Frost prediction method and apparatus, electronic device, and computer-readable storage medium

Abstract

The embodiment of the application discloses a kind of frost prediction method, device, electronic equipment and computer readable storage medium, belong to frost prediction technical field.The method comprises: obtaining the sample multi-dimensional operating parameter of sample equipment and the evaporator image of sample equipment;Determine sample frost index based on evaporator image;Based on sample multi-dimensional operating parameter and sample frost index, generate target training data;Based on target training data, initial model is trained, and frost prediction model is obtained.Such, by constructing high-precision training data based on image recognition in research and development stage and training prediction frost prediction model, only needs to collect equipment operating parameter in mass production stage, and frost prediction model can be used to realize accurate frost prediction, to effectively avoid the high cost and structural complexity problem caused by adding camera, simultaneously significantly reduce refrigeration energy consumption, improve refrigeration efficiency and operating stability.

Description

Technical Field

[0001] This application relates to the field of frost prediction technology, specifically to a frost prediction method, apparatus, electronic device, and computer-readable storage medium. Background Technology

[0002] Refrigeration equipment such as refrigerators and air conditioners are usually equipped with evaporators. Frosting on the evaporator will reduce cooling efficiency and increase energy consumption.

[0003] To avoid evaporator frost formation, current methods mainly include timed defrosting and indirect detection based on temperature or humidity. However, timed defrosting cannot adapt to changing operating environments and is prone to energy waste or insufficient defrosting. Indirect detection methods are greatly affected by environmental interference and have limited accuracy. Direct detection solutions such as cameras are costly and structurally complex, making them difficult to promote and apply in mass-produced models.

[0004] Therefore, there is an urgent need for a low-cost, high-accuracy, and highly adaptable method for predicting frost formation. Summary of the Invention

[0005] This application provides a frosting prediction method, device, electronic device, and computer-readable storage medium. By constructing high-precision training data based on image recognition and training a frosting prediction model during the research and development stage, the frosting prediction model can accurately predict the degree of frost on the evaporator in the mass production stage without a camera.

[0006] In a first aspect, embodiments of this application provide a method for predicting frost formation, including: Acquire multidimensional operating parameters of the sample device and an image of the evaporator of the sample device; Based on the evaporator image, the frost index of the sample is determined; Based on the multidimensional operating parameters of the sample and the frost index of the sample, target training data is generated; The initial model is trained based on the target training data to obtain a frosting prediction model; the frosting prediction model is used to predict frosting on the evaporator.

[0007] In one embodiment, determining the sample frost index based on the evaporator image includes: Image recognition is performed on the evaporator image to obtain the frost coverage and frost thickness; The frost index of the sample is determined based on the frost coverage and the frost thickness.

[0008] In one embodiment, the multidimensional operating parameters of the samples include first sample operating parameters; the step of generating target training data based on the multidimensional operating parameters of the samples and the sample frost index includes: Feature extraction is performed on the operating parameters of the first sample to obtain the sample time series features; The target training data is generated by using the multidimensional operating parameters of the sample and the time series features of the sample as training samples, and using the frost index of the sample as the sample label.

[0009] In one embodiment, the multidimensional sample operating parameters further include a second sample operating parameter; the first sample operating parameter includes one or more of the following: sample ambient temperature, sample ambient humidity, sample evaporator temperature, sample refrigerator temperature, sample freezer temperature, and cumulative running time of the sample compressor; the second sample operating parameter includes one or more of the following: number of times the sample door is opened, total sample door opening time, and sample set temperature.

[0010] In one embodiment, after training the initial model based on the target training data to obtain the frost prediction model, the method further includes: Obtain the current multidimensional operating parameters of the target device; The current multidimensional operating parameters are input into the frost prediction model, and the frost prediction model outputs the current frost index of the evaporator of the target device.

