Early leak detection method and device configured to implement such a method

Abstract

The present invention relates to a method (100) for early leak detection on an object (10) by means of a leak detection device (1), the method (100) including the following steps: -establishing (S2) a pressure difference in at least one portion of the tested object (10), respectively with respect to the outside of the object (10); - measuring (S3), during a test cycle (CL) having a predetermined duration, a physical quantity (R) linked to the leak level (FU) in the at least one portion of the object (10); - comparing (S4) the measured value of the physical quantity (R) with a first reference extreme value (Rmax, Rmin); - determining (S5) whether the object (10) has a leak depending on the result of the comparison of the measured value with the reference value (Rmax, Rmin); and - interrupting (S6) the test cycle (CL) if it is determined that the tested object (10) has a leak.

Description

[0001]ANTICIPATED LEAK DETECTION METHOD AND DEVICE CONFIGURED TO IMPLEMENT SUCH A METHOD The present invention relates to the field of leak detection methods for checking the leaktightness of an object, and more particularly to methods for detecting leaks by pressure difference. The present invention also relates to leak detection devices configured to implement such methods. The present invention further relates to a method for filtering leak test cycles on a typology of object to be tested for the construction of a database making it possible to determine at least one reference value linked to the leak test for said typology of object, and in particular for the anticipated leak detection method according to the invention. Various systems and methods are known for detecting leaks, for example tracer gases, the soap bubble method, etc.In the present case, the invention relates more particularly to the detection of leaks by pressure difference (or variation). Thus, when it is desired to detect a leak by pressure difference, the object to be tested, the level of leaktightness of which is to be checked, undergoes a controlled pressure difference. This means in particular that there is a variation in the pressure (overpressure or depression) of a given volume internal to the object (called direct method), or to a closed volume surrounding said object (called indirect method), for example in relation to the outside. Then, after a determined time, there is a measurement of the evolution of the pressure value in a characteristic volume during a predefined time interval, in order to determine whether the object has a leak, and also sometimes to be able to quantify the leak level of the tested object.Indeed, a leak results in a variation in pressure over a given period of time, more specifically the variation in pressure per unit of time can for example be linked to a leak level by the following mathematical relationship: in which, F is the leak generally expressed in standard cubic centimeters per minute (or scm. 3 / min), ΔP is the pressure variation in Pascal (Pa) measured in a relevant (or characteristic) volume, Δt the time interval (in seconds) of the measured pressure variation ΔP, V the relevant volume to be considered (for example the internal volume of the object) generally expressed in cubic centimeters (cm 3 ), and k a multiplicative constant (in Pa -1). It should be noted that leakage can also be expressed in other ways, for example in the form of a mass flow rate. The formula that links leakage to a physical quantity can therefore take different forms depending on the measurement method used and the physical quantity studied. Thus, whatever the object or sub-element of the object, it is possible to check whether it has leaks and to determine its level of tightness (or its level of leakage). The object to be tested may be an electronic device, packaging, a container, etc. The tolerances on the level of tightness (or leakage) can therefore vary greatly depending on the object to be tested, its volume, its shape and / or its function. However, when testing the tightness of an object,environmental parameters and / or temporary and various parasitic phenomena can make it difficult to measure the evolution of pressure over time and / or limit the repeatability of such measurements. This problem is all the more pronounced when leak detection is carried out in an industrial environment, such as a factory, i.e. an environment where the temperature and / or pressure can vary locally and / or over time depending on the operations carried out on the objects to be tested or in the vicinity of the leak detection device. In addition, the detection device may be intended to test the leaktightness of objects on a production or manufacturing line. In this case, it is necessary for the leaktightness test to be as rapid as possible,exact and repeatable as possible so as not to disrupt (or slow down) the production or manufacturing line. The present invention proposes to remedy at least one of the problems mentioned above by proposing a new method for early detection of leaks on an object by means of a leak detection device, said method comprising the following steps: – establishing a pressure difference in at least one part of the object tested or in an enclosure surrounding the object to be tested, this respectively in relation to the outside of the object or in relation to at least one part of the object tested; – measuring, during a test cycle having a predetermined duration,a physical quantity related to the leakage level in said at least one part of the object tested or in an enclosure surrounding the object to be tested; – comparing the measured value of the physical quantity with a first reference extreme value; – determining whether the object has a leak based on the result of the comparison of the measured value with the first reference extreme value; – interrupting the test cycle if it is determined that the object tested has a leak. The method according to the invention thus makes it possible to determine in advance, i.e. before the entire test cycle time has elapsed, whether the object tested has a leak or a leakage level greater than a predetermined threshold value. Thus, with the method according to the invention, reductions in the test cycle time of between 10 and 70% are experimentally observed. According to a possible characteristic,said method comprises the following steps: – comparing the measured value of the physical quantity with a second reference extreme value; – determining whether the object does not have a leak based on the result of the comparison of the measured value with the second reference extreme value; – interrupting the test cycle if it is determined that the tested object does not have a leak. Having a second reference extreme value makes it possible to frame the measured value,a first extreme reference value making it possible to determine that the object has a leak in advance and a second extreme reference value making it possible to determine in advance that the object does not have a leak. This results in a reduction in the average test time for determining whether an object has a leak or whether it meets the sealing requirements previously defined