KR102631970B1 - Measuring apparatus and method for viscosity of liquids - Google Patents

Abstract

One embodiment of the present invention outputs the viscosity of the test object according to the dissolution rate of the liquid test object through a sensor that measures the voltage that changes according to the speed at which the test object for viscosity testing spreads (melts) within the test object. Invent a system and method for estimating liquid viscosity.

Description

Liquid viscosity estimation system and method {MEASURING APPARATUS AND METHOD FOR VISCOSITY OF LIQUIDS}

The present invention relates to a liquid viscosity estimation system and method that can estimate the liquid viscosity so that the taste of the liquid can be classified. More specifically, the viscosity of the liquid test object is calculated based on machine learning based on the speed at which the liquid dissolves. It relates to an apparatus and method for estimating with a learning model.

The content described below is merely for the purpose of providing background information related to embodiments of the present invention, and does not necessarily constitute prior art.

Taste sensors are being developed for the purpose of quality inspection in the food industry. Most of the taste sensors developed to date detect taste substances using the principle of detecting taste substances by electrical or optical methods using simple chemical materials such as lipid membranes, metals, and polymer materials. It has poor selectivity and sensitivity, and has limitations in that it cannot reproduce the human taste system.

In particular, the taste classification of liquids recognized by humans is greatly influenced not only by the sensation felt by the taste organ but also by the viscosity of the tactile organ. Therefore, humans can perceive the same drink as a completely different taste or different food depending on its viscosity. However, current taste sensors have limitations in reflecting the viscosity of liquid because they reflect voltage values.

In addition, taste data does not respond only to a single sensor, but responds complexly to multiple sensors, so it is necessary to classify it according to viscosity and taste.

The above-mentioned background technology is technical information that the inventor possessed for deriving the present invention or acquired in the process of deriving the present invention, and cannot necessarily be said to be known art disclosed to the general public before filing the application for the present invention.

Prior Art 1: Korean Patent No. 10-1684620 (registered on December 8, 2016)

One embodiment of the present invention provides a liquid viscosity estimation system and method that can classify data on taste according to the viscosity of the liquid.

Additionally, an embodiment of the present invention provides a liquid viscosity estimation system and method that can classify types of liquids according to liquid viscosity.

The object of the present invention is not limited to the problems mentioned above, and other objects and advantages of the present invention that are not mentioned can be understood through the following description and will be more clearly understood by examples of the present invention. Additionally, it will be appreciated that the objects and advantages of the present invention can be realized by means and combinations thereof as indicated in the patent claims.

According to an embodiment of the present invention, a method of estimating liquid viscosity based on the viscosity of the liquid involves introducing a liquid test object into a liquid test object containing a sensor, and changing the test object according to the dissolution rate of the test object. Measures the voltage of the test object and reflects the dissolution rate of the test object in a machine learning-based solubility estimation model trained to output the degree of dissolution of the test object based on the measured voltage of the test object to determine the viscosity of the test object. It can be performed through the process of outputting .

According to an embodiment of the present invention, the dissolution rate of the test object is the change in concentration of the test object measured from the sensor from the point of introduction of the test object into the test object, and the change in concentration of the test object is the solubility of the test object. It may be a slope representing the voltage change of the test object extracted based on .

According to an embodiment of the present invention, the solubility estimation model may be trained with training data labeled with the dissolution rate of the test object.

According to an embodiment of the present invention, the training data for training the solubility estimation model can be generated, and the viscosity of the test object whose solubility degree is collected at regular time intervals can be sampled.

According to an embodiment of the present invention, when outputting the viscosity of the test object, any one level corresponding to the dissolution rate of the test object can be output from a list that is preset based on the solubility estimation model and classifies taste. .

The solubility estimation model of the embodiment of the present invention is based on a synthetic neural network, and the dissolution rate of the test object is related to the degree of change in concentration of the test object processed in the fully connected layer included in the solubility estimation model. It can be trained to output gradients.

The solubility estimation model according to an embodiment of the present invention may be a machine learning-based learning model generated separately for each liquid.

When outputting the viscosity of the test object according to an embodiment of the present invention, the liquid test object can be classified based on the viscosity of the test object output by the solubility estimation model.

