Electronic device and control method thereof

Abstract

This electronic device comprises: a memory storing at least one instruction; a communication device; and at least one processor connected to the communication device and the memory so as to control the electronic device, wherein the at least one processor: acquires a plurality of datasets including temperature information corresponding to each of at least one storage compartment type from among a plurality of storage compartment types, on the basis of input data received via the communication device; and, on the basis of a first type dataset including temperature information corresponding to each of the plurality of storage compartment types among the plurality of datasets, acquires augmented data for input to at least one neural network model in order to train the at least one neural network model so as to output a predicted temperature corresponding to at least one of the plurality of storage compartment types by augmenting a second type dataset including temperature information corresponding to some of the plurality of storage compartment types from among the plurality of datasets.

Description

Electronic device and control method thereof

[0001] The present disclosure relates to an electronic device and a method for controlling the same, and more specifically, to an electronic device and a method for controlling the same that augments training data for training a neural network model for temperature prediction.

[0002] With the advancement of electronic technology, the use of electronic products capable of predicting the state of electronic devices in advance based on their state history is increasing.

[0003] In particular, recently in the field of home appliances such as refrigerators, artificial intelligence models capable of predicting the temperature of storage compartments by utilizing temperature records are being developed.

[0004] In the case of such artificial intelligence models, they can be trained to predict the temperature when a specific event occurs by using the usage history and status history of home appliances, such as refrigerators, as training data.

[0005] An electronic device according to one embodiment of the present disclosure comprises a memory for storing at least one instruction, a communication device, and at least one processor connected to the communication device and the memory to control the electronic device, wherein the at least one processor acquires a plurality of datasets containing temperature information corresponding to each of at least one storage room type among a plurality of storage room types based on input data received through the communication device, and acquires augmentation data to input to the at least one neural network model to train the at least one neural network model to output a predicted temperature corresponding to at least one of the plurality of storage room types by augmenting a second type dataset containing temperature information corresponding to some of the plurality of storage room types among the plurality of datasets based on a first type dataset containing temperature information corresponding to each of the plurality of storage room types among the plurality of datasets.

[0006] The above plurality of storage room types include a refrigerator room, a freezer room, and a variable temperature room, and some of the above plurality of storage room types may include at least one of the refrigerator room and the freezer room.

[0007] The above at least one processor can identify a first section in the first type dataset where a temperature change event of the temperature room occurs based on temperature information included in the first type dataset, identify a second section in the second type dataset that has a pattern similar to the temperature information included in the identified section, and record the temperature information corresponding to the temperature room in the first section as the temperature information corresponding to the temperature room in the second section.

[0008] The above at least one processor can identify a first interval in which the temperature change event occurred based on at least one bit corresponding to whether the temperature change room is defrosting among the temperature information included in the first type dataset.

[0009] The temperature information included in the first type dataset includes a plurality of statistical values ​​arranged in a time-series according to a plurality of time-intervals for each of the plurality of storage rooms, and the at least one processor identifies a plurality of time-intervals that have changed by more than a threshold value from the statistical value corresponding to the previous time-interval among the plurality of statistical values ​​corresponding to the variable temperature room, and based on the identified plurality of time-intervals, identifies a plurality of time-intervals in which the temperature change event occurred, and can identify a section including the plurality of time-intervals in which the temperature change event occurred as a first section.

[0010] The above at least one processor can identify a first section in which a first temperature change event of the variable temperature room occurs among the temperature information included in the first type dataset, identify the second section among the second temperature information included in the second type dataset, identify a third section in which a second temperature change event different from the first temperature change event occurs among the temperature information included in the first type dataset, and record temperature information corresponding to the refrigerator room and freezer room among the third section as temperature information corresponding to the refrigerator room and freezer room among the second type dataset.

[0011] The above at least one processor can identify a fourth section in which a temperature change event occurs in at least one of the refrigerator room and freezer room among the temperature information included in the first type dataset, and record the temperature information included in the fourth section as temperature information corresponding to at least one of the refrigerator room and freezer room in the second type dataset.

[0012] The above at least one processor can identify a mode corresponding to at least one of the refrigerator and freezer based on temperature information included in a section excluding the fourth section among the temperature information included in the first type dataset, record the identified mode as temperature information corresponding to at least one of the refrigerator and freezer among the second type dataset, and record the temperature information included in the fourth section as temperature information corresponding to at least one of the refrigerator and freezer among the section excluding the section where the mode is recorded among the second type dataset.

[0013] The electronic device further includes a plurality of temperature sensors for detecting the temperature inside a plurality of storage rooms, and the at least one processor acquires a plurality of partial datasets including a plurality of temperature statistical values ​​corresponding to each of the at least one storage room type based on the input data and sensing data acquired through the plurality of temperature sensors, and can cluster the plurality of datasets according to data similarity based on the plurality of statistical values ​​included in each of the plurality of partial datasets.

[0014] The above temperature statistical values ​​include a mean, a variation, and a range, and each of the mean, variation, and range is a value calculated based on a temperature measurement value included in the input data or the sensing data, and the at least one processor can cluster into the same cluster if the data distance between two of the plurality of partial datasets is less than a predetermined distance.

[0015] A method for controlling an electronic device according to one embodiment of the present disclosure comprises the steps of: acquiring a plurality of datasets including temperature information corresponding to each of at least one storage room type among a plurality of storage room types based on input data; and acquiring augmentation data to input to at least one neural network model to train the neural network model to output a predicted temperature corresponding to at least one of the plurality of storage room types by augmenting a second type dataset including temperature information corresponding to some of the plurality of storage room types among the plurality of datasets based on a first type dataset including temperature information corresponding to each of the plurality of storage room types among the plurality of datasets.

[0016] The above plurality of storage room types include a refrigerator room, a freezer room, and a variable temperature room, and some of the above plurality of storage room types may include at least one of the refrigerator room and the freezer room.

[0017] The step of acquiring the augmented data may include identifying a first section in the first type dataset where a temperature change event of the temperature-changing room occurred based on temperature information included in the first type dataset, identifying a second section among the temperature information included in the second type dataset that has a pattern similar to the temperature information included in the identified section, and recording the temperature information corresponding to the temperature-changing room in the first section as the temperature information corresponding to the temperature-changing room in the second section.

[0018] The step of identifying the second section may include identifying the first section in which the temperature change event occurred based on at least one bit corresponding to whether the temperature change room is in operation among the temperature information included in the first type dataset.

[0019] The temperature information included in the first type dataset includes a plurality of statistical values ​​arranged in a time-series according to a plurality of time-intervals for each of the plurality of storage rooms, and the step of identifying the second interval may include the step of identifying a plurality of time-intervals that have changed by more than a threshold value from the statistical value corresponding to the previous time-interval among the plurality of statistical values ​​corresponding to the variable temperature room, and the step of identifying a plurality of time-intervals in which the temperature change event occurred based on the identified plurality of time-intervals, and identifying the interval including the plurality of time-intervals in which the temperature change event occurred as the first interval.

[0020] The step of identifying the second section includes identifying the first section in which a first temperature change event of the variable temperature room occurred among the temperature information included in the first type dataset, and identifying the second section among the second temperature information included in the second type dataset; and the step of acquiring the augmented data may include identifying the third section in which a second temperature change event different from the first temperature change event occurred among the temperature information included in the first type dataset, and recording the temperature information corresponding to the refrigerator room and freezer room among the third section as the temperature information corresponding to the refrigerator room and freezer room among the second type dataset.

[0021] The step of acquiring the augmented data may include identifying a fourth section in which a temperature change event occurs in at least one of the refrigerator and freezer rooms among the temperature information included in the first type dataset, and recording the temperature information included in the fourth section as temperature information corresponding to at least one of the refrigerator and freezer rooms in the second type dataset.

[0022] The step of recording temperature information included in the fourth section may include: identifying a mode corresponding to at least one of the refrigerator and freezer based on temperature information included in a section excluding the fourth section among the temperature information included in the first type dataset; recording the identified mode as temperature information corresponding to at least one of the refrigerator and freezer among the second type dataset; and recording the temperature information included in the fourth section as temperature information corresponding to at least one of the refrigerator and freezer among the sections excluding the section where the mode is recorded among the second type dataset.

[0023] The step of acquiring multiple datasets may include the step of acquiring multiple partial datasets including multiple temperature statistical values ​​corresponding to each of the at least one storage room type based on the input data and sensing data, and the step of clustering the multiple datasets according to data similarity based on the multiple statistical values ​​included in each of the multiple partial datasets.

[0024] A non-transient computer-readable recording medium storing computer instructions that cause the electronic device to perform an operation when executed by a processor of an electronic device according to one embodiment of the present disclosure, wherein the operation comprises: a step of acquiring a plurality of datasets corresponding to each of a plurality of device types classified according to a plurality of storage room types based on received input data; and a step of acquiring augmentation data to input to at least one neural network model to train the neural network model to output a predicted temperature corresponding to at least one of the plurality of storage room types by augmenting a second type dataset, which includes temperature information corresponding to some of the plurality of storage room types among the plurality of datasets, based on a first type dataset containing temperature information corresponding to each of the plurality of storage room types among the plurality of datasets.

[0025] FIG. 1 is a drawing for explaining the operation of an electronic device and an external electronic device according to one or more embodiments of the present disclosure.

[0026] FIG. 2 is a block diagram for explaining the configuration of an electronic device according to one or more embodiments of the present disclosure.

[0027] FIG. 3 is a flowchart for explaining the operation of an electronic device according to one or more embodiments of the present disclosure.

[0028] FIG. 4 is a drawing for explaining an operation to augment learning data according to one or more embodiments of the present disclosure.

[0029] FIG. 5 is a drawing for illustrating a dataset according to one or more embodiments of the present disclosure.

[0030] FIG. 6 is a drawing for illustrating a dataset according to one or more embodiments of the present disclosure.

[0031] FIG. 7 is a drawing for explaining an operation to augment learning data according to one or more embodiments of the present disclosure.

[0032] FIG. 8 is a drawing for illustrating a dataset according to one or more embodiments of the present disclosure.

[0033] FIG. 9 is a drawing for illustrating training data and test data according to one or more embodiments of the present disclosure.

[0034] FIG. 10 is a drawing for illustrating training data and test data according to one or more embodiments of the present disclosure.

[0035] FIG. 11 is a drawing for illustrating a plurality of clusters in one or more embodiments of the present disclosure.

[0036] FIG. 12 is a flowchart illustrating a method for controlling an electronic device according to one or more embodiments of the present disclosure.

[0037] The various embodiments of the present disclosure and the terms used therein are not intended to limit the technical features described in the present disclosure to specific embodiments, and should be understood to include various modifications, equivalents, or substitutions of said embodiments.

