Machine for making liquid, semiliquid or semisolid food products and related control method

Abstract

To provide a machine for making liquid or semiliquid or semisolid food products, and a method for controlling the machine.SOLUTION: There is provided a machine for making liquid or semiliquid or semisolid food products, comprising: a processing container 2 for processing base products; a stirrer 3 adapted to mix products inside the processing container that processes the base products; a thermal treatment system 4 configured to exchange heat with the processing container that processes the base products; a control unit 5 connected to the thermal treatment system and to the stirrer to drive them; at least one sensor S configured to detect an operating parameter of the machine; a first artificial intelligence control module configured to define a pre-trained classifier having at least one input and at least one output, where the at least one input is connected to the sensor; and a drive module connected to the output of the first control module and at least to the thermal treatment system and / or to the stirrer.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a machine for manufacturing liquid or semi - liquid foods, in particular products in the ice - cream, pastry or catering industries, and to a method for controlling such a machine.

Background Art

[0002] Known in the prior art of the related industry is a machine for thermally and / or mechanically processing a base product preferably having at least one liquid or semi - liquid component to produce a final product of the liquid or semi - liquid or semi - solid type.

[0003] Particularly known in the ice - cream industry is a machine comprising a processing container, an agitator mounted inside the container, and a heat system associated with the container for heat - exchanging with the walls of the container.

[0004] In this type of machine, the base product is thus batch - frozen inside the container, i.e., cooled by the heat system and simultaneously agitated by the agitator, and the base product is converted into the final ice - cream product.

[0005] Generally speaking, in a machine for manufacturing traditional artisanal gelato, the batch - freezing process is achieved by using a combination of a freezing cycle and mechanical agitation. The base food is batch - frozen at a temperature that varies depending on the type of process and the composition of the mixture itself (usually included in the range of - 11 < T(°C) < - 6). This stage of the process is characterized by the rapid cooling of the agitated mixture, which makes it possible to disperse the solid crystal phase and the gas phase into the ice - cream structure without separating the solutes.

[0006] As is known, ice - cream is a fluffy emulsion, characterized by a very specific colloidal structure, which gives ice - cream its unique texture and sensory properties.

[0007] More precisely, gelato, a variant of ice cream, is a solution characterized by a liquid dispersed phase and at the same time, It is an emulsion (containing fat globules), It is a colloid (unevenly dispersed) Bubbles (including air bubbles).

[0008] Furthermore, it should be noted that ice cream requires all three states of matter (solid, liquid, and gas) to exist simultaneously.

[0009] Furthermore, ice cream is known to be a non-Newtonian fluid.

[0010] From the above, it is clear that ice cream is a highly complex food product, consisting of multiple ingredients bound by various interactive forces and exhibiting a variety of physical states that affect the quality, stability and properties of the final product.

[0011] The ingredients of the starting mixture are water, fat, low-fat milk solids, sugars, stabilizers, and emulsifiers, which, as they are gradually cooled and subjected to the mixing action of a mixer, transform into the resulting product, "ready ice cream," with each ingredient dispersed into a distinct phase.

[0012] The structure of ice cream is formed by air being dispersed in a frozen liquid that is roughly two-thirds water. The aqueous phase contains trapped air, fat, and tiny particles of ice crystals, as well as sucrose, lactose, other sugars, and mineral salts.

[0013] Thus, broadly speaking, four main constituent phases can be identified: ice crystals, air bubbles, fat globules and the unfrozen viscous liquid matrix (a solution of monosaccharides, polysaccharides, proteins and emulsifiers), also commonly called buttermilk.

[0014] Close and accurate monitoring of the four phases during processing has a major impact on the physical, structural and organoleptic properties.

[0015] Furthermore, buttermilk is the continuous phase that interacts with the other phases present in ice cream and maintains the structure of the ice cream as a whole, the properties of which depend on the composition and temperature of the mixture.

[0016] Proper consistency and rheology of ice cream are fundamental aspects for its quality: in addition to being a non-Newtonian fluid and exhibiting thixotropic behavior (i.e., its apparent viscosity decreases with time under applied force), ice cream can in fact be considered as a pseudoplastic fluid, whose behavior can only be tentatively described macroscopically with certain mathematical models.

[0017] On the other hand, the amount and size of the ice crystals influence many properties of the final product, such as spoonability, hardness, melt rate, parameters that can be influenced by the aggregation of fat globules in clusters.

[0018] The role of the air bubbles is to provide a light and fluffy consistency.

[0019] From a practical standpoint, it is of fundamental importance to determine when the mixture has reached just the right consistency (ie texture) to obtain an optimum end product.

[0020] Finally, given the complexity of ice cream, and based on what has been said above, it must be emphasized that even slight variations in the mix formulation can cause various problems in the final product.

[0021] Regarding methods for measuring or quantifying the quality of the finished ice cream, please note the following:

[0022] Several conventional methods can be used to define the characteristics of the final product (in terms of quality), such as determination of overrun, evaluation of temperature, melting, i.e. melt rate and panel tests (sensory analysis).

[0023] In particular, overrun is determined by comparing the weight of the mixture with the weight of ice cream in a suitable calibrated container. Overrun refers to the percentage of air contained within the ice cream and provides an indication of consistency, softness and appearance.

[0024] Melting or melt rate is generally quantified by measuring the mass that drips from the product through a mesh screen as a function of time at a given temperature, and correlates with the ice crystal properties and therefore the structure of the ice cream.

[0025] In panel testing, specially qualified panelists with specific experience evaluate the general acceptability of a product, i.e. its quality based on various aspects of sight, smell and taste.

[0026] The production of ice cream takes place through a complex interplay between different process parameters and conditions. Indeed, structural and physical changes can affect the sensory and melting properties, which means that it is not easy to ascertain exactly when the product is "ready".

[0027] Prior art machines implement various strategies that have been developed to monitor the production process and define when the ideal consistency has been reached and the ice cream is ready to be removed from the batch freezing cylinder.

[0028] In the first type of prior art machines, a timer is used to control the production time. This type of machine has the obvious drawback that not all ice cream mixtures can obtain the same consistency and viscosity in the same amount of time, and as a result, the optimization of the ice cream production depends heavily on the experience of the operator who uses the machine and manages the recipe.

[0029] In order to overcome this apparent drawback, the second type of machine for producing ice cream is equipped with a product temperature sensor so as to be able to ascertain, based on this parameter, when the ice cream has reached the optimum consistency.

[0030] It should be noted that product temperature is a quantity that is difficult to determine uniformly within the processing chamber and, moreover, it is not possible to specify an optimum value for all types of ice cream.

[0031] That means, again, that the optimization of ice cream production depends heavily on the experience of the operator who uses the machine and manages the recipe.

[0032] A third type of machine includes a device configured to detect the resistance that the ice cream offers to the rotation of the agitator (thereby indirectly measuring the hardness and viscosity of the ice cream).

[0033] This device can be mechanical or electronic, i.e. based on the power input of the agitator motor.

[0034] In such an electronic device, current and voltage values ​​are processed to provide an indirect measurement of the torque delivered by the agitator drive motor.

[0035] In yet another prior art solution, the ice cream machine is equipped with a level sensor that can detect the presence of the mixture inside the batch freezing chamber and can perform electrical conductivity measurements to verify if the ice cream has reached the appropriate consistency.

[0036] The principle on which this solution is based is the fact that as the temperature of the batch freezing chamber decreases, the moisture present in the mixture solidifies during the process, reducing the concentration of the solutes and thus the electrical conductivity which can be detected by conductance measurements.

[0037] The applicant has found that the machines for producing liquid or semi-liquid food products existing up to now, whatever their type, are not able to act substantially in real time on the ice cream production process currently applied, in order to overcome the following drawbacks: - long process times, - the inability to batch freeze unbalanced mixtures (i.e. mixtures in which the ratio of certain ingredients, e.g. sugar, moisture or fat, differs from the optimum ratio for the process); - Heat exchange problems with mixtures containing a high fat content; - the possibility of ice cream lumps forming inside the mixer, - Reproducibility issues in defining the exact consistency.

[0038] Furthermore, with the machines currently available for producing liquid or semi-liquid food products, it is not possible to produce an end product of optimum quality in terms of ingredient balance under all mixing conditions.

[0039] The need is therefore particularly keenly felt in the industry concerned for a machine for processing liquid or semi-liquid products, in particular ice cream, which is able to complete the batch freezing process in an optimal manner and at the same time produce a product of optimal quality. Summary of the Invention

[0040] The object of the present invention is to meet the above-mentioned needs and overcome the above-mentioned drawbacks of the prior art by providing a machine for processing liquid and / or semi-liquid products, in particular ice cream, which makes it possible to optimize the batch freezing process and at the same time produce a product of optimal quality, and a method for controlling said machine.

[0041] Another object of the invention is to provide a machine for processing liquid and / or semi-liquid products, and a method for controlling said machine, which allows batch freezing of ice cream in the shortest possible time, without any adverse effect on the quality of the final product.

[0042] It is yet another object of the present invention to provide a machine for processing liquid, semi-liquid or semi-solid products, which is able to prevent potential problems during the processing cycle, and a method for controlling said machine.

[0043] Yet another object of the invention is to provide a machine for processing liquid, semi-liquid or semi-solid products, which makes it possible to accurately identify the right time to extract the ice cream, i.e. when the mixture has reached the optimum consistency.

