EP4182654A1 - Procede de caracterisation de fuite - Google Patents
Procede de caracterisation de fuiteInfo
- Publication number
- EP4182654A1 EP4182654A1 EP21754814.8A EP21754814A EP4182654A1 EP 4182654 A1 EP4182654 A1 EP 4182654A1 EP 21754814 A EP21754814 A EP 21754814A EP 4182654 A1 EP4182654 A1 EP 4182654A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- vibro
- leak
- acoustic
- fluid network
- sensors
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M3/00—Investigating fluid-tightness of structures
- G01M3/02—Investigating fluid-tightness of structures by using fluid or vacuum
- G01M3/04—Investigating fluid-tightness of structures by using fluid or vacuum by detecting the presence of fluid at the leakage point
- G01M3/24—Investigating fluid-tightness of structures by using fluid or vacuum by detecting the presence of fluid at the leakage point using infrasonic, sonic, or ultrasonic vibrations
- G01M3/243—Investigating fluid-tightness of structures by using fluid or vacuum by detecting the presence of fluid at the leakage point using infrasonic, sonic, or ultrasonic vibrations for pipes
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M3/00—Investigating fluid-tightness of structures
- G01M3/02—Investigating fluid-tightness of structures by using fluid or vacuum
- G01M3/26—Investigating fluid-tightness of structures by using fluid or vacuum by measuring rate of loss or gain of fluid, e.g. by pressure-responsive devices, by flow detectors
- G01M3/28—Investigating fluid-tightness of structures by using fluid or vacuum by measuring rate of loss or gain of fluid, e.g. by pressure-responsive devices, by flow detectors for pipes, cables or tubes; for pipe joints or seals; for valves ; for welds
- G01M3/2807—Investigating fluid-tightness of structures by using fluid or vacuum by measuring rate of loss or gain of fluid, e.g. by pressure-responsive devices, by flow detectors for pipes, cables or tubes; for pipe joints or seals; for valves ; for welds for pipes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Definitions
- This presentation relates to a method for characterizing a leak in a fluid network, making it possible to characterize the seriousness of a leak in a fluid network by determining its type and/or its flow rate.
- Such a method can in particular be used to characterize leaks within a water distribution network. However, it could also be used for gas, fuel or any other type of fluid, liquid or gaseous networks. Such a method can also be applied to different sizes of networks.
- the vibro-acoustic listening methods aim to listen locally, using a microphone for example, the signals emitted by the leaks within the pipes.
- a technique is quite effective but it requires a large number of listening points, i.e. a large number of sensors or, in the case of a mobile configuration, a full-time expert operator moving the along the network, in order to cover the entire network.
- they are highly subject to acoustic disturbances from the environment of pipelines, for example road traffic.
- these methods make it possible to locate the leaks but they do not make it possible to characterize them.
- the sectorization methods aim to sectorize the network into small isolated areas and to compare the inlet and outlet flows of each zone in order to detect the presence of a leak flow.
- a method is not sufficient on its own since it does not make it possible to locate the location of the leak precisely enough, in particular on a mesh network which would require too many sensors.
- the sectorization does not make it possible to discriminate the seriousness of each leak when several leaks are present in the same sector. In any case, when an estimate of the severity is possible, this estimate can only be obtained a posteriori, after the repair of the leak, which prevents any prioritized maintenance.
- This presentation relates to a method for training a statistical learning model intended for the characterization of leaks in a fluid network, in which the fluid network is equipped with a plurality of vibro-acoustic sensors configured to provide vibro-acoustic signals, comprising the construction of a database associating, at least for one plurality of documented leaks, at least one leak characterization datum actually determined from among the type of leak and the leak rate with at least one vibro-acoustic signal obtained directly or indirectly from at least one vibro-acoustic sensor, and comprising training the statistical learning model on the database thus constructed.
- the term "vibro-acoustic sensor” means a sensor coupled to any type of liquid or solid medium and capable of recording a displacement, a speed, an acceleration or even a time derivative of higher order in one or more directions, and in particular in the three directions of space. It may therefore in particular be an accelerometer, a seismometer, a geophone, a microphone, or even a hydrophone, to cite only these examples. These may be permanently mounted sensors and/or mobile sensors temporarily applied by an operator.
- the fluid network is provided with digital mapping comprising at least the geometry of the fluid network and the location of said vibro-acoustic sensors.
- This digital cartography can conform to the real cartography or even include simulated cartographic elements. These simulated cartographic elements make it possible to complete the digital cartography when certain elements of the real cartography are unknown. These simulated cartographic elements can also make it possible to simulate alternative scenarios by substituting a real cartographic element by a simulated cartographic element, for example by virtually modifying the diameter of a pipeline, or by simulating fictitious pipelines in the extension of the existing network.