According to an embodiment of the present invention, a device for estimating liquid viscosity based on the viscosity of a liquid includes a memory electrically connected to the processor and storing at least one code executed by the processor, When executed through the processor, the memory reflects the dissolution rate of the test object based on a machine learning-based solubility estimation model in which the processor is trained to output the degree of dissolution of the test object based on the measured test object voltage. Codes that cause the viscosity of the inspection object to be output may be stored.

Other aspects, features, and advantages in addition to those described above will become apparent from the following drawings, claims, and detailed description of the invention.

An embodiment of the present invention is to output the viscosity of the test object according to the dissolution rate of the liquid test object through a sensor that measures the voltage that changes according to the speed at which the test object for viscosity testing spreads (melts) within the test object. You can. In particular, the viscosity of the test object can be estimated based on a solubility estimation model trained with learning data generated by labeling the dissolution rate of the test object.

Additionally, embodiments of the present invention enable classifying test objects by liquid according to the estimated viscosity of the test object.

Furthermore, it is possible to classify test objects by viscosity and classify taste according to liquid viscosity.

The effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

1 is a diagram schematically explaining the environment of a liquid viscosity estimation system according to an embodiment of the present invention.
Figure 2 is a block diagram showing the configuration of a server for estimating liquid viscosity according to an embodiment of the present invention.
Figure 3 is a block diagram showing the configuration of a sensor connected to communication with a server for estimating liquid viscosity according to an embodiment of the present invention.
Figure 4 is a diagram illustrating an example of generating a solubility estimation model for estimating liquid viscosity according to an embodiment of the present invention.
Figures 5 and 6 are flowcharts showing a method for estimating liquid viscosity according to an embodiment of the present invention.

Hereinafter, embodiments invented in the present specification will be described in detail with reference to the accompanying drawings. However, identical or similar components will be assigned the same reference numbers regardless of reference numerals, and duplicate descriptions thereof will be omitted. The suffixes “module” and “part” for components used in the following description are given or used interchangeably only for the ease of preparing the specification, and do not have distinct meanings or roles in themselves. Additionally, in describing the embodiments invented in this specification, if it is determined that a detailed description of related known technologies may obscure the gist of the embodiments invented in this specification, the detailed description will be omitted. In addition, the attached drawings are only for easy understanding of the embodiments invented in the present specification, and the technical idea invented in the present specification is not limited by the attached drawings, and are included in the spirit and technical scope of the present invention. It should be understood to include all modifications, equivalents and substitutions.

Terms containing ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

When a component is said to be "connected" or "connected" to another component, it is understood that it may be directly connected to or connected to the other component, but that other components may exist in between. It should be. On the other hand, when it is mentioned that a component is “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between.

Singular expressions include plural expressions unless the context clearly dictates otherwise.

In this application, terms such as “comprise” or “have” are intended to designate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

1 is a diagram schematically illustrating an environment for estimating liquid viscosity according to an embodiment of the present invention.

The liquid viscosity estimation system 1 can be implemented in a device that can measure taste, such as an electronic tongue or an electronic tongue equipped with a taste sensor. Additionally, it can be implemented as an electronic device (e.g., personal terminal, server) connected to a sensor mounted on the electronic tongue and wired or wireless communication.

In an embodiment of the present invention, the liquid viscosity estimation system 1 is implemented through the server 100, and the server 100 and the electronic tongue are communicated to determine the viscosity of the liquid through the liquid viscosity estimation model stored in the server 100. Let us explain with an example of estimation.

Specifically, in one embodiment, the liquid viscosity estimation system 1 can estimate the viscosity of the liquid and classify the taste of the liquid according to viscosity based on the estimated viscosity of the liquid.

As described, the liquid viscosity estimation system 1 according to an embodiment of the present invention can be implemented through an electronic tongue. For example, the liquid viscosity estimation system 1 may be configured to accommodate drinkable liquid. In particular, the test object 20 for estimating viscosity may be provided in a liquid form, and the test object 20 may be placed within the liquid test object 10 into which the liquid test object 20 is placed. A sensor 50 may be provided that measures the change in concentration of the test object 10 as a voltage according to the speed of spreading.