[0038] In relation to the description of the drawings, similar reference numerals may be used for similar or related components.

[0039] The singular form of the noun corresponding to the item may include one or multiple items, unless the relevant context clearly indicates otherwise.

[0040] In the present disclosure, each of the phrases such as “A or B”, “at least one of A and B”, “at least one of A or B”, “A, B or C”, “at least one of A, B and C”, and “at least one of A, B, or C” may include any one of the items listed together in the corresponding phrase, or all possible combinations thereof.

[0041] The term "and / or" includes a combination of multiple related described components or any of the multiple related described components.

[0042] Terms such as "first," "second," or "first" or "second" may be used simply to distinguish a component from another corresponding component and do not limit the components in other aspects (e.g., importance or order).

[0043] Additionally, terms such as 'front,' 'rear,' 'top,' 'bottom,' 'side,' 'left,' 'right,' 'top,' and 'bottom' used in this disclosure are defined based on the drawings, and the shape and location of each component are not limited by these terms.

[0044] Terms such as "include" or "have" are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in this disclosure, and do not preclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0045] When it is said that a component is "connected," "combined," "supported," or "in contact" with another component, this includes not only cases where the components are directly connected, combined, supported, or in contact, but also cases where they are indirectly connected, combined, supported, or in contact through a third component.

[0046] When it is said that a component is located "on" another component, this includes not only cases where one component is in contact with the other, but also cases where another component exists between the two components.

[0047] Meanwhile, the various elements and areas in the drawings are depicted schematically. Accordingly, the technical concept of the present invention is not limited by the relative sizes or spacing depicted in the attached drawings.

[0048] Hereinafter, embodiments according to the present disclosure are described in detail with reference to the attached drawings so that those skilled in the art can easily implement them.

[0049] FIG. 1 is a drawing for explaining the operation of an electronic device and an external electronic device according to one or more embodiments of the present disclosure.

[0050] According to FIG. 1, an electronic device (100), external electronic devices (10-1, 10-2, ...), and a server device (20) are illustrated. Each of the electronic device (100), external electronic devices (10-1, 10-2, ...), and server device (20) may be implemented in the form shown in FIG. 1, but is not necessarily limited thereto and may be implemented in various forms.

[0051] The electronic device (100) can be implemented as a refrigerator. The electronic device (100) implemented as a refrigerator may include a main body, an interior, an insulating material, and a plurality of storage compartments.

[0052] The "main body" may include an inner body, an outer body positioned on the outside of the inner body, and an insulating material provided between the inner body and the outer body.

[0053] The "inner body" may include at least one of a case, plate, panel, or liner forming a storage chamber. The inner body may be formed as a single body or may be formed by assembling multiple plates. The "outer body" may form the exterior of the main body and may be coupled to the outer side of the inner body so that an insulating material is disposed between the inner body and the outer body.

[0054] The "insulating material" can insulate the interior and exterior of the storage room so that the temperature inside the storage room is maintained at a set appropriate temperature without being affected by the external environment. According to one embodiment, the insulating material may include a foamed insulating material. The foamed insulating material can be formed by injecting and foaming urethane foam, which is a mixture of polyurethane and a foaming agent, between the inner and outer layers.

[0055] According to one embodiment, the insulation material may additionally include a vacuum insulation material in addition to a foam insulation material, or the insulation material may consist solely of a vacuum insulation material instead of a foam insulation material. The vacuum insulation material may include a core material and an outer shell material that accommodates the core material and seals the interior under vacuum or near-vacuum pressure. However, the insulation material is not limited to the foam insulation material or vacuum insulation material described above and may include various materials that can be used for insulation.

[0056] The "storage room" may include a space defined by an internal structure. The storage room may further include an internal structure defining a space corresponding to the storage room. Various items such as food, medicine, and cosmetics may be stored in the storage room, and the storage room may be formed so that at least one side is open to allow for the retrieval and retrieval of items.

[0057] A refrigerator may include one or more storage compartments. When two or more storage compartments are formed in a refrigerator, each storage compartment may have a different use and may be maintained at a different temperature. To this end, each storage compartment may be partitioned from one another by a partition containing insulation.

[0058] The storage room may be provided to be maintained within an appropriate temperature range according to its intended use and may include a "refrigeration room," "freezing room," or "variable temperature room" distinguished according to its intended use and / or temperature range. The refrigerator room may be maintained at a temperature suitable for refrigerated storage of goods, and the freezer room may be maintained at a temperature suitable for frozen storage of goods. "Refrigeration" may mean cooling goods to a temperature that does not freeze them; for example, the refrigerator room may be maintained within a range of 0 degrees Celsius to 7 degrees Celsius. "Freezing" may mean cooling goods to freeze them or to maintain them in a frozen state; for example, the freezer room may be maintained within a range of -20 degrees Celsius to -1 degree Celsius. The variable temperature room may be used as either a refrigerator room or a freezer room, with or without the user's choice.

[0059] Storage rooms may be referred to by various names, such as "vegetable room," "fresh room," "cooling room," and "ice-making room," in addition to terms like "refrigeration room," "freezing room," and "variable temperature room." The terms "refrigeration room," "freezing room," and "variable temperature room" used below should be understood as encompassing storage rooms with corresponding uses and temperature ranges.

[0060] Although the above description describes an example where the electronic device (100) is implemented as a refrigerator, it is not necessarily limited to this.

[0061] For example, the electronic device (100) is a smartphone, tablet PC, desktop PC, laptop PC, PC, set-top box, OTT service (Over-the-top media service) server, console (video game console), Blu-ray player, DVD player, home automation control panel, security control panel, media box (e.g., Samsung HomeSync). TM , Apple TV TM , or Google TV TM ), game console (e.g., Xbox) TM PlayStation TM It can be implemented as at least one of ). However, it is not limited to this.

[0062] Meanwhile, the external electronic device (10-1, 10-2, 쪋) connected to the server device (20) can be implemented as a refrigerator, just like the electronic device (100), but is not necessarily limited thereto and can also be implemented as a smartphone, tablet PC, desktop PC, laptop PC, etc. as described above.

[0063] However, for the convenience of explanation, the following description will be explained by assuming that each of the electronic device (100) and the external electronic device (10-1, 10-2, ...) is implemented as a refrigerator having multiple storage compartments.

[0064] The electronic device (100) is connected to a server device (20) connected to an external electronic device (10-1, 10-2, ...) and can receive various types of data necessary for the operation of the electronic device (100).

[0065] For example, the electronic device (100) may receive data regarding the usage history of each external electronic device (10-1, 10-2, 쪋) from the server device (20). Here, the data regarding the usage history may include records of measuring the temperature of each of the multiple storage rooms in the external electronic devices (10-1, 10-2, 쪋).

[0066] However, it is not limited to this, and the usage history data may include the history of operation and status of external electronic devices (10-1, 10-2, ���), such as records of door opening of the storage room and records of defrosting operations. Here, the defrosting operation may correspond to the process of removing frost (ice) formed inside the freezer and refrigerator rooms.

[0067] Meanwhile, the data provided to the electronic device (100) from the server device (20) may correspond to training data for training a neural network model. Here, the neural network model is a computer system or software module for implementing human-level intelligence, and has the characteristic that the machine learns and makes judgments on its own, and the recognition rate improves as it is used.

[0068] For example, the neural network model may correspond to a model trained to predict the state of the electronic device (100). Here, the state of the electronic device (100) may correspond to the temperature (internal or external temperature) and humidity of the storage room, etc. However, it is not limited thereto.

[0069] The neural network model may be stored in the electronic device (100) and trained to predict the state of the electronic device (100). However, it is not limited thereto, and the neural network model may be stored in a server device (20) or an external electronic device (10-1, 10-2, 쪋) outside the electronic device (100) and correspond to a model trained to predict the state of the electronic device (100).

[0070] Here, if the neural network model corresponds to a model trained to predict the internal temperature of a storage room, the temperature records provided by the server device (20) can be used as training data for the neural network model. Here, the temperature records may include temperature records over time and statistical values ​​for each of the multiple storage rooms.

[0071] Here, the training data may include temperature records, etc. acquired by external electronic devices (10-1, 10-2, 쪋). That is, the neural network model described above can be trained using temperature records, etc. acquired by each of the external electronic devices (10-1, 10-2, 쪋) and the electronic device (100) as training data.

[0072] For example, if both the external electronic device (10-1, 10-2, 쪋) and the electronic device (100) are implemented as refrigerators, the neural network model may be trained using the temperature records acquired by the external electronic device (10-1, 10-2, 쪋) and the electronic device (100) as training data to predict the temperature of the storage room of the electronic device (100). However, it is not limited thereto.

[0073] The server device (20) may receive data acquired by external electronic devices (10-1, 10-2, 쪋) and provide it to the electronic device (100), or provide data received from the electronic device (100) to the external electronic devices (10-1, 10-2, 쪋). Here, the data may include the usage history, etc. described above. However, it is not limited thereto.

[0074] An electronic device (100) may receive data including usage history, etc. from a server device (20) and use it as training data. If some of the data received by the electronic device (100) cannot be used as training data, the electronic device (100) may augment the remaining data using some of the data received (or data acquired by the electronic device (100)).

[0075] Here, augmentation refers to the process of creating new data by transforming existing data in various ways. Through this augmentation, the learning performance of neural network models can be improved using the newly generated data.

[0076] For example, if the characteristics of the data received by the electronic device (100) and the types of each of the multiple storage chambers included in the electronic device (100) are different, the electronic device (100) cannot use the received data as training data.

[0077] The characteristics of the data received here may include information regarding each of the multiple storage rooms included in the data. For example, if some of the received data includes temperature records for the refrigerator room and the freezer room, while the electronic device (100) includes the refrigerator room, the freezer room, and the variable temperature room, the received data cannot be used as training data to predict the temperature of the variable temperature room.

[0078] In this case, the electronic device (100) may use some of the received data, including temperature information for the variable temperature room, as training data. At this time, the electronic device (100) may use some of the data to augment the remaining data (data including temperature information for the freezer and refrigerator rooms). A detailed explanation of this will be described in detail below in FIG. 4.

[0079] However, it is not limited to this, and if the data received by the electronic device (100) is insufficient to train a neural network model, the data received by the electronic device (100) can be augmented to generate new data.

[0080] If the received data is insufficient to train a neural network model, training the model solely with the received data may reduce the predictive accuracy of the neural network model.

[0081] At this time, in order to improve the prediction accuracy of the neural network model, the received data can be augmented to generate new data, which can then be used as training data. A detailed explanation of this will be described in detail below in FIG. 4.

[0082] The specific operation steps for the electronic device (100) to augment learning data will be described in detail below in Fig. 2.