- the training method comprises, for at least one documented leak, a step of measuring the leak rate.
- this leak rate can be obtained by a direct measurement at the level of the leak. Such a measurement can in particular be carried out by an operator just before repairing the leak.
- the fluid network is equipped with at least one flow sensor providing sectorization data.
- These sectorization data can also comprise pressure measurements taken by at least one pressure sensor. The location of these sectorization sensors can be recorded in the digital map of the network.
- the training method comprises, for at least one documented leak, a step of determining the leak rate using the sectorization data. For some leaks, depending on their number, their location and the number of nearby flowmeters, this determination can be made directly, while the leak exists. For other leaks, if this direct determination is not possible, the determination can be made indirectly, for example by comparing the sectorization data before and after the repair of the leak. Such a determination step renders the direct measurement of the leak rate by an operator at the level of the leak superfluous. Consequently, it is possible to include minor leaks in the database for which a repair is not considered a priority.
- the determination step is carried out automatically by comparing the sectorization data before and after the repair of the leak in question.
- the database comprises, for at least one documented leak, vibro-acoustic signals recorded by at least certain vibro-acoustic sensors of the fluid network.
- each signal is recorded in the database in association with the identification and/or location of the vibro-acoustic sensor at the origin of the signal.
- a documented leak a plurality of signals are thus associated in the database, which increases the size of the database and therefore reinforces the training of the statistical learning model.
- signals that have undergone attenuation and/or alterations during their propagation along the fluid network are associated, which helps the statistical learning model to characterize a leak even in the event of attenuation and/or alteration. alteration of vibro-acoustic signals. Indeed, any physical element constituting a pipe network can have an impact on wave propagation and thus cause signal alterations.
- the training method comprises, for at least one documented leak, a step of simulating at least one virtual vibro-acoustic sensor having a virtual location recorded in the digital mapping of the fluid network and a vibro-acoustic signal simulated from the vibro-acoustic signals actually measured by real vibro-acoustic sensors and geometric data from the digital mapping of the fluid network. Thanks to such a simulation step, one can artificially increase the number of signals available for training the statistical learning model while keeping a reasonable number of real sensors within the fluid network. This reinforces the training of the statistical learning model without the additional cost of additional equipment.
- Such a simulated vibro-acoustic signal can in particular be obtained by digital simulation. In particular, such a simulation can be based on the use of transfer functions, associated with each element of the fluid network.
- the step of simulating at least one virtual vibro-acoustic sensor is based on the real mapping of the network.
- the step of simulating at least one virtual vibro-acoustic sensor is based on a simulated cartography comprising at least one simulated cartographic element.
- this simulated mapping may comprise at least one fictitious pipeline.
- the number of virtual vibro-acoustic sensors is at least 2 times greater than the number of real vibro-acoustic sensors.
- the training method comprises a step of locating the leak based on the vibro-acoustic signals from the vibro-acoustic sensors and geometric data from the digital mapping of the fluid network. We thus obtain information concerning the location of the leak, which makes it possible, if necessary, to send a maintenance agent directly to the right address in order to characterize and/or repair the leak.
- the database comprises, for at least one documented leak, the vibro-acoustic signal actually recorded near the leak.
- the vibro-acoustic signal recorded with the leak that is to say before any attenuation or any alteration due to its propagation along the fluid network: such a signal thus constitutes a signature primary of the leak, particularly useful for training the statistical learning model.
- This leak signal can in particular be recorded by a maintenance agent just before repairing the leak. Depending on the accessibility of the leak, this signal will be measured within 50 m of the leak, preferably within 10 m of the leak, preferably within 1 m of the leak.
- the training method comprises, for at least one documented leak, a step of reconstructing the vibro-acoustic signal at the level of the leak from the vibro-acoustic signals of the vibro-acoustic sensors and geometric data from the digital mapping of the fluid network.
- This is another method for obtaining an approximation of the vibro-acoustic signal at the leak, especially when it is not possible, or desirable, to record the true signal directly at the leak.
- Such a signal thus constitutes a primary signature of the leak, particularly useful for training the model statistical learning.
- Such a signal reconstructed at the leak can in particular be obtained by simulating a virtual sensor at or near the leak.
- the training method comprises at least one leak generation step during which a leak is artificially caused within the fluid network.
- a leak is artificially caused within the fluid network.
- the training method comprises, for at least one documented leak, a noise step during which noise is added to at least one vibro-acoustic signal actually measured.
- a noise step during which noise is added to at least one vibro-acoustic signal actually measured.
- the added noise may in particular include conventional colored noises, white or pink for example, and/or noises specific to water networks, such as the noise of a car passing by, a mechanical meter rotating or even people talking nearby, to name but a few examples.