That is, if the viscosity of the liquid test object 20 is high, the speed at which the test object 20 spreads on the test object 10 may be slow. On the contrary, the lower the viscosity of the test object 20, the faster the spreading speed in the test object 10 may be. Therefore, the viscosity of the test object 20 can be estimated according to the speed at which the test object 20 melts in the test object 10.

At this time, a plurality of sensors 50 provided on the test object 10 may be provided, and the sensor 50 samples the viscosity of the test object 20 at regular time intervals and corresponds to a preset taste classification (list). Therefore, the level of taste can be classified according to viscosity.

Meanwhile, the liquid viscosity estimation system 1 may include a solubility estimation model. As the test object 20 dissolves in the test object 10, a change in concentration of the test object 10 may occur. This change in concentration of the test object 10 can be measured as a voltage value through the sensor 50. The model that learned the measured voltage value may be a solubility estimation model. In other words, the solubility estimation model allows predicting the viscosity of the test object 20 according to the change in voltage value that occurs as the test object 20 spreads on the test object 10.

Specifically, the solubility estimation model can be created by learning the slope that changes until the sensor 50 converges to a preset voltage value. Here, the solubility estimation model can be learned by adding a fully connected layer to a deep neural network, but the embodiment is not limited to this.

Figure 2 is a block diagram showing the configuration of a server for estimating liquid viscosity according to an embodiment of the present invention, and Figure 3 is a configuration of a sensor connected to communication with a server for estimating liquid viscosity according to an embodiment of the present invention. This is the block diagram shown.

Before explaining the drawings, for convenience of explanation, the device for estimating liquid viscosity will be described as an example implemented in an electronic tongue, and the sensor 50 will be described as an example included in the liquid test object 10. Do this.

Additionally, a learning model for estimating liquid viscosity can be downloaded from the server 100, and the downloaded learning model can be implemented through the processor of the electronic tongue. Alternatively, the learning model may be stored in the processor of the electronic tongue, and the viscosity of the liquid may be measured based on the stored learning model.

Referring to FIG. 2 , the server 100 may estimate the viscosity of the liquid test object 20 based on the voltage value measuring the change in concentration of the test object 10 measured by the sensor 50.

This server 100 may be configured to include a communication unit 120, memory 160, learning processor 150, and processor 180.

The communication unit 120 of the server 100 is configured to be connected to the sensor 50 for communication, and receives the change in concentration of the test object 10 according to the degree of melting of the test object 20 acquired by the sensor 50. You can.

Memory 160 may include a model storage unit 161. The model storage unit 161 may store the pre-learning or learned model 161a through the learning processor 150. Learning models, including learned models or pre-learning models, are not particularly limited in type, such as those based on deep neural networks.

The learning processor 150 may train the learning model 161a using training data. The learning model can be used while mounted on the server 100 or sensor 50.

Learning models can be implemented in hardware, software, or a combination of hardware and software. When part or all of the learning model is implemented as software, one or more instructions constituting the learning model may be stored in the memory 160.

The learning model of the present invention is a solubility estimation model of a liquid viscosity estimation system based on machine learning, and the solubility estimation model is a solubility estimation model that begins when the test object 20 flows into the test object 10. It is dissolved in 10) and can be trained with training data that labels the change concentration of the test object 10 with the change slope of the input voltage.

Specifically, the degree of dissolution of the test object 20 applied to the solubility estimation model depends on the speed at which the test object 20 spreads on the test object 10 from the time the test object 20 is introduced into the test object 10. Accordingly, it may be a change in concentration of the test object 10. Here, the concentration change may be a voltage value measured according to the concentration change of the test object 10, and the voltage value may converge to a preset value (eg, 1 or 0). Additionally, the change in concentration of the test object 10 may be a slope representing the change in voltage of the test object 10 measured according to the solubility of the test object 20.

For example, when the test object 20, which is a liquid to be measured, is dissolved in the test object 10, the concentration of the liquid in the test object 10 may increase as the test object 20 spreads. Accordingly, the measurement value measured by the sensor 50 may increase. In other words, the slope of the value measured by the sensor 50 can be said to be the speed at which the inspection object 20 to be measured spreads from the inspection object 10.

Here, as the viscosity of the test object 20 increases, the test object 20 spreads more slowly within the test object 10, so the change in slope measured by the sensor 50 may be smaller. In other words, the slope measured by the sensor 50 can be said to reflect the viscosity of the inspection object 20.

