FR3154186A1 - METHOD FOR STANDARDIZING AN ACOUSTIC SIGNAL - Google Patents

Abstract

A method for characterizing a liquid metal, the method implementing a probe (1), comprising: a measurement transmitter (11) configured to emit an incident ultrasonic wave into the liquid metal; a receiver (21) configured to detect an ultrasonic wave reflected or diffracted by the liquid metal, following the emission of the incident ultrasonic wave; the method comprising normalizing the ultrasonic wave detected by the receiver by a normalizing amplitude. The normalizing amplitude is determined experimentally, from ultrasonic waves detected while the metal is considered to be free of detectable inclusions. Figure 1B.

Description

METHOD FOR STANDARDIZING AN ACOUSTIC SIGNAL

The technical field of the invention is the characterization, by ultrasound, of a casting of liquid metal, for example a casting of an aluminum alloy.

Prior art

In the field of metal casting, it is necessary to control the quality of the liquid metal, in order to detect inclusions. This is particularly the case for aluminum. Some aluminum alloy products, for example, strip for casing applications, have high requirements in terms of inclusion health. These inclusions are generated during the melting, production and casting of these alloys when they are in the liquid phase. The quality of the molten metal is determined by the quantity and size of inclusions present in the metal. These can be oxides, nitrides, carbides, fluorides or borides. Inclusions commonly encountered in molten aluminum are, for example, particles of Al ₂ O ₃ , MgAl ₂ O ₄ , Al ₄ C ₃ , MgO, CaO, CaF ₂ , TiB ₂ , TiVB ₂ , SiO ₂ . Their size ranges from a few microns to a few hundred microns.

It is possible to monitor the inclusion quality of a casting online using ultrasonic measurement techniques. In this type of measurement, an ultrasonic acoustic wave propagates through the liquid metal. In the presence of inclusions, part of the emitted acoustic wave is reflected or diffracted and can be detected. The amplitude of the reflected wave can be used to determine the size of the inclusions in the metal. The ultrasonic technique allows characterization of a portion of the liquid metal, typically 2% when the inspection is carried out in the chute transporting the liquid metal from the furnace to the casting pit.

Probes for carrying out measurements, by ultrasound, on castings, have been described in patents EP3204764B1, EP3204763B1.

Document EP1194772B1 describes a method for calibrating a probe, in which reflectors are placed at the end of rods, the latter being arranged in liquid metal. The reflectors can be of different sizes. It is thus possible to establish a correlation between the size of each reflector and the amplitude of the detected acoustic wave. Such a method needs to be applied for each probe. In addition, it assumes the arrangement of calibrated reflectors in a liquid metal, which can be complex.

Another possibility for estimating the size of inclusions is to use a correlation curve between measurements resulting from an ultrasonic probe and measurements resulting from a method considered to be a reference method. The reference method may, for example, be a “LIMCA” (Liquid Metal Cleanliness Analysis) type analysis method, based on an analysis of the electrical resistance of a sample of liquid metal. It may also be the method described in EP1194772B1, using calibrated reflectors. It is thus possible to establish a correlation function between the size of inclusions, measured by the reference method, and the amplitude of the ultrasonic signal.

A difficulty with using ultrasonic probes is that their response can vary from one probe to another. However, it is difficult to envisage calibrating each probe with a reference method, such as LIMCA or as described in EP1194772B1. Such calibration requires time.

The variability of the response between two probes with identical structures is due in particular to differences that can affect the electronic circuits or the materials used. Faced with identical inclusions, two different probes with identical designs can generate detection signals of different amplitudes, due to the variability of the probes' response.

It is preferable, even necessary, to take such variability into account, for example if one wishes to apply the same correlation curve to measurements resulting from different ultrasonic probes.

The inventors propose a method for taking into account the variability of responses of two different probes. The objective is that two probes, of identical design, generate identical signals, or signals that can be considered as such, in the presence of identical inclusions.