[0044] The technical features of the present invention in accordance with the above-mentioned objects are clearly disclosed in the following claims, and its advantages will become more apparent in the following detailed description, taken in conjunction with the accompanying drawings which merely represent exemplary embodiments. [Brief description of the drawings]

[0045] [Figure 1] 1 shows diagrammatically a machine for producing a liquid, semi-liquid or semi-solid product according to the invention; [Diagram 2] 1 shows diagrammatically a machine for producing a liquid, semi-liquid or semi-solid product according to the invention; [Diagram 3] 2 shows diagrammatically details of a control unit of the machine of FIG. 1; [Figure 4] 2 shows a schematic representation of the control flow executed by the control unit of the machine of FIG. 1; [Diagram 5] 2 shows a schematic representation of the architecture of the control module of the machine 1 of the invention in an embodiment including a neural network. [Figure 6] 1 shows a schematic representation of the composition of some of the mixtures used during the training step. [Figure 7-9] 2 shows a schematic representation of the trends of some of the parameters during a processing cycle in the machine of the invention during the training step, thus when the control module has not yet been trained. [Figure 10] 2 represents diagrammatically the processing cycle in the machine of the invention during the training step, thus when the control module has not yet been trained. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0046] The invention provides a machine 1 for producing liquid, semi-liquid or semi-solid food products.

[0047] It should be noted that with the machine 1 it is mainly possible to produce ice cream and cold dessert products.

[0048] More preferably, the machine 1 is capable of producing ice cream, artisanal gelato and / or soft serve ice cream, sorbet, cream, granita.

[0049] Machine 1 is - a treatment vessel 2 for treating the base product, - a stirrer 3 adapted to mix the product inside the treatment vessel 2 of the base product and rotating inside the treatment vessel 2; a heat treatment system 4 configured to heat treat the products inside the container 2; a control unit 5 connected to and driving the heat treatment system 4 and the stirrer 3, at least one sensor S configured to detect operating parameters of the machine and / or the state of parts of the machine;

[0050] According to the invention, the control unit 5 comprises: a first artificial intelligence control module 6 configured to define a pre-trained classifier CL having at least one input (I1, I2, I3, I4, I5, I6, I7) and at least one output U11, the at least one input I1 being connected to a sensor S; a drive module 7 connected to the output U11 of the first control module 6 and to at least the heat treatment system 4 and / or the agitator 3 and which adjusts at least one operating parameter of the heat treatment system 4 and / or the agitator 3 as a function of the value of the at least one output U11 of the first control module 6.

[0051] According to one embodiment, the first control module 6 comprises a number of inputs (I1, I2, I3, I4, I5, I6, I7).

[0052] According to yet another embodiment, at least one output U11 of the first control module 6 represents a classification of the mixture of base products.

[0053] Essentially, the first control module 6 is configured to classify the mixture.

[0054] It should be noted that in one embodiment, the first control module 6 comprises a classifier CL that is based on a machine learning (artificial intelligence) solution.

[0055] It should be noted that in one embodiment, the first control module 6 comprises a neural network RN comprising a number of layers (S1, S2, S3, S4, S5, S6) and having a number of inputs (I1, I2, I3, I4, I5, I6, I7).

[0056] Without loss of generality, described in detail below is an embodiment in which the first control module 6 comprises a neural network RN.

[0057] However, other embodiments are described below which do not comprise a neural network RN as the first control module 6.

[0058] Preferably, according to another embodiment, the machine 1 comprises a dispenser 21 connected to the first container 2 for providing the product.

[0059] According to another embodiment, the machine 1 is preferably equipped with an interface U connected to the control unit 5 and enabling a user to send commands / settings to the control unit 5 .

[0060] Preferably, the machine 1 comprises an electric motor connected to the agitator 3 for driving the latter in rotation.

[0061] Preferably, the neural network RN comprises at least two hidden layers (preferably with ReLU or tanh or LeakyReLU activation functions) with a plurality of neurons, preferably more than 10 neurons.

[0062] Preferably, the last layer of neurons S6, ie the layer at the output of the neural network RN, has a sigmoid activation function (for two classes) or a softmax activation function (for more than two classes).

[0063] Regarding the activation functions of the neurons that make up the various layers (S1, S2, S3, S4, S5, S6), it is assumed that these activation functions are of the following types: - Sigmoid (compresses input values ​​between 0 and 1), - ReLU (Rectified linear unit), a linear activation function defined as f(x) = max(0,x), - softmax (preferably used in the output state: outputs sum to 1 and can advantageously represent probability distributions), - Tanh (hyperbolic tangent, which has the shape of an S-curve like a sigmoid function but compresses the input values ​​between -1 and 1, commonly used in hidden layers).

[0064] It should be noted that vessel 2 and agitator 3 together define a batch refrigeration unit.

[0065] A neural network RN, as is known, is a computational model whose purpose is to determine an output value on the basis of an input value.

[0066] The neural network RN is made up of multiple neuron layers (in the example of Figure 5, there are six layers, namely S1, S2, S3, S4, S5, and S6, it being understood that the number of layers may be more or less than the number shown), each of which performs a weighted sum operation on its input and passes the result to an activation function, which serves to keep it within a certain range or approximate it.

[0067] According to one embodiment, neurons in a hidden layer propagate their results to neurons in the next layer.

[0068] The result is given by a linear combination of the outputs (activations) of the last hidden layer.

[0069] According to one embodiment, at least some of the neural network layers S2, S3, S4, S5 are hidden layers and the neural network is a deep neural network.

[0070] According to another embodiment, the neural network RN is a recurrent neural network (RNN).

[0071] According to yet another embodiment, the neural network RN is a convolutional neural network (CNN).

[0072] Alternatively, the neural network RN is any variant of the RNN and CNN types, or any hybrid version given by a combination of RNN and CNN and their variants.

[0073] According to yet another embodiment, the neural network RN is assisted by a component that performs what is known as "attention", according to the known Transformer architecture.

[0074] It should be noted that in this case, the first control module 6 comprises a neural network RN and other artificial intelligence sub-modules configured to perform what is known as "attention", according to the known Transformer architecture.

[0075] In experimental tests with neural network RN, carried out by the Applicant on machine 1, the best results in terms of performance were observed with networks of the CNN type.

[0076] The structure of the neural network RN, shown diagrammatically in FIG. 5, will be explained in more detail below, in particular with regard to its inputs and outputs.

[0077] It should be noted that the number of neurons per layer, as well as the number of layers, can be varied without departing from the scope of the inventive concept.

[0078] Preferably, the network consists of more than four layers.

[0079] Focusing on the architecture of the control unit 5, the following should be noted:

[0080] According to one embodiment, the machine 1 comprises an additional second control module 8 configured to perform processing based on data of at least one sensor S and / or to detect anomalies.

[0081] The second module 8 may be used with a different kind of first control module 6, i.e. a first artificial intelligence control module 6 that is not composed of only one neural network. Thus, what has been described with respect to the second module 8 also explicitly applies to a first artificial intelligence control module 6 that is not composed of only one neural network.

[0082] The second module 8 thus receives data from the at least one sensor S.

[0083] The second control module 8 has at least a first output U21 connected to the drive module 7 so as to be able to transmit control information to the drive module 7.

[0084] The control information is information derived by the second control module 8 based on data provided from at least one sensor S as input to the second control module 8 .

[0085] Furthermore, according to another embodiment, the drive module 7 is configured to adjust at least one operating parameter of the heat treatment system 4 and / or the agitator 3 as a function of the value of at least one output U11 of the first control module 6 and / or the value of the output U21 of the second control module 8.

[0086] This concept is explained in more detail below.

[0087] Basically, the drive module 7 can operate in two different modes, i.e. adjust at least one operating parameter of the heat treatment system 4 and / or the agitator 3 as a function of: 1) only the value of output U11 of the first control module 6, 2) It can be adjusted as a function of the values ​​of both the first control module 6 and the second control module 8 outputs (U11, U21).

[0088] Obviously, the present invention uses a first control module 6 and the value of only the output U21 of the second control module 8 for the drive module 7 can only be used in certain special cases and at an initial stage, for reasons that will be explained further below.

[0089] It should therefore be noted that during normal machine operation, the drive module 7 may drive the heat treatment system 4 and / or the agitator 3 based on only one output (U11 or U21) or based on both.

[0090] It should also be noted that the values ​​of the output U11 of the first control module 6 and the output U21 of the second control module 8 already take into account (i.e. are a function of) the value of the signal of at least one sensor S.

[0091] It should be noted that in this disclosure, the term "output" is used to mean a software variable or parameter that provides a value (numeric and / or alphanumeric).

[0092] In fact, the control unit 5 is made up of hardware and software elements which define the aforementioned modules (6, 7, 8).

[0093] According to another embodiment, the second control module 8 comprises a second output U22 connected to an input I2 of one of the inputs of the first layer of the layers S1, S2, S3, S4, S5, S6 of the neural network RN for transmitting control information to the first neural network control module 6.

[0094] Essentially, according to this embodiment, the second control module 8 is coupled to the first control module 6 for exchanging information therewith.

[0095] In particular, the output U22 of the second control module 8 constitutes the input I2 of the first control module 6.

[0096] With specific reference to the case in which the first control module 6 is a neural network RN, it should be noted that the output U22 of the second control module 8 constitutes the input I2 of the first layer S1 of the layers S1, S2, S3, S4, S5, S6 of the neural network RN.

[0097] According to one embodiment, the at least one sensor S is configured to detect a value relating to the power supply of a drive motor 20 connected to the agitator 3 for driving the agitator 3 in rotation.

[0098] In other words, the sensor S can detect the current value and / or the voltage value and / or the power value of the motor 20 that drives the agitator 3.

[0099] The sensor S may be of any type that is compatible and capable of detecting one or more of the current and / or voltage and / or power of the motor 20 that drives the agitator 3 .

[0100] It may also be integrated into an additional control and drive unit (driver) for the drive motor of the stirrer 3 .

[0101] According to another embodiment, the second control module 8 is configured to derive a parameter related to the consistency and / or viscosity of the product being processed in the vessel 2 from a value related to the power supply of the motor 20 driving the agitator 3.

[0102] Preferably, the parameters relating to the consistency and / or viscosity of the product being processed in the vessel 2 are comprised between a minimum value corresponding to the product to be processed and a maximum value corresponding to the final product.

[0103] Preferably, the parameters relating to the consistency and / or viscosity of the product being processed in vessel 2 are values ​​between 0 and 100, where 0 corresponds to the product at the start of the cycle, i.e. when the base ingredient has not yet been batch frozen, and 100 corresponds to the finished ice cream product at the end of the process.