[0011] In one embodiment, after inputting the current multidimensional operating parameters into the frosting prediction model and outputting the current frosting index of the evaporator of the target device through the frosting prediction model, the method further includes: When the current frost index is greater than the index threshold, the target device is controlled to perform defrosting and defrosting data is acquired. Based on the current multidimensional operating parameters, the current frost index, and the defrosting data, a frost prediction log is generated; The frost prediction log is uploaded to the cloud, and the model update parameters returned by the cloud are received. The model update parameters are determined by the cloud training the cloud model based on the frost prediction log to obtain the updated cloud model parameters. The model parameters of the frost prediction model are updated based on the model update parameters.

[0012] In one embodiment, training the cloud model based on the frost prediction log to obtain updated model parameters for the cloud model includes: The frost prediction log is parsed to obtain the current multidimensional operating parameters, the current frost index, and the defrosting data; Based on the current frost index and the defrosting data, frost prediction and evaluation information is determined; When the frost prediction evaluation information meets the evaluation criteria, a retraining dataset is generated based on the current multidimensional operating parameters and the current frost index. Based on the retraining dataset, the cloud model is trained to obtain the updated model parameters of the cloud model.

[0013] Secondly, embodiments of this application provide a frosting prediction device, the device comprising: The sample acquisition module is used to acquire multidimensional operating parameters of the sample device and an image of the evaporator of the sample device. An index determination module is used to determine the frosting index of a sample based on the evaporator image; The data generation module is used to generate target training data based on the multidimensional operating parameters of the sample and the frost index of the sample; The model training module is used to train the initial model based on the target training data to obtain the frost prediction model; the frost prediction model is used to predict frost formation on the evaporator.

[0014] In one embodiment, the index determination module includes: An image recognition submodule is used to perform image recognition on the evaporator image to obtain the frost coverage rate and frost thickness. The index determination submodule is used to determine the frost index of the sample based on the frost coverage and the frost thickness.

[0015] In one embodiment, the multidimensional sample operating parameters include first sample operating parameters; the data generation module includes: The feature extraction submodule is used to extract features from the running parameters of the first sample to obtain the time series features of the sample; The data generation submodule is used to generate the target training data by using the multidimensional operating parameters of the sample and the time series features of the sample as training samples and the frost index of the sample as sample labels.

[0016] In one embodiment, the multidimensional sample operating parameters further include a second sample operating parameter; the first sample operating parameter includes one or more of the following: sample ambient temperature, sample ambient humidity, sample evaporator temperature, sample refrigerator temperature, sample freezer temperature, and cumulative running time of the sample compressor; the second sample operating parameter includes one or more of the following: number of times the sample door is opened, total sample door opening time, and sample set temperature.

[0017] In one embodiment, the frost prediction device further includes: The parameter acquisition module is used to acquire the current multi-dimensional operating parameters of the target device; The frost prediction module is used to input the current multidimensional operating parameters into the frost prediction model and output the current frost index of the evaporator of the target device through the frost prediction model.

[0018] In one embodiment, the frost prediction device further includes: The defrosting module is used to control the target device to perform defrosting when the current frost index is greater than the index threshold, and to acquire defrosting data. The log generation module is used to generate a frost prediction log based on the current multidimensional operating parameters, the current frost index, and the defrosting data; The data transceiver module is used to upload the frost prediction log to the cloud and receive the model update parameters returned by the cloud. The model update parameters are determined by the cloud training the cloud model based on the frost prediction log to obtain the updated cloud model parameters. The parameter update module is used to update the model parameters of the frost prediction model based on the model update parameters.

[0019] In one embodiment, training the cloud model based on the frost prediction log to obtain updated model parameters for the cloud model includes: The frost prediction log is parsed to obtain the current multidimensional operating parameters, the current frost index, and the defrosting data; Based on the current frost index and the defrosting data, frost prediction and evaluation information is determined; When the frost prediction evaluation information meets the evaluation criteria, a retraining dataset is generated based on the current multidimensional operating parameters and the current frost index. Based on the retraining dataset, the cloud model is trained to obtain the updated model parameters of the cloud model.

[0020] Thirdly, embodiments of this application also provide an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps in the above-described frosting prediction method.

[0021] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in the above-described frosting prediction method.