• un émetteur de mesure, configuré pour émettre une onde ultrasonore incidente dans le métal liquide;
• un récepteur, configuré pour détecter une onde ultrasonore réfléchie ou diffractée par une inclusion dans le métal liquide, suite à l’émission de l’onde ultrasonore incidente ;

A first object of the invention is a method for characterizing a liquid metal, the method implementing a probe, comprising:

• a measuring transmitter, configured to emit an incident ultrasonic wave into the liquid metal;
• a receiver, configured to detect an ultrasonic wave reflected or diffracted by an inclusion in the liquid metal, following the emission of the incident ultrasonic wave;

1. disposition de la sonde, de telle sorte que l’émetteur de mesure et le récepteur soient disposés dans le métal liquide ou face au métal liquide ;
2. alimentation de l’émetteur de mesure, de façon que l’émetteur de mesure émette une onde ultrasonore incidente se propageant dans le métal liquide ;
3. détection, par le récepteur, d’une onde ultrasonore réfléchie ou diffractée par une inclusion dans le métal liquide sous l’effet de l’onde incidente ;
4. réitération des étapes b) à c) durant différents instants de mesure, les instants de mesure formant une plage temporelle de mesure, et mesure d’une amplitude de l’onde ultrasonore détectée par le récepteur en chaque instant de mesure ;
5. sélection, dans la plage temporelle de mesure, d’une période de normalisation, la période de normalisation regroupant des instants de mesure consécutifs ;
6. calcul d’une amplitude de normalisation en fonction des amplitudes mesurées aux instants de la période de normalisation ;
7. normalisation des amplitudes mesurées durant la plage temporelle de mesure, durant chaque étape c), par l’amplitude de normalisation calculée lors de l’étape f).

the method comprising the following steps:

1. arrangement of the probe, such that the measuring transmitter and the receiver are arranged in the liquid metal or facing the liquid metal;
2. power supply to the measuring transmitter, so that the measuring transmitter emits an incident ultrasonic wave propagating in the liquid metal;
3. detection, by the receiver, of an ultrasonic wave reflected or diffracted by an inclusion in the liquid metal under the effect of the incident wave;
4. repeating steps b) to c) during different measurement times, the measurement times forming a measurement time range, and measuring an amplitude of the ultrasonic wave detected by the receiver at each measurement time;
5. selection, within the measurement time range, of a normalization period, the normalization period grouping together consecutive measurement times;
6. calculation of a normalization amplitude as a function of the amplitudes measured at the times of the normalization period;
7. normalization of the amplitudes measured during the measurement time range, during each step c), by the normalization amplitude calculated during step f).

• assignation d’une valeur de normalisation à l’amplitude de normalisation ;
• multiplication des amplitudes normalisées par la valeur de normalisation.

According to one possibility, step f) comprises:

• assigning a normalization value to the normalization amplitude;
• multiplication of the normalized amplitudes by the normalization value.

The normalization period can correspond to a predefined part of the measurement time range.

The normalization period can correspond to the last moments of the measurement time range.

The normalization period may extend into the last few minutes of the measurement time range.

• ei) analyse des amplitudes respectivement mesurées durant la plage temporelle de mesure ;
• eii) identification d’une période d’amplitude minimale, comportant différents instants de mesure successifs, durant laquelle les amplitudes mesurées sont minimales, par rapport aux intensités mesurées en dehors de la période d’intensité minimale ;
  de façon que la période de normalisation comprenne tout ou partie de la période d’amplitude minimale.

Step e) may include:

• ei) analysis of the amplitudes respectively measured during the measurement time range;
• (eii) identification of a period of minimum amplitude, comprising different successive measurement times, during which the measured amplitudes are minimal, compared to the intensities measured outside the period of minimum intensity;
  so that the normalization period includes all or part of the minimum amplitude period.

• la sous-étape ei) comporte un calcul d’une moyenne glissante ou d’une médiane glissante des amplitudes respectivement mesurées durant la plage temporelle de mesure, la moyenne ou la médiane étant calculée selon différentes fenêtres temporelles successives ;
• lors de la sous-étape eii), la période d’amplitude minimale correspond à une période dans laquelle la moyenne ou la médiane glissante est minimale.

According to one possibility,

• sub-step ei) comprises a calculation of a moving average or a moving median of the amplitudes respectively measured during the measurement time range, the average or the median being calculated according to different successive time windows;
• during sub-step eii), the period of minimum amplitude corresponds to a period in which the moving average or median is minimal.

• le métal liquide s’écoule le long d’une goulotte ;
• l’émetteur et le récepteur sont disposés face à la goulotte.

According to one possibility,

• the liquid metal flows along a chute;
• the transmitter and receiver are arranged facing the chute.

• l’émetteur de mesure est couplé à un premier guide d’onde, immergé dans le métal liquide ;
• le récepteur est couplé à un deuxième guide d’onde, immergé dans le métal liquide.

According to one possibility,

• the measuring transmitter is coupled to a first waveguide, immersed in the liquid metal;
• the receiver is coupled to a second waveguide, immersed in the liquid metal.