[0104] According to yet another embodiment, the drive module 7 is configured to adjust at least one operating parameter of the thermal treatment system 4 and / or the agitator 3 as a function only of the value of the output U21 of the second control module 8 for a preset length of time at the start of a treatment cycle.

[0105] Essentially, according to this embodiment, for a preset length of time at the start of a processing cycle (preferably less than 30 seconds), the drive module 7 is active only based on the value of the output U21 of the second control module 8.

[0106] Advantageously, the fact that only the value of the output U21 of the second control module 8 is used as the basis for the drive module 7 makes it possible to capture multiple samples of the signal of the sensor S, which, together with other information, constitute the values ​​for the input of the neural network RN. That is to say, only at the moment when multiple input values ​​have been sampled is it possible to use the first control module 6 (by itself or in combination with the second module 8) to transmit control information to the drive module 7.

[0107] It should be noted that in practice the first control module 6 must have as input a certain number of values ​​captured at different times so as to be able to calculate the value of at least one output U11.

[0108] Essentially, when a sufficient predetermined amount of information (i.e., number of data samples) is statistically available, it is possible to provide input data to the control module 6 so that it can calculate at least one output U11 (so that the drive module 7 can use the output U11 so that it can adjust the heat treatment system 4 and / or the agitator 3 as necessary).

[0109] In relation to this aspect, the following should be noted.

[0110] It should be noted that the first control module 6 preferably allows to perform a fast classification by iterative refinement. In other words, without waiting for the end of a processing cycle (unlike what is performed by most prior art classifiers), an early classification based on different time intervals is performed, and the first control module 6 can perform the classification at different time intervals as the number of data samples captured by the sensor S gradually increases. This has the advantage of performing an early classification at the beginning of the batch freezing process, giving the machine the discretion to correct possible mixture imbalances, for example by implementing a slower but more "robust" production process.

[0111] Essentially, as more time passes after the start of the process, and therefore as the number of data samples captured by the sensor S and applied as input increases, the first control module 6 can increase the reliability of its classification, although it is active substantially from the start of the cycle and does not need to wait for the end of the cycle.

[0112] According to another embodiment, the thermal treatment system 4 is a thermodynamic system.

[0113] Preferably, the thermal treatment system 4 is a thermodynamic type system.

[0114] In other words, the heat treatment system 4 is composed of a circuit that uses a heat exchange fluid, and includes an evaporator 10 , a compressor 11 , a throttling element 12 and a condenser 13 .

[0115] Preferably, the circuit implements a thermodynamic vapor compression cycle.

[0116] The heat exchange fluid is a refrigerating fluid.

[0117] According to another embodiment, the drive module 7 is configured to adjust one or more of the following operating parameters: the driving force of the compressor 11, - the opening of the throttle element 12, - rotation speed of the agitator 3, - the processing cycle time (i.e. the moment when the processing cycle stops).

[0118] It should be noted that according to the above described architecture, the settable cycle for controlling the machine 1 can use the output of the control module 6 of the first neural network as input for the drive module 7 .

[0119] As will be made clearer below, when describing a particular mode of training the neural network, it is advantageously possible to train the machine 1 to effectively complete an ice cream production cycle even with mixtures that are not perfectly balanced, i.e. mixtures containing one or more ingredients present in proportions that deviate from their optimal (theoretical) proportions.

[0120] According to one embodiment, the at least one sensor S is configured to detect at least one of the following parameters: the operating pressure of the heat exchange fluid at a given point in the circuit of the thermal treatment system 4, the operating temperature of the heat exchange fluid at a given point in the circuit of the thermal treatment system 4, - operating temperature of vessel 2, the operating temperature of the heat exchange fluid leaving the evaporator 10 associated with the vessel 2, the operating temperature of the heat exchange fluid entering the evaporator 10 associated with the vessel 2, - the operating pressure of the heat exchange fluid leaving the compressor 11, - the operating pressure of the heat exchange fluid leaving the evaporator 10, - processing cycle time, - the temperature of the surrounding air in the vicinity of the machine 1.

[0121] More generally, it should be noted that the machine 1 may comprise a number of sensors S, each configured to detect one of the aforementioned parameters.

[0122] In particular, the machine 1 may therefore be equipped with a pressure sensor S configured to detect the operating pressure of the heat exchange fluid at a given point in the circuit of the thermal treatment system 4 .

[0123] According to another embodiment, the machine 1 may include a temperature sensor S configured to detect the operating temperature of the heat exchange fluid at a given point in the circuit of the thermal treatment system 4 .

[0124] According to yet another embodiment, the machine 1 may comprise a sensor S configured to detect the operating temperature of the heat exchange fluid exiting the evaporator 10 (associated with the vessel).

[0125] According to yet another embodiment, the machine 1 may be equipped with a sensor S configured to detect the operating temperature of the container 2 .

[0126] According to yet another embodiment, the machine 1 may include a sensor S configured to detect the operating temperature of the heat exchange fluid exiting the evaporator 10 associated with the vessel 2 .

[0127] According to yet another embodiment, the machine 1 may include a sensor S configured to detect the operating temperature of the heat exchange fluid flowing into the evaporator 10 (associated with the vessel 2).

[0128] According to yet another embodiment, the machine 1 may include a sensor S configured to detect the operating pressure of the heat exchange fluid exiting the compressor 11 .

[0129] According to yet another embodiment, the machine 1 may include a sensor S configured to detect the operating pressure of the heat exchange fluid exiting the evaporator 10 .

[0130] According to another embodiment, the machine 1 may be equipped with a sensor S configured to detect the processing cycle time (such a sensor could also be a software variable of the control unit 5).

[0131] According to yet another aspect, the at least one sensor S is configured to detect an acceleration value of at least a portion of the machine.

[0132] According to another embodiment, the inputs I1, I2, I3, I4, I5, I6, I7 include one or more of the following inputs: a first input (I1) relating to (i.e. representing) a value representative of the consistency or viscosity of the product being processed in vessel 2, a second input I2 related to (i.e. representing) a value representative of the operating temperature of vessel 2, a third input I3 related to (i.e. representing) a value representative of the operating temperature of the heat exchange fluid entering the evaporator 10 associated with the vessel 2; a fourth input I4 related to (i.e. representing) a value representative of the operating temperature of the heat exchange fluid leaving the evaporator 10 associated with the vessel 2; a fifth input I5 related to (i.e. representing) a value representative of the operating pressure of the heat exchange fluid leaving the compressor 11; a sixth input I6 related to (i.e. representing) a value representative of the operating pressure of the heat exchange fluid leaving the evaporator 10; a seventh input I7 which is related to (i.e. represents) a value representative of the opening of the throttling element 12.

[0133] If the first module 6 consists of a neural network RN, these inputs are the inputs of a first of the layers S1, S2, S3, S4, S5, S6 of the neural network RN.

[0134] It should be noted that these inputs I1, I2, I3, I4, I5, I6, I7 were selected because they were considered to be the most important parameters for controlling and monitoring the processing cycle of the ice cream industry, i.e. of a machine 1 suitable for producing ice cream.

[0135] It will be understood that this specification covers embodiments in which the machine 1 has a neural network RN with only one input I1, or with N of the above mentioned inputs selected in any manner from two to seven.

[0136] In particular, it is emphasized that, among the above-mentioned inputs I1, I2, I3, I4, I5, I6, I7, those which have the most importance or weight (i.e., influence) in determining at least one output U11 of the first control module 6 (determined based on the training tests performed) are the first input I1 and the sixth input I6.

[0137] According to another embodiment, the first control module 6 is configured to classify the mixture being processed in the vessel 2 based on a balance of the components of the mixture, and at least one output U11 of the first control module 6 represents the classification of the mixture being processed in the vessel 2.

[0138] In other words, according to this embodiment, at least one output U11 represents the probability that the mixture being processed is balanced with respect to one or more base components or overall base components.

[0139] As is known, balanced ingredients are essential to obtain an optimal quality ice cream. Information about the (non-)balanced ingredients is therefore essential to be able to take corrective action when necessary (i.e. to control the heat treatment system 4 or the mixer 3 accordingly) so that the base mixture can be processed in an optimal manner.

[0140] Advantageously, according to another aspect, classifying the type of unbalanced mixture is fundamental to being able to identify the optimum freezing point, thus increasing the likelihood of completing the batch freezing process.

[0141] Firstly, for the operation of the control unit it is sufficient to ascertain whether the mixture being processed is balanced or unbalanced with respect to one or more base components or the entire base component.

[0142] According to another embodiment, the first control module 6 is configured to perform a multi-class classification of the mixture being processed in the vessel 2 based on the balance of components in the mixture.

[0143] According to this embodiment, the first control module 6 is configured with a number of outputs U11, U12, U13, U14, U15, U16, U17, U18, U19.

[0144] According to this embodiment, when the first control module 6 is composed of a neural network RN, the last layer of the layers S1, S2, S3, S4, S5, S6 of the neural network RN is composed of a number of outputs U11, U12, U13, U14, U15, U16, U17, U18, U19.

[0145] Each output U11, U12, U13, U14, U15, U16, U17, U18, U19 relates to (i.e. represents) a mixture class in which the base components are either balanced or unbalanced in a predetermined manner.

[0146] In a basic embodiment, the first control module 6 has only one output U11, which represents the degree to which one or all of the base components are balanced or unbalanced.

[0147] More precisely, each output U11, U12, U13, U14, U15, U16, U17, U18, U19 provides a numerical value between a minimum and a maximum value (preferably between 0 and 1) representing the probability that the product being processed in vessel 2 corresponds to a pre-set class.

[0148] According to one embodiment, the aforementioned outputs U11, U12, U13, U14, U15, U16, U17, U18, U19 of the first control module 6 are taken together to relate to (i.e., represent) two or more of the following classes of mixtures being processed: - the product being treated in vessel 2 is balanced with respect to at least one base component; - the product being treated in vessel 2 is unbalanced with respect to its water content; - the product being processed in vessel 2 is unbalanced in terms of fat, - the product being processed in vessel 2 is unbalanced with respect to sugars.