for the type of object considered. According to another possible characteristic, the first and second extreme reference values ​​define a reference interval of values, the comparison of the measured value with respect to said reference interval making it possible to determine, as soon as the measured value is outside said reference interval, whether or not the object has a leak, as long as the measured value is between the limits of the reference interval,the leak test continues until the test cycle is completed. According to another possible characteristic, the first reference extremal value RMAX is defined by the following formula: RMAX = RL + MC + k1 ID; where RL is a leak threshold, MC is a measure of central tendency of a compensation value, I, D is a dispersion indicator and k1 is a positive real constant. According to another possible characteristic, the second extreme reference value R MIN is defined by the following formula: R MIN = R L + M C – k2I D ; where RL is a leakage threshold, MC is a measure of central tendency of a compensation value, ID is an indicator of dispersion and k2 is a positive real constant. Note that the leakage threshold R Lthus corresponds to the leak value defined (or predetermined) to consider the object as leakproof or as having a leak. In the context of a normal test cycle, i.e. without an early leak detection method, after a defined time, and beyond transient phenomena altering the measurement, the variation of the physical quantity measured per unit of time is identified (directly or indirectly) with the leak level of the tested object and is compared with the leak threshold R Lto determine whether the object is leaktight or not. It should also be noted that the measured value of the physical quantity is identified, for example, with a leak level or a leak rate, for example expressed in Pa / sec, but this measurement may have a shift, for example a shift that varies during the day or year due to environmental parameters (temperature, humidity, season, etc.) that must be taken into account and compensated for in establishing the first value and / or the second extreme reference value, in order to be able to determine as precisely as possible whether the tested object has a leak or a leak level greater than a predetermined threshold value, and / or on the contrary whether the tested object can be considered leaktight (or at least that it has a leak level less than a predetermined value, or leak threshold).According to another possible feature, the value of the central tendency measurement and the dispersion indicator are determined from several tests of objects of the same type as the object to be tested (i.e. identical or similar objects). Advantageously, the values ​​of the central tendency measurement and the dispersion indicator are empirical values ​​determined on a large number of identical or similar objects structurally and / or in terms of their behavior with respect to the leak detection method, this in order to determine the most correct central tendency measurement and dispersion indicator values ​​possible and to be sure that the method according to the invention indicates in advance with a high level of precision and accuracy whether the tested object has a leak or is leakproof.According to another possible characteristic, the measure of central tendency is an arithmetic mean or a median, while the indicator of dispersion is respectively a standard deviation or an interquartile range. Depending on the type of distribution of the possible measured values ​​of the physical quantity, the measure of central tendency is an arithmetic mean or a median, the indicator of dispersion being then respectively a standard deviation or an interquartile range. More specifically, when the distribution of the measured values ​​of the physical quantity is a normal distribution, that is to say when the statistical distribution of values ​​is according to a normal law, then the measure of central tendency is an arithmetic mean and the indicator of dispersion is a standard deviation.Whereas if the distribution of the measured values ​​of the physical quantity is a non-normal distribution then the central tendency measurement is a median and the dispersion indicator is an interquartile range. According to another possible characteristic, the constants k1 and k2 are equal to each other, said constants k1 and k2 being between 1 and 5, and preferably equal to 3. The constant values ​​k1 and k2 make it possible to establish a level of confidence on the determination of whether the object has a leak or is leaktight, generally the greater the value of the constants k1 and k2, the greater the accuracy of the method according to the invention, but at the expense of the time saved by the method according to the invention on the test cycle. It will be noted that the greater the values ​​of the constants k1 and k2, the less transient phenomena (draft, human intervention, etc.) are taken into account or impacting on the anticipated decision, but to the detriment of the time saving provided by said method. According to another possible characteristic, if the measured value of the physical quantity at the start of the test is substantially equal to one of the last values ​​measured during the previous test cycle (and preferably the last value measured during the previous test cycle, the chosen measured value having to be sufficiently representative of the values ​​measured at the end of the test cycle), then the comparison of the measured values ​​with one or more extreme values ​​is interrupted and the test cycle is completed, the leak level being determined and compared to a predetermined threshold value. During leak tests in an industrial environment, for example on a production line, unexpected and varied situations may occur and lead to testing the same object twice.In this situation, it is necessary to deactivate the anticipated leak detection method and to complete the test cycle in order to determine whether the object under test has a leak or a leak level greater than a predetermined threshold value (or a leak threshold R. L), among other things, because the tested object is already in a stabilized state and the measured values ​​can no longer be compared to reference extremal values ​​in a relevant manner. According to another possible characteristic, said method comprises the following steps: – calculation of a function on the basis of two distinct measured values ​​of the physical quantity; – comparison of said calculated function with a first reference extremal function; depending on the result of the comparison of the calculated function with said first reference extremal function, either the anticipated leak detection method continues to apply, or the test cycle is carried out to its conclusion, and is for example accompanied by an alert indicating an anomaly.Calculating a function based on two measured values ​​of the physical quantity, whether consecutive or not, but distinct, and comparing it to a reference function makes it possible to determine whether the test cycle (and the leak test on the object) is carried out without anomalies that could invalidate the anticipated leak detection method. The calculated