[0022] Fifthly, embodiments of this application also provide a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the various optional implementations described in the embodiments of this application.

[0023] In summary, in this embodiment, multi-dimensional operating parameters of the sample device and an image of the evaporator of the sample device can be obtained; based on the evaporator image, the sample frost index is determined; based on the sample multi-dimensional operating parameters and the sample frost index, target training data is generated; based on the target training data, the initial model is trained to obtain a frost prediction model; the frost prediction model is used to predict frost formation on the evaporator. Thus, by constructing high-precision training data based on image recognition and training the frost prediction model during the R&D phase, only the device operating parameters need to be collected during the mass production phase to achieve accurate frost prediction using the frost prediction model. This effectively avoids the high cost and structural complexity issues associated with adding cameras, while significantly reducing cooling energy consumption and improving cooling efficiency and operational stability. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic flowchart of a frosting prediction method provided in an embodiment of this application; Figure 2 This is a schematic diagram of a specific embodiment of generating target training data provided in this application; Figure 3 This is a schematic diagram of a specific embodiment of the present application for predicting the current frost index; Figure 4 This is a schematic diagram of a specific embodiment of the updated frost prediction model provided in this application; Figure 5 This is a schematic diagram of the structure of a frosting prediction device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0026] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0027] It should be noted that frost formation on the evaporator surface is a common phenomenon in the refrigeration equipment industry during operation. The presence of frost severely hinders heat exchange between the evaporator and the air inside the refrigerator, leading to decreased refrigeration efficiency, increased compressor load, and higher energy consumption. Therefore, timely and effective defrosting is crucial for ensuring refrigerator performance and energy efficiency. Traditional defrosting control strategies mainly include the following: The first method is timed defrosting, which starts the defrosting program at preset fixed time intervals regardless of the actual frost condition. While this method is simple to control and inexpensive, it cannot adapt to different regional climates, seasonal changes, and varying usage conditions such as the frequency of door opening and closing and the amount of items stored. This can easily lead to unnecessary defrosting when the frost layer is still thin (over-defrosting) or failure to start defrosting when the frost layer is already thick (under-defrosting). The former results in energy waste, while the latter affects the cooling effect.

[0028] Secondly, there are indirect detection methods based on single or a few sensor signals, such as monitoring the temperature difference between the evaporator inlet and outlet, and changes in humidity inside the chamber. These methods attempt to infer the degree of frost formation by observing changes in related physical quantities. However, these signals are not only related to frost formation but are also highly susceptible to interference from various factors such as ambient temperature fluctuations, sensor accuracy drift, and airflow organization within the chamber. This results in insufficient reliability of the detection results, poor environmental adaptability, and difficulty in achieving precise control.

[0029] In recent years, some studies have attempted to use more direct detection methods, such as installing cameras or laser sensors near the evaporator to directly obtain information on frost coverage and thickness through image recognition technology. This method is highly accurate, but it requires adding dedicated image acquisition hardware to each product, which not only significantly increases material costs and structural complexity, but also imposes stringent requirements on installation location, anti-fogging, and lighting. Furthermore, it involves embedded implementation of image processing algorithms and power consumption issues, making it difficult to implement in cost-sensitive mass-produced home appliances.

[0030] In summary, there is an urgent need in this field for a technical solution that can accurately and adaptively predict the degree of frost formation on the refrigerator evaporator without increasing high hardware costs, and then execute intelligent frost control accordingly.

[0031] To address the issues of low detection accuracy and high hardware cost in current frost detection methods, this application aims to provide a frost prediction method. This method acquires multi-dimensional operating parameters and evaporator images of a sample device, and determines the sample frost index based on the evaporator image. It then generates target training data based on the sample's multi-dimensional operating parameters and frost index, and trains an initial model using this target training data to obtain a frost prediction model. Thus, by constructing a high-precision frost database and training the prediction model based on image recognition during the R&D phase, accurate frost prediction can be achieved in the mass production phase using only existing sensor data. This method effectively avoids the high cost and structural complexity associated with adding cameras, overcomes the shortcomings of poor accuracy and insufficient adaptability of traditional timed or indirect detection methods, and enables on-demand defrosting, thereby significantly reducing refrigerator energy consumption and improving cooling efficiency and operational stability.