- the database includes, for at least one documented leak, structural data of the pipe at the level of the leak. These structural data may in particular include the material of the pipe, its nominal diameter, its thickness, its depth or even the material of the surrounding soil or the type of backfill. This data is useful to help the statistical learning model take into account the signal variations that may appear for a leak of a given type and flow rate depending on the physical properties of the pipe carrying the leak. In this way, the statistical learning model will have an easier time characterizing the leaks if these additional contextual variables are given to it. [0030] In certain embodiments, the database comprises, for at least one documented leak, contextual repair data among the type of backfill used, the flooding state around the leak and/or a photograph of the the escape. These data can in particular be used for training the statistical learning model.
- the database includes data from documented leaks from a single fluid network.
- the statistical learning model is then specialized in the characterization of the leaks of this particular fluid network; it nevertheless remains capable of characterizing leaks from other fluid networks if necessary, with lower precision.
- the database includes data from documented leaks from multiple fluid networks. This increases the size of the database, which strengthens the training of the statistical learning model. Greater weight may however be given to documented leaks from a particular fluid network when the statistical learning model is intended to be applied to that particular fluid network.
- At least certain vibro-acoustic sensors are correlating sensors directly or indirectly sharing a common clock. Such sensors make it possible to follow the propagation of a sound wave, or a vibration signal, along the fluid network. Thus, thanks to such sensors it is possible to estimate the distance separating the sensor from the leak and therefore, using several sensors, to determine the position of the leak.
- At least one vibro-acoustic sensor is provided at least every 200 m, preferably at least every 100 m, more preferably at least every 50 m, along the fluid network .
- the finer the mesh the greater the probability of having a large number of sensors close to the leak, and therefore of having a large number of usable signals. Consequently, the finer the mesh, the more different signals the database contains for a given leak and therefore the more efficient the training of the statistical learning model.
- the distance between sensors can be modulated according to the type of material making up the pipe; in particular a finer mesh is preferable for plastic networks.
- At least one sensor is installed at a keyhole of the fluid network.
- the sensor may in particular comprise a measuring head located at the bottom of the extension tube, in contact with the pipe.
- the sensors can be mounted on or within any element of the network.
- the training method includes a normalization step resulting in converting the raw vibro-acoustic signal from at least one vibro-acoustic sensor into a standardized vibro-acoustic signal having a predetermined format.
- a normalization step resulting in converting the raw vibro-acoustic signal from at least one vibro-acoustic sensor into a standardized vibro-acoustic signal having a predetermined format.
- This normalization step can use transfer functions, determined theoretically or empirically, for each type or model of sensor used. This step can also artificially introduce a loss of quality with respect to the raw vibro-acoustic signal.
- This normalization of the signal can be, in particular, a compression, a spectral folding, or other signal processing techniques.
- the predetermined common format of the standardized vibro-acoustic signals is a sound format, for example of the WAVE type.
- the training method includes a pre-processing step resulting in qualifying and cleaning the vibro-acoustic signal from at least one vibro-acoustic sensor. This step makes it possible in particular to check whether the signal is not corrupted or polluted by noise. covering the signal, such as a passing vehicle for example, and therefore whether the signal can be used by the statistical learning model. It also makes it possible to clean the signal of any interference using one or more filters.
- the training method includes a step of transforming the representation of the vibro-acoustic signals into other mathematical spaces.
- the training method comprises a decomposition step resulting in decomposing the vibro-acoustic signal from at least one vibro-acoustic sensor.
- This may in particular be a Fourier decomposition.
- the result of this decomposition can be saved in the database and can be used for training the statistical learning model.
- the digital mapping of the fluid network includes location data for valves and/or other equipment. In this way, it is possible to predict and take into account the alterations of the vibro-acoustic signals during their propagation through this equipment. This increases the accuracy of the simulations and/or signal reconstructions.
- a transfer function can be associated with each type of equipment.
- the digital map of the fluid network includes structural data of the fluid network.
- These structural data may in particular include the material of each pipe, its nominal diameter, its thickness, its depth or even the material of the surrounding soil or the type of backfill. This data is useful for increasing the accuracy of simulations and/or reconstructions of signals within the fluid network. They can also be used to help the statistical learning model to take into account the signal variations that may appear for a leak of a given type and flow rate depending on the physical properties of the pipe carrying a given vibro-acoustic sensor. . In this way, the statistical learning model will have an easier time characterizing the leaks if these additional contextual variables are given to it.
- the type of leak is determined at least from three different types, preferably at least from five different types.
- the type of leak can be determined from among the following types: pipe leak, pipe break, pipe branch leak, connection leak, collar leak, fire hydrant leak, meter leak, joint leak, leak on the valve, leak on the suction cup, leak on the flange, leak on the stuffing box, leak on another nearby network (eg sanitation, gas, etc.), to name but a few examples.