function is advantageously relative or indicative of a parameter relating to the shape of the curve of the physical quantity, i.e. the evolution over time of the measured value of the physical quantity.Otherwise, the anticipated leak detection method is deactivated and the test cycle is carried out to its end with the issuance of an alert indicating an anomaly, so that the operator in charge of monitoring the leak detection method knows that he must carry out checks and possibly retest the object later (otherwise the object may be considered as not meeting the predefined sealing requirements). According to another possible characteristic, there is a comparison of the calculated function with a second reference extremal function, if the result of the comparison of the calculated function with said second reference extremal function is positive then the anticipated leak detection method continues to apply, otherwise the test cycle is carried out to its end, and is for example accompanied by an alert indicating an anomaly.Having a first and a second reference extremal function makes it possible to determine an interval in which the calculated function is considered correct, or normal, and that there are no anomalies invalidating the leak test. According to another possible characteristic, said first reference extremal function is calculated according to the following formula: F. MAX = F m + k3I F ; where Fm is a measure of central tendency of the calculated function, IF is a dispersion indicator relative to said calculated function, k3 is a positive real constant. According to another possible characteristic, said second reference extremal function is calculated according to the following formula: FMIN = Fm – k4 IF ; where F m is a measure of central tendency of the calculated function, I Fis a dispersion indicator relative to said calculated function, k4 is a positive real constant. As previously, the central tendency measure of the calculated function is an arithmetic mean or a median (of said function), while the dispersion indicator relative to said calculated function is respectively a standard deviation or an interquartile range. The choice of the type of central tendency measure and dispersion indicator for the calculated function is advantageously a function of the distribution type of said calculated function (normal or non-normal distribution). In addition, the constants k3 and k4 are advantageously equal and have, for example, a value greater than or equal to 2, and preferably greater than or equal to 3, 4 or 5 depending on the expected level of accuracy or precision. It should be noted that the greater the values ​​of the constants k3 and k4, the less transient phenomena (draft, human intervention,etc.) are taken into account or impacting on the anticipated decision, but to the detriment of the time saving provided by said method. According to another possible characteristic, the calculated function is a difference and / or a ratio between two measured values ​​of the physical quantity. Advantageously, the calculated function of two measured values ​​of the physical quantity, consecutive or not, but distinct, relates to the derivative of the curve or to the evolution over time of the measured value of the physical quantity. The invention also relates to a method for filtering leak test cycles on a typology of object to be tested for the construction of a database making it possible to determine at least one reference value linked to the leak test for said typology of object, and in particular for the method of early detection of leak detection mentioned above,a test cycle comprising the measurement of a physical quantity linked to the leakage level in said at least one part of the object tested or in an enclosure surrounding the object to be tested, said method comprising at least the following steps: – determination of the difference, called total difference, between measured values ​​of the physical quantity at the start and at the end of the test cycle; – comparison of this total difference with a total difference reference value, then optionally, determination whether the test cycle can be integrated into said database for the determination of at least one reference value. The filtering method according to the invention allows, by calculating a difference between the measured values ​​of the physical quantity at the start and at the end of the test cycle,to quickly eliminate any aberrant leak test cycle which leads to the generation of a database which does not correspond to the usual and / or average test conditions of the object considered. It will be noted that by start and end of cycle, we mean values ​​representative of the physical quantity measured at the start of the test and at the end of the test, this can be values ​​taken at a given time (for example after empirical determination of its stability and its representativeness) or an average value of the first and last values ​​measured. According to a possible characteristic of the filtering method, the total deviation reference value is a measure of central tendency calculated from the test cycles of said database. Advantageously,the values ​​of the central tendency measurement and the dispersion indicator are empirical values ​​determined on a large number of identical or similar objects structurally and / or in terms of their behavior with respect to the leak detection method, in order to determine the most correct central tendency measurement and dispersion indicator values ​​possible and to be sure that the filtering method allows the creation of a database for efficiently sorting the test cycles and obtaining reference values ​​for the proper functioning of the previously mentioned early leak detection method. According to another possible characteristic of the filtering method, a dispersion index is calculated for the total deviation reference value from the test cycles in said database. According to another possible characteristic of the filtering method,the comparison of the total deviation of the test cycle is carried out by verifying that said total deviation is included in an interval defined by the calculated central tendency measure and dispersion index. According to another possible characteristic of the filtering method, the central tendency measure of the total deviation is an arithmetic mean or a median, while the associated dispersion indicator is respectively a standard deviation or an interquartile range. Depending on the type of distribution (normal or not) of the possible measured values, the total deviation, the central tendency measure is an arithmetic mean or a median, the dispersion indicator then being respectively a standard deviation or an interquartile range. According to another possible characteristic of the filtering method, said method further comprises the following steps: – determination of at least two deviations, called intermediate deviations,between two distinct successive measured values ​​of the physical quantity during a test cycle; – comparison of each of said at least two intermediate deviations with a reference value associated with each of said intermediate deviations; then optionally, determination of whether the test cycle can