[0083] FIG. 2 is a block diagram for explaining the configuration of an electronic device according to one or more embodiments of the present disclosure.

[0084] According to FIG. 2, the electronic device (100) may include a memory (110), a communication device (120), and at least one processor (130).

[0085] Since the electronic device (100) has been described in detail in FIG. 1, the detailed configuration of the electronic device (100) will be described in detail below.

[0086] The memory (110) is electrically connected to at least one processor (130) and can store data necessary for various embodiments of the present disclosure. For example, the memory (110) may be implemented as an internal memory such as ROM (e.g., EEPROM (electrically erasable programmable read-only memory)) or RAM included in the processor (130), or it may be implemented as a memory separate from at least one processor (130).

[0087] Depending on the purpose of data storage, the memory (110) may be implemented in the form of a memory embedded in the electronic device (100) or in the form of a memory that can be attached to and detached from the electronic device (100). For example, data for operating the electronic device (100) may be stored in a memory embedded in the electronic device (100), and data for the expansion function of the electronic device (100) may be stored in a memory that can be attached to and detached from the display device (210). When implemented as memory embedded in a display device (210), the memory (110) may be at least one of volatile memory (e.g., DRAM (dynamic RAM), SRAM (static RAM), or SDRAM (synchronous dynamic RAM), non-volatile memory (e.g., OTPROM (one time programmable ROM), PROM (programmable ROM), EPROM (erasable and programmable ROM), EEPROM (electrically erasable and programmable ROM), mask ROM, flash ROM, flash memory (e.g., NAND flash or NOR flash), hard drive, or solid state drive (SSD).

[0088] Meanwhile, although the illustrated example shows the electronic device (100) being composed of a single memory, when distinguishing between volatile memory and non-volatile memory, the electronic device (100) may be described as including multiple memories.

[0089] A memory (110) according to one or more embodiments may store at least one instruction. Here, the at least one instruction may correspond to at least one command for the electronic device (100) to acquire augmented data. In addition, the memory (110) may store information necessary for the operation of the electronic device (100).

[0090] According to one or more embodiments, the memory (110) can store at least one neural network model.

[0091] Neural network models consist of machine learning (deep learning) technology that uses algorithms to self-classify and learn the features of input data, and elemental technologies that utilize machine learning algorithms to mimic functions such as cognition and judgment of the human brain.

[0092] The elemental technologies may include, for example, at least one of linguistic understanding technology that recognizes human language / characters, visual understanding technology that recognizes objects like human vision, reasoning / prediction technology that judges information to logically reason and predict, and knowledge representation technology that processes human experience information into knowledge data.

[0093] For example, the neural network model may correspond to a model trained to output a predicted temperature corresponding to at least one of a plurality of storage room types. Here, the neural network model may correspond to the aforementioned training data (e.g., temperature information by storage room type).

[0094] For example, the training data here may correspond to data obtained by the electronic device (100) through a server device, etc. Additionally, the training data may correspond to data obtained through a server device, etc., or data obtained by the electronic device (100) detecting the temperature of a storage room. Additionally, the training data may correspond to data obtained by augmenting such training data as described above. However, it is not limited thereto.

[0095] Such a neural network model may be stored in the memory (110) of the electronic device (100), but is not limited thereto, and may be stored in an external device, etc. and correspond to a model stored in the electronic device (100). However, it is not limited thereto.

[0096] The neural network model is not limited to the example described above, and the neural network model can be implemented with various models trained to perform the actions necessary for the electronic device (100) to acquire augmented data.

[0097] The communication device (120) is configured to communicate with various types of external devices according to various types of communication methods. The communication device (120) may include a Wi-Fi module, a Bluetooth module, an infrared communication module, and a wireless communication module, etc. Here, each communication module may be implemented in the form of at least one hardware chip.

[0098] Wi-Fi modules and Bluetooth modules can perform communication using Wi-Fi and Bluetooth methods, respectively. When using a Wi-Fi module or a Bluetooth module, various connection information, such as SSID and session key, is transmitted and received first; after establishing a communication connection using this information, various information can be transmitted and received.

[0099] The infrared communication module performs communication according to infrared communication (IrDA, Infrared Data Association) technology, which uses infrared rays located between visible light and millimeter waves to wirelessly transmit data over short distances.

[0100] In addition to the communication method described above, the wireless communication module may include at least one communication chip that performs communication according to various wireless communication standards such as Zigbee, 3G (3rd Generation), 3GPP (3rd Generation Partnership Project), LTE (Long Term Evolution), LTE-A (LTE Advanced), 4G (4th Generation), and 5G (5th Generation).

[0101] In addition, the communication device (120) may include at least one wired communication module that performs communication using a LAN (Local Area Network) module, an Ethernet module, a pair cable, a coaxial cable, a fiber optic cable, or an UWB (Ultra Wide-Band) module, etc. Such a communication device (120) may also be referred to as a transceiver.

[0102] According to one or more embodiments, the electronic device (100) can communicate with an external electronic device (such as a server device) through a communication device (120). For example, the electronic device (100) can receive input data from a server device.

[0103] For example, the input data may correspond to temperature data obtained by an external electronic device recording temperature information for each of the multiple storage room types. Here, the temperature data may include temperature information recorded over time. Here, the temperature information may include recorded (or measured) temperature values ​​and statistical values ​​(mean, variance, etc.) calculated from the multiple temperature values.

[0104] Meanwhile, the electronic device (100) can provide augmented data obtained through the communication device (120) to an external electronic device. Here, the external electronic device may correspond to a device in which a neural network model is stored.

[0105] Here, the neural network model may correspond to a model for predicting temperature according to the aforementioned storage room types. External electronic devices may use the provided augmented data as training data to train the neural network model. However, it is not limited to this.

[0106] At least one processor (130) can perform overall control operations of the electronic device (100). Specifically, at least one processor (130) functions to control the overall operation of the electronic device (100).

[0107] At least one processor (130) may be implemented as a digital signal processor (DSP) that processes digital signals, a microprocessor, or a time controller (TCON). However, it is not limited thereto, and may include or be defined by one or more of a central processing unit (CPU), a micro controller unit (MCU), a micro processing unit (MPU), a controller, an application processor (AP), a graphics-processing unit (GPU), a communication processor (CP), or an ARM processor. Additionally, at least one processor (130) may be implemented as a System on Chip (SoC) or large-scale integration (LSI) with a built-in processing algorithm, or may be implemented in the form of a Field Programmable Gate Array (FPGA). Furthermore, at least one processor (130) may perform various functions by executing computer executable instructions stored in memory. Meanwhile, in FIG. 2, only one processor is included in the electronic device (100). Although it has been described, when implementing it, it may include multiple processors (e.g., CPU + GPU, CPU + DSP).

[0108] According to one or more embodiments, at least one processor (130) can acquire a plurality of datasets based on input data received through a communication device (120).

[0109] Here, a plurality of datasets may include temperature information corresponding to each of at least one storage room type among a plurality of storage room types. For example, each of the plurality of datasets may include temperature information classified according to the storage room type. Here, the temperature information may include the temperature measurement values ​​or statistical values ​​described above.

[0110] According to one embodiment, a plurality of storage room types may include a refrigerator room, a freezer room, and a variable temperature room. Here, the variable temperature room may correspond to a space where the temperature can be adjusted as desired by the user.

[0111] Here, the variable temperature chamber can be used in various modes such as freezing, refrigeration, and fresh storage. The electronic device (100) can maintain freshness by setting the optimal temperature for various types of food ingredients through the variable temperature chamber.

[0112] Meanwhile, at least one processor (130) can classify input data including temperature measurement records to obtain multiple datasets.

[0113] Specifically, at least one processor (130) can collect external electronic devices and temperature record data acquired by the electronic devices, and add identification information (labels) of a freezer, a refrigerator, and a variable temperature room to each record.

[0114] Here, classification information may correspond to information indicating which category or group each data point belongs to. For example, classification information may correspond to criteria for classifying data for the freezer, refrigerator, and variable temperature compartments, respectively.

[0115] Next, at least one processor (130) can separate the temperature records by filtering the data according to the label for the freezer, refrigerator, and temperature-controlled room. Here, data filtering may correspond to a process of selecting data that meets the conditions and extracting only the necessary parts.

[0116] Next, at least one processor (130) can store the filtered data as a separate dataset for each space (freezer, refrigerator, temperature-controlled room) and calculate statistics (average, range, etc.) based on this.

[0117] Meanwhile, the electronic device (100) may further include a plurality of temperature sensors for detecting the temperature inside a plurality of storage rooms. Here, the temperature sensors can measure the temperature of each storage room in real time. The electronic device (100) can control operation based on the temperature measured through the temperature sensors.

[0118] According to one example, at least one processor (130) can acquire multiple partial datasets based on input data and sensing data acquired through a plurality of temperature sensors. Here, the electronic device (100) can acquire sensing data by measuring the temperature for each of the plurality of storage chambers included in the electronic device (100) through the temperature sensors.

[0119] Here, the partial dataset may include multiple temperature statistical values ​​corresponding to each of at least one storage room type. Here, the temperature statistical values ​​may include a mean, variation, and range.

[0120] Here, the mean, variance, and range may each correspond to a value calculated based on the temperature measurement included in the input data or sensing data. This will be described in detail below in FIG. 5.

[0121] According to one example, at least one processor (130) can cluster multiple datasets based on data similarity based on multiple statistical values ​​included in each of the multiple partial datasets.

[0122] Here, data similarity can refer to a numerical value representing the degree of similarity between data points. Data similarity can be used in clustering to group close (similar) data together. For example, here, close data may refer to data where the coordinates of two points are close.

[0123] Clustering may correspond to a technique for classifying data with similar characteristics by grouping them into groups. For example, at least one processor (130) may group points that show similar temperature fluctuation patterns over a certain period. For example, temperature changes on days when doors are frequently opened and closed and on days when they are not may be classified into different clusters.

[0124] According to one example, at least one processor (130) can cluster multiple partial datasets into the same cluster if the data distance between two partial datasets is less than a predetermined distance.

[0125] Here, data distance can correspond to a numerical value that measures the difference or interval between two data points. Here, a data point can correspond to a coordinate representing a single data point in a multidimensional space.

[0126] For example, the data distance may correspond to the Euclidean distance. The Euclidean distance may correspond to the straight-line distance between two data points, and the electronic device (100) can obtain the similarity between two data points by calculating the Euclidean distance. This will be explained in detail in FIG. 9.

[0127] The distance defined here may serve as a criterion for clustering subsets into the same cluster when the distance between two subsets is less than or equal to a specific value (or threshold). The distance defined here may be set by user input, but is not limited thereto.