• un émetteur de mesure (11) configuré pour émettre une onde ultrasonore incidente dans le métal liquide;
• un récepteur (21) configuré pour détecter une onde ultrasonore réfléchie ou diffractée par une inclusion dans le métal liquide, suite à l’émission de l’onde ultrasonore incidente ;
• une unité de traitement (32), configurée pour :
• mesurer une amplitude de l’onde ultrasonore détectée par le récepteur en différents instants de mesure, les différents instant de mesure formant une plage temporelle de mesure;
• mettre en œuvre les étapes e) à g) d’un procédé selon le premier objet de l’invention à partir des amplitudes mesurées durant la plage temporelle de mesure.

A second object of the invention is an ultrasonic probe, comprising:

• a measuring transmitter (11) configured to emit an ultrasonic wave incident into the liquid metal;
• a receiver (21) configured to detect an ultrasonic wave reflected or diffracted by an inclusion in the liquid metal, following the emission of the incident ultrasonic wave;
• a processing unit (32), configured to:
• measuring an amplitude of the ultrasonic wave detected by the receiver at different measurement times, the different measurement times forming a measurement time range;
• implementing steps e) to g) of a method according to the first subject of the invention from the amplitudes measured during the measurement time range.

The invention will be better understood by reading the description of the exemplary embodiments presented in the remainder of the description, in conjunction with the figures listed below.

FIGURES

There FIG. 1 represents a first embodiment of a probe allowing implementation of the invention.

There FIG. 1 represents another embodiment of a probe allowing an implementation of the invention.

There FIG. 2 schematizes acoustic waves detected at a measurement time.

There FIG. 3 schematizes the main stages of a process according to the invention.

There FIG. 3 schematizes a configuration in which the normalization period is located at the end of a measurement time range.

There FIG. 3 schematizes a configuration in which the normalization period is located in the middle of a measurement time range.

There FIG. 4 represents a correlation curve between detected wave amplitudes and an inclusion dimension.

There FIG. 5 shows amplitudes measured at different measurement times.

There FIG. 5 shows the amplitudes represented on the FIG. 5 after applying a normalization function.

Presentation of specific embodiments

We have represented, in connection with the FIG. 1 , a first example of an ultrasonic probe, allowing an implementation of the invention. The probe comprises a measurement transmitter 11 and a receiver 21. In this example, the measurement transmitter 11 and the receiver 21 are formed by the same acoustic transducer. The acoustic transducer is for example a piezoelectric transducer, known to those skilled in the art.

The transducer is, for example, a pellet of piezoelectric material emitting an acoustic wave at a frequency of 15 MHz. Generally speaking, the frequency of the emitted acoustic wave is preferably between 1 MHz and 20 MHz.

The probe is intended for detecting inclusions in a liquid medium 2 to be analyzed, and more precisely in a liquid metal, the metal being for example an aluminum alloy. In Figures 1A and 1B, the dotted horizontal line represents an interface between the liquid metal 2 and an ambient medium 3, the latter being for example air.

The probe 1 comprises a waveguide 13, intended to be immersed in the liquid metal 2. The waveguide is configured to facilitate propagation of an acoustic wave between the transducer, acting as a transmitter-receiver, and the liquid metal 2. The waveguide 13 is for example formed from a material of the silicon nitride or silicon oxynitride (SiAlON) type, or more generally a material, of the refractory metal or ceramic type, considered to be inert to the liquid metal 2. The use of such a waveguide has been described in EP3204763B1 or in EP1194772B1.

In Figures 1A and 1B, an ultrasonic field 5 is shown diagrammatically, which corresponds to a part of the liquid metal inspected by the probe 1. When an inclusion 4 passes through the ultrasonic field, a part of the acoustic wave emitted by the measuring transmitter 11 is reflected or diffracted towards the receiver 21. The characteristics of the reflected wave, in particular its amplitude, allow the detection of an inclusion and its characterization, for example an estimation of its size.

Preferably, the probe may comprise a power transmitter 12, configured to emit an acoustic wave of power, i.e. greater than 10W, and preferably greater than 100W, for example 120W, as described in EP3204764. Indeed, the application of a high-power acoustic wave makes it possible to obtain the wetting of the first waveguide 13 by the liquid metal 2. The power transmitter 12 is activated at regular intervals, so as to allow good wetting of the waveguide 13 by the liquid metal 2 to be maintained.