[0149] According to a particular embodiment, the aforementioned outputs U11, U12, U13, U14, U15, U16, U17, U18, U19 of the first control module 6 relate to (i.e., represent) two or more of the following classes of mixtures being treated: - the product being treated in vessel 2 is balanced with respect to at least one base component; - the product being treated in vessel 2 is unbalanced with respect to its water content; - the product being processed in vessel 2 is unbalanced with an excess of fat, - the product being processed in vessel 2 is unbalanced due to a lack of fat, - the product being processed in vessel 2 is unbalanced due to a lack of sugars, - the product being processed in vessel 2 is unbalanced due to an excess of sugars.

[0150] According to one embodiment, the control unit 5 is configured to transmit data to at least one input I1 of the first control module 6 at preset time intervals, preferably less than 30 seconds (more preferably less than 15 seconds, or advantageously less than 5 seconds).

[0151] The above is a description of the structure of the control unit 5, particularly with regard to the first control module 6.

[0152] Now, what is described below is the training procedure of the neural network RN, i.e. how the neural network described above was trained.

[0153] The following is applicable to the training of a first control module 6 composed of a neural network RN, but it should be taken into account that some aspects of the procedure described, in particular with regard to the type of mixture used, are also applicable to a control module 6 composed of another type of artificial intelligence classifier CL.

[0154] Training a neural network RN means finding the weights of its activation functions to give the best possible performance.

[0155] Generally speaking, training a neural network involves adjusting the weights of the connections between neurons in order to minimize the prediction error of the neural network model.

[0156] As is known, the process of training a neural network takes place in several steps: -1) Weight initialization: the weights of the connections between neurons are initialized with random values; -2) Feedforward: An input is presented to the neural network, which processes the input by passing it through various neuronal layers, each of which has a weighted activation function of the input; - 3) Calculation of the error: the error is calculated by comparing the output generated by the neural network with the expected value; -4) Backpropagation: This is the step in which the error is propagated back through the neural network and the weights of the connections between neurons are modified to reduce the error.

[0157] Steps 2 to 4 are repeated a fixed number of times or until the error reaches a predefined tolerance.

[0158] Preferably, after being trained, the neural network RN is validated on an additional data set (test data) to evaluate its performance on new data.

[0159] To train the neural network RN, the machine 1 was operated, i.e. processing cycles were performed, with different types of mixtures fed into the container 2, being balanced to different degrees, i.e. with different classifications in terms of the balance of the components, as described in more detail below.

[0160] For example, for one type of mixture, whether balanced or unbalanced with respect to its base components, multiple tests are performed (preferably more than five for each type of mixture), and during training, outputs associated with the desired class of that mixture are assigned a probability of 1, and outputs associated with the classes of the other mixtures are assigned a probability of 0.

[0161] The mixtures used for training (considering that for each type of mixture, several batches were used to train machine 1 on slight variations in the mixture's composition) were: - a balanced mixture, - An unbalanced mixture of excess fat, -An unbalanced mixture with too much sugar (sucrose), - An unbalanced mixture lacking sugar (sucrose), -An unbalanced mixture with too much water.

[0162] More precisely, the following nine mixtures were used for training: - a balanced mixture, - an unbalanced mixture with excess fat (preferably cream) by a first predetermined percentage (preferably corresponding to three times the optimal balance), - an unbalanced mixture with excess fat (preferably cream) by a second predetermined percentage (preferably corresponding to 4.4 times the optimal balance), - an unbalanced mixture with an excess of sugar (sucrose) by a first predetermined percentage (preferably corresponding to +1 / 3 of the optimal balance), - an unbalanced mixture with an excess of sugar (sucrose) by a second predetermined percentage (preferably corresponding to +2 / 3 of the optimal balance), - an unbalanced mixture with a third predetermined percentage of excess sugar (sucrose) (preferably corresponding to 100% of the optimal balance); - an unbalanced mixture lacking sugar (sucrose) by a first predetermined percentage (preferably corresponding to -1 / 3 of the optimal balance), - an unbalanced mixture lacking sugar (sucrose) by a second predetermined percentage (preferably corresponding to -2 / 3 of the optimal balance); - An unbalanced mixture with excess water (preferably corresponding to +10% relative to the optimum balance) was used.

[0163] FIG. 6 shows, for example, the composition of the test mixtures (in numerical percentages) for the following mixtures, respectively: Balanced mixture (first row), An unbalanced mixture with too much water (row 2); An unbalanced mixture (row 3) with three times more fat (cream) than optimal balance. An unbalanced mixture (row 4) with 4.4 times more fat (cream) than optimal balance. An unbalanced mixture (row 5) that is 2 / 3 of the optimal balance and lacks sugar (sucrose); An unbalanced mixture (row 6) that is one-third below the optimum balance and lacks sugar (sucrose); An unbalanced mixture (line 7) with too much sugar (sucrose) - 1 / 3 of the optimal balance. An unbalanced mixture (line 8) with 2 / 3 of the optimum balance and too much sugar (sucrose) This is an unbalanced mixture (line 9) with too much sugar (sucrose), 100% more than the optimal balance.

[0164] The acronym "LMS" stands for lean milk solids.

[0165] It should be noted that greater accuracy may be obtained by using more mixes, for example multiple unbalanced mixes that are over-hydrated and / or multiple unbalanced mixes that are under-fat.

[0166] As can be easily understood, a test mix was placed in the container 2 and the machine 1 was started to perform an ice cream making cycle on the test mix.

[0167] Thus, during this cycle, the neural network RN receives as input the values ​​from the sensor S, i.e. at its input as described above, and at the output of the neural network RN a label corresponding to the correct classification of the mixture being processed in the container 2 has been set.

[0168] It should be noted that the neural network RN is essentially configured to act as a classifier.

[0169] Preferably, training is performed for a number of epochs between 100 and 600, and even more preferably less than 250.

[0170] Preferably, the training is done via the Adam optimizer.

[0171] More preferably, training is performed in batches of 12 samples each.

[0172] In tests carried out, it was observed that sugars have the effect of lowering the freezing point of the mixture being treated.

[0173] As a result, mixtures that are only lightly sweetened (low in sugar) tend to freeze faster and risk forming ice blocks before the desired consistency is reached, while mixtures with more sugar tend to freeze at a lower temperature and therefore remain more liquid.

[0174] It should be noted that by understanding the type of mix, and in particular the sugar balance, through classification of the mix via the first control module 6, as will be described in more detail below, it is possible to implement measures aimed at completing the batch refrigeration cycle in an optimal manner as a function of the type of product being processed.

[0175] In practice, as mentioned above, in the case of sugars for example, understanding through the first module 6 whether the mixture is unbalanced (preferably, but not necessarily, in excess or deficiency) makes it possible to take appropriate corrective measures.

[0176] FIG. 7 shows the trends of some parameters captured by sensor S during a manufacturing process using a balanced mixture.

[0177] More precisely, in the graph shown in FIG. 7, R1 is the trend over time (X-axis) of a parameter related to the consistency and / or viscosity of the product being processed in vessel 2, obtained from values ​​related to the power of the motor of agitator 3, this parameter being represented by values ​​on the y-axis, varying from 0 to 100.

[0178] In FIG. 7, R2 denotes the temperature at the inlet side of the evaporator 10, R3 denotes the temperature at the outlet side of the evaporator 10, and R4 denotes the difference between these two temperatures.

[0179] As can be inferred from the graph, the parameters related to the consistency and / or viscosity of the product being processed in vessel 2 tend to increase during processing until they reach a maximum value (corresponding to 100), when the product is ready.

[0180] Experiments have shown that parameters related to the consistency and / or viscosity of the product being processed in vessel 2 have different trends as a function of the type of mixture regarding the balance of one or more components or as a function of anomalies that may occur during processing (e.g. frozen chunks).

[0181] 8, derived during supervised training by sampling trends in parameters related to the consistency and / or viscosity of the product being processed in vessel 2 for different mixtures. - Mixture M1 is unbalanced with 3 times the amount of cream, - Mixture M2 is unbalanced with 4.4 times the amount of cream, -Mixture M3 is unbalanced with 10% water. - It is noted that mixture M4 is balanced.

[0182] As can be observed, for different types of mixtures in terms of balance, the parameters related to the consistency and / or viscosity of the product being processed in container 2 reach a consistency corresponding to the final product (i.e. substantially equal to 100) at different times.

[0183] As can be noted, the unbalanced mixtures (M1, M2, M3) reach maximum consistency in a shorter time than the balanced mixture (M4).

[0184] As can be easily understood from this example, therefore, by knowing the classification of the mix via the control module 6, the drive module 7 is able to control the machine in an optimal manner so as to complete the batch refrigeration cycle with the highest quality product, regardless of the balance of ingredients, based on a combination of the output of the first control module 6 (classification of the mix) and the output of the second control module 8 (values ​​of parameters related to product consistency and / or viscosity).

[0185] 9, derived during supervised training by sampling trends in values ​​of the pressure of the heat exchange fluid of the thermodynamic system 4 downstream of the compressor 11. - Mixture M1 is unbalanced with 3 times the amount of cream, - Mixture M2 is unbalanced with 4.4 times the amount of cream, -Mixture M3 is unbalanced with 10% water. - It can be observed that mixture M4 is balanced.

[0186] As is evident from FIG. 9, the pressure of the heat exchange fluid in the thermodynamic system 4 downstream of the compressor varies for the various mixtures along the cycle.

[0187] For the first control module 6, this input therefore constitutes a value which can certainly be used for classification purposes.

[0188] The graph also shows that for the water unbalanced mixture M3, the pressure during the treatment cycle is lower than for the water balanced mixture M4.

[0189] FIG. 10 shows an unbalanced mixture (B) compared to a balanced mixture (A). - Curve X1 represents the consistency or viscosity of the mixture measured over time. Curve X2 represents the opening of the throttle element 12. Curve X3 represents the temperature of the heat exchange fluid leaving the evaporator 10 of the thermodynamic system 4. Curve X4 represents the temperature of the heat exchange fluid entering the evaporator 10 of the thermodynamic system 4.

