EP4676283A1

EP4676283A1 - Method for recognizing a capsule for preparing a drink, in particular coffee

Info

Publication number: EP4676283A1
Application number: EP24716460.1A
Authority: EP
Inventors: Antonio DE MOLINER; Christian Bortoluzzi; Nicolas TORTORA; Roberto Oboe; Gian Antonio SUSTO; Federico ZOPPAS
Original assignee: IRCA SpA Industria Resistenze Corazzate e Affini
Current assignee: IRCA SpA Industria Resistenze Corazzate e Affini
Priority date: 2023-03-07
Filing date: 2024-03-06
Publication date: 2026-01-14
Also published as: IT202300004137A1; WO2024184820A1

Abstract

A method for recognizing a capsule for preparing a drink, in particular a capsule containing powder for preparing the drink, arranged in a machine for preparing drinks provided with at least one measurement device (2, 3) adapted to measure the flow rate or pressure of liquid in a conduit (4) of the machine, and adapted to generate a signal associated with said flow rate or said pressure; wherein the recognition is performed as a function of said signal generated by said measurement device (2, 3) during a predetermined time interval, in particular when the pump (5) of the machine is active for preparing the drink; wherein the recognition is performed as a function of at least one feature of said signal, selected among mean value, variance, maximum peak, minimum peak and derivative.

Description

METHOD FOR RECOGNIZING A CAPSULE FOR PREPARING A DRINK, IN PARTICULAR COFFEE

Field of the invention

The present invention relates in particular to a method for recognizing a capsule for preparing a drink, e.g. for recognizing a capsule containing coffee powder. Background art and field of the invention

Machines for preparing coffee using a capsule are very common.

A machine of this type comprises a tank, a pump, a heater and a seat for a capsule. The capsule contains coffee powder, although the capsule can contain powder for preparing another drink.

Some machines are configured so that the coffee is dispensed in several successive steps: pre-infusion step, in which the pump is activated and the capsule is filled with liquid; pause step, in which the machine pump remains inactive; and the actual dispensing step, in which the pump is activated again, so that the liquid (i.e. liquid coffee) comes out of the capsule and is dispensed.

Some machines are instead configured so that, to prepare the coffee, the pump remains continuously active. This type of machine can be called a continuous dispensing machine.

There is a wide variety of capsules for preparing coffee or other drinks. The capsules may differ from one another due to various features. In particular, the capsules may differ from one another due to the materials of which they are made, e.g. plastic or aluminum. Further, the capsules may differ from one another due to the features of the powder they contain. In particular, both the amount of powder contained and the features of the powder itself may vary. For example, the powder may be made by a blend of types of coffee, or aromas may be added.

It is desirable for the machine for preparing coffee, or another drink, to be able to recognize the type of capsule that is inserted therein.

Summary of the invention

An object of the present invention is to enable the recognition of the type of capsule arranged in a machine for preparing a drink, e.g. for preparing coffee.

In particular, an object of the present invention is to provide a method that allows said recognition, which can be implemented by a machine that is easy to manufacture and at low costs.

The present invention achieves at least one of these objects and other objects that will become clear in light of the present description, by a method for recognizing a capsule for preparing a drink, in particular a capsule containing powder for preparing the drink, arranged in a machine for preparing drinks provided with at least one measurement device adapted to measure the flow rate or pressure of liquid in a conduit of the machine, and adapted to generate a signal associated with said flow rate or said pressure; wherein the recognition is performed as a function of said signal generated by said measurement device during a predetermined time interval, in particular when the pump of the machine is active for preparing the drink; wherein the recognition is performed as a function of at least one feature of said signal selected among mean value, variance, maximum peak, minimum peak and derivative; preferably by means of a machine learning algorithm.

In particular, at least one of said features is preferably used as input data for the machine learning algorithm. Preferably, at least two of said features and more preferably at least three of said features are used as input data of said machine learning algorithm.

The inventors have experimentally observed that to be able to perform the recognition, preferably by the machine learning algorithm, mean value, variance, maximum peak, minimum peak and derivative of the signal are significant and allow a reliable recognition.

Among these, to obtain a reliable recognition, mean value, variance, maximum peak and derivative of the signal are particularly significant; and mean value, variance and maximum peak are more greatly significant.