[0032] The following sections provide detailed descriptions of each example. It should be noted that the order in which the embodiments are described is not intended to limit the priority of the embodiments.

[0033] Figure 1 The illustration shows a schematic flowchart of a frosting prediction method according to an embodiment of this application. The entity executing the frosting prediction method can be a frosting prediction device, which can be integrated into any electronic device with data processing, network communication, and program execution functions. The electronic device can be a server or a terminal, etc.

[0034] The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, network acceleration services (Content Delivery Network, CDN), as well as big data and artificial intelligence platforms.

[0035] The terminal can be a smartphone, tablet, laptop, desktop computer, etc., but is not limited to these. The terminal and the server can be connected directly or indirectly through wired or wireless communication, which is not limited herein.

[0036] In this embodiment, the description will be from the perspective of the frost prediction device, which can be integrated into a server or terminal. To facilitate the explanation of the frost prediction method of this application, the following will describe the frost prediction device integrated into the terminal in detail, that is, the terminal will be used as the execution subject for detailed explanation.

[0037] Reference Figure 1The diagram shows a flowchart of a frosting prediction method according to this application. The method may specifically include steps S101 to S103, as follows: S101: Obtain the multidimensional operating parameters of the sample equipment and the evaporator image of the sample equipment.

[0038] In this embodiment, the sample device refers to the sample cooling device used during the model training phase, which is equipped with an image acquisition device. The image acquisition device includes a camera, a laser ranging module, an infrared imaging module, etc.; the sample cooling device includes refrigerator prototypes, freezer prototypes, air conditioner prototypes, etc., and these sample cooling devices can be of the same model or different models.

[0039] In this embodiment, the multidimensional operating parameters of the sample refer to the set of multidimensional data related to the frosting process collected by various built-in or external sensors during the operation of the sample equipment. These multidimensional operating parameters reflect the operating status of the equipment and the environmental conditions in which it is located.

[0040] In this embodiment, the evaporator image refers to visual data acquired by an image acquisition device installed near the evaporator of the sample device, which clearly reflects the condition of the frost layer on the evaporator surface.

[0041] In this embodiment, during the data preparation stage of model training, conventional sensors (such as temperature and humidity sensors) configured on the sample equipment are used to continuously sample multidimensional operating parameters. At the same time, specially installed image acquisition devices are used to periodically or continuously capture images of the evaporator. This ensures that the operating parameter set on each time slice corresponds to a real frosting state derived from image analysis, providing an accurate data foundation for subsequent supervised learning.

[0042] S102: Determine the frost index of the sample based on the evaporator image.

[0043] In this embodiment, the sample frost index is an indicator parameter used to quantify the severity of frost formation on the evaporator surface of the sample device. This sample frost index is obtained based on the analysis of evaporator images; a higher sample frost index indicates more severe frost formation on the sample device.

[0044] In practical implementation, after acquiring the evaporator image, the terminal can apply image processing algorithms (such as image segmentation, edge detection, pixel statistics, etc.) to the acquired evaporator image to identify the frosting condition of the sample device, thereby evaluating the sample frosting index. This sample frosting index can digitize visual information and serve as a target value for model learning.

[0045] S103: Generate target training data based on sample multidimensional operating parameters and sample frost index.

[0046] In this embodiment, the target training data refers to the structured dataset prepared for the initial model training, wherein the target training data uses the multidimensional running parameters of the samples as input features and the corresponding sample frost index as the label of the expected output.

[0047] In this embodiment, the multidimensional operating parameters of each group of samples are paired with the frost index of the samples determined at the same time to form a target training data. Then, by collecting the target training data at all time points, a target training dataset can be constructed.

[0048] S104: Train the initial model based on the target training data to obtain the frost prediction model.