[0190] In FIG. 10, the zones labeled P1 and P2 correspond to the shut-down of compressor 11, which corresponds to the end of the batch refrigeration cycle.

[0191] It should be noted that if ice cream is not removed from the container 2 while the compressor 11 is stopped, the machine 1 will restart the compressor 11 and the agitator 3 (zones P3 and P4, which are not relevant here) in order to keep the ice cream at the required temperature.

[0192] It should be noted that FIG. 10 corresponds to tests carried out on the machine 1 during training of the control module 6 of the first neural network.

[0193] With reference to the unbalanced mixture "B" and the line LO representing the optimum consistency, it is noted that the consistency curve X1, i.e. the viscosity of the mixture, does not reach the optimum consistency LO and therefore the ice cream produced is of sub-optimal quality.

[0194] In such a case, the presence of a special control cycle by the first control module 6 makes it possible for the drive module 7 to stop the treatment cycle at point P5, which in the case of an unbalanced mixture "B", corresponds to a consistency definitely thicker than the consistency at point P2.

[0195] The above is clear evidence of the increased technical performance made possible by the machine 1 of the invention.

[0196] As non-limiting examples, the control actions implemented through the drive module 7 are described below.

[0197] Firstly, it is specified that the expression "adjusting at least one operating parameter relating to the heat treatment system (4) and / or the agitator (3)" is used to mean that the drive module 7 acts on and controls the functioning of the heat treatment system 4 and / or the agitator 3 (in particular the electric motor coupled to the agitator 3).

[0198] It should be noted that the drive module 7 can also act on other parts of the machine 1 .

[0199] It should also be noted that the drive module 7 can advantageously act on one or more of the following components of the thermal treatment system 4: the compressor 11 and / or the throttling element 12 and / or the condenser 13 (cooling fan and / or cooling water pump).

[0200] Described below are some cases that may arise during operation of the machine 1 as a result of the composition of the mixture and / or the specific process conditions.

[0201] In this context, a method is described in which the control unit 5 adjusts the operating parameters for the heat treatment system 4 and / or the agitator 3.

[0202] Certain conditions during processing can result in a frozen lump of ice cream in the center of the mixer 3.

[0203] The control module 8 may be configured to detect potential anomalies of this kind, i.e. including frozen ice cream chunks in the center of the mixer 3, by one or more of the following sensor signals: - a reduction in values ​​related to the consistency of the product, and / or increased superheat at the suction end of the refrigeration compressor.

[0204] Under such circumstances, if the first control module 6 classifies the mixture as an unbalanced mixture (preferably fatty and unbalanced if the classification is multi-class), the drive module 7 may be configured to command one or more of the following actions (to allow the batch refrigeration cycle to be completed in an optimal manner): - increasing the opening of the pressure reducing element (throttle valve) to reduce superheat at the inlet of the compressor 11; - driving the motor of the stirrer 3 intermittently, preferably in pulse mode (preferably at high speed); - Reversing the direction of rotation of the motor of the stirrer 3.

[0205] Advantageously, thanks to these actions, the processing cycle can be completed. It should be noted that the fact that the first control module 6 has classified the mixture as an unbalanced mixture advantageously confirms the anomaly detected by the drive module 7. Indeed, experiments have shown that an unbalanced mixture generally tends to result in the formation of a frozen ice cream lump in the middle.

[0206] As such, the corrective action taken can allow the processing cycle to be effectively completed, resulting in the production of an optimum quality product, regardless of whether the base components are balanced or not.

[0207] In another example, under certain conditions, a layer of ice may form on the interior walls of the container 2 .

[0208] The control module 8 may be configured to detect potential anomalies of this kind, including a layer of ice on the inner wall of the container 2, by one or more of the following sensor signals: - the change in temperature at the inlet and outlet of the evaporator 10 (i.e. the value differs from the expected, desired or previous value).

[0209] Under such circumstances, if the first control module 6 classifies the mixture as an unbalanced mixture (preferably unbalanced with a sugar and / or fat deficiency if the classification is multi-class), the drive module 7 may be configured to command one or more of the following actions (to allow the batch refrigeration cycle to be completed in an optimal manner): - adjusting (reducing) the driving force of the compressor 11; - activating a mode of injecting hot gas into the evaporator 10, i.e. a mode of heating the evaporator 10;

[0210] It should be noted that in order to inject hot gas into the evaporator 10 to heat it, i.e., to operate the evaporator 10 as an exchanger that supplies heat (heats) instead of absorbing heat (cools) for a certain length of time, the heat treatment system 4 must be configured to allow the evaporator 10 to operate in two alternative modes. - (1) A configuration known as the high-temperature gas cycle (not shown, as it is known to anyone skilled in the art) - or (2) act like a heat pump (not shown, as known to anyone skilled in the art).

[0211] In these configurations, the thermal treatment system 4 comprises one or more valves operable by the control unit 5 to allow switching between a cooling mode of the evaporator 10 and an operating mode by injection of hot gas, i.e. by heating the evaporator 10. In another example, the product being processed may reach an optimal consistency in too short a time than the expected set time.

[0212] This generally means that the final temperature of the ice cream is not correct, i.e. it is not as cold as would be expected for good quality ice cream.

[0213] The control module 8 may be configured to detect potential anomalies of this kind, i.e. anomalies involving a consistency value reached in less than a predefined set time, by one or more of the following sensor signals: - product consistency signal (detected for example indirectly via the power input value of the motor 20 of the agitator 3) relative to the processing cycle start time, - a temperature signal of the process vessel 2 which differs from the expected time values ​​or time profile.

[0214] Under such circumstances, if the first control module 6 classifies the mixture as an unbalanced mixture (preferably unbalanced with less sugar and / or more fat if the classification is multi-class), the drive module 7 may be configured to command one or more of the following actions (to allow the batch refrigeration cycle to be completed in an optimal manner): - increasing the consistency value associated with the product, which corresponds to the end of the processing cycle (e.g. to be able to stop at a consistency value higher than the standard level), increasing the opening of the throttle element 12 (which advantageously allows the inner wall of the container 2 to become less cold and / or reduces the effort required of the stirrer 3 to scrape the ice off the surface of the inner wall of the container 2, which could cause the consistency value to change); by reducing the refrigeration capacity of the thermal system 4, in particular by reducing the power of the compressor 11.

[0215] In another example, the values ​​detected related to the consistency of the product, which, as is known, is one of the main parameters that affect the quality of ice cream products, can become low while the product is being processed.

[0216] The control module 8 may be configured to detect potential anomalies of this kind, including a decrease in product consistency relative to an expected value or trend over an expected (set) time, via one or more of the following sensor signals: A signal relating to the input of power (voltage and / or current) of the motor 20 of the stirrer 3.

[0217] Under such circumstances, if the first control module 6 classifies the mixture as an unbalanced mixture, the drive module 7 may be configured to command one or more of the following actions (to allow the batch refrigeration cycle to be completed in an optimal manner): reducing the refrigeration capacity of the thermal system 4, in particular reducing the power of the compressor 11, Injecting hot gas into the evaporator 10, i.e. heating the evaporator 10.

[0218] Regarding the configuration of the thermal system 4 to enable the injection of hot gas into the evaporator 10, what was stated above for the frozen ice cream mass in the center of the mixer 3 also applies in this case.

[0219] In another example, the values ​​detected related to the consistency of the product, which, as is known, is one of the main parameters affecting the quality of an ice cream product, and which may not change over time while the product is processed.

[0220] The control module 8 may be configured to detect potential anomalies of this kind, including product consistency that does not change over time, against an expected value or trend over a predicted (set) time, via one or more of the following sensor signals: - a signal regarding the input of power (voltage and / or current) of the motor of the stirrer 3, a temperature signal representative of the temperature of the refrigerating fluid entering and / or leaving the evaporator 10.

[0221] Under such circumstances, if the first control module 6 classifies the mixture as an unbalanced mixture, the drive module 7 may be configured to command one or more of the following actions (to allow the batch refrigeration cycle to be completed in an optimal manner): - adjusting (reducing) the refrigeration capacity of the thermal system 4, in particular reducing the power of the compressor 11, - Injecting hot gas into the evaporator 10.

[0222] Regarding the configuration of the thermal system 4 to enable the injection of hot gas into the evaporator 10, what was stated above for the frozen ice cream mass in the center of the mixer 3 also applies in this case.

[0223] In another example, the temperature of the heat exchange fluid flowing out of the evaporator 10 may drop suddenly, i.e., the value of the temperature of the heat exchange fluid flowing out of the evaporator 10 may drop below a threshold value in less than a preset time.

[0224] The control module 8 may be configured to detect such potential anomalies, including a drop in the temperature of the heat exchange fluid exiting the evaporator 10 below a pre-set value for less than a pre-set time, via one or more of the following sensor signals: - a signal regarding the input of power (voltage and / or current) of the motor of the stirrer 3, a temperature signal representative of the temperature of the refrigerating fluid entering and / or leaving the evaporator 10.

[0225] Under such circumstances, if the first control module 6 classifies the mixture as an unbalanced mixture, the drive module 7 may be configured to command one or more of the following actions (to allow the batch refrigeration cycle to be completed in an optimal manner): reducing the refrigeration capacity of the thermal system 4, in particular reducing the power of the compressor 11, - Injecting hot gas into the evaporator 10, i.e. operating the evaporator 10 in heating mode.

[0226] Regarding the configuration of the thermal system 4 to enable the injection of hot gas into the evaporator 10, what was stated above for the frozen ice cream mass in the center of the mixer 3 also applies in this case.

[0227] In another example, the processing time may be too long.

[0228] The control module 8 may be configured to detect potential anomalies of this type, i.e., anomalies involving processing for a period of time greater than a set or expected period of time, via one or more of the following sensor signals: - a processing time signal (obtainable, for example, via a timer on machine 1), a signal representative of the trend over time of the consistency of the product being processed in vessel 2, the pressure value of the refrigeration fluid downstream of the compressor 11, a temperature signal representative of the air temperature of the external environment close to machine 1, - the temperature signal at the outlet of the compressor 11.