- the database also includes data, in particular vibro-acoustic signals, corresponding to leak-free scenarios. Thanks to such leak-free scenarios, we introduce the possibility for the statistical learning model, once trained, to conclude that there is no leakage when no leakage is present in the network.
- the statistical learning model is of the classifier type.
- the leak flow rate is determined from among predetermined flow rate ranges. These ranges can have constant or variable widths. Preferably, the width of each range is less than or equal to 10 m3/h, more preferably less than or equal to 5 m3/h.
- the statistical learning model is of the regressor type.
- the leak rate is determined as accurately as possible, with a certain margin of error.
- this margin of error is less than or equal to 10% or even less than or equal to 10 m3/h or 5 m3/h.
- the fluid network is a water network. It is preferably a water distribution network, drinking water, in charge.
- the statistical learning model is a neural network.
- the neural network is of the convolutional type with preferably at least two convolutional layers of temporal filter applied to the vibration signals and at least two layers not convolutional.
- Non-convolutional layers can be applied to results from convolutional layers and/or contextual data.
- the convolutional layers of the neural network contain temporal filters of size between 25 and 100 ms applied to the vibration signals. Convolutional layers are organized such that the number of filters increases with each layer. The simple layers are organized in such a way that the number of neurons can decrease until the final estimate. To avoid overlearning, the dropout technique can be used between 30 and 70%. To obtain an estimate of uncertainty, a Bayesian network can be used.
- the statistical learning model is a forest of decision trees, a vector support machine, or else a nonlinear regression.
- This presentation also relates to a method for characterizing leaks in a fluid network, in which the fluid network is equipped with a plurality of vibro-acoustic sensors configured to supply vibro-acoustic signals, and in which a statistical learning model receives as input at least one vibro-acoustic signal obtained directly or indirectly from at least one vibro-acoustic sensor and provides as output at least one leak characterization datum among the type of leak and the leak rate.
- the statistical learning model receives as input the vibro-acoustic signals from several vibro-acoustic sensors. acoustics.
- the statistical learning model can receive the signals from all the vibro-acoustic sensors of the fluid network, or from only some of them, for example the sensors closest to the leak or those providing a signal with a level and/or quality above a certain threshold.
- the fluid network is provided with digital mapping comprising at least the geometry of the fluid network and the location of said vibro-acoustic sensors.
- This digital cartography can conform to the real cartography or even include simulated cartographic elements. These simulated cartographic elements make it possible to complete the digital cartography when certain elements of the real cartography are unknown. These simulated cartographic elements can also make it possible to simulate alternative scenarios by substituting a real cartographic element with a simulated cartographic element, for example by virtually modifying the diameter of a pipe, or even by simulating fictitious pipes in the extension of the existing network.
- the characterization method comprises a step of simulating at least one virtual vibro-acoustic sensor having a virtual location recorded in the digital mapping of the fluid network and a simulated vibro-acoustic signal from vibro-acoustic signals from the real vibro-acoustic sensors and geometric data from the digital mapping of the fluid network.
- a simulated vibro-acoustic signal can in particular be obtained by digital simulation of acoustic propagation.
- such a resolution can be based on the use of transfer functions, determined for example using laboratory tests or in real conditions, associated with each element of the fluid network.
- the neural network receives at least one simulated vibro-acoustic signal as input.
- the step of simulating at least one virtual vibro-acoustic sensor is based on the actual mapping of the network. In certain embodiments, the step of simulating at least one virtual vibro-acoustic sensor is based on a simulated cartography comprising at least one simulated cartographic element. In particular, this simulated mapping may comprise at least one fictitious pipeline. In some embodiments, the number of virtual vibro-acoustic sensors is at least twice the number of real vibro-acoustic sensors.
- the characterization method comprises a step of locating the leak from the vibro-acoustic signals from the vibro-acoustic sensors and geometric data from the digital mapping of the fluid network. This provides information about the location of the leak, which makes it possible, if necessary, to send a maintenance agent directly to the right address in order to repair the leak. The location of the leak can also be taken into account in the evaluation of the priority of repairing this leak.
- the characterization method comprises a step of reconstructing the vibro-acoustic signal at the level of the leak from the vibro-acoustic signals of the vibro-acoustic sensors and the geometric data of the digital mapping of the network of fluid.
- An approximation of the vibro-acoustic signal generated at the level of the leak is thus obtained, that is to say before any attenuation or any alteration due to its propagation along the fluid network: such a signal, constituting a primary signature of the leak, brings a significant amount of information to the statistical learning model, which facilitates characterization.
- Such a signal at the source can in particular be obtained by digital simulation of the acoustic propagation.
- such a resolution can be based on the use of transfer functions associated with each element of the fluid network.
