ES2836853A1 - METHOD AND SYSTEM FOR THE CHARACTERIZATION AND CERTIFICATION OF THE EVOLUTION OF FOOD PROCESSES (Machine-translation by Google Translate, not legally binding) - Google Patents

Abstract

Method and system for the characterization and certification of the evolution of food processes. The present invention is directed to a method and a system for characterizing and certifying food processes by detecting gaseous substances emitted by food and measurable changes by sensors during said food processes. The detection of substances makes it possible to characterize the temporal evolution of the olfactory signal to determine and certify food processes and to classify said signals into categories of food processes with different characteristics. (Machine-translation by Google Translate, not legally binding)

Description

DESCRIPTION

METHOD AND SYSTEM FOR THE CHARACTERIZATION AND CERTIFICATION OF THE

EVOLUTION OF FOOD PROCESSES

OBJECT OF THE INVENTION

The present invention refers to a method and a system to characterize and certify the evolution of food processes.

BACKGROUND OF THE INVENTION

The control of food production in the industry is carried out to guarantee the health and quality aspects of the products, and in some cases to grant them designations / certifications that can finally determine the consumer's choice. Different manual or automated inspection protocols are used for this task. In some of these protocols, quantitative measurement devices are used, such as chemical analysis probes, electronic noses / tongues, pH meters, immersion / penetration probes or colorimeters, among many others.

In most of the current applications that require checking the quality status of a food, specific or instantaneous checks are carried out, which are often invasive (for example by making incisions in a fruit). As a consequence, the test result is sometimes incomplete or inaccurate and also implies the need to dispose of the product that has been subjected to an invasive test.

DESCRIPTION OF THE INVENTION

The present invention proposes a solution to the above problems by means of a method to characterize and certify the evolution of food processes according to claim 1, a system to characterize and certify the evolution of food processes according to claim 17, a data processing system according to claim 24, a computer program according to claim 25 and a computer-readable medium according to claim 26. Preferred embodiments of the invention.

In a first inventive aspect, a method is provided to characterize and certify the evolution of food processes by a system that comprises:

• a detection module comprising at least one element for measuring gaseous components configured to generate at least one signal of the time evolution of at least one detected gaseous component;

• an analysis module configured to characterize the at least one signal generated by the detection module based on its temporal evolution and to determine if said at least one signal corresponds to the development of a specific food process; where the method comprises the following stages:

a) generating at least one signal of the time evolution of at least one gaseous component detected by the at least one measuring element of the detection module during a predetermined period of time;

b) receiving, by the analysis module, the generated signal;

c) characterize the signal based on its temporal evolution, by the analysis module; Y

d) determine by the characteristics of the signal if its temporal evolution corresponds to a specific food process.

The method of the present invention provides a mechanism for the characterization and certification of food processes by detecting and measuring the time evolution of gaseous components. Throughout this document, it will be understood that certifying the evolution of a specific food process implies determining that said food process has occurred, in particular that it has taken place in pre-established terms.

As a food process, any type of food processing and preservation carried out in the different parts of the production, transport, storage and sale chain carried out with the aim of guaranteeing the quality and hygiene of said foods must be understood. Examples of food processes are ripening, curing, aging, fermenting, smoking, cultivating, growing, baking, or roasting.

During a food process, food produces substances and gases of organic origin of varied composition that are released into the environment. The method and system of the present The invention makes it possible to characterize the temporal evolution of these components and, thus, determine if a specific food process has taken place. The present invention has applications in the characterization and certification of food processes that evolve over time, such as ripening of fruits, vegetables or meats, smoking of foods, curing of sausages or cheeses, fermentation of dairy products and beverages, baking of pastries and bread, roasted coffee beans, etc. For example, the method and system of the present invention make it possible to detect changes in the evolution of the ethylene concentration in fruits and vegetables, to certify the quality of a sausage curing process, and / or to detect the release of alcohols in processes of fermentation.

As has been commented, while in the state of the art specific or instantaneous checks are usually carried out, the present invention characterizes the evolution of the food process by means of the continuous sensorization of the gaseous components detected. The information on the temporal evolution, not only the specific measurement during the food process of a product, is important to identify that the food process has been carried out in the desired terms, for example to check the correct maturation of a food. Furthermore, the present invention does not require invasive measures, that is, it does not damage the product to obtain its status information.

In the context of the invention, determining that the temporal evolution of the signal corresponds to a specific food process is to be understood as the validation that said food process has taken place and has elapsed, in particular according to pre-established terms.

The particular food process can be understood as belonging to a predefined category of food processes. Examples of food process categories are:

- ripening process; or

- curing process; or

- aging process; or

- fermentation process; or

- smoking process; or

- cultivation process; or

- growing process; or

- baking process; or

- roasting process; or

- accelerated and / or intervened food process; or

- unhealthy food process; or

- a combination of any of the above.