[0128] According to one or more embodiments, at least one processor (130) can obtain temperature information of a second type dataset based on a first type dataset.

[0129] The second type dataset may include temperature information corresponding to some of the multiple storage room types. For example, while the first type dataset includes temperature information for all storage room types, the second type dataset may include temperature information for only some of the total storage room types.

[0130] The temperature information of the second type dataset may be temperature information corresponding to at least one of a plurality of storage room types in the second type dataset. Here, the temperature information may be information to be used as training data for a neural network model (temperature prediction model) described later.

[0131] For example, the temperature information of the second type dataset may be data generated from information included in the first dataset.

[0132] For example, at least one processor (130) can augment a second type dataset based on a first type dataset. For example, at least one processor (130) can augment a second type dataset with relatively fewer types of information from a first type dataset with relatively more types of information (storage room types).

[0133] Here, the first type dataset may include temperature information corresponding to each of the multiple storage room types among the multiple datasets. The second type dataset may include temperature information corresponding to some of the multiple storage room types among the multiple datasets.

[0134] For example, the first type dataset may include temperature information corresponding to each of the refrigerator compartment, the freezer compartment, and the variable temperature compartment. For example, the second type dataset may include temperature information corresponding to each of the refrigerator compartment and the freezer compartment.

[0135] Meanwhile, at least one processor (130) can identify a first type first section based on temperature information included in a first type dataset.

[0136] Here, the first interval may correspond to the interval in the first type dataset where a temperature change event of the variable temperature room occurred. Here, the interval may correspond to at least one of a plurality of time intervals distinguished according to the time at which the temperature of the variable temperature room was measured.

[0137] Here, a temperature change event can refer to the moment when the temperature inside the refrigerator changes abruptly. For example, a temperature change event can occur during cooling operation or when opening and closing the door. In this context, a temperature change event may correspond to an event where the internal temperature temporarily rises when the refrigerator door is opened.

[0138] According to one embodiment, at least one processor (130) can identify a first interval in which a temperature change event occurred based on at least one bit corresponding to whether the temperature change room is defrosted among the temperature information included in the first type dataset.

[0139] Here, at least one bit may correspond to a bit recorded with a different value depending on whether a defrost operation is performed. For example, if a defrost operation is performed during a specific time interval in which the temperature is measured, this bit may be recorded as 1. On the other hand, if a defrost operation is performed during a specific time interval in which the temperature is measured, the bit may be recorded as 0.

[0140] Here, at least one bit corresponding to whether or not the operation is performed may be recorded by an external electronic device or an electronic device (100). However, it is not limited thereto. A detailed explanation thereof will be described in detail below in FIG. 5.

[0141] Meanwhile, the temperature information included in the first type dataset may include multiple statistical values ​​arranged chronologically according to multiple time intervals for each of the multiple storage rooms.

[0142] In this case, at least one processor (130) can identify a plurality of time intervals that have changed by more than a threshold value from the statistical value corresponding to the previous time interval among a plurality of statistical values ​​corresponding to the temperature change room.

[0143] Here, the threshold value may correspond to a reference value for determining a temperature change event when a statistical value changes. For example, if the average temperature of a certain time interval changes by 5 degrees or more, at least one processor (130) can set this as a threshold value and identify the interval where a temperature change event occurred.

[0144] Accordingly, at least one processor (130) can identify multiple time intervals in which a temperature change event occurred based on multiple identified time intervals.

[0145] For example, at least one processor (130) can identify an rising time interval in which the average temperature rises by 5 degrees or more and a falling time interval in which the average temperature falls by 5 degrees or more. At this time, at least one processor (130) can identify a plurality of intermediate time intervals between the rising time interval and the falling time interval.

[0146] At this time, at least one processor (130) can identify that a temperature change event has occurred during an upward time interval, a plurality of intermediate time intervals, and a downward time interval.

[0147] In this case, at least one processor (130) can identify a section including a plurality of time intervals in which a temperature change event occurred as a first section. For example, the first section may include a rising time interval, a plurality of intermediate time intervals, and a falling time interval.

[0148] Meanwhile, at least one processor (130) can identify a second section that is similar in pattern to the temperature information included in the identified section among the temperature information included in the second type dataset.

[0149] Here, a pattern may refer to the manner in which statistical values, such as the mean or variance calculated for each time interval, change over time. For example, if the mean or variance values ​​calculated for each interval exhibit a specific rule or variability, the pattern may include multiple statistical values ​​(e.g., mean, variance, etc.) exhibiting such rule or variability.

[0150] Here, at least one processor (130) may detect temperature change events of the refrigerator through this pattern.

[0151] Meanwhile, the second interval may correspond to an interval that has temperature information and patterns similar to those included in the identified interval (e.g., the first interval). Here, similarity in patterns may refer to cases where the patterns of change in statistical values ​​according to the time interval are similar to each other.

[0152] For example, if the averages of two temperature records rise equally over a certain interval and the variances fluctuate by a similar magnitude, the two patterns can be considered similar.

[0153] Here, at least one processor (130) can identify two patterns as similar if the statistical values ​​of the two patterns differ within a threshold. Here, the threshold may correspond to a boundary (criterion) value that determines that the two are identical or similar when comparing the differences in statistical values, if the difference is below a certain standard.

[0154] For example, at least one processor (130) can identify that the temperature change patterns of the two sections are similar if the average included in each of the two sections changes within a threshold value set to 2 degrees or less. However, it is not limited to this.

[0155] Afterwards, at least one processor (130) can record temperature information corresponding to the temperature room in the first section as temperature information corresponding to the temperature room in the second section.

[0156] For example, at least one processor (130) can record the statistical values ​​of the temperature room included in the first section of the first type dataset as the statistical values ​​of the temperature room in the second section of the second type dataset. A detailed explanation of this will be provided later in FIG. 5.

[0157] Meanwhile, at least one processor (130) can identify a first section in which a first temperature change event of a variable temperature room occurs among the temperature information included in the first type dataset. Subsequently, at least one processor (130) can identify a second section among the second temperature information included in the second type dataset.

[0158] Next, at least one processor (130) can identify a third section. Here, the third section may correspond to a section identified as having occurred in the first type dataset where the second temperature change event occurred.

[0159] Here, the second temperature change event may correspond to an event different from the first temperature change event among the temperature information included in the first type dataset. For example, if the first temperature change event corresponds to a door opening event, the second temperature change event may correspond to an event in which the refrigerator's defrosting operation is performed.

[0160] Afterwards, at least one processor (130) can record temperature information corresponding to the refrigerator and freezer in the third section as temperature information corresponding to the refrigerator and freezer in the second type dataset.

[0161] Accordingly, if the second type dataset does not contain statistical values ​​of the temperature-changing room, at least one processor (130) can use the statistical values ​​included in the first type dataset to record them in the second type dataset. Through this, at least one processor (130) can augment the second type dataset into a training dataset containing temperature-changing room data.

[0162] That is, even if a second type dataset is generated based on data obtained by an external electronic device that does not include a temperature control room, at least one processor (130) can augment it into a dataset including temperature control room data using a pattern similar to that of the first type dataset.

[0163] Meanwhile, at least one processor (130) can identify a fourth section of temperature information included in the first type dataset. Here, the fourth section may correspond to a section in which at least one of the refrigerator and freezer sections has a temperature change event.

[0164] Next, at least one processor (130) can record temperature information included in the fourth section as temperature information corresponding to at least one of the refrigerator room and freezer room in the second type dataset.

[0165] Meanwhile, at least one processor (130) can identify a mode based on temperature information included in a section excluding the fourth section among the temperature information included in the first type dataset. Here, the mode may correspond to a mode corresponding to at least one of the refrigerator room and the freezer room.

[0166] For example, the mode may refer to the statistical value that appears most frequently among the temperature information corresponding to the refrigerator compartment and refrigerator in the first type dataset.

[0167] Subsequently, at least one processor (130) can record the identified mode as temperature information corresponding to at least one of the refrigerator and freezer rooms in the second type dataset.

[0168] For example, at least one processor (130) can record the mode in a section of the second type dataset where no temperature change event occurred (e.g., a section where the temperature is constant or a section where the temperature fluctuation is below a threshold value).

[0169] However, this is not limited thereto, and at least one processor (130) may record a mode in a section where a temperature change event (e.g., a door opening event) different from the temperature change event (e.g., a defrost operation event) corresponding to the fourth section described above occurs. Here, the mode may correspond to the mode corresponding to the section where the defrost operation event occurred among the first type datasets.

[0170] For example, if the first type dataset includes temperature information corresponding to the refrigerator, freezer, and temperature-controlled room, respectively, the statistical values ​​of the refrigerator and freezer rooms may be affected by the temperature-controlled room (e.g., opening of the temperature-controlled room door, defrosting operation of the temperature-controlled room).

[0171] In this case, if at least one processor (130) acquires training data by augmenting a second type dataset using the first type dataset as is, the temperature prediction accuracy of the neural network model for the refrigerator and freezer may decrease.

[0172] For example, there may be cases where the neural network model is a model for predicting the temperature of the refrigerator or freezer compartment of a refrigerator that is not equipped with a variable temperature compartment. In this case, if at least one processor (130) augments the second type dataset without removing the influence of the variable temperature compartment and uses it as training data to train the neural network model, the prediction accuracy of the refrigerator and freezer compartments may be reduced.

[0173] If the temperature information of the sections excluding the aforementioned fourth section (temperature information corresponding to the refrigerator and freezer sections) is replaced with the mode from the aforementioned first type dataset, the influence of the variable temperature section in the first type dataset can be eliminated.

[0174] Accordingly, since the augmented second type dataset does not reflect the influence of the variable temperature room among the first type datasets, when this second type dataset (training dataset) is used, the prediction accuracy of the neural network model for the refrigerator and freezer rooms can be improved.

[0175] Next, at least one processor (130) can record temperature information included in the fourth section as temperature information of the section excluding the section where the mode of the second type dataset is recorded.

[0176] For example, the interval where the mode is recorded here may correspond to the interval excluding the interval identified as having a temperature change event in the second type dataset.

[0177] However, this is not limited to this, and the interval where the mode is recorded may also correspond to an interval where a temperature change event different from the temperature change event corresponding to the fourth interval (e.g., door opening event) (e.g., defrosting operation event) occurs.

[0178] The temperature information recorded here may correspond to temperature information corresponding to at least one of the refrigerator and freezer compartments among the second type dataset. A detailed explanation of this will be described in detail below in FIG. 7.

[0179] According to one or more embodiments, at least one processor (130) may acquire augmented data to be input as training data to at least one neural network model. Here, the training data may correspond to data for training at least one neural network model.

[0180] Here, at least one neural network model may correspond to a model for outputting a predicted temperature corresponding to at least one of a plurality of storage room types.