- a signal reconstructed at the leak can in particular be obtained by simulating a virtual sensor at or near the leak.
- the statistical learning model receives as input at least the vibro-acoustic signal reconstructed at the level of the leak.
- at least some vibro-acoustic sensors are correlating sensors directly or indirectly sharing a common clock. Such sensors make it possible to follow the propagation of a sound wave, or of a vibration signal, along the fluid network. Thus, thanks to such sensors it is possible to estimate the distance separating the sensor from the leak and therefore, using several sensors, to determine the position of the leak.
- At least one vibro-acoustic sensor is provided at least every 200 m, preferably at least every 100 m, more preferably at least every 50 m, along the fluid network .
- the finer the mesh the greater the probability of having a large number of sensors near the leak, and therefore of having a large number of usable signals. Consequently, the finer the mesh, the more different signals the database contains for a given leak and therefore the more efficient the training of the statistical learning model.
- the distance between sensors can be modulated according to the type of material making up the pipe; in particular a finer mesh is preferable for plastic networks.
- At least one sensor is installed at a keyhole of the fluid network.
- the sensor may in particular comprise a measuring head located at the bottom of the extension tube, in contact with the pipe.
- the sensors can be mounted on or within any element of the network.
- the characterization method comprises a normalization step resulting in converting the raw vibro-acoustic signal from at least one vibro-acoustic sensor into a normalized vibro-acoustic signal having a predetermined format.
- a normalization step resulting in converting the raw vibro-acoustic signal from at least one vibro-acoustic sensor into a normalized vibro-acoustic signal having a predetermined format.
- the predetermined common format of the standardized vibro-acoustic signals is a sound format, for example of the WAVE type.
- the characterization method includes a pre-processing step resulting in qualifying and cleaning the vibro-acoustic signal from at least one vibro-acoustic sensor. This step makes it possible in particular to check whether the signal is not corrupted or polluted by a noise covering the signal, such as a passing vehicle for example, and therefore whether the signal can be exploited by the statistical learning model. It also cleans the signal of any interference using one or more filters.
- the characterization method comprises a step of transforming the representation of the vibro-acoustic signals into other mathematical spaces.
- the characterization method comprises a decomposition step resulting in decomposing the vibro-acoustic signal from at least one vibro-acoustic sensor.
- This may in particular be a Fourier decomposition.
- the result of this decomposition can be provided as input to the statistical learning model, instead of or in addition to the original signal. This can facilitate the characterization by the statistical learning model.
- the digital mapping of the fluid network includes location data for valves and/or other equipment. In this way, it is possible to predict and take into account the alterations of the vibro-acoustic signals during their propagation through this equipment. This increases the accuracy of the simulations and/or signal reconstructions.
- a transfer function can be associated with each type of equipment.
- the digital map of the fluid network includes structural data of the fluid network.
- These structural data may include, but are not limited to, the material of the pipe, its nominal diameter, its thickness, its depth or even the surrounding material. This data is useful to help the statistical learning model take into account the signal variations that may appear for a leak of a given type and flow rate depending on the physical properties of the pipe carrying the leak. In this way, the statistical learning model has an easier time characterizing leaks, regardless of the size or material of the pipes involved.
- the type of leak is determined at least from 3 different types, preferably at least from 5 different types.
- the type of leak can be determined from among the following types: pipe leak, pipe break, pipe branch leak, connection leak, collar leak, fire hydrant leak, meter leak, joint leak, leak on the valve, leak on the suction cup, leak on the flange, leak on the stuffing box, leak on another nearby network (eg sanitation, gas, etc.), to name but a few examples.
- the statistical learning model is of the classifier type.
- the leak flow rate is determined from among predetermined flow rate ranges. These ranges can have constant or variable widths. Preferably, the width of each range is less than or equal to 10 m3/h, more preferably less than or equal to 5 m3/h.
- the statistical learning model is of the regressor type.
- the leak rate is determined as accurately as possible, with a certain margin of error.
- this margin of error is less than or equal to 10% or even less than or equal to 10 m3/h or 5 m3/h.
- the fluid network is a water network. It is preferably a water distribution network, drinking water, in charge.
- the neural network is of the convolutional type with preferably at least two convolutional layers of temporal filter applied to the vibration signals and at least two non-convolutional layers.
- Non-convolutional layers can be applied to results from convolutional layers and/or contextual data.
- the convolutional layers of the neural network contain temporal filters of size between 25 and 100 ms applied to the vibration signals. Convolutional layers are organized such that the number of filters increases with each layer. The simple layers are organized in such a way that the number of neurons can decrease until the final estimate.
- the dropout technique can be used between 30 and 70%.
- a Bayesian network can be used.
- the statistical learning model is a forest of decision trees, a vector support machine, or else a nonlinear regression.