In the context of the invention, characterizing a food process has to be understood as the identification of elements and patterns that make it possible to describe specific properties of said food process by monitoring its evolution over time.

Throughout the document, an analysis module is to be understood as a set of media with the capacity to process, transmit and store information; preferably a computer comprising a processor, a communication device and a memory. How to characterize the evolution in time of a signal has to be understood to establish the attributes of the signal that make it unique so that it is distinguishable from other signals. Preferably, the signal is characterized based on its magnitude and / or temporal structure, using parameters such as the amplitude of the signal, maxima and minima of the signal, the value of the derivative and / or the correlation of the signal with one or more additional signals. This characterization makes it possible to determine if the signals correspond to a specific food process.

The characterization of the signal can be implemented with discrete variables in time (corresponding to specific characteristics and / or labels that occur at a given moment in the temporal evolution of the signals) and / or with continuous variables (the signal itself or those corresponding to operations whose result is a continuous function that takes values during the acquisition process, such as the calculation of the derivative of the signal, the calculation of the integral of the signal, the combination of two or more signals or the correlation of two or more signals ). A label is to be understood as a value assigned to the signal at a given moment of time to distinguish it from the rest of the moments of time and / or to contextualize said value, each label being able to come from an automatic labeling process. The same label can be assigned to more than one signal at the same time to contextualize these signals.

In one embodiment, step d) of determining whether the temporal evolution of the signal corresponds to a specific food process comprises performing a classification, a regression or another process to associate the temporal structure of the measured signal to a process concrete food.

In one embodiment, step (c) comprises identifying at least one temporal event in the signal and step (d) is performed based on the sequentiality of the identified temporal events.

In the context of the present invention, "temporal event" will be understood as a particular event characterizing the signal that may correspond to a point or to a larger region of the signal. Preferably, said temporal events are the presence of maximums or minimums, the Slope values measured in the signal, the exceeding of predefined thresholds, and / or sequential sets of the above that define a specific temporal structure. Each temporal event is identified based on the discrete and / or continuous variables that characterize one or more signals. .

In this embodiment, once the temporal events have been identified, it is determined whether the signals correspond to the development of a specific food process based on the temporal structure of said temporal events. For example, a time structure of time events could be a sequence of local highs followed by a specific threshold being exceeded. According to that example, if that sequence of temporal events occurs, the method determines that a specific food process has taken place, in particular that it has taken place in pre-established terms.

The identification of temporal events and their sequentiality allow the classification of the signals as a specific food process (for example, the ripening process of a fruit or the curing process of a cheese) or as a family of food processes (for example, accelerated food process , intervened or unhealthy).

In one embodiment, step (c) comprises using at least one algorithm for identifying temporal events and their temporality, which uses, for example, the computation of the derivatives of the signals, detection of maximums / minimums, detection of exceeding thresholds and / or correlation between signals. Thus, the signals are characterized in relation to the sequentiality and temporality of the identified temporal events.

As signals are to be understood those produced by the sensors and / or inputs of the system, including those that detect gaseous chemical substances from the food process, as well as other types of additional signals that can be considered, for example, signals related to environmental conditions and / or measured and / or monitored quantities.

In a particular example, the identification of temporary events in the signal additionally comprises a step of labeling said temporary events. As tagging, is to be understood a process of assigning a time event a distinctive tag of said time event.

In a particular embodiment, if by the method it is determined that a food process has taken place, the method additionally comprises a step of classifying the signals into at least one food process sub-category. In this embodiment, once it has been determined that a specific food process of a specific category has taken place, a more specific categorization is performed into subcategories of food processes. Below are some examples of specific food processes and their possible subcategories:

- Category: ripening of bananas. Subcategories: natural, accelerated with calcium carbide, inhibited by cooling, etc.

- Category: cheese curing. Subcategory: with designation of origin, cured with unhealthy results, etc.

- Category: cured ham. Subcategory: cured with gender located in the same geographical area, cured with sufficient ventilation, etc.

- Category: beverage fermentation. Subcategory: natural, fermentation with unhealthy results, etc.

- Category: greenhouse fruit growth. Subcategory: natural, accelerated with CO ² injection, etc.

- Category: vegetable growing. Subcategory: natural, accelerated with artificial light, etc.

- Category: vegetable cultivation. Subcategory: irrigated with clean or polluted water, adequate air quality, etc.

- Category: dried fruit. Subcategory: with natural sunlight, in oven, etc.

During the course of food processes, the gaseous component measurement elements generate signals associated with the detection of gaseous components. In these signals it is possible to distinguish and quantify different temporal events that define a characteristic temporal structure for each of said food processes.

In a preferred embodiment, the system comprises a set of measurement elements, configured to detect a plurality of gaseous components.