be integrated into said database for the determination of at least one reference value. Advantageously, there is an accumulation of several criteria making it possible to determine the relevance or not of the integration of a test cycle into the database. According to another possible characteristic of the filtering method, the reference value of an intermediate deviation is a measure of central tendency calculated from the test cycles of said database. According to another possible characteristic of the filtering method,a dispersion index is calculated for each of the intermediate deviation reference values ​​from the test cycles of said database. According to another possible characteristic of the filtering method, the comparison of the intermediate deviation of the test cycle is carried out by verifying that each of the intermediate deviations is included in an interval defined by the associated calculated central tendency measure and dispersion index. According to another possible characteristic of the filtering method, the central tendency measure of each of the intermediate deviations is an arithmetic mean or a median, while the associated dispersion indicator is respectively a standard deviation or an interquartile range. Depending on the type of distribution (normal or not) of the possible measured intermediate deviation values, the central tendency measure is an arithmetic mean or a median,the dispersion indicator then being respectively a standard deviation or an interquartile range. According to another possible characteristic of the filtering method, said method further comprises the following steps: – comparison of the measured value of the physical quantity at the start of the test cycle with the measured value of the physical quantity at the end of the previous test cycle, said test cycles being consecutive test cycles; – determination whether the test cycle can be integrated into said database for the determination of at least one reference value, the test cycle not being integrated into said database if said measured values ​​are substantially equal. Advantageously, several parameters are checked to determine the relevance of storing a test cycle in said database, here the aim is to avoid testing the same object twice. In this situation,it is necessary not to store two test cycles concerning the same object, in order to preserve the relevance of the database thus constituted. According to another possible characteristic of the filtering method, said comparison of said measured values ​​of the physical quantity between two consecutive test cycles is carried out according to a central tendency measurement and a dispersion index associated with each of said measured values. According to another possible characteristic of the filtering method, the central tendency measurement is a mean or a median, the dispersion index associated with the comparison of measured values ​​between two consecutive test cycles being a standard deviation or an interquartile deviation. The invention also relates to a leak detection device configured to implement the anticipated leak detection decision method as defined above. The invention further relates to an electronic device,such as a computer, or a leak detection device, configured to implement the leak test cycle filtering method as defined above. The invention will be better understood, and other aims, details, characteristics and advantages thereof will appear more clearly during the following description of particular embodiments of the invention, given solely for illustrative and non-limiting purposes, with reference to the appended drawings, in which: – Figure 1, referenced [Fig.1], is a very schematic representation of a leak detection device according to the invention; – Figure 2, referenced [Fig.2], is a graph illustrating an example of the variation in pressure during a leak detection procedure carried out by the device of Figure 1; – Figure 3, referenced [Fig. 3], is a flowchart illustrating the leak detection method according to the invention; – Figure 4, referenced [Fig.4],[Fig. 1] is thus a very schematic representation of a leak detection device 1 for checking the tightness of an object 10. The tested object can be any object whose tightness must be checked, packaging, heat exchanger, mobile phone, an electric traction battery of a motor vehicle, etc. Said device 1 thus comprises: – a system 5 for pressurizing or vacuumizing a characteristic volume relating to the tested object 10, that is to say that it can be a volume internal to the object (direct method) or a closed volume surrounding said object (indirect method); – ​​a first pressure sensor 7 configured to measure the pressure variations of said characteristic volume, said sensor 7 making it possible to check the tightness of the object 10; – a second pressure sensor 17, which is an optional sensor,configured to measure the pressure applied in said characteristic volume by said device 5; – air connections 11, such as conduits or pipes, configured to connect the pressurizing system 5 to the object 10 and to a reference 13; – an electronic entity 15, such as an electronic circuit, connected to the different pressure sensors 7 and 17 and configured to recover the pressure values ​​measured by said sensors 7 and 17. Advantageously, the pressurizing or vacuuming system 5 comprises a pressure source 51 (or depression) which may be for example a pump, a compressor, a supply of compressed air or pressurized gas. The first pressure sensor 7 is preferably a differential pressure sensor. While the second sensor 17 is advantageously an absolute pressure sensor. In an alternative embodiment not shown,the first pressure sensor 7 is an absolute pressure sensor, and the leak detection device does not include a reference 13. A leak detection on an object 10 carried out by the device 1 of [Fig. 1] is considered as a test cycle CL and can be split into four main stages, more particularly illustrated in [Fig. 2]: – a filling stage I of the characteristic volume of the tested object with compressed air (or any gas, preferably inert, such as nitrogen), the pressure increasing to a desired pressure value P1; – a stabilization stage II, after having put the characteristic volume of the tested object under pressure, it is necessary for it to return to thermal and mechanical equilibrium generally for a predefined duration, also called stabilization time, in order to limit disturbances of the leak measurement by transient phenomena; – a test stage III,at the end of which there is measurement of the pressure variation as a function of time, in the characteristic volume, to determine a leak level of the tested object; – an emptying step IV, during which the pressurized characteristic volume of the tested object is brought back to atmospheric pressure. It will be noted that the leak detection can also be carried out in vacuum (or depression), that is to say that instead of increasing the pressure during the first step I, the pressure prevailing in the characteristic volume of the object to be tested is lowered to a predetermined value. Steps II and III remain unchanged. While the fourth step (or step IV) consists of increasing the pressure prevailing in the