[0181] For example, at least one processor (130) can obtain such training data by writing new data to a second type dataset based on a first type dataset.

[0182] Accordingly, at least one neural network model (temperature prediction model) can be trained using this training data to train a model with higher accuracy compared to the case where the received data is simply used as training data.

[0183] Specifically, there may be cases where the input data received by the electronic device (100) is biased toward the data of a specific model (a refrigerator model equipped with a variable temperature chamber).

[0184] In this case, if the temperature prediction model is trained using the received input data as is, the trained prediction model may be biased toward a specific refrigerator model. Consequently, the prediction model may not be universally applicable to other refrigerator models.

[0185] In this case, at least one processor (130) can use the augmented data as training data to train a generalized model that is not biased toward a specific model by augmenting the data of the remaining model (a refrigerator model without a variable temperature room) based on the data of a specific model.

[0186] Although the electronic device (100) in FIG. 2 is illustrated as including only basic components (i.e., memory, communication device, processor), the electronic device (100) may include various additional components in addition to the components described above.

[0187] FIG. 3 is a flowchart for explaining the operation of an electronic device according to one or more embodiments of the present disclosure.

[0188] According to FIG. 3, the electronic device (100) receives input data, extracts data characteristics, and then clusters the data to augment the training data, showing the operation in steps.

[0189] The electronic device (100) can extract data characteristics based on input data (S310). Here, the data characteristics may include each type of a plurality of refrigerators that have acquired input data, etc. Here, the refrigerator type may correspond to one of a plurality of types distinguished based on the storage compartment type.

[0190] Here, the plurality of types may include at least one of a first type having a refrigerator room, a freezer room, and a variable temperature room, and a second type having a refrigerator room and a freezer room.

[0191] The electronic device (100) can first separate data by refrigerator type to extract data characteristics. Subsequently, the electronic device (100) can separate data by refrigerator type and by multiple storage room types (e.g., refrigerator room, freezer room, and variable temperature room, etc.).

[0192] Here, the electronic device (100) can separate data by adding classification information (labels) for the refrigerator compartment, freezer compartment, and variable temperature compartment to each of the data for each refrigerator type and performing data filtering. Since a detailed explanation of this has been described above in FIG. 2, a redundant explanation will be omitted.

[0193] Subsequently, the electronic device (100) can calculate the temperature change amount, average, and variance for each storage room type. At this time, the electronic device (100) can perform noise removal on the data for each storage room type. This may refer to a process of controlling by filtering out abnormal or temporary outliers during the noise removal temperature recording.

[0194] Accordingly, the electronic device (100) can extract data characteristics including refrigerator type information, storage room type information, and statistical values ​​(mean, variance, etc.) from input data (and data acquired by the electronic device (100)).

[0195] Through this, even if the input data, etc. are biased toward a characteristic model (e.g., a refrigerator type equipped with a variable temperature chamber), the electronic device (100) can separate the data into a subdivided dataset according to the characteristics.

[0196] The electronic device (100) can be used to extract a portion of a segmented dataset and use it for training a model, and can augment the extracted dataset based on another dataset.

[0197] Additionally, the electronic device (100) can augment some of the segmented datasets based on other datasets. Through this, the electronic device (100) can train a neural network model that can be used universally without distinguishing between refrigerator types based on the augmented data.

[0198] Next, the electronic device (100) can perform clustering based on extracted data characteristics (S320). For example, the electronic device (100) can perform clustering based on the distribution of temperature change amount, mean, variance, maximum value, etc. based on extracted data characteristics.

[0199] Specifically, the electronic device (100) can obtain multiple datasets in which temperature statistical values ​​according to multiple storage room types are listed in a time series based on extracted data characteristics. Here, each dataset may include multiple sub-datasets.

[0200] Here, multiple subsets may include statistical values ​​such as temperature change, mean, and variance for each of the multiple storage room types. In this case, each subset may correspond to a vector in which each statistical value is a component.

[0201] Here, the electronic device (100) can cluster into multiple clusters based on data similarity. Here, data similarity may refer to the degree of closeness of vector distance. Here, vector distance may correspond to the data distance described above (e.g., Euclidean distance). Since this has been explained in detail in FIG. 2, a redundant explanation will be omitted.

[0202] Accordingly, the electronic device (100) can cluster multiple partial datasets into multiple clusters based on statistical characteristics.

[0203] Next, the electronic device (100) can be separated into training data and test data by cluster (S330).

[0204] Here, the training data may correspond to data provided for an artificial intelligence model to learn patterns. The test data may correspond to data used to evaluate the performance of a model that has completed training. The detailed process of the electronic device (100) separating the training data and the test data will be explained in detail in FIG. 9.

[0205] Next, the electronic device (100) can augment data by converting refrigerator data (second type dataset) (S340). The electronic device (100) can augment data classified as training data.

[0206] Here, the electronic device (100) can augment another type of dataset (e.g., second type dataset) based on a specific type of dataset (e.g., first type dataset) among the training data. This will be explained in detail below in FIG. 4.

[0207] FIG. 4 is a drawing for explaining an operation to augment learning data according to one or more embodiments of the present disclosure.

[0208] According to FIG. 4, the electronic device (100) classifies input data according to characteristics and then augments the learning data in steps.

[0209] The electronic device (100) can check the presence or absence of R, F, and CV compartments for each refrigerator and classify the input data based on the presence or absence of R, F, and CV compartments (S410). The electronic device (100) can receive the input data in the form of a file in which temperature measurement records are stored.

[0210] Here, R room may mean a Refrigerator, F room may mean a Freezer, and CV room may mean a Convertible. The electronic device (100) can identify the presence or absence of R, F, and CV rooms by identifying information about the storage room type included in the input data.

[0211] When the electronic device (100) receives input data in the form of a file, it can identify whether the classified data is the last file (S420). For example, when the electronic device (100) receives multiple files as input data, it can sequentially classify the input data contained in each file (classification based on the presence or absence of R, F, and CV).

[0212] If the file that has been classified is identified as not being the last file (i.e., if there are files remaining to be classified), classification can be performed on the next file.

[0213] On the other hand, if the electronic device (100) is identified as having classified the input data of the last file, the electronic device (100) can extract data of a refrigerator type in which a CV room exists (e.g., a first type dataset) (S430).

[0214] Next, the electronic device (100) can extract the section where the user door opens (S440). For example, when the door opens, outside air is introduced, and the temperature inside the storage room may rise rapidly. In this case, the electronic device (100) can identify the section where the temperature rises rapidly at a specific point in time among the classified data and identify it as the section where the door opened.

[0215] Specifically, the electronic device (100) can set a threshold value (e.g., a 2-degree increase) for the range in which the temperature rises above a certain standard when the door is opened.

[0216] In this case, the electronic device (100) can identify the time when the door was opened by finding a section in the temperature record where the temperature rises rapidly above this threshold. Based on the identified time, the electronic device (100) can identify the section where the door was opened.

[0217] Meanwhile, the electronic device (100) may also extract the door opening occurrence interval by utilizing door opening sensing data. Here, the door opening sensing data may correspond to data that detects the time when the refrigerator door is opened and closed and records the time information.

[0218] In this case, the electronic device (100) or the external electronic device can record the time when the door is opened and closed. The electronic device (100) can identify the interval during which the door opening occurred by analyzing the data acquired by the electronic device (100) or the external electronic device together with the temperature record.

[0219] Next, the electronic device (100) can extract data of a refrigerator type in which a CV room does not exist (e.g., a second type dataset) (S450). Here, a refrigerator type in which a CV room exists may correspond to a refrigerator type that includes at least one of an R room or an F room.

[0220] Next, the electronic device (100) can extract a subset of data with similar temperature information from the data of a refrigerator type that does not have a CV room (S460). For example, the electronic device (100) can extract a subset of a section of the data of a refrigerator that does not have a CV room that has a similar temperature pattern to the data of a refrigerator that does have a CV room.

[0221] Here, a subset may refer to two intervals with similar temperature patterns, and each of the two intervals may include multiple time intervals. Here, the two intervals may correspond to a portion of the data included in the refrigerator type that does not have a CV room and a portion of the data included in the refrigerator type that does not have a CV room.

[0222] Each of the multiple time intervals here may include multiple statistical values ​​classified by multiple storage room types. This is explained in detail in FIG. 5.

[0223] Next, the electronic device (100) can augment the CV room information of the subset data (S470). For example, the electronic device (100) can substitute the CV room information of the subset data of the refrigerator type data that has a CV room into the CV room information of the subset data of the refrigerator type data that does not have a CV room.

[0224] In this context, substitution can mean recording new information as CV room data within a subset of data for refrigerator types that do not have a CV room.

[0225] Next, the electronic device (100) can extract defrost section data from refrigerator type data in which a CV room exists (S480).

[0226] For example, the electronic device (100) can identify a section in which a defrosting operation occurs among data of a type in which a CV chamber exists, and extract data included in this section.

[0227] Specifically, the electronic device (100) can analyze the average temperature rise pattern. Since cooling stops during the defrosting period, the average temperature calculated in each time period can gradually rise.

[0228] For example, if the average temperature rises continuously above a threshold (e.g., 0.5 degrees or higher) during multiple time intervals, the electronic device (100) can identify the interval as a interval where a defrosting operation is performed.

[0229] Meanwhile, the electronic device (100) can analyze the pattern of reduced variance. For example, during the defrosting operation, the temperature change of the storage room becomes constant, and the variability of the temperature may be reduced. In this case, the electronic device (100) can identify the section as a defrosting section if the temperature variance per section falls below a threshold value (e.g., the variance in a specific section continues to decrease to 0.1 or less).

[0230] Next, the electronic device (100) can augment data (second type data) by applying defrosting section data to a section where there is no temperature change in the R and F chambers (or a section where the temperature change is below a threshold value) (S490). Here, the section where there is no temperature change may correspond to a section included in the type of data where the CV chamber does not exist.

[0231] Accordingly, the electronic device (100) can augment other types of data through data included in the section where not only door opening but also defrosting operation is performed among the types of data where the CV room exists.

[0232] Through this, even if the data with relatively little information (e.g., the second type dataset) does not actually contain temperature information according to the door opening or defrosting operation interval, the second type dataset, etc., can be augmented through the first type dataset. Accordingly, the electronic device (100) can improve the prediction accuracy of the model by training the temperature prediction model through the augmented data.

[0233] Subsequently, the electronic device (100) can version and store the augmented data. Here, data versioning may correspond to a method of managing the history of changes (augments) and various versions of the data. Through data versioning, the electronic device (100) can track and reuse data at a specific point in time during the learning process, etc.