In one embodiment, the gaseous component measurement element is an olfactory sensor. Preferably, the olfactory sensor is configured to detect substances emitted by a food. Preferably the olfactory sensor is configured to detect at least one of: Volatile Organic Compounds (VOCs), ethylene, sulfides, isovaleric acid, butyric acid, alcohols, flavorings, oxidases and ammonia. In a more particular preferred embodiment, the system comprises a gaseous component measurement element configured to detect Volatile Organic Compounds (VOCs).

The aggregation of olfactory sensors in a device is commonly referred to as an "artificial nose" or "electronic nose." Applications of artificial noses are often geared towards threshold detection, odor discrimination / classification, warning systems, etc., and are typically based on point measurements. On the contrary, the present invention is based on the analysis of the temporal structure of the information recorded by the sensors during a prolonged measurement.

In one embodiment, the olfactory sensor is of any of the following types: chemoresistive, chemocapacitive, potentiometric, gravimetric, optical, acoustic, thermal, polymeric, amperometric, chromatographic, spectrometric, or field effect.

In a more particular preferred embodiment, the system comprises at least one chemoresistive olfactory sensor.

In one embodiment the detection module further comprises at least one additional sensor, preferably an ambient condition sensor, a particle sensor, a contact sensor, a positioning sensor and / or an image sensor. In this embodiment the method additionally comprises a step of obtaining measurements of at least one magnitude by means of the at least one additional sensor and of identifying temporal events of the signal associated with said at least one magnitude, for example a temporal event associated with a detected change. in magnitude, or exceeding a predefined magnitude threshold. These identified temporal events are used as additional information during the signal characterization stage.

In one embodiment at least one additional sensor is an ambient condition sensor, preferably a humidity, temperature, atmospheric pressure, air velocity, and / or salinity sensor.

In a preferred embodiment, the system comprises at least one sensor for environmental conditions. In a more particular preferred embodiment, the ambient condition sensor is configured to measure humidity.

In one embodiment at least one additional sensor is a particle sensor, preferably a spore and / or microplastic sensor.

In one embodiment at least one additional sensor is a contact sensor, preferably a sensor for lipids, hardness, weight, mass, volume, glucose, sucrose, phosphates, and / or isomaltol.

In one embodiment at least one additional sensor is a positioning sensor, for example based on GPS, Galileo, and / or GLONASS. Advantageously, this embodiment makes it possible to know the location of the product, which may be of interest when monitoring food processes that take place during the transfer of food, for example to verify if the transport of a food has been carried out under conditions that do not pose a risk. sanitary.

In one embodiment at least one additional sensor is an image sensor, such as a camera for detecting and / or measuring stains, color and / or gloss on food.

In one embodiment, the identification of temporary events associated with environmental conditions and / or other quantities measured by additional sensors comprises a step of labeling said temporary events.

Advantageously, the detection of environmental conditions and / or the recording of the evolution of other magnitudes provide additional information that makes it possible to contextualize the evolution of the recorded signal, as well as to correlate temporary events detected in the signal with the information obtained by the additional sensors. This contextual information makes it possible to determine with better results whether a specific food process has occurred and / or to obtain a more precise classification of signals.

In one embodiment the system further comprises a conditioning module configured to condition the signal generated by the detection module. In this embodiment the method further comprises, between steps (a) and (b), the steps of receiving and conditioning the signal by the conditioning module. In this embodiment, step (c) is carried out on the generated and conditioned signal. In one embodiment, the step of conditioning the signal comprises filtering and / or resampling said signal.

In one embodiment, steps (c) and (d) are performed by means of a previously trained machine learning algorithm with training signals that correspond to at least one specific food process. In a preferred embodiment, in said training signals, in addition, a series of time events and predefined labels are identified. When the method is to classify the signal into different subcategories, the classification can also be performed by a previously trained machine learning algorithm with training signals that are classified into at least one predefined food process subcategory.

In the embodiment where stages (c) and (d) are performed by means of a machine learning algorithm, there is a previous training phase, in which signals obtained in the repeated registration of a food process are used to train the learning algorithm automatic, preserving the individual and joint temporal structure of the signals. In a preferred embodiment, the machine learning algorithm is supervised and comprises a neural network and / or a random forest and / or vector support machine with the coding and temporality requirements established by the type of classifier and by the characteristics of the signals. produced by the device. The result of the training phase is an algorithm capable of extracting and comparing the temporal structure of events in the recorded signals with the representation of a food process. Once the machine learning algorithm has been built, it can be used to characterize the recorded signals according to their temporal structures in what is called the exploitation phase. The result of the exploitation phase is the determination of whether a signal corresponds to the development of a specific food process. In all processing stages, the temporal structure of both the individual signals and their integrated representation are preserved, since the sequentiality in the evolution of temporal events is essential to characterize the food process. The temporality and / or sequentiality of the temporal events in relation to the determination of the food process, in general are not known a priori, and are represented in the machine learning algorithm with the acquired information

1

through training.