There FIG. 1 represents a preferred configuration, in which the probe comprises a measuring transmitter 11 and a receiver 21. The measuring transmitter 11 is connected to a first waveguide 13. The receiver 21 is connected to a second waveguide 23, preferably identical to the first waveguide 13 (same material, same dimensions), but distinct from the latter. The length of each waveguide is for example between 10 cm and 50 cm. The first and second waveguides are preferably rectilinear, and extend respectively around a first central axis Δ1 and a second central axis Δ2. The central axes Δ1 and Δ2 are intersecting, and inclined relative to each other at an angle α. The angle is for example between 25° and 35°. The angle α is for example equal to 28°. Such a configuration has been described in EP3204764B1.

In this embodiment, the probe may comprise a first power emitter 12, and a second power emitter 22, as described in connection with the first embodiment. The first and second power emitters are activated periodically, so as to enhance the wetting of the first waveguide 13 and the second waveguide 23, respectively, by the liquid metal 2.

The probe comprises a control unit 31, configured to allow control of the emission of an acoustic wave by the measurement transmitter 11, and to allow control of the emission of a power acoustic wave by the power transmitter 12 when the probe comprises such a transmitter.

Regarding the acoustic waves emitted by the measuring transmitter 11, they are generally emitted in series. In practice, each series of acoustic waves is emitted by the measuring transmitter 11 at a regular frequency, for example 10 Hz. A series comprises at least one acoustic wave, and preferably several successive acoustic waves. Within the same series, the emission frequency of the acoustic waves can be between 100 Hz and 300 Hz, which corresponds to the excitation frequency of the piezoelectric transducer of the measuring transmitter. In Figure 2, three waves detected by the receiver 21 are shown diagrammatically, following the emission of a series of three waves by the measuring transmitter 11. The bracket indicates that the three waves belong to the same series. In Figure 2, the ordinate axis corresponds to the detected signal by receiver 21 while the abscissa axis corresponds to time t.

The probe comprises a processing unit 32, configured to calculate an amplitude of the waves detected by the receiver 21 in response to a series of acoustic waves emitted by the measuring transmitter 11. The calculated amplitude, or measured amplitude, may be an average of the maximum amplitudes of each detected acoustic wave, or the maximum amplitude of all the acoustic waves detected following the emission of a series of waves. Subsequently, the calculated amplitude is noted . denotes a measurement instant, which corresponds to the detection of a series of acoustic waves by the receiver 21.

The processing unit 32 may comprise one or more microprocessors or electronic circuits, configured to determine the measured amplitude and implement the steps described below.

The presence of a 4 inclusion, of significant size, in the liquid metal, causes the measured amplitude to vary . The measured amplitude can make it possible to estimate an inclusion size, from an experimentally established correlation curve, using a reference method, for example the LIMCA method or the method based on the use of calibrated reflectors, these methods being described in connection with the prior art.

The same correlation curve can be used on different ultrasound probes of identical design, provided that the amplitudes resulting from said ultrasound probes are comparable. However, as previously indicated, it is necessary to take into account the variability of the probes, even if they are similar to each other: same dimensions and same materials used.

The processing unit 32 comprises an amplifier, intended to amplify the signal resulting from the receiver 21, and whose amplification gain G ₀ is generally fixed in the factory, following the manufacture of the probe. The gain is generally not modified during the procedure. It can be subject to an adjustment depending on the conditions of use of the probe.

The probe 1 is intended to be immersed in a liquid metal 2 flowing from a furnace. The liquid metal may, for example, flow from a furnace, or be present in a ladle or crucible. For example, the liquid metal flows into a chute. The probe's waveguide(s) are immersed in the chute. The probe may be placed directly at the outlet of the furnace, or further downstream, at the outlet of a filter or degasser. This may be a deep bed filter or a CFF (Ceramic Foam Filter) or any other type of filter. When the probe is immersed in a melt downstream of a CFF filter, the presence of bubbles in the liquid metal 2 may require adjustment of the amplification gain G ₀ to take into account the attenuation effect induced by the bubbles. The probe can also be placed downstream of a degasser and in a crucible.

For reasons of cost and time, it is not realistic to perform, for each probe, a correlation curve with a reference method. However, it is necessary to take into account the variability of the response of each probe. The inventor proposes a method making it possible to obtain a comparable response for each probe, so as to be able to use a correlation curve as described below, in connection with the FIG. 4 .

The main steps of the process are illustrated in the FIG. 3 .

Step 100 : Carrying out measurements at different measurement times.