[0049] In this embodiment, the initial model refers to a machine learning model that is either untrained or has a basic structure. This initial model can employ linear regression, multinomial regression, random forest, neural networks, or similar models. The frost prediction model refers to the model obtained after training the initial model using the target training data. This frost prediction model is used to predict frost formation on the evaporator. Specifically, the frost prediction model outputs the current frost index of the target device based on its current multidimensional operating parameters. The target device is a refrigeration unit without image acquisition devices such as cameras.

[0050] In this embodiment, after generating a sufficient amount of target training data, the terminal will iteratively train the initial model using the target training data. Through iterative optimization (such as adjusting the neural network weights and constructing a decision tree structure), the initial model can continuously reduce the error between the frost index predicted by the model and the actual sample frost index, and finally obtain a frost prediction model that can accurately infer the frost index from new and unseen operating parameters.

[0051] In this embodiment, after the frost prediction model is trained, it can be deployed to the target device so that the frost prediction model can output the current frost index of the target device based on the current multidimensional operating parameters of the target device.

[0052] The technical solution of this application, by utilizing image recognition technology to obtain high-precision sample frost indices during the R&D phase and combining them with synchronously acquired multi-dimensional operating parameters of the samples to train an initial model, successfully transfers the direct visual detection capability of frost conditions to mass-production equipment that relies solely on conventional sensors. This method avoids installing costly and complex image acquisition devices on each mass-production device, significantly reducing hardware costs and installation difficulty. Simultaneously, compared to traditional timed or single-parameter threshold methods, it more accurately reflects the actual frost dynamics, enabling on-demand defrosting, effectively reducing unnecessary defrosting cycles or avoiding defrosting delays, ultimately achieving the beneficial effects of reducing equipment energy consumption and improving cooling efficiency and stability.

[0053] In one feasible implementation, the step of determining the sample frost index based on the evaporator image may specifically include: performing image recognition on the evaporator image to obtain the frost coverage and frost thickness; and determining the sample frost index based on the frost coverage and frost thickness.

[0054] In this embodiment, frost coverage refers to the ratio of the pixel area or region area covered by frost within the effective area of ​​the evaporator image to the total pixel area or total reference area of ​​the evaporator. Frost coverage can be expressed as a percentage, reflecting the extent of frost formation on the evaporator.

[0055] In this embodiment, the evaporator image can first be preprocessed (e.g., noise reduction, contrast enhancement), and then an image segmentation algorithm (e.g., threshold-based segmentation, edge detection, or machine learning segmentation model) can be used to distinguish the frost-covered areas from the non-frost-covered areas (e.g., metal fins, background) in the image. Subsequently, the area of ​​the frost-covered area is obtained by pixel statistics or area calculation, and then divided by the total area of ​​the reference area of ​​the evaporator image to calculate the frost coverage rate.

[0056] In this embodiment, frost thickness refers to an estimate of the average height or degree of accumulation of frost on the evaporator surface in the direction perpendicular to the fins or tube wall. This frost thickness reflects the depth of frost formation on the evaporator.

[0057] In this embodiment, the frost thickness can be estimated using information such as the shadow features, texture variation features, and edge contour contrast of the frost layer in the image. Specifically, the frost thickness can be determined by analyzing the local gray-level gradient or texture statistical features (such as Gabor filters or local binary patterns) of the evaporator image, which can cause local texture coarsening, edge blurring, or changes in surface reflectivity due to frost accumulation. Alternatively, visual features such as shadow features, texture variation features, and edge contour contrast can be input into a pre-trained thickness prediction model, and the frost thickness can be obtained by outputting the thickness prediction model.

[0058] In this embodiment, the sample frost index is a single numerical index that integrates information on frost coverage and frost thickness, and is used to more comprehensively characterize the overall frost load.

[0059] In this embodiment, it is considered that the effects of frost coverage and thickness on heat exchange efficiency are not simply linearly additive. Low coverage but large thickness, versus high coverage but thin thickness, may have different degrees of impact on refrigeration performance. Therefore, a comprehensive calculation model can be pre-defined. This comprehensive calculation model can adopt a linear weighted summation model or a nonlinear function.