tested object to a pressure value corresponding to atmospheric pressure. It can therefore be considered that there is an “interversion” of the filling and emptying steps I and IV between the leak detection procedures under pressure and under vacuum. The invention,which is an anticipated method of leak detection by pressure difference using the device 1, therefore advantageously fits in particular at the level of stabilization step II and / or test step III (described above) of a test cycle C, L, one of the aims of the invention being to reduce the test cycle time and to be able to carry out leak detection as quickly as possible, thus being able to determine in an anticipated manner and in the most accurate way possible whether the tested object should be considered as having a leak or not. Said method according to the invention, more particularly illustrated in [Fig.3], thus comprises the following steps: – connecting S1 the object 10 to be tested to the leak detection device 1; – varying the pressure S2 in at least one part of the object tested or in an enclosure surrounding the object 10 to be tested, and thus establishing a pressure difference between a part of the object 10 and the exterior of the object 10, or between an enclosure surrounding the object 10 and at least one part of the object 10; – measuring S3 a physical quantity R linked to the leak level FU in said at least one part of the object 10 tested or in an enclosure surrounding the object 10 to be tested; – comparing S4 the measured value of the physical quantity R with a first extreme reference value R. MAX and / or to a second extreme reference value R MIN ; – determine S5 whether the object 10 has a leak Fu based on the result of the comparison of the measured value R with the first reference extreme value R MAXand / or determine whether the object does not have a leak based on the result of the comparison of the measured value with the second reference extreme value RMIN; – interrupt S6 the test cycle if it is determined that the tested object has a leak or if the tested object does not have a leak (i.e. if the object can be considered leaktight). It will be noted that the physical quantity may be a quantity homogeneous to a pressure, to a flow rate or to a pressure per unit of time (in particular the pressure variation per unit of time). According to alternative embodiments of the invention not shown, the measurement of the physical quantity over time is obtained by means of a suitable leak detection device, such as the device 1, i.e. configured to measure these quantities and comprising for example a pressure sensor, a flow meter, etc.The first and second reference extreme values ​​RMAX and RMIN define a reference interval IR of values, the comparison of the measured value R with respect to said reference interval I. R to determine whether or not the object has a leak as soon as the measured value R is outside the said reference interval IR, as long as the measured value R is between the RMAX and RMIN limits of the IR reference interval, the leak test continues until the test cycle is completed. The first value and the second extreme reference value RMAX and RMIN are therefore values ​​intended to frame the measured value R: – the first extreme reference value R MAX being defined by the following formula: R MAX = R L + M C + k1I D ; where RL is a leakage threshold, MC is a measure of central tendency of a compensation value, I Dis a dispersion indicator and k1 is a positive real constant; – the second extreme reference value RMIN being defined by the following formula: R MIN = R L + M C – k2I D ; where R L is a leakage threshold, M C is a measure of central tendency of a compensation value, ID is an indicator of dispersion and k2 is a positive real constant. More specifically, the leakage threshold R L corresponds to the leak threshold level (previously determined based on the desired level of sealing) from which the tested object 10 is considered to have a leak. That is to say, if the leak level of the object is higher than the predetermined leak threshold level R L, then the tested object is considered to have a leak, whereas if the leak level of the object is lower than the predetermined leak threshold level RL, then the tested object is considered not to have a leak (or to be leakproof). The measured value of the physical quantity R is identified for example with a release level or a leak rate, for example expressed in Pa / sec, but this measurement can have a lag, for example variable during the day due to environmental parameters (temperature, humidity, etc.), but also simply, because the pressure (which has varied) is not yet stabilized.It is therefore necessary to take this offset into account and compensate for it (by means of the MC compensation value) in establishing the first value and / or the second extreme reference value RMAX and / or RMIN, in order to be able to determine as precisely as possible whether the object under test has a leak or a leak level greater than the predetermined threshold value R. L , and / or on the contrary if the tested object can be considered as leaktight or if it has a leak level lower than the predetermined threshold value RL. It will be noted that the measurement step S3 of a physical quantity R linked to the leak level FU can be repeated several times, for example at regular intervals, as long as the value of the physical quantity R is between the first and second extreme reference values ​​R MAX and R MIN, that is, as long as the measured R value does not allow it to be determined whether the object is watertight or has a leak. It is therefore necessary to determine first and second extreme reference values ​​R MAXi and R MINi for each of the measured values ​​R i during a measurement step S3. This determines, for example empirically, the components of the first and second extreme reference values ​​R MAXi and R MINifor each of the measured values ​​Ri, i.e. the central tendency measurement of a compensation value MCi and the associated dispersion indicator IDi. To make this determination, it is necessary to carry out several leak tests, over a complete test cycle, on objects similar or identical to the tested object (by similar we mean an object exhibiting behavior close to or identical to their behavior with respect to the leak detection method), in order to determine the most probable values ​​of the measured values ​​at a given time, and thus to determine the central tendency measurement MC of a compensation value and the dispersion indicator I Dassociated with a measurement step S3 at a given time. Note that the measured values ​​of central tendency measurement MC is an arithmetic mean or a median, while the associated dispersion indicator ID is respectively a standard deviation or an interquartile range. Indeed, when creating a database of measured values ​​Ri during a measurement step S3, we then obtain a statistical distribution of the measured values ​​R i , if the distribution is normal (i.e. the said distribution follows a normal law) then the measure of central tendency MC of compensation value corresponds to an arithmetic mean and the dispersion indicator I D corresponds to a standard deviation. However, if the