The recognition, in particular by machine learning algorithm, was particularly reliable using at least two of said features of said signal, preferably at least three of said features selected from mean value, variance, maximum peak, minimum peak and derivative.

The recognition was particularly reliable using at least mean value, variance and maximum peak of said signal. It is particularly preferable that said algorithm is selected from Logistic regression, Support Vector Machines, Multi Layer Perceptron, and neural network, preferably from Logistic regression, Support Vector Machines, and neural network.

The Inventors have found that the flow rate and pressure of the liquid vary when the type of capsule used is changed, whereas they remain substantially constant for the same type of capsule. In particular, the Inventors have found that the flow rate and pressure are affected by features of the powder contained in the capsule, e.g. quantity and/or type of powder.

Therefore, an idea at the basis of the present invention is to analyze the flow rate and/or the pressure, in particular one or more features of flow rate and/or pressure, in order to be able to recognize, or classify, a capsule arranged in a machine for preparing drinks, in particular for preparing coffee.

Capsules of the same type means capsules that are substantially identical to one another, in particular containing the same amount and type of powder.

The recognition of the type of capsule can have various advantageous applications. For example, the recognition of the type of capsule may be exploited to optimize the configuration of the machine as a function of the type of capsule, or the recognition may be exploited to allow the preparation of the coffee with a capsule recognized as original and so as to inhibit the operation of the machine if the capsule is recognized as non-original, i.e. compatible.

Advantageously, the method according to the invention allows the type of capsule to be recognized or classified (e.g. original or compatible) from the sole analysis of the flow rate, or in other words of the flow, and/or pressure of the liquid during the dispensing operation or, in other words, when the pump is active.

The invention further relates to a machine for preparing drinks by means of a capsule according to claim 16.

The invention further relates to an electronic control unit for a machine for preparing drinks by means of a capsule according to claim 19.

Further features and advantages of the invention will become more apparent in light of the detailed description of exemplary but not exclusive embodiments.

The dependent claims describe particular embodiments of the invention.

Brief description of the drawings The description of the invention refers to the accompanying drawings, which are provided by way of example and not limitation, in which:

Figure 1 illustrates a diagram of an example of machine according to the invention;

Figure 2 illustrates a diagram of another example of machine according to the invention;

Figure 3 illustrates a diagram of another example of machine according to the invention;

Figure 4 illustrates a trend chart of the flow rate over time in a first type of machine when an original capsule type is used;

Figure 5 illustrates a trend chart of the mean of the flow rate over time, related to five types of original capsules and two types of compatible capsules; each curve drawn with a solid line represents the mean trend of the flow rate over time of the dispensing operations carried out with each type of original capsule and each curve drawn with a broken line represents the mean trend of the flow rate over time of the dispensing operations carried out with each type of compatible capsule;

Figure 6 illustrates values related to dispensing operations carried out with original capsules (circles) and values related to dispensing operations carried out with compatible capsules (“x”), in a three-axis reference system, said axis being mean value, variance and maximum peak value of the flow rate, respectively;

Figure 7 illustrates the trend of the flow rate over time in a second type of machine using a type of capsule;

Figure 8 illustrates the trend of the flow rate over time in the second type of machine using another type of capsule;

The same elements or parts have the same reference numerals.

Description of example embodiments of the invention

With reference to the Figures, embodiments of a method, of a machine and of an electronic control unit according to the invention are described.

In all the embodiments, the invention relates to a method for recognizing a capsule (not illustrated) for preparing a drink, in particular a capsule containing powder for preparing the drink, arranged in a machine for preparing drinks provided with at least one measurement device 2, 3 adapted to measure the flow rate or pressure of liquid in a conduit 4 of the machine, and adapted to generate a signal associated with said flow rate or said pressure; wherein the recognition is performed as a function of said signal generated by said measurement device 2, 3 during a predetermined time interval, in particular when the pump 5 of the machine is active for preparing the drink; wherein the recognition is performed as a function of at least one feature of said signal selected among mean value, variance, maximum peak, minimum peak and derivative; preferably by means of a machine learning algorithm.

In particular, at least one of said features is preferably used as input data for the machine learning algorithm.