[0234] FIG. 5 is a drawing for illustrating a dataset according to one or more embodiments of the present disclosure.

[0235] According to FIG. 5, a first type dataset (510) and a second type dataset (520) are illustrated.

[0236] The first type dataset (510) may include a section (511) (e.g., first section) in which a temperature change event (e.g., door opening, defrosting operation) occurs. This section may include temperature information corresponding to the refrigerator and freezer compartments and temperature information (512) corresponding to the variable temperature compartment.

[0237] Here, temperature information corresponding to each storage room may include a first feature value (feature 1), a second feature value (feature 2), and an operator value. The first feature value and the second feature value may each correspond to at least one of a plurality of types of statistical values ​​(e.g., mean, variance, maximum value, minimum value, mode, etc.). Here, the operator value may correspond to a value in which the status of a specific operation (e.g., defrosting operation, door opening operation) of each storage room is recorded.

[0238] However, the above-described example is merely an example, and the first type dataset (510) (or the second type dataset (520)) may include various types of information in addition to the first feature value (feature 1), the second feature value (feature 2), and the operator value.

[0239] The second type dataset (520) may likewise include temperature information corresponding to the refrigerator and freezer compartments and temperature information (522) corresponding to the variable temperature compartment. Here, the second type dataset (520) may correspond to data generated based on data obtained by a refrigerator equipped with a refrigerator and a freezer compartment, but not equipped with a variable temperature compartment.

[0240] Here, the second type dataset (520) may include temperature information (522) corresponding to a variable temperature room, but this temperature information may be invalid data. Here, invalid data may mean data that cannot be used because it does not match the device or environment.

[0241] For example, in the second type dataset (520), temperature information (522) corresponding to the temperature room may all be stored as '-1000'. Here, '-1000' may correspond to a missing value. Here, a missing value may refer to a value where information required for data analysis is empty, such that a specific value is not recorded or is missing in the dataset.

[0242] If the second type dataset (520) is used as training data, the accuracy of the prediction model may be reduced because the training data contains invalid missing values.

[0243] Meanwhile, the electronic device (100) can analyze the pattern of the section (511) where a temperature change event occurred in the first type dataset (510) and search for a section containing a pattern similar to this pattern in the second type dataset (520). Here, the section (511) where a temperature change event occurred may include temperature information corresponding to the refrigerator and freezer, respectively.

[0244] Meanwhile, the electronic device (100) can sequentially search for temperature information corresponding to the refrigerator and freezer rooms among the second type dataset (520).

[0245] Here, the electronic device (100) can identify whether a specific section (521) includes the aforementioned pattern from the beginning (i.e., from the first time interval) of the second type dataset (520). The electronic device (100) can search for the specific section (521) in the second type dataset by sequentially changing it.

[0246] The electronic device (100) can identify a pattern identical or similar to the pattern analyzed in the section where a temperature change event occurred in the first type dataset (510).

[0247] In this case, the electronic device (100) can augment the second type dataset through information (512) corresponding to the temperature-changing room among the first type datasets. This will be explained in detail in the following section.

[0248] FIG. 6 is a drawing for illustrating a dataset according to one or more embodiments of the present disclosure.

[0249] According to FIG. 6, a first type dataset (610) and a second type dataset (620) are illustrated.

[0250] The first type dataset (610) may include temperature information (611) corresponding to the refrigerator and freezer, respectively, and temperature information (612) corresponding to the temperature change room, in the interval where the temperature change event occurred.

[0251] If the electronic device (100) identifies a pattern identical or similar to the pattern of the section where a temperature change event occurred in the second type dataset (620), it can identify a section (621) containing the pattern.

[0252] Here, a similar pattern may mean a pattern in which the difference between statistical values ​​(e.g., mean, variance, etc.) of two intervals (parts included in each of the first type dataset (610) and the second type dataset (620)) is within a set threshold when compared.

[0253] The electronic device (100) can identify the patterns of two intervals as similar patterns if the statistical difference does not exceed a threshold. Here, the statistical difference may correspond to a numerical value indicating how different the values ​​are by comparing the numerical difference between the statistical values ​​(e.g., mean, variance, etc.) of the two data intervals.

[0254] If the electronic device (100) identifies a section (621) containing a similar pattern, the temperature information (612) corresponding to the temperature room among the first type dataset can be substituted into the second type dataset.

[0255] For example, the electronic device (100) can substitute temperature information (612) corresponding to a variable temperature room in a section where a temperature change event occurred in a 1-type dataset with temperature information (622) corresponding to a section containing a similar pattern in a 2-type dataset.

[0256] Here, the temperature information (622) corresponding to the section containing a similar pattern may correspond to the information corresponding to the variable temperature room among the second type datasets.

[0257] In this way, the electronic device (100) can identify a subset (specific section) of similar temperature information corresponding to the refrigerator and freezer compartments in each of the first type dataset (510) and the second type dataset (520), and augment the second type dataset (520) based on this.

[0258] Accordingly, the electronic device (100) can enhance the second type dataset by efficiently reflecting the characteristics of the first type dataset into the second type dataset by applying the variable temperature room data of the first type dataset to a section with similar characteristics (refrigerator room, freezer room temperature pattern) instead of applying it to an arbitrary section.

[0259] FIG. 7 is a drawing for explaining an operation to augment learning data according to one or more embodiments of the present disclosure.

[0260] According to FIG. 7, the electronic device (100) classifies input data and then, based on data of a refrigerator type having a variable temperature room, shows the operation of increasing temperature information corresponding to each of the refrigerator room and the freezer room in steps.

[0261] The electronic device (100) can check whether there are R, F, and CV compartments for each refrigerator and classify input data based on the presence or absence of R, F, and CV compartments (S710). Subsequently, the electronic device (100) can identify whether the file that has been classified is the last file (S720).

[0262] Afterwards, if the electronic device (100) is identified as having classified the input data of the last file, the electronic device (100) can extract data of a refrigerator type in which a CV room exists (e.g., a first type dataset) (S730).

[0263] As the process of classifying input data collected in the form of multiple files by the electronic device (100) as described above has been specifically explained in FIG. 4, a redundant explanation will be omitted.

[0264] Next, the electronic device (100) can remove CV actual data from the extracted data (e.g., first type dataset) (S740).

[0265] Next, the electronic device (100) can replace all data of the defrosting sections of the R and F rooms affected by the operation of the CV room with the mode (S750). The mode may refer to the statistical value that appears most frequently among the temperature information corresponding to the refrigerator room and the refrigerator among the extracted data. Here, the operation of the CV room may refer to a temperature change event of the CV room. For example, a temperature change event of the CV room may correspond to the defrosting operation of the CV room and the opening of the door of the CV room.

[0266] Here, if the electronic device (100) uses the temperature information corresponding to each of the R and F rooms included in the extracted data as is, the prediction model learned through this may have reduced prediction accuracy.

[0267] Specifically, the first type dataset may include temperature information corresponding to rooms R and F. Here, the temperature information corresponding to rooms R and F may include temperature information affected by a temperature change event in the CV room.

[0268] The affected temperature information here may refer to the temperature information of rooms R and F that has changed due to temperature change events in the CV room. For example, if a door opening event occurs in the CV room, the temperatures of rooms R and F may rise sharply and then fall during that period (or subsequent periods).

[0269] For example, if the second type dataset does not include temperature information of the CV room, and the second type dataset is augmented with temperature information affected by the CV room, the augmented second type dataset may affect the accuracy of the prediction model.

[0270] The electronic device (100) can perform data augmentation by removing other influences (e.g., opening of the CV room door, defrosting operation of the CV room, etc.) received by the R and F rooms by the CV room.

[0271] The electronic device (100) can remove the influence of temperature change events in the CV room so that temperature information caused by temperature change events in the CV room (e.g., defrosting operation or door opening, etc.) in the first type dataset is not reflected in the second type dataset.

[0272] Specifically, the electronic device (100) can eliminate the effect by replacing the R and F room temperature information included in the section where a temperature change event occurred in the first type dataset (e.g., a defrosting operation section or a door opening section, etc.) with the mode.

[0273] Next, the electronic device (100) can augment the data excluding the extracted data (e.g., a second type dataset) with data including user door opening events and defrosting operation section data for rooms R and F (S760). For example, the electronic device (100) can augment the second type dataset, etc., which includes only user door opening events and temperature information of the defrosting operation section occurring in rooms R and F. That is, the electronic device (100) can augment the second type dataset by removing the influence of the CV room (the influence received by rooms R and F due to door opening, defrosting operation, etc. of the CV room in the first type dataset).

[0274] Meanwhile, the data received by the electronic device (100) and the data acquired by the electronic device (100) may be biased toward data of a specific refrigerator type. The biased data may refer to the data that accounts for the largest proportion of the total data.

[0275] At this time, the electronic device (100) can train a prediction model using the augmented data by using information included in the biased data (e.g., temperature statistical information for each of the temperature room, refrigerator room, and freezer room), so that even if the data collected by the electronic device (100) is insufficient to train a prediction model, the prediction model can be trained using the augmented data.

[0276] Through this, the electronic device (100) can train the existing temperature prediction model into a more accurate temperature prediction model.

[0277] FIG. 8 is a drawing for illustrating a dataset according to one or more embodiments of the present disclosure.

[0278] According to FIG. 8, a first type dataset (810) and a second type dataset (820) are illustrated. Here, the first type dataset may include a section (811) in which a temperature change event occurred. For example, the temperature change event may correspond to a door opening event.

[0279] The electronic device (100) can input temperature information included in the section (811) where a door opening event occurred in the first type dataset (810) into the second type dataset.

[0280] Specifically, the electronic device (100) can substitute temperature information corresponding to each of the R room and F room included in the section (811) where the door opening event occurred into the temperature information corresponding to each of the R room and F room in the second type dataset.

[0281] The second type dataset may correspond to data generated based on data obtained from a refrigerator of a type that does not include a variable temperature chamber (e.g., a second type refrigerator).

[0282] Specifically, the electronic device (100) can receive data as input data that the first type of refrigerator detects and records the temperatures of the refrigerator compartment and the freezer compartment. The electronic device (100) can classify this input data to obtain a dataset such as the second type dataset (821).

[0283] When the electronic device (100) receives input data, the input data may have a relatively higher proportion of data obtained by the first type refrigerator compared to data obtained by the second type refrigerator.

[0284] In this case, when the electronic device (100) classifies input data to obtain a first dataset (810) and a second dataset (820), the information included in the second type dataset (820) (temperature information of rooms R and F) may be insufficient to train a temperature prediction model.

[0285] For example, the second type dataset (820) may not include temperature information (temperature measurement records or statistical values, etc.) regarding a door opening event (or defrosting operation event). In this case, the second type dataset (820) alone may not be sufficient to train a neural network model, etc., to predict the temperature when a specific temperature change event occurs.