During casting, measurements are regularly taken, for example at measuring times repeated at a frequency for example between 0.1 Hz and 20 Hz. All measurement times forms a measurement time range . Each measurement instant is associated with a measured amplitude. . As described in connection with the FIG. 2 , the measured amplitude is calculated by the processing unit 32 from one or a series of waves detected by the receiver 21 following the emission of one or a series of acoustic waves by the measurement transmitter 11.

Step 110 : Selecting a normalization period.

An important aspect of the invention is that during the measurement time range , we can select a normalization period , comprising different measurement times consecutive. The normalization period corresponds to measurement moments to which the amplitude measured is the lowest. Indeed, during a casting, it is considered that at a given position, occupied by probe 1, along the casting, a time period can be defined, called the period of minimum amplitude , during which the measured amplitude is minimal. The minimum amplitude period corresponds to a period during which the casting is considered "clean", in the sense that the metal contains a minimum proportion of inclusions. The normalization period corresponds to all or part of the minimum amplitude period . During the normalization period , the measured amplitude is then minimal, compared to the amplitude measured during previous measurement times and/or subsequent to the standardization period. During the standardization period , the average amplitude is lower than the average amplitude calculated during the measurement time range .

The period of normalization can be defined a priori, depending on the position occupied by probe 1 with respect to the casting, in particular when probe 1 is placed downstream of a filter. It is considered that it is at this position that the liquid metal has a minimum proportion of inclusions compared to the other positions of the casting. At this position, the normalization period may correspond to the last moments of the casting, for example the last minute or minutes of the casting, knowing that the duration of a casting is generally of the order of 1 to 3 hours or even more. The normalization period then corresponds to the last measurement moments of the measurement time range . This scenario is illustrated in the FIG. 3 .

Alternatively, the normalization period is identified a posteriori, following an analysis of the amplitudes measured during the measurement times . Based on the measurements, the minimum amplitude period is defined. The normalization period is selected in the minimum amplitude period. . This corresponds to the case illustrated in Figure 3C. The minimum amplitude period can be identified by calculating a running average or a running median of the amplitudes respectively measured during the measurement time range, the average or median being calculated according to different successive time windows, offset from each other by approximately 1 minute. The minimum amplitude period corresponds to a period in which the moving average or median is minimal.

When the probe is placed at the outlet of the furnace, depending on the processes implemented, it can be considered that the liquid metal contains a high quantity of inclusions at the start and end of the casting.

Step 120 : Determining a normalization amplitude

After the normalization period has been selected, a normalization amplitude is determined , representative of the normalization period. It can be an average or a median of the amplitudes measured during the normalization period .

Step 130 : Normalization

• correspond à la fonction de normalisation.
• est l’amplitude normalisée
• peut être par exemple à 20. Dans ce cas, l’amplitude normalisée est exprimée en %. La valeur de 20% correspond à une coulée considérée comme propre., c’est-à-dire dont les éventuelles inclusions ne produisent pas de signaux ultrasonores significatifs.
• est l’amplitude de normalisation précédemment définie.

During this stage, the amplitudes respectively measured at each measurement time are normalized by the normalization amplitude . Preferably, a normalization value, noted , is assigned to the normalization amplitude _. Thus, the standardization is carried out in such a way that:
(1)
Or :

• corresponds to the normalization function.
• is the normalized amplitude
• can be for example 20. In this case, the normalized amplitude is expressed in %. The value of 20% corresponds to a casting considered clean, that is to say, any inclusions of which do not produce significant ultrasonic signals.
• is the previously defined normalization amplitude.

An advantage of the method is that the value of the normalized amplitude corresponding to the normalization value , for example 20%, represents a clean flow, regardless of the probe used. Normalization makes it possible to overcome the variability affecting the probes: the normalized amplitudes of the different probes are comparable with each other, because the normalization amplitude is established, for each probe, under comparable conditions, in this case in the presence of a liquid metal considered to be clean.

Standardization assumes that it is accepted that the liquid metal, considered clean, that is to say free of detectable inclusions, generates acoustic waves whose amplitude is assimilated to noise. This amplitude corresponds to the standardization amplitude

Normalization makes it possible to use a correlation curve as shown in Figure 4. The correlation curve is established experimentally with a probe, after applying the normalization function determined for said probe. Thus, the ordinate axis corresponds to , Or corresponds to measured amplitudes correlated with known inclusion dimensions. The x-axis corresponds to a dimension, for example the diameter, measured by the reference method.

On the FIG. 4 , we have represented points, each point corresponding to an experimental measurement. We have also plotted a calibration curve, in dotted lines, obtained by adjusting the experimental points.