In a particular example, there is an automatic labeling process for all types of temporal events (those corresponding to signals associated with gaseous components, changes in environmental conditions and / or other magnitudes). The generated labels provide relevant information about the signal itself and about the context of its events to improve the results of the learning algorithms.

In a particular example, the results of the machine learning algorithm (determination of the particular food process that has taken place) are also automatically tagged during the exploitation phase. In this way, the automatic learning algorithm itself is fed back with said automatic labels of the results to, advantageously, improve the precision of the algorithm in subsequent executions of the method.

In one embodiment, step (d) of the method comprises comparing the signal with at least one reference signal, preferably a reference signal encoded in the result of the learning algorithm used. For example, in a neural network once trained, the representation of the reference signal is reflected in the weights of the network connections. Additionally or alternatively, comparison with one or more reference signals can be used to classify the signal into at least one subcategory of food processes.

In one embodiment, the steps of the method are repeated periodically, where the repetition period is a predefined value, to monitor and characterize the evolution of the detected and / or classified food processes. Advantageously, the method makes it possible to detect changes in food processes that should be reproduced periodically.

In one embodiment, if the method detects that the food process has not followed the desired evolution, an alert is generated so that measures can be taken to correct said food process. In this way, it is achieved that the final food is in the desired conditions.

In a second inventive aspect, a system is provided to characterize and certify the evolution of food processes, which comprises:

- a detection module comprising at least one element for measuring gaseous components configured to generate at least one signal of the time evolution of at least one detected gaseous component;

- an analysis module configured to characterize the at least one signal generated by the detection module based on its temporal evolution, to determine if said at least one signal corresponds to the development of a specific food process; and wherein the analysis module is configured to carry out steps (b) to (d) of the method according to any of the embodiments of the first inventive aspect.

Throughout the document, an analysis module is to be understood as a set of media with the capacity to process, transmit and store information; preferably a computer comprising a processor, a communication device and a memory.

In one embodiment the system further comprises a conditioning module configured to condition the signal generated by the detection module.

In one embodiment, the analysis module of the system is further configured to classify the at least one signal in at least one sub-category of food processes.

In one embodiment the system comprises at least one additional sensor. Preferably, the additional sensor is an ambient sensor, a particle sensor, a contact sensor, a positioning sensor and / or an image sensor.

In one embodiment, the system comprises a plurality of additional sensors, preferably sensors of different types for the measurement and / or monitoring of a plurality of quantities.

In one embodiment at least one additional sensor is a sensor for ambient conditions, preferably humidity, temperature, atmospheric pressure, air velocity, and / or salinity. In a preferred embodiment, the system comprises at least one sensor for environmental conditions. In a more particular preferred embodiment, the ambient condition sensor is configured to measure humidity.

In one embodiment at least one additional sensor is a particle sensor, preferably a spore and / or microplastic sensor.

In one embodiment at least one additional sensor is a contact sensor, preferably a sensor for lipids, hardness, weight, mass, volume, glucose, sucrose, phosphates, and / or isomaltol.

In one embodiment at least one additional sensor is a positioning sensor.

In one embodiment at least one additional sensor is an image sensor.

In one embodiment, the at least one gaseous component measurement element is an olfactory sensor. Preferably, the olfactory sensor is configured to detect substances emitted by a food. Preferably, the olfactory sensor is configured to detect at least one of the following substances: Volatile Organic Compounds (VOCs), ethylene, sulfides, isovaleric acid, butyric acid, alcohols, flavorings, oxidases and ammonia.

In one embodiment, the olfactory sensor is of any of the following types: chemoresistive, chemocapacitive, potentiometric, gravimetric, optical, acoustic, thermal, polymeric, amperometric, chromatographic, spectrometric, or field effect.

In a preferred embodiment the system comprises a plurality of measurement elements configured to detect a plurality of gaseous components. Preferably, the plurality of measuring elements is configured to detect one or more of the following substances: Volatile Organic Compounds (VOCs), ethylene, sulfides, isovaleric acid, butyric acid, alcohols, flavorings, oxidases and ammonia. Preferably, the measuring elements are olfactory sensors. In a more particular preferred embodiment, the system comprises a gaseous component measurement element configured to detect Volatile Organic Compounds (VOCs).

In a third inventive aspect there is provided a data processing system comprising means for carrying out steps (b) to (d) of the method according to any of the embodiments of the first inventive aspect.

In a fourth inventive aspect a computer program is provided comprising instructions which, when the program is executed by a computer, causes the computer to carry out steps (b) to (d) of the method according to any of the embodiments.

1

of the first inventive aspect.

In a fifth inventive aspect a computer-readable medium is provided comprising instructions which, when executed by a computer, cause the computer to carry out steps (b) to (d) of the method according to any of the embodiments of the first aspect. inventive.