In an embodiment, the slope of the change in voltage measured by the sensor 50 may be added to the fully connected layer of the deep neural network, and specific examples are described in detail below.

When the sensor 50 measures a change in concentration of the test object 10, the measured change in concentration of the test object 10 can be input into a preset solubility estimation model. At this time, the solubility estimation model is trained with training data labeled with the degree of change in concentration of the test object 10 according to the speed at which the test object 20 is dissolved in the test object 10.

Specifically, the training data collects the voltage change (slope) according to the concentration change of the test object 10 measured by the sensor 50 described above at regular time intervals or in real time/quasi-real time, and the collected test object 10 It can be sampled with a slope that represents the change in voltage according to the change in concentration.

That is, the training data of the present invention can be, for example, the sensor 50 collects the change in concentration of the test object 10 as a voltage value at 1-second intervals for 5 minutes, and converts the collected data into a voltage value that changes every second. It can be generated based on the slope of .

Thereafter, training data may be generated until the voltage value measured by the sensor 50 converges to a preset voltage value. For example, assuming that the convergence value of the sensor 50 is a voltage value of 10, training data can be generated until the voltage value measured by the sensor 50 converges to 10.

On the other hand, the solubility estimation model according to the embodiment of the present invention adds the slope value that changes until the voltage value measured by the sensor 50 converges to a preset value to the fully connected layer in the deep neural network to estimate the value measured by the sensor 50. It may be a model trained to output a value related to the speed at which the test object 20 spreads in relation to the voltage value.

The solubility estimation model generated in this way may be a learning model generated in response to the taste corresponding to the test object whose viscosity is to be tested. In an embodiment, the learning model spreads from the time the test object to be tested is introduced into the test object, and may map the viscosity of the test object using the change in concentration of the test object as an input with a list that classifies taste.

For example, when a person actually drinks the taste list corresponding to the viscosity degree of the first test object whose viscosity is to be tested in accordance with the viscosity degree of the first test object whose viscosity is to be tested in response to the dissolution rate of the test object. It is possible to output tastes that can be felt.

At this time, training data on the solubility of various test objects can be generated to ensure the accuracy of the solubility estimation model. For example, after mapping and storing the dissolution rate of the test object for which training data was generated and the preset taste information, the learning model is trained to estimate the viscosity of the test object, and then the test object for which training data has not been generated is inspected. When the test object flows into the body, the slope of the voltage value that changes with respect to the speed at which the test object spreads in the test object is input, and the learning model can be trained again by adding it to the training data.

The processor 180 may infer a result value for new input data using a learning model and generate a response or control command based on the inferred result value.

The processor 180 may evaluate the viscosity of the inspection object by applying a learning model to information about the speed at which the inspection object spreads on the inspection object received from the sensor 50 through the communication unit 120.

Referring to FIG. 3, the sensor 50 may include a communication unit 52, a sensing unit 54, a memory 56, and a processor 58.

The communication unit 52 may be connected to the communication unit 120 of the server 100. Specifically, when the test object 10 flows into the test object 20, the communication unit 52 transmits a voltage value according to the concentration change of the test object 10 according to the solubility of the test object 20 to the server 100. do.

The sensing unit 54 is configured to sense a change in concentration of the test object 10 as the test object 20 melts in the test object 10. Specifically, the sensing unit 54 can sense the voltage value that measures the change in concentration of the test object 10, and in the process of spreading the test object 20 flowing into the test object 10, the test object ( 10) The change in concentration itself can also be sensed. Hereinafter, in an embodiment of the present invention, the sensing unit 54 will be described as an example of sensing the voltage value at which the test object 10 measures a change in concentration.

At this time, the sensing unit 54 can collect the change in concentration of the test object 10 as a voltage value at 1-second intervals for 5 minutes, and send the collected data to the server 100 as the slope of the voltage value that changes every second. Can be sent.

The memory 56 may store the test object viscosity measurement conditions for measuring the viscosity of the test object 20 through the change in concentration of the test object 10 in the sensing unit 54. That is, the memory 55 can store data and commands for the operation of the sensing unit 54 that can measure the viscosity of the inspection object.