The machine learning algorithm comprises in particular a machine learning classifier, in particular a trained machine learning classifier.

The machine learning classifier is preferably a Logistic regression classifier, a Support Vector Machines classifier, a Multi Layer Perceptron classifier, or neural network classifier, in particular adapted to indicate whether at least one of said features of the signal corresponds to a signal feature stored in a database.

In particular, at least one of said features is used as input data for said machine learning algorithm, in particular as input data for the machine learning classifier.

The method, in particular, can be performed when the capsule is arranged in the machine.

The aforesaid predetermined time interval can have, by way of example, a maximum duration that may start from the activation of the pump 5 for preparing the drink until the complete dispensing of the drink. However, the predetermined time interval may last from 1 to 60 s, which may be for example a part of the maximum duration.

The aforesaid signal may be for example an analog or digital signal.

In particular, the method comprises the steps of: a) arranging a capsule in the machine; b) activating the pump, in particular for preparing the drink; c) by means of said measurement device, measuring the flow rate or pressure of liquid in the conduit of the machine during said predetermined time interval, generating said signal associated with said flow rate or said pressure, and sending said signal to an electronic control unit, in particular of the machine; d) by means of said electronic control unit, performing the recognition of the capsule as a function of said signal.

Advantageously, the recognition is performed as a function of at least one feature of said signal selected among mean value, variance, maximum peak, minimum peak and derivative.

By way of example, if the recognition is performed as a function of a feature of the flow rate, e.g. the mean value in said predetermined interval, it can be recognized that the capsule in use is of a specific type, e.g. of the “original capsule” type, if the value of said feature, e.g. if said mean value, is greater than or less than a predetermined value.

Preferably, the recognition is performed as a function of at least two, more preferably at least three, of said features selected from mean value, variance, maximum peak, minimum peak and derivative, in particular used as input data for the machine learning algorithm as already explained.

Preferably, the recognition is carried out as a function of at least mean value, variance and maximum peak of said signal, in particular used as input data for the machine learning algorithm as already explained.

The number of features as a function of which the recognition is performed may increase the precision of the recognition.

The recognition may be performed as a function of the signal generated by said measurement device 2, 3, e.g. during a step of filling the capsule with liquid, in particular with the liquid exiting the pump 5 of the machine. In other words, said signal is generated during said step of filling the capsule with liquid, e.g. during all or part of said filling step. The step during which the capsule is filled with liquid is also known as pre-infusion step.

Further, or alternatively, the recognition can be carried out as a function of said signal generated by said measurement device 2, 3 during a step in which liquid flows out of the capsule, e.g. during all or part of the step in which liquid flows out of the capsule. In other words, said signal is generated during said step in which liquid flows out of the capsule. Therefore, the recognition can be carried out as a function of the signal generated during the pre-infusion step and/or during the step in which liquid flows out of the capsule.

The method can be carried out with a machine configured so that the dispensing of the coffee is performed in several steps that are subsequent to one another: preinfusion step 11 (Fig. 4), in which the pump 5 is activated and the capsule is filled with liquid; pause step 12, in which the pump 5 of the machine remains inactive; and actual dispensing step 13, in which the pump 5 is activated again, whereby the liquid (i.e. the liquid coffee) flows out of the capsule and is dispensed. The measurement of the flow rate and/or pressure of liquid as a function of which the recognition is carried out can be performed during the pre-infusion step 11 and/or during the dispensing step 13.

Furthermore, the method can be carried out with a machine configured so that, to prepare the drink, the pump 5 remains active in continuous mode. This type of machine can be called a continuous dispensing machine.

By way of mere example, the recognition can be carried out as a function of said signal generated by means of said measurement device 2, 3 starting from the activation (in particular from the first activation) of the pump 5, or after a predetermined time from the activation of the pump 5, e.g. after about 5.5 s from the activation of the pump 5. The time is for example selected so as to use the signal detected when the flow rate is substantially stable over time.

Preferably, the recognition is carried out by at least one machine learning algorithm, preferably a supervised machine learning algorithm. Preferably, said at least one algorithm is selected from Logistic regression, Support Vector Machines, Multi Layer Perceptron, and neural network.

“Supervised” means in particular that it is based on data previously classified by a user.