All features and / or method steps described herein (including claims, description, and drawings) may be combined in any combination, except combinations of such mutually exclusive features.

DESCRIPTION OF THE FIGURES

These and other characteristics and advantages of the invention will become more apparent from the detailed description that follows of a preferred embodiment, given solely by way of illustrative and non-limiting example, with reference to the accompanying figures. .

Figure 1 This figure shows a schematic representation of the system to characterize and certify food processes according to an embodiment of the invention.

Figure 2 This figure shows a schematic representation of the method to characterize and certify food processes according to an embodiment of the invention.

Figure 3 This figure shows an example of the signals collected in a banana ripening process for two different conditions.

DETAILED EXHIBITION OF THE INVENTION

The present invention proposes a method and a system to characterize and certify food processes.

The system (1) to characterize and certify food processes comprises a module of detection (2) and an analysis module (3). The detection module (2) comprises at least one element for measuring gaseous components (2.1). The measuring elements (2.1) detect gaseous substances produced by a food and generate signals (2.2) of the temporal evolution of the gaseous components detected. The analysis module (3) receives the signals (2.2) generated by the detection module (2), characterizes the signals (2.2) based on their temporal evolution and, according to the characterization, determines whether the temporal evolution of the signal corresponds to a specific food process.

If it is determined that the temporal evolution of a specific food process has occurred, in particular in pre-established terms, said food process is certified.

In one embodiment, the analysis module (3) is further configured to classify the signals (2.2) in at least one sub-category of food processes.

Figure 1 shows a schematic representation of the food process characterization and certification system according to an embodiment of the invention. In this embodiment, the detection module (2) comprises a plurality of measurement elements (2.1) of gaseous components and a plurality of additional sensors (2.3, 2.4, 2.5, 2.6, 2.7). The measuring elements (2.1) of gaseous components are olfactory sensors. In a preferred embodiment, the olfactory sensors are of the metal oxide (MOX) and nondispersive infrared (NDIR) type and are configured to record substances such as Volatile Organic Compounds (VOCs), ethylene, sulfides, isovaleric acid, butyric acid, alcohols, flavorings, oxidases and ammonia.

The additional sensors (2.3, 2.4, 2.5, 2.6, 2.7) provide data related to other quantities detected and / or measured during the course of the food process. The data from the additional sensors (2.3, 2.4, 2.5, 2.6, 2.7) allow contextualizing the signals obtained by the olfactory sensors to determine with better results if a specific food process has occurred.

In this embodiment the additional sensors include sensors for environmental conditions (2.3), in particular temperature and humidity sensors, a particle sensor (2.4), a contact or proximity sensor (2.5), a positioning sensor (2.6) and an image sensor (2.7). From the data of these additional sensors, the signals obtained by the olfactory sensors are contextualized, automatically generating labels that

1

provide relevant information on the origin and context of the events; for example, a label reporting an increase in temperature or the concentration of Volatile Organic Compounds (VOCs).

In the embodiment of Figure 1 the analysis module (3) includes a processor (7), in particular a microcontroller, configured to receive (120) the signals (2.2) generated by the detection module (2), to characterize ( 131) the signals (2.2) based on their temporal evolution and to determine (132) if a specific food process has occurred based on the characterization of the signals (2.2). If it is determined that the temporal evolution of a specific food process has occurred, in particular in pre-established terms, said food process is certified.

In one embodiment, the processor (7) is further configured to classify the signals (2.2) into at least one subcategory of food processes and / or to command the detection module (2) to generate the signals (2.2).

In the embodiment of Figure 1 the analysis module (3) additionally includes a wireless transmission system (8) and a memory (9). The wireless transmission system (8) allows the exchange of data with external processing and / or storage systems. In a preferred embodiment, the system is configured to transmit the data to a centralized server accessible via the web and a REST API, for storage. The memory (9) allows the storage of data in the system itself (1).

As a REST API, a software architecture that acts as a communication interface between two systems using HTTP must be understood.

The system (1) of Figure 1 additionally includes a conditioning module (4) configured to condition the signal (2.2) generated by the detection module (2). In a preferred embodiment, the conditioning module (4) includes one or more operational amplifiers and one or more RLC filters.

In the embodiment shown in figure 1, the detection module (2) and the analysis module (3) are implemented as parts of a single device and are housed inside an enclosure or casing (10) together with the other elements of the system (1). However, in other embodiments the detection module (2) and the analysis module (3) can

1

be implemented as separate devices.

Preferably, the additional sensors (2.3, 2.4, 2.5, 2.6, 2.7) are integrated forming a single device, thus facilitating the portability of the device. However, in other embodiments the additional sensors (2.3, 2.4, 2.5, 2.6, 2.7) can be independent peripheral elements that can connect / disconnect from the system via cable, and / or wirelessly, using technologies such as WiFi, Bluetooth, LoRA , Zigbee or any type of mobile connectivity (3G, 4G, 5G ...).