The processor 58 may control the overall operation of the sensor 50 in addition to operations related to the application program stored in the memory 56. The processor 58 processes signals, data, information, etc. input or output through the components discussed above or runs an application program stored in the memory 56 to provide or process appropriate information or functions as a liquid viscosity estimation system. make it possible

Figure 4 is a diagram illustrating an example of generating a solubility estimation model for estimating liquid viscosity according to an embodiment of the present invention.

Prior to the description of the drawing, the embodiment of the present invention processes the viscosity of the inspection object 20 in a deep neural network by entering a value reflecting the viscosity of the inspection object 20 as an input value in addition to the voltage value measured by the sensor 50. Let it be used as a piece of information to be used. This allows classification that is close to the taste that humans intuitively feel.

Specifically, if we look at the process of generating training data with reference to the drawing, liquid can be measured by dissolving it in a solution to measure it with a sensor. That is, the embodiment of the present invention can measure the change value in which the concentration of the test object 10 changes over time by placing the test object 20 in the test object 10. That is, the sensor 50 can measure the concentration of the test object 10 that changes with time as a voltage value, and the measured voltage value can be extracted as a slope that changes with time. The slope for the extracted voltage value may be an input value input to a deep neural network to estimate the viscosity of the test object.

Unlike this, the input value input to the deep neural network may be change information on the test object 10 in addition to the concentration change measured by the sensor 50, and may be information on changes in the test object 10 measured by the sensor 50 and the voltage value other than the voltage value. Training data can be generated using sensor measurements as input.

Specifically, the sensor 50 can collect the concentration change of the test object 10 as a voltage value at 1-second intervals for 5 minutes, and the collected data can be generated based on the slope of the voltage value that changes every second. there is.

Thereafter, training data may be generated until the voltage value measured by the sensor 50 converges to a preset voltage value. For example, assuming that the convergence value of the sensor 50 is a voltage value of 10, training data can be generated until the voltage value measured by the sensor 50 converges to 10.

Figures 5 and 6 are flowcharts showing a method for estimating liquid viscosity according to an embodiment of the present invention.

Referring to the drawing, the viscosity of the test object can be measured as the liquid test object flows into and dissolves the liquid test object. Specifically, the viscosity of the test object may change when the test object flows into the test object, and the concentration of the test object may change as the test object dissolves in the test object. The change in concentration of the test object can be measured as a voltage value at regular intervals (S110 S120).

Here, the concentration change of the test object can be collected at 1 second intervals for a certain period of time, until the voltage value measured by the sensor 50, which is when the test concentration of the test object is higher than the preset concentration, converges to the preset voltage value. It can be assumed that

By adding the measured voltage value to the fully connected layer of the deep neural network, a value related to the speed at which the test object 20 spreads can be output for the voltage value measured by the sensor 50. In other words, the learning model created in response to the taste corresponding to the test object whose viscosity is to be tested can be said to be a solubility estimation model.

As the liquid test object 20 whose viscosity is to be measured is introduced into the liquid test object 20, the speed at which the test object 20 spreads on the test object 10 can be mapped to a list that separates tastes. . According to the mapped conditions, the viscosity of the test object 20 can be output according to the liquid viscosity of the test object 20 (S130).

In particular, when the degree of dissolution in the test object of a drinkable liquid test object is learning data, not only the viscosity of the test object but also the taste characteristics according to the liquid viscosity are classified through the dissolution rate of the test object measured by the server 100. You may.

In other words, the viscosity can be compared with the taste list corresponding to the test object to be tested (S122). In an embodiment, the viscosity of the test object is estimated by mapping the viscosity of the test object, which spreads from the time the test object to be tested flows into the test object and takes the change in concentration of the test object as an input, with a list that distinguishes taste, and the viscosity of the test object is estimated. Flavors can be printed (S124, S126).

Specifically, the collected learning data extracted by viscosity can be used for learning (training), and based on the data extracted by viscosity, the taste of the same drinkable test object can be classified according to viscosity. As a result, the accuracy of the model for estimating the solubility of liquids for each viscosity can be evaluated.

The learning data generated in this way can be used to train a learning model on a personal device (e.g., a mobile terminal, tablet, smart watch, etc.) or can be transmitted to a server and used to train a learning model.