statistical distribution of said measured values ​​R iis not normal, then the central tendency measure MC of compensation value corresponds to a median and the dispersion indicator ID corresponds to an interquartile range. The constants k1 and k2 are, for their part, advantageously equal to each other, said constants k1 and k2 being between 1 and 5, and preferably equal to 3. The first and second extreme values ​​R MAX and R MINallow to determine an interval of values ​​corresponding to a confidence interval, (i.e. according to the value of the constants k1 and k2) in which there is a percentage chance that the measured value R corresponds to the real value. Thus, when the constants k1 and k2 are equal to 3, the probability that the measured value R is correct is 99.73%, while if the constants k1 and k2 are equal to 4, then the probability that the measured value R is correct is 99.993%. In an alternative embodiment illustrated in [Fig. 4] of the method 100, the method comprises a step S7 of comparing the value of the physical quantity R measured at the start of the test of an object, therefore at cycle Ci, with the measured value R at the end of the test of the previous object, therefore at cycle C 1-1 , or C cycles i etc 1-1are two consecutive cycles. Step S7 is more particularly illustrated in [Fig.5] by the representation of two consecutive cycles, so if the measured value R in cycle C i is substantially equal to the measured value R in cycle C 1-1 , that is to say that these values ​​present a difference of less than 10%, and preferably less than 5%, then the comparison of the measured values ​​with one or more extreme values, therefore step S1, is interrupted and a so-called "long" test cycle C Lis completed, after a given (or predetermined) time the leak level being determined and compared to a predetermined threshold value. There is therefore deactivation of the anticipated leak detection method 100 and the test cycle CL is completed in order to determine whether the tested object has a leak or a leak level greater than a predetermined threshold value RL. It will be noted that step S7 is advantageously carried out as early as possible during the test cycle C L, in order to optimize time savings in the event of non-compliance. In another variant embodiment of the anticipated leak detection method according to the invention, this comprises at least one calculation of a function F, step referenced S8, between two distinct measured values ​​of the physical quantity R and their comparison S9 with one or more associated reference functions FMIN and / or FMAX to this calculated function F. This variant of method 100'' is more particularly illustrated in the form of a flowchart in [Fig.6], this variant thus comprises steps S8 and S9 in addition to the steps of method 100' illustrated in [Fig.5]. Said calculated function F(R i , R i-1 ) is for example a difference between two measured values ​​R i , R i- 1 of the physical quantity R, consecutive or not, but distinct. The function F(R i , R i-1) is advantageously proportional to the derivative of the curve or to the evolution over time of the measured value of the physical quantity R. The function F(Ri, Ri-1) is for example of the form k(Ri - Ri- 1), where k is a multiplicative constant. The calculation S8 of a function F can be based on the same measurements of the physical quantity R, from the moment that there have been at least two measurements of physical quantities (therefore that there has been an iteration of step S3), or correspond to complementary measurements of the physical quantity R, that is to say additional parallel measurements to those carried out to determine whether the tested object 10 has a leak (as a function of steps S4 and S5). Said calculated function F(R i , R i-1 ) is therefore compared S9 to a first extremal reference function F MAXand / or to a second reference extremal function FMIN. It will be noted that said first reference extremal function is calculated according to the following formula: F MAX = F m + k3I F ; where Fm is a measure of central tendency of the calculated function, IF is an indicator of dispersion relative to said calculated function, k3 is a positive real constant. While said second reference extremal function is calculated according to the following formula: FMIN = Fm – k4 IF; where F m is a measure of central tendency of the calculated function, I Fis a dispersion indicator relative to the said calculated function, k4 is a positive real constant. As before, the central tendency measure Fm of the calculated function is an arithmetic mean or a median, while the dispersion indicator IF relative to the said calculated function is respectively a standard deviation or an interquartile range. The choice of the type of central tendency measure Fm and dispersion indicator I F for the calculated function is a function of the distribution type of said calculated function (normal or non-normal distribution). In addition, the constants k3 and k4 are advantageously equal and have, for example, a value greater than or equal to 2, and preferably greater than or equal to 3 depending on the expected level of accuracy or precision. Depending on the result of the comparison of the calculated function F(R i , R i-1 ) with said first extremal function F MAXreference and / or the second reference extremal function FMIN, either the anticipated 100'' leak detection method continues to apply, or the CL test cycle is carried out to its conclusion and is for example accompanied by an indicative alert for the operator. More specifically, if the calculated function F(Ri, Ri-1) is included in the reference interval [FMIN; FMAX], it is considered that there are no anomalies invalidating the leak test and in this case the detection method continues to apply, whereas if the function F(R i , R i-1 ) is not included in the reference interval [FMIN; FMAX], an anomaly occurred during the leak test, and the object must be tested according to a test cycle C Llong to ensure the accuracy of the leak test. It will be noted that the different steps S3 to S5, S7 and / or S8 and S9 can be implemented simultaneously or one after the other depending on the environmental conditions of the leak test and / or the object tested. Furthermore, the present invention also relates to a filtering method 200 of leak test cycles on a typology of object to be tested for the construction of a database making it possible to determine at least one reference value linked to the leak test for said typology of object. Said filtering method 200 makes it possible in particular to design a database BDD making it possible to determine (or calculate) the different extreme reference values ​​R MAX , R MIN and / or extremal reference functions F MAX, FMIN for the different measurements of the physical quantity R and thus allow optimal implementation of the early leak detection method 100, 100' and 100'' described above. Said method 200 comprises at least the following steps: – selection / study E1 of a test cycle CL; – determination E2 of the difference ∆ T , called total deviation, between measured values ​​of the physical quantity R at the start and end of the test cycle C L , respectively R ini and Rfin; – comparison