The casing (10) has accesses through which the entry and exit of air is allowed. In figure 1 these accesses have been represented by dashed lines. In this embodiment, the system additionally includes an air capture and conditioning system (5), which allows controlling and modifying variables such as temperature, humidity and / or air flow that enters the interior of the casing (10). In this embodiment, the system additionally includes an air diffusion system (6) towards the olfactory sensors (2.1) and / or towards the additional sensors (2.3, 2.4, 2.5, 2.6, 2.7). Some of these sensors, such as humidity and temperature sensors, require continuous access to outside air in order to perform the measurement properly. This access to outside air may not be necessary for other sensors.

Figure 2 shows a schematic representation of the method (100) of characterization and certification of food processes according to an embodiment of the invention. This method (100) comprises the following steps:

a) generating (110) at least one signal (2.2) of the time evolution of at least one gaseous component detected by the at least one measuring element (2.1) of the detection module (2) during a predetermined period of time;

b) receiving (120), by the analysis module (3), the generated signal (2.2);

c) characterize (131) the signal (2.2) based on its temporal evolution, by the analysis module (3), and

d) determine (132) by the characteristics of the signal (2.2) if its temporal evolution corresponds to a specific food process.

Thus, we start from the record of the temporal evolution of the olfactory sensor signals (2.1) as described in relation to figure 1. If the system (1) includes additional sensors (2.3, 2.4, 2.5, 2.6 , 2.7), the additional data is optionally used obtained from these sensors, temporarily synchronizing this additional data with the signals of the olfactory sensors (2.1). As previously described, the additional data provided by additional sensors (2.3, 2.4, 2.5, 2.6, 2.7), make it possible to report on the temporal evolution of the food process, and thus contextualize the evolution of olfactory signals in the course of the food process. so that more precise results are obtained in the characterization of food processes and / or in the classification of said signals.

In this embodiment, the olfactory signal is labeled based on the detection of events such as the detection of a sudden increase of a component, for example Volatile Organic Compounds (VOCs), or the detection of visible signs on the surface of the food, for example example the appearance of spots on a fruit detected by a camera. Furthermore, additionally or alternatively, the olfactory signal can be labeled based on a sensed environmental condition, such as a change in temperature, pressure, humidity or salinity.

The analysis module (3) receives (120) the generated signal (2.2) and characterizes it (131) based on its temporal evolution. In this embodiment, the analysis module (3) determines (132) if said signal (2.2) corresponds to the evolution of a specific food process, classifying it in at least one category of food processes. In a preferred embodiment, first of all, the olfactory signals are synchronized in time and resampled, adapting to the original sampling and the type of data obtained (continuous / discrete measurements, labels, etc.). Next, the temporal events in the structure of the signals and the temporality and sequentiality of said temporal events are identified, by computing the derivative of the signal, detection of maximums and / or minimums in the signal, exceeding predefined thresholds by part of the signal, correlation between signals, etc.

In a preferred embodiment, the characterization (131) of the signal (2.2), the determination (132) of whether the signal (2.2) corresponds to the development of a specific food process and the possible classification of said signal (2.2) are carried out by means of a pre-trained machine learning algorithm. In the pre-training phase, the machine learning algorithm is fed with training signals classified in predefined food process categories, as well as with additional data and labels obtained during the repeated recording of said predefined food process. In this way the machine learning algorithm is trained, preserving the individual and joint temporal structure

1

of the information obtained from the sensors and their labels. In a preferred embodiment, the machine learning algorithm is supervised and is based on a neural network and / or a random forest and / or vector support machine with established coding and temporality requirements. The result of the training is an algorithm capable of extracting the temporal structure of events in the registered olfactory signals and comparing it with its representation of predefined food processes. Once the machine learning algorithm has been trained, by providing it with a registered olfactory signal and optionally additional data, the machine learning algorithm is able to determine (132) if a signal corresponds to the development of a specific food process through its classification, that is, classifies that it belongs to a certain category of food processes or identifies that it does not correspond to any of these predefined food processes. Additionally, training can be improved by feeding back the machine learning algorithm with the tags automatically generated by said algorithm.

In one embodiment, data acquisition is performed during the time the food process is taking place.

With the method of the invention described in figure 2, among other applications, it can be determined that the signals correspond to different categories of food processes:

- food processing accelerated or intervened in a way other than desirable by the consumer, for example, drinks with accelerated aging by ultrasound;

- accelerated or intervened food process that violates the legislation, for example, accelerated ripening of the fruit using acetylene;

- food process with guarantee seal;

- unhealthy food production process;

- food process of maturation, curing or fermentation that has followed a certain pattern;

- food process of maturation, curing or fermentation of inferior or superior quality to that of other similar products;

- food process of maturation, curing or fermentation with designation of origin.