[0229] Under such circumstances, if the first control module 6 classifies the mixture as an unbalanced mixture (if the classification is multi-class, particularly unbalanced with too much fat or sugar), the drive module 7 may be configured to command one or more of the following actions (to allow the batch refrigeration cycle to be completed in an optimal manner): - adjusting (increasing) the refrigeration capacity of the thermal system 4, in particular increasing the power of the compressor 11, In the case of an air condenser, this is to adjust (increase) the flow rate of cooling air going to the condenser 13; in the case of a water condenser, this is to adjust (increase) the flow rate of water going to the condenser 13.

[0230] Advantageously, as can be gathered from the examples given above, the machine 1 of the invention is able to quickly identify and correct the main anomalies that may have a negative effect on the quality of the ice cream, so that ice cream made from any kind of mixture, whether balanced or not with regard to one or more of its ingredients, can be handled with confidence, obtaining optimal results at all times and under all conditions.

[0231] Advantageously, the machine 1 of the invention is also able to estimate the right time to remove the ice cream and / or give the signal "ice cream ready", i.e. when the texture of the mixture is optimal.

[0232] Also, under certain conditions, the ice cream maker should consider using an intentionally unbalanced mixture (because certain recipes require putting into the container 2 certain ingredients, be they sugar, water or fat, in non-optimal proportions). Thus, under such conditions, the machine 1 is able to produce ice cream with optimal properties even though the mixture that has to be used is not perfectly balanced.

[0233] Thus, the machine 1 of the invention is able to produce products of optimum quality and at the same time offers the possibility of creating new products in order to expand the product range for the ice cream industry.

[0234] It should also be considered that the machine 1 of the invention advantageously allows the ice cream to be processed in the shortest possible time and allows the batch freezing cycle to be stopped when the ice cream is really ready, i.e. when its consistency and texture are just right in order to optimize the quality of the ice cream.

[0235] Furthermore, the machine 1 of the invention is advantageously able to prevent several problems that may arise during processing and to take prompt measures to solve them.

[0236] Furthermore, the machine 1 of the invention is extremely easy to use, as it does not require any special settings, and it is able to compensate for any mistakes that the operator may make when balancing the ingredients, still always producing ice cream of optimum quality.

[0237] According to another aspect, there is also defined a method for controlling a machine 1 for producing liquid or semi-liquid food products, said method comprising the following steps: a machine 1 for producing liquid or semi-liquid food products, - a treatment vessel 2 for treating the base product, - a stirrer 3 adapted to mix the product inside the treatment vessel 2 in which the base product is treated; a heat treatment system 4 configured to exchange heat with a treatment vessel 2 for treating a base product; providing a machine 1 having at least one sensor S configured to detect an operating parameter of the machine; providing a first artificial intelligence control module 6 configured to define a classifier having at least one input I1, I2, I3, I4, I5, I6, I7 and at least one output U11, the at least one output U11 being connected to at least one sensor S; training a first artificial intelligence control module 6 on the basis of the processing of different mixtures inside the container 2 to obtain a first trained control module 6; - feeding the mixture into the container 2 and starting the treatment cycle; - driving at least the heat treatment system 4 and / or the agitator 3 so as to adjust at least one operating parameter thereof as a function of the value of at least one output U11 of the first control module 6.

[0238] According to one embodiment, the first control module 6 comprises a neural network RN.

[0239] According to another embodiment, the at least one output U11 represents a classification of the mixture of base products, and the method includes a step of classifying, via the first control module 6, the mixture being processed in the vessel 2 into one or more pre-defined categories with respect to the balance of base components.

[0240] According to another embodiment, the inputs I1, I2, I3, I4, I5, I6, I7 of the first control module (6) consist of one or more of the following inputs: a first input I1 related to a value representative of the consistency and / or viscosity of the product being processed in the vessel 2, a second input I2 related to a value representative of the operating temperature of vessel 2, a third input I3 related to a value representative of the operating temperature of the heat exchange fluid entering the evaporator 10 associated with the vessel 2; a fourth input I4 related to a value representative of the operating temperature of the heat exchange fluid leaving the evaporator 10 associated with the vessel 2; a fifth input I5 related to a value representative of the operating pressure of the heat exchange fluid leaving the compressor 11, a sixth input I6 related to a value representative of the operating pressure of the heat exchange fluid leaving the evaporator 10; a seventh input I7, which is related to a value representative of the opening of the throttle element 12.

[0241] According to yet another aspect, the method includes monitoring a signal of at least one sensor S to identify an event corresponding to an anomaly in the manufacturing process.

[0242] According to this embodiment, the step of driving at least the heat treatment system 4 and / or the agitator 3 includes a step of driving at least the heat treatment system 4 and / or the agitator 3 to adjust at least one operating parameter thereof as a function of the mixture classification value corresponding to the at least one output U11 of the first control module (6) and of the mechanical process faults previously identified during the monitoring step.

[0243] That is, according to this embodiment, a combination of the identified process anomaly and the classification performed by the first control module 6 adjusts one or more control parameters.

[0244] According to another embodiment, the step of training the first control module 6 in a supervised mode, based on the processing, inside the container 2, of mixtures with different balances of base components, so as to obtain a first control module 6 trained to classify said mixtures of base products, comprises the following steps: - processing mixtures of the following types: - a balanced mixture with respect to the base components; - unbalanced mixtures with respect to the water content, - an unbalanced mixture with respect to fats, - processing a mixture unbalanced in terms of sugars; - during training, providing the first control module 6 with labels associated with the desired categories for each type of mixture being processed.

[0245] It should be noted that the label effectively corresponds to the type of mixture being processed, i.e., for a balanced mixture, the assigned label is used to set an output (e.g. U11) to a value that corresponds to the desired classification of the type of mixture being processed.

[0246] According to yet another embodiment, the step of driving at least the heat treatment system 4 and / or the agitator 3 to adjust at least one operating parameter thereof as a function of the classification value of the mixture comprises the following steps: - checking whether the identified mixture classification value corresponds to an abnormal state of the mixture, i.e. an unbalanced mixture; - giving instructions to the heat treatment system 4 and / or to the stirrer 3 also based on the result of the check.

[0247] According to another embodiment, the step of driving at least the heat treatment system 4 and / or the agitator 3 to adjust at least one operating parameter thereof comprises the following steps: - comparing the detected defects with the classification values ​​of the identified mixture; - giving instructions to the heat treatment system 4 and / or to the agitator 3 also based on the result of the comparison.

[0248] According to another embodiment, the step of driving at least the heat treatment system 4 and / or the agitator 3 to adjust at least one operating parameter thereof as a function of the classification value of the mixture comprises one or more instructions to: - varying the opening of a throttle element 12 forming part of the heat treatment system 4 (thereby adjusting the throttle pressure), - adjusting the power of the compressor 11 forming part of the heat treatment system 4, - adjusting the flow rate of cooling fluid, air or water, in the condenser 13 forming part of the heat treatment system 4, - switching the thermal treatment system 4 between a cooling mode for cooling the vessel 2 and a heating mode for heating the vessel 2, the heating mode including an operating mode in which the thermal treatment system 4 performs what is known as a hot gas cycle or a heat pump cycle by injection of hot gas, - adjusting the speed and direction of rotation of the agitator 3.

[0249] The control method of the present invention can prevent a number of problems that may occur during the processing of products, particularly ice cream, which may include, for example: - the ice block in the center of mixer 3, - Ingredients that have been frozen too much, - A final product of the wrong consistency and texture.

[0250] Essentially, the proposed control method advantageously allows the most effective corrective action to be implemented as a function of the characteristics of the product being processed, thus not only allowing potential problems to be dealt with quickly, but also ensuring that the product being manufactured is of optimal quality, regardless of whether the base mix is ​​well-balanced or not.

[0251] As pointed out at the beginning, ice cream is a very complex product, the production of which requires a very sophisticated control system, which is also based on the composition of the mixture itself. Advantageously, the proposed method makes it possible to control the ice cream production process in a very effective and reliable manner.

[0252] According to another embodiment, the machine 1 comprises a sensor S configured to detect the amount of base product placed in the treatment vessel 2 .

[0253] Alternatively, the machine 1 may be equipped with a user interface configured to detect the amount of base product placed in the treatment vessel 2 .

[0254] It should therefore be noted that one of the inputs (I1, I2, I3, I4, I5, I6, I7) of the neural network RN, for example an additional input (not shown in the drawing), may be related to the value of the quantity of the base product detected by the sensor S.

[0255] It should also be noted that, according to another embodiment, the training procedure may include training the machine to recognize different levels of the amount of base product inside the container 2. Indeed, in that way, the first control module 6 can more accurately classify the mixture even if the amounts dosed in the container 2 are different.

[0256] Described below is another embodiment of the first control module 6, also based on machine learning, but alternatively where the first control module 6 is a neural network.

[0257] Wherever possible, everything that has been said above in relation to the first control module 6 (either in general or with specific reference to the neural network RN) is also applicable to these other embodiments.

[0258] In one embodiment, the first control module 6 is comprised of one or more decision trees.

[0259] According to this embodiment, for example, the first control module 6 is configured with a Random Forest (RF) model.

[0260] A random forest model is a model composed of any number of decision trees. It should be noted that in such cases, the output class assignment, i.e., classification, is done by combining all the outputs (i.e., decisions or "votes") of all the decision trees to obtain a single result.

[0261] Essentially, the class that the first control module 6 provides as output is the one that gets the majority vote.

[0262] When training the first control module 6, what is called a "time series forest" is preferably used.

[0263] It should be noted that a time series of this kind is in fact defined by a sequence of (sampled) data values ​​obtained from at least one sensor S (or several sensors S).

[0264] "Time Series Forests" allows training a random forest model using features extracted from time series, e.g., mean, standard deviation, slope, rather than the raw data.