Preferably, the recognition is performed using a database where information is stored relating to at least one type of capsule; in particular wherein one or more features of said signal are compared with respective signal features associated with the flow rate or pressure of said type of capsule stored in said database. For example, information can be stored in the database related to one or more types of original capsules.

The recognition may optionally be carried out as a function of at least two (preferably three or at least three) of said features of said signal, and in this case optionally the recognition can be carried out with respect to a surface defined by a threshold value for each of said features, said surface being in a reference system having a number of coordinates equal to the number of said features.

For example, said surface can be defined by a mean value threshold value, a variance threshold value and a maximum peak threshold value.

The method may be carried out with a machine provided with one or more measurement devices 2 and/or provided with one or more measurement devices 3.

Each measurement device 2 is for example a flow meter and each measurement device 3 is for example a pressure sensor or a pressure transducer.

For example, the method can be carried out with a machine provided with at least one first measurement device 2 adapted to measure the flow rate and adapted to generate a first signal associated with the flow rate of liquid in said conduit 4 of the machine, the machine also being provided with at least one second measurement device 3 adapted to measure the pressure of liquid in the conduit 4 of the machine and adapted to generate a second signal associated with the pressure; and the recognition can be performed both as a function of said first signal and as a function of said second signal, in particular generated during said predetermined time interval, in particular when the pump 5 of the machine is active for preparing the drink.

Therefore, when one or more measurement devices 2 and/or one or more measurement devices 3 are provided, the recognition can be carried out as a function of a plurality of signals, i.e. as a function of the signal generated by each of the measurement devices 2 and by each of the measurement devices 3.

By way of example, the machine comprises in particular a tank 8 for the liquid, typically water, for preparing the drink; the pump 5 adapted to pump the liquid contained in the tank 8; a heating device 6 (or heater), adapted to heat the liquid coming from the pump 5 and a seat 9 for the capsule, also known as the infusion chamber. In particular, the pump 5 is downstream of the tank 8; the heating device 6 is downstream of the pump 5 and the seat 9 is arranged downstream of the heating device 6.

At least one measurement device 2 can be provided upstream of the pump 5 and/or at least one measurement device 2, 3 downstream of the pump 5.

The measurement device 3 can optionally be arranged downstream of the heating device 6.

The measurement device 2, 3 (or the measurement devices 2, 3) is, in particular, arranged downstream of the tank 8.

The machine also comprises an electronic control unit 7, 7’, 7” configured to perform the method according to the invention.

The electronic control unit 7, 7’, 7” is in particular connected to said at least one measurement device 2, 3, in particular so as to be able to receive and analyze the aforesaid signal.

The electronic control unit 7, 7’, 7” is preferably also connected to the pump 5 and/or to the heating device 6, in particular so as to be able to control the operation thereof.

Fig. 1 illustrates a diagram of an example of a machine according to the invention wherein the machine is provided with a measurement device 2 (in particular only one measurement device) adapted to measure the flow rate of liquid in the conduit 4. The measurement device 2 is, in particular, a flow meter.

In this example, the measurement device 2 is preferably arranged upstream of the pump 5, although it can be arranged downstream of the pump 5.

Fig. 2 illustrates a diagram of another example of a machine according to the invention, wherein the machine is provided with a measurement device 3 (in particular only one measurement device) adapted to measure the flow rate of liquid in the conduit 4. The measurement device 3 is, in particular, a pressure sensor or pressure transducer.

In this example, the measurement device 3 is in particular arranged downstream of the pump 5.

Fig. 3 illustrates a diagram of another example of a machine according to the invention, wherein the machine is provided with a measurement device 2 (in particular only one measurement device 2) adapted to measure the flow rate of liquid in the conduit 4 and a measurement device 3 (in particular only one measurement device 3) adapted to measure the pressure of liquid in the conduit 4. The measurement device 2 is, in particular, a flow meter and the measurement device 3 is in particular a pressure sensor or pressure transducer.

In this example, the measurement device 2 is arranged upstream of the pump 5 (although it could be arranged downstream of the pump 5) and the measurement device 3 is arranged downstream of the pump 5.