[0060] In this embodiment, the frost index of the sample can be determined based on frost coverage and frost thickness using the following linear weighted summation model: CI = α × S + β × T (1); Where CI represents the sample frost index; S represents the frost coverage rate; T represents the frost thickness; α represents the first weight set for the frost coverage rate; and β represents the second weight set for the frost coverage rate.

[0061] In this embodiment, the frost index of the sample can also be determined based on the frost coverage and frost thickness using the following nonlinear function: CI=S^γ×T^δ(2; Where CI represents the sample frost index; S represents the frost coverage rate; T represents the frost thickness; γ represents the third weight set for the frost coverage rate; and δ represents the fourth weight set for the frost coverage rate.

[0062] In this embodiment, by introducing image recognition technology, the visual state of the evaporator surface is precisely decomposed into two quantifiable key parameters: frost coverage and frost thickness. These are further combined and calculated into a sample frost index. Compared to judging solely by coverage or thickness, this sample frost index comprehensively reflects the spatial distribution and accumulation of frost, thus more accurately characterizing the actual impact of frost on heat exchange performance. This provides crucial, high-quality labeled data for training a high-precision frost prediction model, thereby improving the model training effect during the training phase and the model prediction effect during mass production.

[0063] In one feasible implementation, refer to Figure 2 The multidimensional operating parameters of the samples include the first sample operating parameters. Based on the multidimensional operating parameters of the samples and the sample frost index, the target training data is generated, which may specifically include steps S201~S202, as follows: S201: Extract features from the running parameters of the first sample to obtain the time series features of the sample.

[0064] In this embodiment, considering that frosting is a dynamic process of accumulation and development, the current degree of frosting depends not only on the current environment and operating status, but also on the changing trend and persistence of these statuses over a period of time. If only instantaneous parameter values ​​are used, this key information will be lost. Therefore, the first sample operating parameters will be extracted from the multidimensional operating parameters of the sample for time series analysis.

[0065] In this embodiment, the first sample operating parameter is the sample operating parameter with temporal characteristics among the multidimensional sample operating parameters. Specifically, the first sample operating parameter refers to the time-series data that is continuously collected on the sample device at a preset frequency and can reflect the device's operating status and environmental conditions. The first sample operating parameter itself can constitute a sequence that changes over time.

[0066] In this embodiment, feature extraction refers to the process of deriving new features from the first sample running parameter data arranged in chronological order through mathematical calculations or statistical methods, which can more effectively characterize the inherent patterns, trends, or periodic information of the data.

[0067] In this embodiment, the sample time series feature refers to a new feature obtained by analyzing and calculating the time series of the first sample's operating parameters. This sample time series feature is not a parameter value at a single moment, but rather a general indicator reflecting the parameter behavior over a time window.

[0068] In practical implementation, the time series features of the sample can specifically include statistical features, trend features, and / or frequency domain features. Statistical features include the mean and standard deviation of the first sample's operating parameters within the target time window; trend features include the slope and intercept obtained by linearly fitting the first sample's operating parameters within the target time window, or the mean and standard deviation of the first-order differences of data at adjacent time points; frequency domain features include frequency features extracted from the first sample's operating parameters through Fourier transform. For example, when the first sample's operating parameters include the sample evaporator temperature, feature extraction of the sample evaporator temperature over the past hour can yield multiple time series features such as the average evaporator temperature, the standard deviation of evaporator temperature fluctuations, and the slope of evaporator temperature decrease. These features collectively describe the behavioral pattern of the sample evaporator temperature over a past period.

[0069] S202: Use the multidimensional operating parameters and time series features of the samples as training samples, and use the frost index of the samples as sample labels to generate target training data.

[0070] In this embodiment, for each time point used to collect multidimensional operating parameters of the samples, the following operations can be performed: obtain the multidimensional operating parameters of the samples at that time point; extract features from the first sample operating parameters based on that time point and the historical first sample operating parameters within a preset time window before that time point to obtain sample time series features; merge the multidimensional operating parameters of the samples and the sample time series features to generate training samples; label the sample frost index determined based on the evaporator image at that time point as a sample label; pair the training sample with the sample label to obtain target training data.