[0265] Alternatively, instead of time-series forests, other algorithms based on tree models can be used, for example XGBооst (XGB) or time-series bag-of-features.

[0266] Alternatively, the first control module 6 comprises a classification configured to define a time series distance measure and a kernel, such as for example k-nearest neighbors (k-NN) or a support vector machine (SVM), respectively.

[0267] In another embodiment, the first control module 6 is configured with an algorithm for learning a model of the "Support Vector Machine" type.

[0268] Advantageously, support vector machines are models used to perform binary classification, but can also be used for multi-class classification (by subdividing the first module 6 into interconnected sub-modules, each configured to perform binary classification).

[0269] In the case of binary classification, a first control module 6, preferably of the support vector machine type, receives as input a training data set of a number of input time series belonging to two classes together with their associated labels.

[0270] Essentially, the training procedure can take as input multiple time series, each associated with a particular class, which may contain balanced mixtures of one or more base components to different degrees, as already discussed above.

[0271] Training input dataset {(X t , y t )}, t=1,2,...,T, where y belongs to {0,1}, the training procedure makes it possible to construct a decision function f(X) that correctly classifies time series that are not included in the training time series.

[0272] During the training procedure, the support vector machine determines a hyperplane that separates the input data belonging to two classes.

[0273] If the training sequence is linearly separable, then the decision function f(X) will be a linear function called the "decision boundary" (a linear SVM classifier).

[0274] The decision boundary is found such that the margin, i.e. the distance between the decision boundary and the nearest data item, is as large as possible.

[0275] For many datasets, a hyperplane is unlikely to be a good classifier, so more geometric shapes are typically employed, mapping the training data into a higher dimensional space and finding a hyperplane in the new space.

[0276] In such a case, the first control module 6 consists of a kernel-based SVM: the classification of points is performed in a multidimensional space using a "kernel function" (typical choices are a polynomial kernel or RBF, i.e. radial basis function).

[0277] In the case of multi-class classification, the first control module 6 is composed of multiple binary SVM classifiers working synergistically, in particular, in the case of N classes, (N(N-1)) / 2 binary classifiers should be used.

[0278] In such a case, the first classification module 6 would therefore consist of multiple binary SVM classifiers.

[0279] This approach is called "one-vs-one multiclass classification", and according to this strategy, each trained SVM classifier is able to separate points into two different classes.

[0280] In another embodiment, the first control module 6 comprises a k-NN (k-nearest neighbor) classifier configured to "include" new samples in the training space before classifying them.

[0281] According to this embodiment, the input of the first control module 6 consists of the k training examples and the output is their class membership.

[0282] Therefore, it should be noted that according to this embodiment, the first control module 6 configured with a k-NN classifier does not include a step of training a model.

[0283] A point is classified by the majority vote of its neighbors, i.e. the class assigned to the point is the most common class among its "k-nearest neighbors".The basic idea in time series classification is that we need to examine the similarity between time series.

[0284] Although "majority voting" strictly requires a majority of more than 50%, this works mainly when there are only two classes: if there are more than two classes, say four classes, then a class label with more than 25% of the votes may be assigned.

[0285] According to another embodiment, the first classification module 6 is configured to implement what is known as a "bag-of-words" (dictionary-based) method.

[0286] According to this method, the time series is discretized into sequences of symbols, then words are extracted from these sequences using a sliding window, and finally the number of all words in the dictionary is counted.

[0287] The first classification module 6 is configured to perform the steps mentioned above.

[0288] In one embodiment, the first classification module 6 is configured to implement what is known as Symbolic Ensemble Approximation (SAX). According to this aspect, a time series is discretized into a sequence of symbols, after which sequence data mining methods are applied.

[0289] In another embodiment, the first classification module 6 is configured to perform a symbolic Fourier approximation: the representation of the sequence is generated by representing the signal as a discrete Fourier transform, thus resulting in a frequency domain representation rather than a time domain representation.

[0290] The above emphasizes that according to the present invention, the first control module 6 can implement different classification strategies (for classifying mixtures) without departing from the specific principles of the invention itself.

Claims

1. A machine (1) for producing liquid, semi-liquid or semi-solid food products, comprising: - a treatment vessel (2) for treating the base product, - an agitator (3) adapted to mix the base product inside the processing vessel (2) in which the product is processed; a heat treatment system (4) configured for heat exchange with said treatment vessel (2) for treating said base product; a control unit (5) connected to and driving said heat treatment system (4) and said stirrer (3), at least one sensor (S) configured to detect an operating parameter of the machine, The control unit (5) a first artificial intelligence control module (6) configured to define a pre-trained classifier (CL) having at least one input (I1, I2, I3, I4, I5, I6, I7) and at least one output (U11), said at least one input (I1) being connected to said sensor (S); a drive module (7) connected to said output (U11) of said first control module (6) and to at least said heat treatment system (4) and / or said agitator (3) and adapted to adjust at least one operating parameter of said heat treatment system (4) and / or said agitator (3) as a function of the value of said at least one output (U11) of said first control module (6).

2. 2. The machine (1) according to claim 1, wherein the first control module (6) comprises a plurality of inputs (I1, I2, I3, I4, I5, I6, I7).

3. 3. The machine (1) according to claim 1 or 2, wherein the at least one output (U11) of the first control module (6) represents a classification of the mixture of base products.

4. 4. The machine (1) according to any one of claims 1 to 3, wherein the first control module (6) is configured to perform the classification at a number of time intervals comprised between the start and end of the process, thereby performing what is known as "early classification" and gradually improving the accuracy of the classification by iteratively refining it at different time intervals.

5. 5. The machine (1) according to claim 1, wherein the first control module (6) comprises a neural network (RN) with a number of layers (S1, S2, S3, S4, S5, S6) and with a number of inputs (I1, I2, I3, I4, I5, I6, I7).

6. 6. The machine (1) according to any one of claims 1 to 5, further comprising a second control module (8) connected to the at least one sensor (S) and configured to process and / or detect defects based on data of the at least one sensor (S), the second control module (8) having at least a first control output (U21) connected to the drive module (7).

7. 7. The machine (1) according to claim 6, wherein the drive module (7) is configured to adjust at least one operating parameter of the heat treatment system (4) and / or the agitator (3) as a function of the value of the output (U11) of the first control module (6) and the value of the output (U21) of the second control module (8).

8. The machine (1) according to any one of claims 1 to 7 and claim 2, wherein the second control module (8) further comprises a second control output (U22) connected to one input (I2) of the inputs (I1, I2, I3, I4, I5, I6, I7) of the first control module (6) for transmitting control information to the first control module (6).

9. 9. The machine (1) according to any one of claims 1 to 8 and claim 6, wherein the drive module (7) is configured to adjust the at least one operating parameter of the heat treatment system (4) and / or the agitator (3) only as a function of the value of the output (U21) of the second control module (8) for a preset length of time at the start of a treatment cycle, and thereafter adjust the at least one operating parameter of the heat treatment system (4) and / or the agitator (3) as a function of the value of the at least one output (U11) of the first neural network control module (6) and the output (U21) of the second control module (8) once the preset length of time has elapsed at the start of the treatment cycle.

10. The machine (1) for producing liquid, semi-liquid or semi-solid food products according to any one of claims 1 to 9, wherein the heat treatment system (4) is a circuit using a heat exchange fluid and comprises an evaporator (10), a compressor (11), a throttling element (12) and a condenser (13).

11. The drive module (7) is configured to adjust one or more of the following operating parameters: - the driving force of said compressor (11), - the opening of said throttle element (12), - the speed and direction of rotation of said stirrer (3), - a processing cycle time.

12. The machine (1) according to any one of claims 1 to 11 and claim 10, wherein the heat treatment system (4) is configured to operate in two different modes, namely a first mode in which it runs a reverse Carnot cycle to cool the vessel (2) and a second mode in which it runs a hot gas or heat pump cycle to heat the vessel (2), and wherein the drive module (7) is configured to regulate the following operating parameters, namely the operating mode of the heat treatment system (4).

13. The at least one sensor (S) is configured to detect at least one of the following parameters: the operating pressure of the heat exchange fluid at a given point in the circuit of the heat treatment system (4), the operating temperature of the heat exchange fluid at a given point in the circuit of the thermal treatment system (4), - the operating temperature of said container (2), the operating temperature of the heat exchange fluid leaving the evaporator (10) associated with the vessel (2), the operating temperature of the heat exchange fluid entering the evaporator (10) associated with the vessel (2), the operating pressure of the heat exchange fluid leaving the compressor (11), the operating pressure of the heat exchange fluid leaving the evaporator (10), the temperature of the ambient air in the vicinity of said machine (1), A machine (1) for producing liquid or semi-liquid or semi-solid food products according to any one of claims 1 to 12, configured to detect at least one of the following:

14. 14. The machine (1) for producing liquid, semi-liquid or semi-solid food products according to any one of claims 1 to 13, wherein the at least one sensor (S) is connected to the stirrer (3) and configured to detect a value relating to the power supply of a motor for driving it.

15. The machine (1) for producing liquid or semi-liquid or semi-solid food products according to any one of claims 1 to 14 and claims 6 and 14, wherein the second control module (8) is configured to derive parameters related to the consistency and / or viscosity of the product being processed in the container (2) from values ​​related to the power supply of the motor.

16. 16. The machine (1) for producing liquid or semi-liquid or semi-solid food products according to claim 15, wherein the parameters relating to the consistency and / or viscosity of the product being processed in the container (2) are comprised between a minimum value corresponding to the product to be processed and a maximum value corresponding to the final product.

17. 17. The machine (1) for producing liquid or semi-liquid or semi-solid food products according to any one of claims 1 to 16, wherein the at least one sensor (S) is configured to detect values ​​of acceleration of at least a part of the machine.