Fig. 4 illustrates a chart of the trend of the flow rate, or flow, expressed in ml/s over time expressed in seconds in a machine configured so that the dispensing of the coffee is performed in various steps subsequent to one another (pre-infusion step 11 , pause 12 and dispensing step 13), as previously described. The sampling frequency was one sample every 0.5 seconds. The capsule used was “original, type 1”.

The invention can be better understood through the experimental tests described below by way of non-limiting example.

Fig. 5 illustrates a chart of the trend of the flow rate over time, in particular with a sampling frequency of one sample every 0.5 seconds.

Dispensing operations were carried out with five types (or sub-types) of original capsules and with two types of compatible capsules. The five types of original capsules and the two types of compatible capsules differed from one another, in particular, due to the type of powder contained inside them. For each capsule of each type, 30 dispensing operations were carried out. The types of capsules were: original, type 1 ; original, type 2; original, type 3; original, type 4; original, type 5; compatible, type 1 ; compatible, type 2.

Each curve drawn with a solid line represents the mean trend of the flow rate over time of the dispensing operations carried out with each type of original capsule and each curve drawn with a broken line represents the mean trend of the flow rate over time of the dispensing operations carried out with each type of compatible capsule. In order to be able to use a supervised machine learning algorithm, in particular based on binary classification, each individual dispensing operation can for example be described as a pair x, y where x is a vector of n components and y is 1 or -1 according to the type of capsule (original or compatible, respectively). To describe the individual dispensing operation as a vector x, flow rate measurements were carried out at a predetermined sampling frequency (e.g. one measurement every 0.5 s) in said predetermined time interval. From the pre-infusion step 11 only, characterizing amounts, or features, were selected, specifically: mean value (i.e. average), variance and maximum peak.

Every single dispensing operation is thus identified by three coordinates (mean value, variance, maximum peak). In Fig. 6, the circles refer to values related to dispensing operations with original capsules and the x’s refer to dispensing operations with compatible capsules.

With reference to Figure 6, in practice, the algorithm finds a surface, or separation plane or separator plane, between the circles and the x’s. Said surface is in particular a virtual surface. When a new dispensing operation is carried out it is transformed into a three-component vector and, if it is above the plane then it will be classified as an x (compatible), and if it is below said separation plane it will be classified as a circle (original).

To find the separator plane two different, mutually alternative, algorithms were used. The first algorithm is called logistic regression and the second algorithm is called SVM (Support Vector Machines).

A third algorithm was also tested, alternative to said two algorithms: neural network. Rather than finding a separator plane (which is a two-dimensional linear space), the neural network finds a two-dimensional non-linear space.

The algorithms have demonstrated an average accuracy of: 85% for logistic regression; 85% for SVM; and 90% for the neural network.

Figures 7 and 8 illustrate the trend of the flow rate over time in a machine of the continuous dispensing type, using two different types of capsule, respectively. In both cases, after an initial peak due to the filling of the capsule with liquid and after a subsequent reduction of the flow rate, the flow rate returns to substantially stable values over time.

It is to be noted that in the period in which the flow rate is substantially stable, the flow rate values are very different between the two capsules. Therefore, the recognition can be carried out, e.g. as a function of the mean value of the flow rate measured for example in a time interval comprised within the step in which the flow rate is substantially stable over time.

Claims

1. A method for recognizing a capsule for preparing a drink, in particular a capsule containing powder for preparing the drink, arranged in a machine for preparing drinks provided with at least one measurement device (2, 3) adapted to measure the flow rate or pressure of liquid in a conduit (4) of the machine, and adapted to generate a signal associated with said flow rate or said pressure; wherein the recognition is performed as a function of said signal generated by said measurement device (2, 3) during a predetermined time interval, in particular when the pump (5) of the machine is active for preparing the drink; wherein the recognition is performed as a function of at least one feature of said signal selected among mean value, variance, maximum peak, minimum peak and derivative.

2. A method according to any claim 1 , wherein the recognition is performed as a function of at least two of said features of said signal.

3. A method according to claim 1 or 2, wherein the recognition is performed as a function of at least three of said features of said signal.

4. A method according to any one of the preceding claims, wherein the recognition is performed as a function of at least the mean value, the variance and the maximum peak of said signal.