[0071] In this embodiment, by extracting time-series features from the operating parameters of the first sample and using these features along with the multi-dimensional operating parameters of the sample as model input, the information dimensions contained in the training data can be greatly enriched. This allows the finally trained frost prediction model to not only respond to immediate environmental and operating conditions but also understand the inertia, cumulative effects, and changing trends of the frost process. For example, the frost prediction model can learn dynamic patterns such as a faster frost rate under sustained low temperature and high humidity conditions, even if the current parameters are the same as at another time. This significantly improves the prediction accuracy and robustness for complex and variable frost scenarios, laying a crucial data foundation for achieving more accurate and forward-looking defrosting control.

[0072] In one feasible implementation, the multidimensional sample operating parameters further include a second sample operating parameter; the first sample operating parameter includes one or more of the following: sample ambient temperature, sample ambient humidity, sample evaporator temperature, sample refrigerator temperature, sample freezer temperature, and cumulative running time of the sample compressor; the second sample operating parameter includes one or more of the following: number of times the sample door is opened, total sample door opening time, and sample set temperature.

[0073] In this embodiment, the sample ambient temperature refers to the air temperature of the external environment where the sample device is located. This parameter directly affects the temperature difference between the evaporator and the environment, and is the fundamental thermodynamic driving force determining the rate of condensation and subsequent frosting. The higher the ambient temperature, the greater the temperature difference between the inside and outside of the refrigerator, and the greater the potential frosting load.

[0074] In this embodiment, the sample environment humidity refers to the relative humidity of the air in the external environment of the sample device. Humidity is a direct measure of the water vapor content in the air and is the source of frost formation. A high humidity environment means that the air entering the chamber carries more moisture, significantly accelerating the frosting process on the evaporator surface.

[0075] In this embodiment, the sample evaporator temperature refers to the temperature of the evaporator fins or coil surface of the sample device. This is a direct condition that determines whether water vapor can condense and freeze on its surface. The lower the evaporator temperature (usually well below 0°C), the easier it is for water vapor to directly sublimate into frost on its surface, and the frost rate is closely related to the evaporator temperature.

[0076] In this embodiment, the sample device can be a refrigerator, freezer, or other device equipped with refrigeration equipment. The sample refrigerator compartment temperature refers to the air temperature of the refrigerator compartment of the sample device, and the sample freezer compartment temperature refers to the air temperature of the freezer compartment of the sample device. The sample refrigerator compartment temperature and the sample freezer compartment temperature, as the air temperatures of each compartment of the sample device, reflect the cooling effect and heat load of the refrigeration system. The set values, actual values, and fluctuations of the sample refrigerator compartment temperature and the sample freezer compartment temperature together determine the compressor start-stop cycle and the evaporator cooling load, thereby affecting the duration and intensity of frosting.

[0077] In this embodiment, the cumulative runtime of the sample compressor refers to the total time the compressor has been operating since the last defrosting. The cumulative runtime of the sample compressor is a cumulative quantity that directly reflects the length of time the evaporator is in a low-temperature cooling state and continuously in contact with humid air. It is a macroscopic time scale for the accumulation of frost over time.

[0078] In this embodiment, unlike the first sample operating parameter which is a sample operating parameter with temporal characteristics among the multidimensional sample operating parameters, the second sample operating parameter is a sample operating parameter without temporal characteristics among the multidimensional sample operating parameters. This second sample operating parameter refers to a sample operating parameter that reflects user habits, device configuration, or discrete events and has an indirect but significant impact on the frosting process.

[0079] In this embodiment, the number of times the sample door is opened refers to the cumulative number of times the sample equipment door has been opened since the last defrosting ended within the current time period. Each time the door is opened, warm and humid air from outside quickly enters the chamber, which is the main source of disturbance that causes a sudden increase in humidity and heat load inside the chamber.