E3 of this total difference ∆ T to a reference value of total deviation ∆ Tref , and preferably to a first extreme reference value ∆ Tref1 and to a second extreme reference value ∆ Tref2 ; – determination whether the test cycle C Lmay or may not be integrated into said database for the determination of at least one reference value. It should be noted that the start and end of the CL cycle are understood to mean values ​​representative of the physical quantity R measured at the start of the test and at the end of the test; this may be values ​​taken at a given time (for example after empirical determination of its stability and its representativeness) or an average value of the first and last values ​​measured. More particularly, from the test cycles C L from the said BDD database, there is calculation: – of the total deviation reference value ∆ Tref which is a measure of central tendency; – a dispersion index I∆ qi is calculated for the total deviation reference value ∆Tref. The measure of central tendency ∆ Trefis an arithmetic mean or a median, while the associated dispersion indicator I∆ is respectively a standard deviation or an interquartile range (this depending on the type of distribution of the total deviation values). Thus, the comparison is advantageously the verification that said total deviation ∆T of a test cycle is included in an interval defined by the central tendency measure ∆ Tref and the dispersion index I ∆ calculated, that is, ∆ Tref1 = ∆ Tref - k e I ∆ ≤ ∆ T ≤ ∆ Tref + k e I ∆ = ∆ Tref2 , where k eis a positive real constant. The constant ke has a value between 1 and 5, and preferably equal to 1. Thus, if the total deviation reference value is included in the interval defined above, then the test cycle CL (i.e. all the measurement points) is integrated into the database BDD allowing the determination of certain reference values ​​for said database and / or for the anticipated leak detection decision method 100, 100' or 100'', this for a given object or type of object. Otherwise, the test cycle C L, and its associated values ​​of measurement of the physical quantity R, is not stored to contribute to said database BDD and there is then a test of another test cycle, for example stored in memory, by the filtering method 200. [Fig.8] illustrates a first variant embodiment of the filtering method of [Fig.7], the filtering method 200' of [Fig.8] further comprising, with respect to the method 200 of [Fig.7], the following steps: – determination E4 of at least two deviations ∆i1 and ∆i2, called intermediate deviations, between two successive distinct measured values ​​of the physical quantity Ri and R i-1 during a C test cycle L ; – comparison E5 of each of said at least two intermediate deviations ∆ i1 and ∆ i2 to one or more reference values ​​∆ref1 and ∆ref2 associated with each of said intermediate deviations ∆ i1 and ∆ i2. It will be noted that steps E4 and E5 can be repeated in a sliding manner on the measured values ​​of the physical quantity R, in order to verify that the test cycle does not present anomalies invalidating its integration into said database BDD. More particularly, from the test cycles CL of said database BDD, there is calculation: – of the reference value ∆ ref1 and ∆ ref2 of each of the intermediate deviations ∆ i1 and ∆i2 which are each a measure of central tendency; – of a dispersion index Ii1 and Ii2 for each of the intermediate deviations ∆1ref and ∆ 2ref . The measure of central tendency ∆ ref1 or ∆ ref2of each of the intermediate deviations ∆i1 or ∆i2 is an arithmetic mean or a median, while the associated dispersion indicator Ii1 and Ii2 is respectively a standard deviation or an interquartile range. More particularly, the comparison E5 is advantageously the verification that said intermediate deviations ∆i1 and ∆i2 of a CL test cycle are respectively included in an interval defined by the central tendency measure ∆ ref1 or ∆ ref2 and the dispersion index k i1 and k i2 calculated values ​​associated with them, i.e. ∆ref1 – ki1 Ii1 ≤ ∆i1 ≤ ∆ref1 + Ii1 and ∆ref2 – ki2 Ii2 ≤ ∆i2 ≤ ∆ref2 + Ii2, where ki1 and ki2 are positive real constants. Thus, if the intermediate deviation value is included in the interval considered then the test cycle C L(i.e. all the measurement points) is integrated into the database BDD, in order to enable the determination of certain reference values ​​for said database, said filtering method and / or for the anticipated leak detection decision method 100, 100' or 100'', this for a given object or type of object. The constants ki1 and ki2 are, for their part, advantageously equal to each other, said constants k i1 and k i2 being between 1 and 5, and preferably equal to 1. Said filtering method 200' may also comprise the following additional steps: – comparison E6 of the measured value of the physical quantity R iniat the start of the test cycle Ci with the measured value of the physical quantity Rfin at the end of the previous test cycle Ci-1, said test cycles Ci and Ci-1 being consecutive test cycles; – determination E7 whether the test cycle Ci can be integrated into said database BDD for the determination of at least one reference value. The test cycle C i is not integrated into said database if BDD said measured values ​​are substantially equal. More particularly, it is also possible that the comparison step E6 of said measured values ​​of the physical quantity R between two consecutive test cycles C i etc 1-1 is carried out according to a measure of central tendency MC and a dispersion index ID associated with each of said measured central tendency values. The measure of central tendency M Cthe comparison of measured values ​​between two consecutive test cycles is a mean or a median, the dispersion index ID associated with being a standard deviation or an interquartile range More particularly, each of the measured values ​​R ini and R fin can therefore be framed as follows: – MC - k ID ≤ Rini ≤ MC + k ID; – M C - k I D ≤ R fin ≤ M C + k I D ; where k is a positive real constant, for example between 1 and 5, preferably equal to 1, M C and I dbeing as previously parameters calculated for each of the values ​​of physical quantity R according to the test cycles stored in the database BDD. The frames or intervals of the measured values ​​Rini and Rfin are compared to determine if there is overlap between said frames, if there is overlap then the test cycle Ci is not stored in the database, if there is no overlap is integrated into the database BDD, this in order to allow the determination of certain reference values ​​for said database BDD, said filtering method and / or for the anticipated leak detection decision method 100, 100' or 100'', this for a given object or type of object.The different steps E2 and E3, E4 and E5, E6 can be carried out sequentially or simultaneously, but the test cycle CL analyzed / tested by means of the filtering method according to the invention must meet the comparison criteria of steps E3, E5 and / or E7 to be stored E7 in the database BDD. It will be noted that the invention also relates to an electronic device, such as a leak detection device 1 or a computer, configured to implement the filtering method 200, 200' of test cycles as defined below.