The correlation function thus established can be applied to normalized amplitudes from different probes. The correlation curve can be implemented using, as ordinates, normalized values .

A probe as described in connection with Figure 1B was implemented in an aluminum casting, the probe being arranged at an outlet of a CFF filter. The duration of the measurement time range was 60.5 minutes. Figure 5A corresponds to measured amplitudes (y-axis) at different measurement times (x-axis – acquisition frequency 10 Hz). Figure 5B shows the amplitudes represented in Figure 5A after normalization , the normalization value N being 16%. In Figures 5A and 5B, the zero amplitudes correspond to times during which the power transmitter 12 was activated. When the power transmitter is activated, the signal detected by the receiver is not taken into account.

The invention can be applied by implementing ultrasonic probes of the same type, that is to say having the same structure, the variability being due to the differences affecting the components of the probe.

If the measurements are to be used for characterizing inclusion dimensions, it is preferable, or even necessary, that for at least one probe, a correlation has been made between the measured amplitudes and measurements of the inclusion dimensions. The latter are obtained by a reference measurement method. It is preferable that the different probes used are then placed in the same position, or in positions considered equivalent, relative to the casting.

Claims (11)

Method for characterizing a liquid metal, the method using a probe (1), comprising:

• a measuring transmitter (11) configured to emit an ultrasonic wave incident into the liquid metal;
• a receiver (21) configured to detect an ultrasonic wave reflected or diffracted by an inclusion in the liquid metal, following the emission of the incident ultrasonic wave;

the method comprising the following steps:

1. arrangement of the probe (1), such that the measuring transmitter and the receiver are arranged in the liquid metal or facing the liquid metal;
2. power supply to the measuring transmitter, so that the measuring transmitter emits an incident ultrasonic wave propagating in the liquid metal;
3. detection, by the receiver, of an ultrasonic wave reflected or diffracted by an inclusion in the liquid metal under the effect of the incident wave;
4. repetition of steps b) to c) during different measurement times ( ), the measurement instants forming a measurement time range ( ), and measurement of an amplitude ( ) of the ultrasonic wave detected by the receiver at each measurement instant;
5. selection, in the measurement time range ( ), of a period of normalization ( ), the standardization period comprising consecutive measurement times;
6. calculation of a normalization amplitude ( ) as a function of the amplitudes measured at the times of the normalization period;
7. normalization of the amplitudes measured during the measurement time range, during each step c), by the normalization amplitude calculated during step f).

Method according to claim 1, in which the normalization period corresponds to measurement times at which the measured amplitude is the lowest. A method according to any preceding claim, wherein step f) comprises

• assignment of a normalization value ( ) to the normalization amplitude;
• multiplication of the normalized amplitudes by the normalization value.

A method according to any preceding claim, wherein the normalization period corresponds to a predefined part of the measurement time range. Method according to claim 4, wherein the normalization period corresponds to the last instants of the measurement time range. The method of claim 4, wherein the normalization period extends during the last minutes of the measurement time range. Method according to any one of claims 1 to 3, in which step e) comprises

• ei) analysis of the amplitudes respectively measured during the measurement time range;
• (eii) identification of a period of minimum amplitude, comprising different successive measurement times, during which the measured amplitudes are minimal, compared to the intensities measured outside the period of minimum intensity;

so that the normalization period includes all or part of the minimum amplitude period. The method of claim 7, wherein:

• sub-step ei) comprises a calculation of a moving average or a moving median of the amplitudes respectively measured during the measurement time range, the average or the median being calculated according to different successive time windows;
• during sub-step eii), the period of minimum amplitude corresponds to a period in which the moving average or median is minimal.

A method according to any preceding claim, wherein

• the liquid metal flows along a chute;
• the transmitter and receiver are arranged facing the chute.

A method according to any preceding claim, wherein

• the measuring transmitter is coupled to a first waveguide (13), immersed in the liquid metal;
• the receiver is coupled to a second waveguide (23), immersed in the liquid metal.

Ultrasound probe, comprising:

• a measuring transmitter (11) configured to emit an ultrasonic wave incident into the liquid metal;
• a receiver (21) configured to detect an ultrasonic wave reflected or diffracted by an inclusion in the liquid metal, following the emission of the incident ultrasonic wave;
• a processing unit (32), configured to
• measure an amplitude ( ) of the ultrasonic wave detected by the receiver at different measurement times ( ), the different measurement moments forming a measurement time range ( );
• implementing steps e) to g) of a method according to any one of the preceding claims from the measured amplitudes ( ) during the measurement time range.