Therefore, the learning data can be subdivided according to the viscosity of the liquid, and the taste can be classified according to the viscosity of the subdivided liquid, allowing the taste and type of food to be classified according to viscosity even in the same beverage.

As a result, it is possible to classify drinks according to the subjective viscosity felt by the drinker, making it possible to more accurately classify taste according to the characteristics of viscosity.

The description of the embodiments of the present invention described above is for illustrative purposes, and a person skilled in the art of the present invention can easily transform it into another specific form without changing the technical idea or essential features of the present invention. You will be able to understand that Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive. For example, each component described as single may be implemented in a distributed manner, and similarly, components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention.

Claims (15)

1. A method wherein at least a portion of each step is performed by a processor, and wherein the method estimates liquid viscosity based on the viscosity of the liquid, comprising:
Introducing a liquid test object into a liquid test object containing a sensor;
Measuring the voltage of the test object that changes according to the dissolution rate of the test object; and
A step of outputting the viscosity of the test object by reflecting the dissolution rate of the test object in a machine learning-based solubility estimation model trained to output the degree of dissolution of the test object based on the measured voltage of the test object,
The dissolution rate of the test object is the change in concentration of the test object measured from the sensor from the point of introduction of the test object into the test object, and the change in concentration of the test object is the change in concentration of the test object extracted based on the solubility of the test object. The slope representing the change in voltage,
Methods for estimating liquid viscosity.
delete According to paragraph 1,
The solubility estimation model is trained with training data labeled with the dissolution rate of the test object,
Methods for estimating liquid viscosity.
According to paragraph 3,
Generating the training data for training the solubility estimation model,
The step of generating the training data is,
Comprising the step of sampling the viscosity of the test object by collecting the degree of dissolution of the test object at regular time intervals,
Methods for estimating liquid viscosity.
According to paragraph 1,
The step of outputting the viscosity of the test object is,
Comprising the step of outputting any one level corresponding to the dissolution rate of the test object from a list that is preset based on the solubility estimation model and classifies taste.
Methods for estimating liquid viscosity.
According to paragraph 1,
The solubility estimation model is based on a synthetic neural network, and the dissolution rate of the test object is such that it outputs a slope for the degree of concentration change of the test object processed in the fully connected layer included in the solubility estimation model. trained,
Methods for estimating liquid viscosity.
According to clause 6,
The solubility estimation model is a machine learning-based learning model created separately for each liquid,
Methods for estimating liquid viscosity.
According to paragraph 1,
After the step of outputting the viscosity of the test object,
Further comprising classifying the liquid test object based on the viscosity of the test object output by the solubility estimation model,
Methods for estimating liquid viscosity.
A device for estimating liquid viscosity based on the viscosity of the liquid, comprising:
It is electrically connected to the processor and includes a memory that stores at least one code executed by the processor,
When the memory is executed through the processor, the processor
After introducing a liquid test object into a liquid test object containing a sensor, the test stand is based on a machine learning-based solubility estimation model trained to output the degree of dissolution of the test object based on the voltage measured from the test object. Stores codes that cause the viscosity of the inspection object to be output by reflecting the dissolution rate of the upper body,
The dissolution rate of the test object is the change in concentration of the test object measured from the sensor from the point of introduction of the test object into the test object, and the change in concentration of the test object is the change in concentration of the test object extracted based on the solubility of the test object. The slope representing the change in voltage,
Liquid viscosity estimation system.
delete According to clause 9,
The solubility estimation model is trained with training data labeled with the dissolution rate of the test object,
Liquid viscosity estimation system.
According to clause 11,
The training data is,
By sampling the viscosity of the test object, the degree of dissolution of the test object was collected at regular time intervals,
Liquid viscosity estimation system.
According to clause 9,
The memory is the processor,
Further storing codes that cause output of one level corresponding to the dissolution rate of the test object among the list that is preset based on the solubility estimation model and classifies taste,
Liquid viscosity estimation system.
According to clause 9,
The solubility estimation model is based on a synthetic neural network, and the dissolution rate of the test object is such that it outputs a slope for the degree of concentration change of the test object processed in the fully connected layer included in the solubility estimation model. trained,
Liquid viscosity estimation system.
According to clause 14,
The solubility estimation model is a machine learning-based learning model created separately for each liquid,
Liquid viscosity estimation system.

Images