- the statistical learning model has been trained using a training method according to any of the embodiments described above.
- the characterization method comprises a verification step during which the type and/or the flow rate of the leak characterized by the statistical learning model is actually verified and then recorded in the database. to complete the training of the statistical learning model.
- This presentation also relates to a module for characterizing leaks in a fluid network, the fluid network being equipped with a plurality of vibro-acoustic sensors configured to supply vibro-acoustic signals comprising a statistical learning model configured to receive as input at least one vibro-acoustic signal obtained directly or indirectly from at least one vibro-acoustic sensor and to supply as output at least one leak characterization data from among the type of leak and the flow rate leak.
- this characterization module stems from the advantages described above for the characterization method. Moreover, this characterization module can present all or part of the additional characterizations described above with regard to the training method and/or the characterization method. This presentation also relates to a fluid network, comprising a plurality of vibro-acoustic sensors configured to supply vibro-acoustic signals, and a characterization module according to any one of the preceding embodiments.
- This presentation also relates to a computer program comprising instructions for executing the steps of the training method or of the characterization method described above when the program is executed by a computer.
- Figure 1 is an overall diagram of a fluid network equipped with a leak characterization module.
- Figure 2 is an overview diagram of a leak characterization module.
- Figure 3 illustrates the detection of a leak within the fluid network of Figure 1.
- Figure 4 illustrates obtaining the vibro-acoustic signal at the leak.
- Figure 5 illustrates a first example of training a neural network.
- Figure 6 illustrates a first example of leak characterization using this neural network.
- Figure 7 illustrates the simulation of virtual vibro-acoustic sensors.
- Figure 8 illustrates a second example of training a neural network.
- Figure 9 illustrates a second example of leak characterization using this neural network.
- Figure 1 shows a fluid network diagram 1, in this case a drinking water distribution network.
- This fluid network 1 has a plurality of pipes 2 on which are mounted a plurality of vibro-acoustic sensors 3, here acoustic sensors of the accelerometer type.
- the fluid network also has a leak characterization module 10 which can be hosted within a computer of the control of the fluid network 1 or within a remote server.
- the fluid network 1 further comprises sectorization sensors, and in particular flow meters and pressure gauges.
- FIG. 2 illustrates the main elements of this leak characterization module 10. It thus comprises a digital map 11 of the fluid network 1, a leak database 12, a neural network 13 (forming a model of statistical learning), a calculation unit 14 and a memory 15; it also includes all the electronic elements making it possible to operate such an electronic module: power supply, user interfaces, etc.
- the digital map 11 includes the geometry of the fluid network 1, that is to say the position, orientation and length of all the pipes 2, as well as the position of all the network equipment, c i.e. valves, junction collars, connections, key boxes etc.
- the digital map 11 also includes the location of all the vibro-acoustic sensors 3 but also of the sectorization sensors.
- the digital map 11 also includes the most complete structural data possible for the entire network 1, and in particular, as far as possible, the material of each pipe, its nominal diameter, its thickness, its depth or even the material of the surrounding ground.
- the database 12 for its part compiles as much data as possible concerning the leaks identified and characterized in the past within the fluid network 1. Its construction will be described in more detail below.
- the neural network 13 is a convolutional network of the regressor type comprising two convolutional layers of temporal filters and two non-convolutional layers.
- the layers of the neural network 13 contain temporal filters of size between 25 and 100 ms.
- Convolutional layers are organized such that the number of filters increases with each layer.
- the single layers are organized in such a way that the number of neurons decreases until the final estimate.
- the dropout technique is used between 30 and 70%.
- a Bayesian network is used to obtain an uncertainty estimate.
- the calculation unit 14 can in particular take the form of a processor: it is in particular programmed to be able to solve digital problems of propagation of a sound wave along the fluid network 1, on the basis of the geometric and structural data from digital cartography 11.
- the memory 15 can take any form of data storage. It includes in particular the theoretical equations for the propagation of sound waves along a pipeline. It also includes a library of transfer functions, established theoretically or empirically using tests in the laboratory or on site, making it possible to simulate the deformation undergone by a vibro-acoustic signal during its passage through a particular piece of equipment of the fluid network 1 , in particular a valve, an elbow, a connection, or even a collar . This library also includes transfer functions making it possible to convert the raw signal from a vibro-acoustic sensor of a given type and model into a common reference format of the sound wave type, for example in the form of a WAVE type sound file.
- any leak generates a characteristic noise which propagates along the pipes 2 and which can therefore be detected and recorded by vibro-acoustic sensors 3 such as microphones, geophones, hydrophones or accelerometers.
- vibro-acoustic sensors 3 such as microphones, geophones, hydrophones or accelerometers.
- the vibro-acoustic sensors 3 of the network 1 each record a signal 21 revealing the presence of the leak 20.