Claims

Claims [Claims 1] Method for early detection of leaks (100; 100'; 100'') on an object (10) by means of a leak detection device (1), said method (100; 100'; 100'') comprising the following steps: – establishing (S2) a pressure difference in at least one part of the tested object (10) or in an enclosure surrounding the object to be tested, this respectively relative to the outside of the object (10) or relative to at least one part of the tested object (10); – measuring (S3), during a test cycle (C L ) having a predetermined duration, a physical quantity (R) linked to the leak level (F U) in said at least one part of the object (10) under test or in an enclosure surrounding the object (10) to be tested; – comparing (S4) the measured value of the physical quantity (R) with a first reference extreme value (Rmax, Rmin); – determining (S5) whether the object (10) has a leak based on the result of the comparison of the measured value with the first reference extreme value (Rmax, Rmin); – interrupting (S6) the test cycle (CL) if it is determined that the object (10) under test has a leak.[Claims 2] Method (100; 100'; 100'') according to the preceding claim, characterized in that said method (100; 100'; 100'') comprises the following steps: – comparing (S4) the measured value of the physical quantity (R) with a second reference extreme value (Rmax, Rmin); – determining (S5) whether the object (10) does not have a leak based on the result of the comparison (S4) of the measured value (R) with the second reference extreme value (Rmax, Rmin); – interrupting (S6) the test cycle (CL) if it is determined that the tested object (10) does not have a leak. [Claims 3] Method (100; 100'; 100'') according to claim 1 and 2, characterized in that the first and second reference extreme values (R. max , R min ) define an interval (I R ) of reference values, the comparison of the measured value (R) with said reference interval (IR) making it possible to determine, as soon as the measured value (R) is located outside said reference interval (IR), whether or not the object (10) has a leak, as long as the measured value (R) is between the limits of the reference interval (I R ), the leak test continues until the test cycle (CL) is completed. [Claims 4] Method (100; 100'; 100'') according to any one of the preceding claims, characterized in that the first reference extreme value RMAX is defined by the following formula: RMAX = RL + MC + k1 ID; where R L is a leakage level, M C is a measure of central tendency of a compensation value, I Dis a dispersion indicator and k1 is a positive real constant. [Claims 5] Method (100; 100'; 100'') according to claim 2 or 3, characterized in that the second reference extreme value RMIN is defined by the following formula: RMIN = RL + MC – k2 ID; where RL is a leakage threshold, MC is a measure of central tendency of a compensation value, I D is a dispersion indicator and k2 is a positive real constant. [Claims 6] Method (100; 100'; 100'') according to claim 4 and / or 5, characterized in that the value of the central tendency measure (M C ) and the dispersion indicator (ID) are determined from several tests of objects of the same type as the object (10) to be tested. [Claims 7] Method (100; 100'; 100'') according to the preceding claim, characterized in that the central tendency measurement (M C ) is an arithmetic mean or median, while the dispersion indicator (ID ) is respectively a standard deviation or an interquartile range. [Claims 8] Method (100'; 100'') according to any one of the preceding claims, characterized in that if the measured value of the physical quantity (Rini) at the start of the test (Ci) is substantially equal to one of the last measured values (R fin ) during the previous test cycle (C 1-1 ), then the comparison of the measured values to one or more values extremes is interrupted and the test cycle (C L ) is completed, the leak level being determined and compared to a predetermined threshold value (RL). [Claims 9] Method (100'') according to any one of the preceding claims, characterized in that said method (100'') comprises the following steps: – calculation of a function (F) on the basis of two distinct measured values of the physical quantity (R i , R i-1); – comparison of said calculated function F(Ri, Ri-1) with a first reference extremal function (FMAX, FMIN); depending on the result of the comparison of the calculated function (F(R i , R i-1 )) with said first reference extremal function (F MAX , F MIN ), either the anticipated leak detection method (100'') continues to apply, or the test cycle (C L ) is carried out to its conclusion. [Claims 10] Method (100; 100'; 100'') according to the preceding claim, characterized in that there is a comparison of the calculated function (F(Ri, Ri-1)) with a second reference extremal function (FMAX, FMIN), depending on the result of the comparison of the calculated function (F(R i , R i-1 )) with said second reference extremal function (FMAX, FMIN), either the anticipated leak detection method (100'') continues to apply, or the test cycle (C L) is carried out to its conclusion. [Claims 11] Method (100'') according to claim 10, characterized in that: – said first reference extremal function is calculated according to the following formula: F MAX = F m + k3I F ; where Fm is a measure of central tendency of the calculated function, IF is a dispersion indicator relative to said calculated function, k3 is a positive real constant. [Claims 12] Method (100'') according to the preceding claim, characterized in that said second reference extremal function is calculated according to the following formula: F MIN = F m – k4I F ; where Fm is a measure of central tendency of the calculated function, IF is a dispersion indicator relating to said calculated function, k4 is a positive real constant. [Claims 13] Method (100'') according to any one of claims 10 to 13, characterized the calculated function (F(Ri, Ri-1)) is a difference and / or a ratio between two measured values of the physical quantity (R). [Claims 14] Leak detection device (1) by pressure difference configured to implement the early leak detection method (100; 100'; 100'') according to any one of the preceding claims.

Images