Figure 3 shows an example of characterization of the temporal evolution of a fruit ripening process, bananas in this particular example. In this example, we start from two groups of fruit from the same bunch, which have been subjected to processes of

1

different preservation prior to monitoring their maturation. In particular, the first group (A) was kept at room temperature for a period of 48 h, and the second group (B) was kept that time refrigerated at 5 ° C.

For the experiment, both groups (A, B) were placed in separate boxes monitored with two identical systems according to an embodiment of the invention. Figure 3 shows the temporal evolution of the signal recorded by two olfactory sensors, identified in the figure as "Sensor type 1" and "Sensor type 2" during a period of 12 days. Simultaneously, to associate the degree of maturation with the evolution of the signals, the system took photographs of both groups at different times, synchronously recording the temporality of the signals and the photographs. In this example the signal corresponding to group A would correspond to an undisturbed maturation process, while the signal corresponding to group B would correspond to an inhibited maturation process. The sampling period of the sensors in this example was 1 minute. The figure shows the different time evolution of the signals collected by the sensors in the two cases, as well as the different time evolution of each of the sensors and the difference in the level of the signals at each stage of maturation. In particular, there is a clear divergence between the signals of each group from the fourth day of maturation.

Therefore, the characteristic evolution of the signals in food processes can be used as a reference for the certification of the ripening process. In particular, in relation to the example shown in figure 3, the olfactory signals make it possible to distinguish a natural maturation process from an inhibited maturation process.

The system and the method of the invention are equally applicable to characterize and certify other types of food processes in addition to the ripening of fruit, such as the curing process, the aging process, the fermentation process, the smoking process, the baking process and / or or roasting process. In the same way, the system and the method of the invention are applicable to characterize and certify the evolution of food processes such as the cultivation process or the growth process of a food. The invention allows identifying that a food process has taken place, as well as classifying it in a specific type of food process, certifying that the process has developed according to predefined conditions and / or detecting the deviation of the food process with respect to said predefined conditions, allowing thus identify accelerated, intervened, fraudulent and / or unhealthy food processes.

2

Claims (26)