[0286] In this case, the electronic device (100) can increase the amount of training data for the temperature prediction model by augmenting the second type dataset (820) using the temperature information of the R and F rooms included in the first type dataset (810).

[0287] Meanwhile, according to FIGS. 5, 6 and 8, the electronic device (100) is described as an example in which the temperature information included in the first type dataset is directly applied to the second type dataset, but it is not necessarily limited to this, and the temperature information included in the first type dataset may be modified and applied to the second type dataset.

[0288] For example, the electronic device (100) can use a jitter technique to add noise to the temperature information included in the first type dataset and input it into the second type dataset.

[0289] In this context, the jitter technique refers to a method of transforming data by adding small noise (random numbers) to the original data. Jitter techniques are primarily used for time-series or coordinate data, and can generate transformed datasets by randomly adding or subtracting small values.

[0290] For example, noise of approximately ±0.1 degrees can be added to each value in the temperature record data to allow the model to be trained on various data.

[0291] Meanwhile, the electronic device (100) can modify the information included in the first type dataset using the Window extend technique and apply it to the second type dataset.

[0292] Here, the Window extend technique can be considered an augmentation method that creates a new dataset by expanding or shrinking the interval (window) of existing data in time series data.

[0293] For example, the electronic device (100) can use the Window extend technique to extend temperature data in 10-minute intervals to 12 minutes or reduce it to 8 minutes so that the model can learn various time series patterns.

[0294] The method by which the electronic device (100) modifies the first type dataset is not necessarily limited to the example described above, and the electronic device (100) may modify the first type dataset in various other ways in order to train a neural network model (temperature prediction model) with more diverse data.

[0295] FIG. 9 is a drawing for illustrating training data and test data according to one or more embodiments of the present disclosure.

[0296] According to FIG. 9, the electronic device (100) can receive input data (910) to obtain learning data (961) and test data (962). Here, the input data (910) may be received in the form of multiple files.

[0297] The electronic device (100) can separate input data (910) into multiple datasets (921, 922, 923) according to refrigerator type. For example, the first dataset (921) may be a dataset obtained by a first type refrigerator, the second dataset (922) may be a dataset obtained by a second type refrigerator, and the third dataset (923) may be a dataset obtained by a third type refrigerator.

[0298] Here, the third type refrigerator is a refrigerator type equipped with a refrigerator compartment and a variable temperature compartment, just like the second type refrigerator. However, the third type refrigerator may correspond to a refrigerator that differs from the second type refrigerator in the number, structure, and performance of each storage compartment.

[0299] The electronic device (100) can be classified by storage room type for each dataset (921, 922, 923).

[0300] For example, the first dataset (921) may be classified into a refrigerator dataset (931-1), a freezer dataset (931-2), and a variable temperature dataset (931-3). Each dataset (931-1, 931-2, 931-3) may include temperature information (temperature measurement records, temperature statistical values, etc.) for each storage room type.

[0301] Meanwhile, the second dataset (922) can be classified into a refrigerator dataset (932-1) and a freezer dataset (932-2).

[0302] Meanwhile, the third dataset (923) can also be classified into a refrigerator dataset (933-1) and a freezer dataset (933-2).

[0303] However, if the third type refrigerator has a different number of storage compartments from the second type refrigerator, the third dataset (923), unlike the second type dataset (922), may have a refrigerator compartment dataset (933-1) or a freezer compartment dataset (933-2) classified by multiple storage compartments (e.g., first freezer compartment, second freezer compartment, first refrigerator compartment, second refrigerator compartment, etc.).

[0304] The electronic device (100) can obtain statistical values ​​such as the mean and variance for each of the multiple storage room type datasets (931-1, 931-2, 931-3, 932-1, 932-2, 933-1, 933-2) obtained by classifying the first dataset (921) to the third dataset (923).

[0305] However, it is not limited to this, and the electronic device (100) can obtain a temperature change amount for each of the multiple storage room type datasets (931-1, 931-2, 931-3, 932-1, 932-2, 933-1, 933-2), and can perform noise removal processing on the data for the obtained values ​​(average, variance, temperature change amount, etc.).

[0306] The electronic device (100) can record the acquired value in each of the multiple storage room type datasets (931-1, 931-2, 931-3, 932-1, 932-2, 933-1, 933-2).

[0307] The electronic device (100) can cluster multiple datasets into multiple clusters (951, 952, 953, 954, 955) based on acquired statistical values, etc. Here, each of the multiple datasets may include multiple statistical values ​​distinguished by multiple storage room types.

[0308] The electronic device (100) can obtain multiple clusters by classifying the datasets into groups with high similarity of statistical values ​​included in each of the multiple datasets. That is, the electronic device (100) can cluster datasets such that datasets with small statistical differences are classified into the same cluster.

[0309] Each cluster can be clustered based on data similarity. Since data similarity and related data distance have been discussed previously, a redundant explanation will be omitted.

[0310] For example, the first cluster (951), the second cluster (952), the third cluster (953), the fourth cluster (954), and the fifth cluster (955) may all have different statistical characteristics.

[0311] Here, statistical characteristics may refer to traits in which the changing patterns of multiple statistical values ​​distinguished over time (i.e., changes in mean, changes in variance, etc. over time) are similar. In this context, these patterns may be referred to as patterns; since the explanation of patterns has already been provided, redundant details will be omitted.

[0312] Meanwhile, although multiple clusters are shown in a two-dimensional vector space in FIG. 9, they are not limited thereto, and the electronic device (100) can obtain multiple clusters by calculating data distances in a vector space of three dimensions or more.

[0313] For example, if each of the aforementioned multiple datasets includes the mean, variance, and change amount (maximum value - minimum value) for each of the refrigerator room, freezer room, and temperature room, each dataset can be represented as a 9-dimensional vector. In this case, the electronic device (100) can calculate the distance (Euclidean distance) between the multiple 9-dimensional vectors and cluster datasets that are close in distance so that they are included in the same cluster.

[0314] Meanwhile, the electronic device (100) can separate a plurality of datasets included in the third cluster (953) among a plurality of clusters into training data (961) and test data (962). Here, since the description of each of the training data (961) and test data (962) has been described above in FIG. 2, a redundant description will be omitted.

[0315] For example, the electronic device (100) may classify multiple datasets included in the third cluster (953) into training data (961) and test data (962) respectively in a 7:3 ratio. However, it is not necessarily limited to this, and multiple datasets may be classified into training data (961) and test data (962) by various other ratios.

[0316] FIG. 10 is a drawing for illustrating training data and test data according to one or more embodiments of the present disclosure.

[0317] According to FIG. 10, a dataset (1000) is illustrated. The dataset (1000) may include temperature information separated according to multiple statistical values ​​for each of a plurality of storage room types.

[0318] For example, the dataset (1000) may include multiple statistical values, namely, the minimum temperature of the refrigerator (Rm), the average temperature of the refrigerator (Ru), the variance of the refrigerator (Rv), the minimum temperature of the freezer (Fm), the average temperature of the freezer (Fu), the variance of the freezer (Fv), the minimum temperature of the variable temperature room (CVm), the average temperature of the variable temperature room (CVu), and the variance of the variable temperature room (CVv).

[0319] Here, each statistical value included in the dataset (1000) can be listed in a time series.

[0320] For example, if each of the multiple datasets included in the third cluster (953) in FIG. 9 is listed chronologically according to the time record of the temperature measurement, indexing can be performed on the listed multiple partial datasets.

[0321] Here, an index can refer to a unique number or similar identifier representing the order of each data point in time-series data. Indexing refers to the process of sorting data in chronological order and assigning a unique index to each data point.

[0322] In this case, the electronic device (100) can classify data corresponding to an index that is a multiple of 3 among a plurality of indices as test data (1030), and other data as training data (1010, 1020).

[0323] However, it is not limited to this, and the electronic device (100) can classify multiple datasets into training data and test data by various ratios or methods in addition to this.

[0324] FIG. 11 is a drawing for illustrating a plurality of clusters in one or more embodiments of the present disclosure.

[0325] According to FIG. 11, the electronic device (100) can obtain multiple clusters (1110) by clustering multiple datasets classified according to data characteristics.

[0326] Here, each of the plurality of clusters (1110) may include a plurality of statistical values ​​arranged in a time series as described above. The electronic device (100) can extract the number (cnt or count) of data (e.g., statistical values) for each of the plurality of clusters (1110).

[0327] The electronic device (100) can obtain data count information (1120) for each of the plurality of clusters (1110) by extracting the number of data for each cluster.

[0328] In this case, the electronic device (100) corresponds to the third cluster (1) that has the largest number of data among the plurality of clusters (1110). st Identify the majority cluster, and the second cluster (2) corresponding to the second largest number of data points. nd Can identify the majority cluster.

[0329] The electronic device (100) can identify a cluster to be augmented based on two identified clusters. Here, the cluster to be augmented may refer to a cluster that requires augmentation. Here, the cluster that requires augmentation may refer, for example, to a cluster where the amount of data is insufficient to train a temperature prediction model.

[0330] Specifically, the electronic device (100) can identify the remaining clusters (first cluster, nth cluster, etc.) excluding the two identified clusters among the plurality of clusters (1110) as clusters to be augmented. In this case, the electronic device (100) can augment the remaining clusters to be augmented so that the number of data matches the number of data of the second cluster.

[0331] Accordingly, the electronic device (100) can update the cluster-specific data count information (1120) to obtain new cluster-specific data count information (1130).

[0332] For example, the electronic device (100) can augment some of the datasets (e.g., a first type dataset and a second type dataset) included in each of the augmentation target clusters based on other datasets. Since data augmentation has been specifically described above, a redundant description will be omitted.

[0333] Accordingly, the electronic device (100) can increase the number of data for the first cluster from 120 to 254, and increase the number of data for the nth cluster from 53 to 254.

[0334] FIG. 12 is a flowchart illustrating a method for controlling an electronic device according to one or more embodiments of the present disclosure.

[0335] The electronic device (100) can acquire a plurality of datasets including temperature information corresponding to each of at least one storage room type (S1210).

[0336] According to one or more embodiments, the electronic device (100) can acquire a plurality of partial datasets including a plurality of temperature statistical values ​​corresponding to each of at least one storage room type based on input data and sensing data acquired through a plurality of temperature sensors.

[0337] According to one or more embodiments, the electronic device (100) can cluster into multiple datasets based on data similarity, based on multiple statistical values ​​included in each of the multiple partial datasets.

[0338] Next, the electronic device (100) can obtain temperature information corresponding to at least one of the multiple storage room types of the second type dataset based on a first type dataset containing temperature information corresponding to each of the multiple storage room types among the multiple datasets (S1220).

