GASES
    同族专利:
    公开号 | 公开日
    EP3097455B1|2019-02-20|
    US10417843B2|2019-09-17|
    SG10201806233RA|2018-08-30|
    WO2015107317A1|2015-07-23|
    SG11201605936PA|2016-08-30|
    US20160340059A1|2016-11-24|
    EP3097455A1|2016-11-30|
    FR3016710B1|2016-01-08|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US20050197805A1|2001-03-01|2005-09-08|Fisher-Rosemount Systems, Inc.|Data presentation system for abnormal situation prevention in a process plant|
    US20030014692A1|2001-03-08|2003-01-16|California Institute Of Technology|Exception analysis for multimissions|
    US20130274964A1|2012-04-16|2013-10-17|Flight Data Services Limited|Flight data monitoring and validation|FR3074590A1|2017-12-04|2019-06-07|Societe Air France|METHOD FOR PREDICTING AN OPERATING FAILURE OF ONE OR MORE EQUIPMENTS OF AN ASSEMBLY|US5719566A|1996-06-28|1998-02-17|Sundstrand Corporation|Method and apparatus for detecting dormant actuator failure|
    US6499114B1|1999-02-17|2002-12-24|General Electric Company|Remote diagnostic system and method collecting sensor data according to two storage techniques|
    WO2002066933A1|2001-02-23|2002-08-29|Arkray, Inc.|Monitoring apparatus and monitoring object apparatus|
    DE10153151A1|2001-10-27|2003-05-15|Airbus Gmbh|Diagnostic system and diagnostic procedures to support aircraft maintenance|
    US8117007B2|2008-09-12|2012-02-14|The Boeing Company|Statistical analysis for maintenance optimization|
    EP2461296A1|2010-12-02|2012-06-06|Techspace Aero S.A.|Device for monitoring aeronautical equipment|US11086750B2|2016-04-07|2021-08-10|University Of Maryland, College Park|Systems and methods for determination of health indicators using rank correlation analysis|
    US10325421B2|2016-10-24|2019-06-18|The Boeing Company|Systems and methods for aircraft message monitoring|
    JP6671314B2|2017-05-19|2020-03-25|シャープ株式会社|Network system, information processing method, server, and electric device|
    US10719772B2|2017-10-27|2020-07-21|The Boeing Company|Unsupervised multivariate relational fault detection system for a vehicle and method therefor|
    US11061414B2|2017-12-20|2021-07-13|General Electric Company|Fleet mission advisor|
    JP6593821B1|2018-11-08|2019-10-23|三菱電機株式会社|Electronic control unit|
    FR3098942B1|2019-07-17|2021-11-19|Safran Aircraft Engines|Method, device, computer program for predicting aircraft failure / anomaly in starting an aircraft|
    法律状态:
    2015-12-18| PLFP| Fee payment|Year of fee payment: 3 |
    2017-01-27| PLFP| Fee payment|Year of fee payment: 4 |
    2017-12-05| PLFP| Fee payment|Year of fee payment: 5 |
    2020-01-28| PLFP| Fee payment|Year of fee payment: 7 |
    2020-12-23| PLFP| Fee payment|Year of fee payment: 8 |
    2022-01-10| PLFP| Fee payment|Year of fee payment: 9 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1450434A|FR3016710B1|2014-01-20|2014-01-20|METHOD FOR PREDICTING AN OPERATIONAL MALFUNCTION OF AN AIRCRAFT OR AN AIRCRAFT FLEET|FR1450434A| FR3016710B1|2014-01-20|2014-01-20|METHOD FOR PREDICTING AN OPERATIONAL MALFUNCTION OF AN AIRCRAFT OR AN AIRCRAFT FLEET|
    EP15705644.1A| EP3097455B1|2014-01-20|2015-01-20|Method for predicting an operational malfunction in the equipment of an aircraft or aircraft fleet|
    SG11201605936PA| SG11201605936PA|2014-01-20|2015-01-20|Method for predicting an operational malfunction in the equipment of an aircraft or aircraft fleet|
    PCT/FR2015/050138| WO2015107317A1|2014-01-20|2015-01-20|Method for predicting an operational malfunction in the equipment of an aircraft or aircraft fleet|
    SG10201806233RA| SG10201806233RA|2014-01-20|2015-01-20|Method for predicting an operational malfunction in the equipment of an aircraft or aircraft fleet|
    US15/112,533| US10417843B2|2014-01-20|2015-01-20|Method for predicting an operational malfunction in the equipment of an aircraft or aircraft fleet|
    [返回顶部]