1. - Method (100) to characterize and certify the evolution of food processes by a system (1) comprising: • a detection module (2) comprising at least one gaseous component measurement element (2.1) configured to generate at least one signal (2.2) of the time evolution of at least one detected gaseous component; • an analysis module (3) configured to characterize the at least one signal (2.2) generated by the detection module (2) based on its temporal evolution and to determine if said at least one signal (2.2) corresponds to the development of a specific food process; wherein the method (100) comprises the following steps: a) generating (110) at least one signal (2.2) of the time evolution of at least one gaseous component detected by the at least one measuring element (2.1) of the detection module (2) during a predetermined period of time; b) receiving (120), by the analysis module (3), the generated signal (2.2); c) characterizing (131) the signal (2.2) based on its time evolution, by the analysis module (3); Y d) determine (132) by the characteristics of the signal (2.2) if its temporal evolution corresponds to a specific food process. 2. - Method (100) according to the preceding claim, wherein the system (1) comprises a conditioning module (4) configured to condition the signal (2.2) generated by the detection module (2); and wherein the method (100) additionally comprises, between stages (a) and (b), the stages of receiving and conditioning the signal (2.2) by the conditioning module (4), and wherein stage (c ) is performed on the generated and conditioned signal. 3. - Method (100) according to the preceding claim, wherein the step of conditioning the signal (2.2) comprises filtering and / or sampling said signal (2.2). 4. - Method (100) according to any of the preceding claims, wherein steps (c) and (d) are performed by means of a machine learning algorithm previously trained with training signals (3.3) that correspond to at least one food process . 5. - Method (100) according to any of the preceding claims, wherein the food process belongs to one of the following categories: - ripening process; or - curing process; or - aging process; or - fermentation process; or - smoking process; or - cultivation process; or - growing process; or - baking process; or - roasting process; or - accelerated and / or intervened food process; or - unhealthy food process; or - a combination of any of the above. 6. - Method (100) according to any of the preceding claims, wherein the method additionally comprises classifying, by the analysis module (3), the at least one signal (2.2) in at least one subcategory of food processes. 7. - Method (100) according to claim 6, wherein the classification of the signal (2.2) is performed by means of a machine learning algorithm previously trained with training signals (3.3) that are classified into at least one predefined subcategory of process food. 8. - Method (100) according to any of the preceding claims, wherein step (c) comprises identifying at least one temporal event (3.2) in the signal (2.2), and where the determination of step (d) is performed based on the sequentiality of the identified temporal events (3.2). 9. - Method (100) according to claims 6 and 8, wherein the classification of the signal (2.2) is carried out based on the sequentiality of the identified temporal events (3.2). 10. - Method (100) according to any of claims 8 to 9, wherein the temporary events (3.2) detected in the signal (2.2) include maximums, minimums, slopes, exceeding thresholds, and / or sequential sets of the above that define a specific time structure characteristic of the evolution of the food process. 11. - Method (100) according to any of the preceding claims, wherein the at least one measurement element (2.1) of gaseous components is an olfactory sensor, said olfactory sensor preferably being configured to detect substances emitted by food; the substance preferably being at least one of: Volatile Organic Compounds (VOCs), ethylene, sulfides, isovaleric acid, butyric acid, alcohols, flavorings, oxidases and ammonia. 12. - Method (100) according to claim 11, wherein the olfactory sensor is of any of the following types: chemoresistive, chemocapacitive, potentiometric, gravimetric, optical, acoustic, thermal, polymeric, amperometric, chromatographic, spectrometric or field effect . 13. - Method (100) according to any of the preceding claims, wherein: - the detection module (2) additionally comprises at least one additional sensor (2.3, 2.4, 2.5, 2.6, 2.7), preferably an ambient conditions sensor, a particle sensor, a contact sensor, a positioning sensor and / or an image sensor; - the method (100) additionally comprises a step of obtaining measurements of at least one magnitude by means of the at least one additional sensor and of identifying temporal events (3.2) of the signal (2.2) associated with said at least one magnitude; - and where said identified temporary events (3.2) are used as additional information during the characterization stage (131) of the signal (2.2). 14. - Method (100) according to the preceding claim, wherein: - at least one additional sensor is a sensor for environmental conditions (2.3), preferably humidity, temperature, atmospheric pressure, air speed, and / or salinity; me - at least one additional sensor is a particle sensor, preferably a spore and / or microplastic sensor; me - at least one additional sensor is a contact sensor, preferably a sensor for lipids, hardness, weight, mass, volume, glucose, sucrose, phosphates, and / or isomaltol; me 2 - at least one additional sensor is a positioning sensor; me - at least one additional sensor is an image sensor. 15. - Method (100) according to any of the preceding claims, wherein step (d) of the method (100) comprises comparing the signal (2.2) with at least one reference signal. 16. - Method (100) according to any of the preceding claims, wherein the steps of the method are repeated periodically, where the repetition period is a predefined value, to monitor at least one food process detected and / or classified. 17. - System (1) to characterize and certify the evolution of food processes, which includes: - a detection module (2) comprising at least one gaseous component measurement element (2.1) configured to generate at least one signal (2.2) of the time evolution of at least one detected gaseous component; - an analysis module (3) configured to characterize the at least one signal (2.2) generated by the detection module (2) based on its temporal evolution, to determine if said at least one signal (2.2) corresponds to the development of a specific food process; wherein the analysis module (3) is configured to carry out steps (b) to (d) of the method according to any of the preceding claims. 18. - System (1) according to claim 17, further comprising a conditioning module (4) configured to condition the signal (2.2) generated by the detection module (2). 19. - System (1) according to any of claims 17 or 18, wherein the analysis module (3) of the system (1) is additionally configured to classify the at least one signal (2.2) in at least one subcategory of processes food. 20. - System (1) according to any of claims 17 to 19, comprising at least one additional sensor (2.3, 2.4, 2.5, 2.6, 2.7), preferably a sensor of environmental conditions, a particle sensor, a sensor contact, a positioning sensor and / or an image sensor. 21. - System (1) according to claim 20, wherein: - at least one additional sensor is a sensor for environmental conditions (2.3), preferably humidity, temperature, atmospheric pressure, air speed, and / or salinity; me - at least one additional sensor is a particle sensor (2.4), preferably a spore and / or microplastic sensor; me - at least one additional sensor is a contact sensor (2.5), preferably a sensor for lipids, hardness, weight, mass, volume, glucose, sucrose, phosphates, and / or isomaltol; me - at least one additional sensor is a positioning sensor (2.6); me - at least one additional sensor is an image sensor (2.7). 22. - System (1) according to any of claims 17 to 21, wherein the at least one measuring element (2.1) of gaseous components is an olfactory sensor, said olfactory sensor preferably being configured to detect substances emitted by a food; the substance preferably being at least one of: Volatile Organic Compounds (VOCs), ethylene, sulfides, isovaleric acid, butyric acid, alcohols, flavorings, oxidases and ammonia. 23. - System (1) according to claim 22, wherein the olfactory sensor is of any of the following types: chemoresistive, chemocapacitive, potentiometric, gravimetric, optical, acoustic, thermal, polymeric, amperometric, chromatographic, spectrometric or field effect . 24. - A data processing system comprising means to carry out steps (b) to (d) of the method (100) according to any of claims 1 to 16. 25. - A computer program comprising instructions that, when the program is executed by a computer, causes the computer to carry out steps (b) to (d) of the method (100) according to any of claims 1 to 16 . 26. - A computer-readable medium comprising instructions that, when executed by a computer, causes the computer to carry out steps (b) to (d) of the method (100) according to any of claims 1 to 16. 2