5. A method according to any one of the preceding claims, wherein the recognition is performed as a function of the signal generated by said measurement device (2, 3) during a step of filling the capsule with liquid, in particular with the liquid exiting the pump (5) of the machine.

6. A method according to any one of the preceding claims, wherein the recognition is performed as a function of said signal generated by said measurement device (2, 3) during a step in which liquid flows out of the capsule.

7. A method according to any one of the preceding claims, wherein the recognition is performed by means of at least one machine learning algorithm, preferably a supervised machine learning algorithm.

8. A method according to claim 7, wherein said at least one feature, or said at least two or at least three features, of said signal are used as input data for said machine learning algorithm; preferably wherein at least the mean value, the variance and the maximum peak of said signal are used as input data for said machine learning algorithm.

9. A method according to claim 7 or 8, wherein said at least one algorithm is selected from Logistic regression, Support Vector Machines, Multi Layer Perceptron, and neural network.

10. A method according to claim 7, 8 or 9, wherein said machine learning algorithm comprises a machine learning classifier, preferably a Logistic regression classifier, a Support Vector Machines classifier, a Multi Layer Perceptron classifier, or neural network classifier, in particular adapted to indicate whether at least one of said features of the signal corresponds to a signal feature stored in a database.

11. A method according to any one of the preceding claims, wherein the recognition is performed using a database where information relating to at least one type of capsule is stored.

12. A method according to claim 11 , wherein one or more features of said signal are compared with respective signal features associated with the flow rate or pressure of said type of capsule stored in said database.

13. A method according to any one of the preceding claims, wherein the recognition is performed as a function of at least two of said features of said signal, wherein the recognition is performed with respect to a surface, or separator plane, in particular a virtual surface, defined by a threshold value for each of said at least two features, said surface being in a reference system having a number of coordinates equal to the number of said features; in particular wherein said signal is transformed into a vector having a number of components equal to the number of said features, and the recognition is performed as a function of the position of said vector with respect to said surface; preferably wherein said surface is defined by said machine learning algorithm.

14. A method according to any one of the preceding claims, wherein the machine is provided with at least one first measurement device (2) adapted to measure the flow rate and adapted to generate a first signal associated with the flow rate of liquid in said conduit (4) of the machine, and wherein the machine is provided with at least one second measurement device (3) adapted to measure the pressure of liquid in the conduit (4) of the machine and adapted to generate a second signal associated with the pressure; wherein the recognition is performed as a function of said first signal and as a function of said second signal, in particular generated during said predetermined time interval, in particular when the pump (5) of the machine is active for preparing the drink.

15. A method according to any one of the preceding claims, comprising the steps of: a) arranging a capsule in the machine; b) activating the pump (5), in particular for preparing the drink; c) by means of said measurement device (2, 3), measuring the flow rate or pressure of liquid in the conduit (4) of the machine during said predetermined time interval, generating said signal associated with said flow rate or said pressure, and sending said signal to an electronic control unit (7, 7’, 7”), in particular of the machine; d) by means of said electronic control unit (7, 7’, 7”), performing the recognition of the capsule as a function of said signal.

16. A machine for preparing drinks by means of a capsule, in particular containing powder for preparing the drink, the machine being provided with a pump (5) and at least one measurement device (2, 3) adapted to measure the flow rate or pressure of liquid in a conduit (4) of the machine, and to generate a signal associated with said flow rate or said pressure; the machine being configured to perform the method according to any one of the preceding claims.

17. A machine according to claim 16, wherein said at least one measurement device (2, 3) is adapted to measure the flow rate of liquid in the conduit (4) of the machine and is arranged so as to be able to measure the flow rate of liquid upstream or downstream of the pump (5); in particular upstream of a heating device (6) of the machine; preferably wherein said at least one measurement device (2) is a flow meter.

18. A machine according to claim 16, wherein said measurement device (3) is adapted to measure the pressure of liquid in the conduit (4) of the machine and is arranged so as to be able to measure the pressure downstream of the pump (5); in particular upstream or downstream of a heating device (6) of the machine; preferably wherein the measurement device (3) is a pressure sensor.

19. An electronic control unit (7) for machine for preparing a drink by means of a capsule, in particular by means of a capsule containing powder for preparing the drink, the electronic control unit (7) being configured to perform the method according to any one of claims 1 to 15.