- each sensor 3 records a slightly different signal 21: in particular, the level of the signal is all the more attenuated the further the sensor 3 is from the leak 20; in addition, the shape of the signal may also be altered during propagation, in particular when passing through certain equipment on network 1.
- the leak 20 Once the leak 20 has been located in this way, it is possible to go to the site to excavate and repair it. On this occasion, it is also possible to determine its type Tf, i.e. to determine whether it is a leak caused by a crack, a puncture or even a defective seal, for example at the level of a collar or branch. Once the leak 20 has been repaired in this way, the calculation unit 14 is capable of automatically determining the flow rate Qf that this leak 20 exhibited by comparing the sectorization data before and after the repair.
- the training of the neural network 13 is then represented in FIG. 5.
- all the data relating to this leak 20 is recorded in the database 12: in In particular, the characterization, including the type of leak Tf and the leak rate Qf, is recorded in association with the vibro-acoustic signal at leak 22.
- Structural data of the pipe 2 carrying the leak 20 are also recorded in the database 12: these structural data include the material of the pipe, its nominal diameter, its thickness, its depth as well as the material of the surrounding soil.
- Contextual repair data such as the type of backfill used, the state of flooding around the leak or even a photograph of the leak, can also be recorded in the database 12.
- the neural network 13 is applied to the database 12 in order to carry out its initial training. Once the initial training is complete, the neural network13 can then be used to automatically characterize new leaks20.
- the leak characterization module 10 permanently receives the signals 21 recorded by the vibro-acoustic sensors 3.
- the fluid network 1 can comprise different types or models of vibro-acoustic sensors 3, all the signals 21 thus recorded are converted, during a normalization step, into a common format using the transfer functions stored in memory 15.
- each signal 21 undergoes a qualification step during which it is verified that the signal 21 is not corrupted and has not been rendered unusable by excessive parasitic noise such as the passage of a vehicle for example.
- the signals 21 thus qualified then undergo a cleaning step during which they are filtered in order to remove most of the interference.
- the leak characterization module 10 detects the appearance of a signal representative of a leak in the vibro-acoustic signal 21 of one or several vibro-acoustic sensors 3. The characterization module 10 then carries out the location of the leak 20 then the reconstruction of the vibro-acoustic signal at the leak 22 as described above.
- the vibro-acoustic signal at the leak 22 is then transmitted as input to the neural network 13: thanks to its training, the neural network 13 is then capable of outputting the characterization of the leak 20, that is to say its type Tf and/or its flow rate Qf.
- the structural data of the pipe 2 carrying the leak 20, resulting from the digital map 11, can also be transmitted as input to the neural network 13 in order to facilitate the characterization and, if necessary, increase the precision of the latter. .
- FIGS. 7 to 9 illustrate another method making it possible to further increase the ease and precision of characterization.
- the fluid network 101 comprises the same vibro-acoustic sensors 103 as in the first example. However, in addition to these real sensors 103, the fluid network 101 now also includes virtual vibro-acoustic sensors 104.
- These virtual vibro-acoustic sensors 104 are positioned in the digital map 11 so as to reduce the distance separating two sensors, real or virtual.
- two virtual sensors 104 can be simulated between two real sensors 103 consecutive.
- the calculation unit 14 is then capable, for each virtual sensor 104, of simulating the vibro-acoustic signal 123 which would actually be recorded if a real sensor were provided at this location.
- This simulation is possible from the vibro-acoustic signals 121 of the real sensors 103 provided near the virtual sensor 104 considered, by solving the digital simulation of acoustic propagation using propagation equations and transfer functions stored in the memory. 15 of characterization module 10.
- each signal 121, 122, 123 recorded in the database can undergo a noise stage during which noise is added to the signal 121, 122, 123 .
- the neural network 113 is able to characterize the new leak 120 more easily and with greater precision, even by supplying it as input only with the vibro signals. -real acoustics 121 for example.