18. The inputs (I1, I2, I3, I4, I5, I6, I7) of the first control module (6) include one or more of the following inputs: a first input (I1) related to a value representative of the consistency and / or viscosity of the product being processed in said vessel (2), a second input (I2) related to a value representative of the operating temperature of said vessel (2), a third input (I3) related to a value representative of the operating temperature of the heat exchange fluid entering the evaporator (10) associated with said vessel (2), a fourth input (I4) related to a value representative of the operating temperature of the heat exchange fluid leaving the evaporator (10) associated with said vessel (2), a fifth input (I5) related to a value representative of the operating pressure of the heat exchange fluid leaving the compressor (11), a sixth input (I6) related to a value representative of the operating pressure of the heat exchange fluid leaving the evaporator (10), - a seventh input (I7) related to a value representative of the opening of the throttle element (12).

19. 19. The machine (1) for producing a liquid, semi-liquid or semi-solid food product according to any one of claims 1 to 18, wherein the first control module (6) is configured to classify the mixture being processed in the vessel (2) based on the balance of the ingredients in the mixture at least between a balanced class corresponding to a mixture being processed that is balanced with respect to one or more ingredients and an unbalanced class corresponding to a mixture being processed that is not balanced with respect to one or more ingredients, and the at least one output (U11) of the first control module (6) represents a classification of the mixture being processed in the vessel (2).

20. 20. The machine (1) for producing liquid or semi-liquid or semi-solid food products according to claim 19, wherein the first control module (6) is configured to perform a multi-class classification of the mixture based on the balance of ingredients in the mixture being processed in the vessel (2), the first control module (6) comprising a plurality of outputs (U11, U12, U13, U14, U15, U16, U17, U18, U19).

21. The outputs (U11, U12, U13, U14, U15, U16, U17, U18, U19) of the first control module (6) relate in their entirety to two or more of the following classes of mixtures being treated, namely: - the product being treated in said vessel (2) is balanced with respect to at least one base component, the product being treated in said vessel (2) is unbalanced with respect to its water content, the product being processed in said vessel (2) is unbalanced in terms of fat, The machine (1) for producing liquid or semi-liquid or semi-solid food products according to claim 20, which relates to two or more of the following classes: the product being processed in the vessel (2) is unbalanced with respect to sugars.

22. The outputs (U11, U12, U13, U14, U15, U16, U17, U18, U19) of the first control module (6) relate in their entirety to two or more of the following classes of mixtures being treated, namely: - the product being treated in said vessel (2) is balanced with respect to at least one base component, the product being treated in said vessel (2) is unbalanced with respect to its water content, the product being processed in said container (2) is unbalanced with an excess of fat, the product being processed in said vessel (2) is unbalanced due to a lack of sugars, The machine (1) for producing liquid or semi-liquid or semi-solid food products according to claim 20, which relates to two or more of the following classes: the product being processed in the container (2) is unbalanced with an excess of sugars.

23. A machine (1) for producing liquid or semi-liquid or semi-solid food products as claimed in any one of claims 20 to 22, wherein each output (U11, U12, U13, U14, U15, U16, U17, U18, U19) provides a numerical value, comprised between a minimum and a maximum value, representative of the probability that the product being processed in the container (2) corresponds to the pre-set class.

24. 24. The machine (1) for producing liquid or semi-liquid or semi-solid food products according to any one of claims 1 to 23, wherein the control unit (5) is configured to transmit data to at least one input (I1) of the inputs of the first control module (6) at preset time intervals, preferably less than 10 seconds.

25. A machine (1) for producing liquid, semi-liquid or semi-solid food products as claimed in any one of claims 1 to 24 and claim 5, wherein at least some of the neural network layers (S2, S3, S4, S5) are hidden layers and the neural network (RN) is a deep neural network.

26. The machine (1) for producing liquid or semi-liquid or semi-solid food products according to any one of claims 1 to 25 and claim 5, wherein the neural network (RN) is an RNN, i.e. a recurrent neural network.

27. The machine (1) for producing liquid or semi-liquid or semi-solid food products according to any one of claims 1 to 26 and claim 4, wherein the neural network (RN) is any RNN or CNN network.

28. A method for controlling a machine (1) for producing liquid or semi-liquid food products, comprising the following steps: - providing a machine (1) for producing a liquid or semi-liquid food product, said machine (1) comprising: - a treatment vessel (2) for treating the base product, - an agitator (3) adapted to mix the base product inside the processing vessel (2) in which the product is processed; a heat treatment system (4) configured for heat exchange with said treatment vessel (2) for treating said base product; at least one sensor (S) configured to detect an operating parameter of said machine; providing a first artificial intelligence control module (6) configured to define a classifier (CL) having at least one input (I1, I2, I3, I4, I5, I6, I7) and at least one output (U11), said at least one input (I1) being connected to said at least one sensor (S); training said first artificial intelligence control module (6) on the basis of the processing of different mixtures inside said container (2) to obtain a first trained control module (6); - feeding the mixture into said vessel (2) and starting the treatment cycle; - driving at least the heat treatment system (4) and / or the agitator (3) to adjust at least one operating parameter thereof as a function of the value of the at least one output (U11) of the first trained control module (6).

29. 29. The method of claim 28, wherein the step of training the first artificial intelligence control module (6) based on the processing of various mixtures comprises training the first artificial intelligence control module (6) based on processing of mixtures having different balances of the base components.

30. The at least one output (U11) of the first control module (6) represents a classification of the mixture of base products, the method comprising the further steps of: The method according to claim 28 or 29, comprising a further step of classifying, via the first artificial intelligence control module (6), the mixture being processed in the vessel (2) into one or more pre-defined categories.

31. 31. The method according to any one of claims 28 to 30, wherein the first artificial intelligence control module (6) comprises a neural network (RN), and the step of training the first artificial intelligence control module (6) based on processing of various mixtures inside the vessel (2) to obtain a first trained control module (6) comprises a step of training the neural network (RN) in a supervised mode based on processing of mixtures in the vessel (2) having different balances of the base components to obtain a trained neural network (RN).

32. The inputs (I1, I2, I3, I4, I5, I6, I7) of the first control module (6) include one or more of the following inputs: a first input (I1) related to a value representative of the consistency and / or viscosity of the product being processed in said vessel (2), a second input (I2) related to a value representative of the operating temperature of said vessel (2), a third input (I3) related to a value representative of the operating temperature of the heat exchange fluid entering the evaporator (10) associated with said vessel (2), a fourth input (I4) related to a value representative of the operating temperature of the heat exchange fluid leaving the evaporator (10) associated with said vessel (2), a fifth input (I5) related to a value representative of the operating pressure of the heat exchange fluid leaving the compressor (11), a sixth input (I6) related to a value representative of the operating pressure of the heat exchange fluid leaving the evaporator (10), - a seventh input (I7) related to a value representative of an opening of the throttling element (12).

33. The method according to any one of claims 28 to 32, comprising a step of monitoring the signal of the at least one sensor (S) in order to ascertain a fault in the mechanical process, and wherein the step of driving at least the heat treatment system (4) and / or the agitator (3) comprises a step of driving at least the heat treatment system (4) and / or the agitator (3) in order to adjust at least one operating parameter for the heat treatment system (4) and / or the agitator (3) as a function of the value of the at least one output (U11) of the first control module (6) and of the ascertained fault in the mechanical process.

34. 34. The method according to any one of claims 28 to 33 and claim 30, comprising the step of waiting at the start of the processing cycle for a preset length of time before classifying, via the trained neural network (RN), the mixture being processed in the vessel (2) into one or more preset categories.

35. The step of training the first artificial intelligence control module (6) in a supervised mode, based on the processing of the various mixtures inside the container (2), so as to obtain a first control module (6) trained to classify the mixtures, comprises the following steps: - processing a mixture of the following types: - a balanced mixture with respect to said base component, - processing a mixture that is unbalanced with respect to at least one of said base components; - during training, providing said first control module (6) with labels associated with desired output categories for each type of mixture being processed.

36. The step of training the first control module (6) in a supervised mode, based on the processing of different mixtures inside the container (2), comprises the following steps: - processing a mixture of the following types: - a balanced mixture with respect to said base component, - mixtures unbalanced with respect to the water content, - an unbalanced mixture with respect to fats, - processing a mixture that is unbalanced with respect to sugars; - during training, providing said first control module (6) with labels associated with desired output categories for each type of mixture being processed.

37. The step of driving at least the heat treatment system (4) and / or the stirrer (3) to adjust at least one operating parameter related to the heat treatment system (4) and / or the stirrer (3) comprises the following steps: - comparing the detected defects with the classification values ​​of the identified mixture; The method according to any one of claims 28 to 36 and claim 33, comprising the step of: - giving commands to the heat treatment system (4) and / or to the stirrer (3) also based on the result of the comparison.

38. The step of driving at least the heat treatment system (4) and / or the stirrer (3) to adjust at least one operating parameter related to the heat treatment system (4) and / or the stirrer (3) comprises the following steps: - checking whether the classification value of said identified mixture corresponds to a defect state, i.e. an unbalanced mixture; - giving commands to the heat treatment system (4) and / or to the stirrer (3) also based on the result of the check.

39. The step of driving at least the heat treatment system (4) and / or the agitator (3) to adjust at least one operating parameter related to the heat treatment system (4) and / or the agitator (3) comprises one or more of the following driving operations, namely: - a driving action for varying the opening of a throttle element (12) forming part of said thermal treatment system (4), a drive operation for adjusting the power of a compressor (11) forming part of said heat treatment system (4); - a drive operation for adjusting the flow rate of a cooling fluid, air or water, in a condenser (13) forming part of said heat treatment system (4); - a driving operation for switching said thermal treatment system (4) between a cooling mode for cooling said vessel (2) and a heating mode for heating said vessel (2), said heating mode including an operating mode in which said thermal treatment system (4) performs what is known as a hot gas cycle or a heat pump cycle; - a driving action adjusting the speed and direction of rotation of the agitator (3).

40. The method according to any one of claims 28 to 39 and claim 30, wherein the step of classifying the mixture being processed in the vessel (2) via the first artificial intelligence control module (6) into one or more preset categories is performed at preset time intervals between the start and end of processing.

Images