[0339] According to one or more embodiments, at least one processor (130) can obtain temperature information of a second type dataset based on a first type dataset.

[0340] For example, the first type dataset may include temperature information corresponding to each of the multiple storage room types among the multiple datasets. The second type dataset may include temperature information corresponding to some of the multiple storage room types among the multiple datasets.

[0341] The second type dataset may include temperature information corresponding to some of the multiple storage room types. For example, while the first type dataset includes temperature information for all storage room types, the second type dataset may include temperature information for only some of the total storage room types. Here, the temperature information may be information intended to be used as training data for a neural network model (temperature prediction model) described later.

[0342] For example, the temperature information of the second type dataset may be data generated from information included in the first dataset.

[0343] For example, at least one processor (130) can obtain augmented data by augmenting a second type dataset based on a first type dataset. For example, at least one processor (130) can augment a second type dataset, which has relatively fewer types of information, from a first type dataset, which has relatively more types of information (storage room types).

[0344] For example, the augmented data may correspond to data to be input to at least one neural network model to train at least one neural network model to output a predicted temperature corresponding to at least one of a plurality of storage room types. Through this, the electronic device (100) can train a neural network model that can be used universally without distinguishing between refrigerator types based on the augmented data. Additionally, data with relatively little information may be augmented through the first type dataset.

[0345] That is, in the case of a new product or a product that has not been purchased by a user for a long time, even if the usage history information of the product itself is relatively insufficient, the electronic device (100) can use not only the existing data but also the data generated (augmented) from the existing data as training data to train a neural network model by augmenting the existing data to suit the characteristics (storage room type, etc.) of the product.

[0346] Accordingly, an electronic device (100) or an external electronic device such as a refrigerator can improve the prediction accuracy of the model by training the temperature prediction model through the training data obtained by the electronic device (100) in this way.

[0347] Meanwhile, in FIG. 12, the order of all steps has been mapped for convenience of explanation, but it goes without saying that the order of steps that are not related to the order or can be performed in parallel is not necessarily limited to that order.

[0348] Meanwhile, methods according to at least some of the various embodiments of the present disclosure described above can be implemented in the form of an application that can be installed on an existing electronic device.

[0349] In addition, methods according to at least some of the various embodiments of the present disclosure described above may be implemented by software upgrades or hardware upgrades alone for existing electronic devices.

[0350] In addition, methods according to at least some of the various embodiments of the present disclosure described above may also be performed through an embedded server equipped in an electronic device, or through at least one external server among the electronic devices.

[0351] Meanwhile, according to one embodiment of the present disclosure, the various embodiments described above may be implemented as software containing instructions stored on a machine-readable storage medium (e.g., a computer). The machine may include an electronic device (e.g., electronic device (A)) according to the disclosed embodiments, which is a device capable of calling instructions stored from the storage medium and operating according to the called instructions. When instructions are executed by a processor, the processor may perform a function corresponding to the instructions directly or by using other components under the control of the processor. Instructions may include code generated or executed by a compiler or an interpreter. The machine-readable storage medium may be provided in the form of a non-transitory storage medium. Here, "non-transitory storage medium" simply means that it is a tangible device and does not contain a signal (e.g., electromagnetic waves), and this term does not distinguish between cases where data is stored semi-permanently and cases where it is stored temporarily in the storage medium. For example, A 'non-transient storage medium' may include a buffer in which data is temporarily stored. According to one embodiment, the method according to the various embodiments disclosed herein may be provided by being included in a computer program product. The computer program product may be traded between a seller and a buyer as a product. The computer program product may be distributed in the form of a device-readable storage medium (e.g., compact disc read-only memory (CD-ROM)), or distributed online (e.g., download or upload) through an application store (e.g., Play Store™) or directly between two user devices (e.g., smartphones).In the case of online distribution, at least a portion of a computer program product (e.g., a downloadable app) may be temporarily stored or temporarily created on a device-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0352] Various embodiments of the present disclosure may be implemented as software comprising instructions stored on a machine-readable storage medium (e.g., a computer). The machine may include an electronic device (e.g., an electronic device (100-1)) according to the disclosed embodiments, which is a device capable of calling instructions stored from the storage medium and operating according to the called instructions.

[0353] When the above-described instruction is executed by a processor, the processor may perform the function corresponding to the instruction directly or by using other components under the processor's control. The instruction may include code generated or executed by a compiler or an interpreter.

[0354] Although preferred embodiments of the present disclosure have been illustrated and described above, the present disclosure is not limited to the specific embodiments described above. It is understood that various modifications can be made by those skilled in the art without departing from the essence of the present disclosure as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present disclosure.

Claims

1. In an electronic device, Memory that stores at least one instruction; communication device; and It includes at least one processor connected to the communication device and memory to control the electronic device; and The above-mentioned at least one processor is, Based on input data received through the communication device, a plurality of datasets are obtained that include temperature information corresponding to each of at least one storage room type among a plurality of storage room types, and An electronic device that obtains temperature information corresponding to at least one of the plurality of storage room types in a second type dataset, which includes temperature information corresponding to some of the plurality of storage room types among the plurality of datasets, based on a first type dataset that includes temperature information corresponding to each of the plurality of storage room types among the plurality of datasets.

2. In Paragraph 1, The above plurality of storage room types include a refrigerator room, a freezer room, and a variable temperature room, and An electronic device in which some of the above-mentioned storage room types include at least one of the above-mentioned refrigerator room and freezer room.

3. In Paragraph 2, The above-mentioned at least one processor is, Based on temperature information included in the first type dataset, a first section in the first type dataset where a temperature change event of the variable temperature room occurred is identified, and among the temperature information included in the second type dataset, a second section having a pattern similar to the temperature information included in the identified section is identified. An electronic device that records temperature information corresponding to the temperature room in the first section as temperature information corresponding to the temperature room in the second section.

4. In Paragraph 3, The above-mentioned at least one processor is, An electronic device that identifies a first interval in which a temperature change event occurred based on at least one bit corresponding to whether the temperature change room is defrosting among the temperature information included in the first type dataset.

5. In Paragraph 3, The temperature information included in the above-mentioned first type dataset is, It includes multiple statistical values ​​arranged chronologically according to multiple time intervals for each of the above multiple storage rooms, and The above-mentioned at least one processor is, Identifying multiple time intervals that have changed by more than a threshold value from the statistical value corresponding to the previous time interval among the multiple statistical values ​​corresponding to the above-mentioned variable temperature chamber, and An electronic device that identifies a plurality of time intervals in which a temperature change event occurred based on the above-identified plurality of time intervals, and identifies a interval including the plurality of time intervals in which a temperature change event occurred as a first interval.

6. In Paragraph 3, The above-mentioned at least one processor is, Identifying the first section in which the first temperature change event of the variable temperature room occurred among the temperature information included in the first type dataset, and identifying the second section among the second temperature information included in the second type dataset, An electronic device that identifies a third section in which a second temperature change event different from the first temperature change event occurs among the temperature information included in the first type dataset, and records the temperature information corresponding to the refrigerator and freezer in the third section as the temperature information corresponding to the refrigerator and freezer in the second type dataset.

7. In Paragraph 2, The above-mentioned at least one processor is, Identifying a fourth section in which at least one temperature change event occurs among the temperature information included in the first type dataset, and An electronic device that records temperature information included in the above-mentioned fourth section as temperature information corresponding to at least one of the refrigerator room and freezer room in the above-mentioned second type dataset.

8. In Paragraph 7, The above-mentioned at least one processor is, Identifying a mode corresponding to at least one of the refrigerator and freezer based on temperature information included in the interval excluding the fourth interval among the temperature information included in the first type dataset above, and The identified mode is recorded as temperature information corresponding to at least one of the refrigerator and freezer in the second type dataset, and An electronic device that records temperature information included in the above-mentioned fourth section as temperature information corresponding to at least one of the refrigerator and freezer sections among the sections excluding the section where the mode is recorded in the above-mentioned second type dataset.

9. In Paragraph 1, It further includes a plurality of temperature sensors for detecting the temperature inside the plurality of storage rooms. The above-mentioned at least one processor is, Based on the input data and sensing data obtained through the plurality of temperature sensors, a plurality of partial datasets including a plurality of temperature statistical values ​​corresponding to each of the at least one storage room type are obtained, and An electronic device that clusters the plurality of datasets according to data similarity based on a plurality of statistical values ​​included in each of the plurality of partial datasets.

10. In Paragraph 9, The above temperature statistical values ​​include the mean, variation, and range, Each of the above mean, variance, and range is a value calculated based on temperature measurements included in the input data or the sensing data, and The above-mentioned at least one processor is, An electronic device that clusters into the same cluster when the data distance between two of the plurality of partial datasets is less than a predetermined distance.

11. In a method for controlling an electronic device, A step of acquiring a plurality of datasets including temperature information corresponding to each of at least one storage room type among a plurality of storage room types based on input data; and A control method comprising: a step of obtaining temperature information corresponding to at least one of the plurality of storage room types in a second type dataset, which includes temperature information corresponding to some of the plurality of storage room types among the plurality of datasets, based on a first type dataset that includes temperature information corresponding to each of the plurality of storage room types among the plurality of datasets.

12. In Paragraph 11, The above plurality of storage room types include a refrigerator room, a freezer room, and a variable temperature room, and A control method in which some of the above plurality of storage room types include at least one of the above refrigerator room and freezer room.

13. In Paragraph 12, The step of acquiring the above-mentioned augmented data is, A step of identifying a first section in the first type dataset where a temperature change event of the variable temperature room occurred based on temperature information included in the first type dataset, and identifying a second section among temperature information included in the second type dataset that has a pattern similar to the temperature information included in the identified section; and A control method comprising: a step of recording temperature information corresponding to the temperature room in the first section as temperature information corresponding to the temperature room in the second section.

14. In Paragraph 13, The step of identifying the second section above is, A control method comprising: a step of identifying a first interval in which a temperature change event occurred based on at least one bit corresponding to whether the temperature change room is defrosted among the temperature information included in the first type dataset.

15. A non-transient computer-readable recording medium storing computer instructions that cause said electronic device to perform an operation when executed by a processor of said electronic device, wherein said operation is, A step of acquiring a plurality of datasets corresponding to each of a plurality of device types classified according to a plurality of storage room types based on received input data; and A non-transient computer-readable recording medium comprising: a step of obtaining a temperature corresponding to at least one of the plurality of storage room types of a second type dataset, which includes temperature information corresponding to some of the plurality of storage room types among the plurality of datasets, based on a first type dataset that includes temperature information corresponding to each of the plurality of storage room types among the plurality of datasets.

Images