Landscapes
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Examining Or Testing Airtightness (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
FR2007434A FR3112610B1 (fr) | 2020-07-15 | 2020-07-15 | Procédé de caractérisation de fuite |
PCT/FR2021/051318 WO2022013502A1 (fr) | 2020-07-15 | 2021-07-15 | Procede de caracterisation de fuite |
Publications (1)
Publication Number | Publication Date |
---|---|
EP4182654A1 true EP4182654A1 (fr) | 2023-05-24 |
Family
ID=74045501
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP21754814.8A Pending EP4182654A1 (fr) | 2020-07-15 | 2021-07-15 | Procede de caracterisation de fuite |
Country Status (4)
Country | Link |
---|---|
US (1) | US20230184620A1 (fr) |
EP (1) | EP4182654A1 (fr) |
FR (1) | FR3112610B1 (fr) |
WO (1) | WO2022013502A1 (fr) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20230152480A1 (en) * | 2021-11-16 | 2023-05-18 | Landmark Graphics Corporation | Random noise attenuation for seismic data |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6389881B1 (en) * | 1999-05-27 | 2002-05-21 | Acoustic Systems, Inc. | Method and apparatus for pattern match filtering for real time acoustic pipeline leak detection and location |
KR100420717B1 (ko) * | 2001-05-03 | 2004-03-02 | (주)동명기술공단종합건축사사무소 | 상수관의 누수 탐지 방법 및 그 시스템 |
WO2014170673A1 (fr) * | 2013-04-19 | 2014-10-23 | Acoustic Sensing Technology (Uk) Ltd | Système d'inspection de tuyauterie et procédés associés |
WO2018049149A1 (fr) * | 2016-09-08 | 2018-03-15 | General Electric Company | Système de surveillance de pipeline |
-
2020
- 2020-07-15 FR FR2007434A patent/FR3112610B1/fr active Active
-
2021
- 2021-07-15 WO PCT/FR2021/051318 patent/WO2022013502A1/fr unknown
- 2021-07-15 US US18/012,692 patent/US20230184620A1/en active Pending
- 2021-07-15 EP EP21754814.8A patent/EP4182654A1/fr active Pending
Also Published As
Publication number | Publication date |
---|---|
WO2022013502A1 (fr) | 2022-01-20 |
US20230184620A1 (en) | 2023-06-15 |
FR3112610A1 (fr) | 2022-01-21 |
FR3112610B1 (fr) | 2022-08-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109357167B (zh) | 一种燃气管道漏点检测装置及检测方法 | |
Ye et al. | Kalman filtering of hydraulic measurements for burst detection in water distribution systems | |
EP1637904A1 (fr) | Méthode pour génerer un modèle de reservoir sur maillage flexible | |
FR2818742A1 (fr) | Methode pour former un module a reseaux neuronaux optimise, destine a simuler le mode d'ecoulement d'une veine de fluides polyphasiques | |
WO2022013502A1 (fr) | Procede de caracterisation de fuite | |
FR3045842A1 (fr) | Procede d'exploitation des hydrocarbures d'un bassin sedimentaire, au moyen d'une simulation de bassin, avec prise en compte des effets geomecaniques | |
JP6966803B2 (ja) | モニタリング対象機器の異常発生予兆検知方法及びシステム | |
FR2848320A1 (fr) | Methode pour modeliser des caracteristiques hydrodynamiques d'ecoulements polyphasiques par reseaux de neurones | |
Pakoksung et al. | The near-field tsunami generated by the 15 January 2022 eruption of the Hunga Tonga-Hunga Ha’apai volcano and its impact on Tongatapu, Tonga | |
EP4217702A1 (fr) | Procédé de caractérisation de fuite | |
JP6834952B2 (ja) | 劣化分析装置、劣化分析方法および劣化分析プログラムおよび記録媒体 | |
FR2970732A1 (fr) | Procede pour mettre a jour un modele de reservoir en temps reel a partir de donnees dynamiques tout en conservant sa coherence avec les observations statiques | |
AU2022234947A1 (en) | Pipeline anomaly detection method and system | |
Yoo et al. | Inertia‐gravity waves revealed in radiosonde data at Jang Bogo Station, Antarctica (74° 37′ S, 164° 13′ E): 2. Potential sources and their relation to inertia‐gravity waves | |
Romano et al. | Burst detection and location in water distribution systems | |
Raynor et al. | Location matters: evaluating Greater Prairie-Chicken (Tympanuchus cupido) boom chorus propagation | |
EP1987376B1 (fr) | Procede et systeme d'identification et d'evaluation de risque de defaillance d'un systeme de confinement geologique | |
EP3356780B1 (fr) | Procede de detection de fuite dans un reseau de fluide | |
Wang | Machine learning model design for IoT-based flooding forecast | |
Work et al. | Nearshore impacts of dredging for beach nourishment | |
Spicer et al. | Coastal windstorms create unsteady, unpredictable storm surges in a fluvial Maine estuary | |
WO2021239646A1 (fr) | Detection de changement de composition physico-chimique d'un liquide | |
Cody | Acoustic Monitoring for Leaks in Water Distribution Networks | |
WO2017220918A1 (fr) | Procede de caracterisation du sous-sol d'une region utilisant des signaux sismiques passifs, et systeme correspondant | |
WO2018020142A1 (fr) | Procede et systeme de detection d'evenement sismique |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
17P | Request for examination filed |
Effective date: 20221213 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
DAX | Request for extension of the european patent (deleted) | ||
RAV | Requested validation state of the european patent: fee paid |
Extension state: MA Effective date: 20230207 |