[0080] In this embodiment, the total sample door opening time refers to the cumulative time the sample equipment door is open during the current time period. The longer the total time, the more sufficient the exchange between the external humid air and the internal cold air, the more moisture is introduced, and the more significant the contribution to frost formation.

[0081] In this embodiment, the sample set temperature refers to the target temperature value set by the user for the sample device. Specifically, the sample set temperature may include the sample refrigerator compartment set temperature and the sample freezer compartment set temperature. The set temperature determines the target of the control system, directly affecting the compressor's operating strategy and the average temperature level of the evaporator, thereby influencing the frosting trend from the source of the control logic.

[0082] In this embodiment, by acquiring sample operating parameters in two dimensions—first sample operating parameters reflecting the physical state of the equipment and second sample operating parameters related to the user and configuration—the comprehensiveness of the multi-dimensional sample operating parameters can be effectively improved. This ensures that the multi-dimensional sample operating parameters not only cover the core thermodynamic variables driving frost (such as temperature and humidity) but also incorporate dynamic disturbance factors reflecting actual usage scenarios (such as door opening behavior) and system configuration factors (such as set temperature). This comprehensive parameter combination allows the subsequently trained frost prediction model to simultaneously learn the physical laws and behavioral patterns of frost, greatly enhancing the model's adaptability and generalization ability to different usage environments and user habits, providing a solid and complete data foundation for achieving accurate personalized frost prediction.

[0083] In one feasible implementation, refer to Figure 3 After training the initial model based on the target training data to obtain the frost prediction model, the frost prediction method may further include steps S301~S302, as follows: S301: Obtain the current multidimensional operating parameters of the target device.

[0084] In this embodiment, the target device refers to mass-produced refrigeration equipment that has been put into actual use but has not yet been equipped with an image acquisition device for directly detecting frost. This target device includes mass-produced refrigerators, freezers, and air conditioners. The hardware configuration of the target device is limited to conventional sensors and control units.

[0085] In this embodiment, the current multidimensional operating parameters refer to the operating data of the target device in its current operating state.

[0086] S302: Input the current multi-dimensional operating parameters into the frost prediction model, and output the current frost index of the evaporator of the target device through the frost prediction model.

[0087] In this embodiment, the current multidimensional operating parameters may specifically include a first operating parameter and a second operating parameter. The first operating parameter and the first sample operating parameter are of the same type, and the first operating parameter may specifically include one or more of the following: ambient temperature, ambient humidity, evaporator temperature, refrigerator compartment temperature, freezer compartment temperature, and cumulative compressor operating time. The second operating parameter and the second sample operating parameter are of the same type, and the second operating parameter may include one or more of the following: number of door openings, total door opening time, and set temperature.

[0088] In this embodiment, the step of inputting the current multidimensional operating parameters into the frosting prediction model and outputting the current frosting index of the evaporator of the target device through the frosting prediction model may specifically include: extracting features from the first operating parameters to obtain the current time series features; inputting the current multidimensional operating parameters and the current time series features into the frosting prediction model and outputting the current frosting index of the evaporator of the target device through the frosting prediction model.

[0089] In this embodiment, the current frost index is the predicted value calculated and output by the frost prediction model after the current multidimensional operating parameters are input into the frost prediction model. This current frost index is used to characterize the comprehensive evaluation result of the frost prediction model on the severity of frost on the evaporator of the target equipment at the current moment. The current frost index can be a dimensionless or a scalar with specific dimensions.

[0090] In this embodiment, data can be periodically (e.g., every 5 minutes) read from various sensors (such as temperature and humidity sensors) of the target device, and status information (such as cumulative compressor running time and door opening / closing records) can be obtained from the system register. Following a predetermined format and rules, this raw data undergoes necessary preprocessing (e.g., filtering, unit conversion) and feature construction (e.g., calculating the average temperature over the past hour), assembling it into a current multidimensional operating parameter vector that is completely consistent with the feature vector structure used during the training of the frost prediction model. This current multidimensional operating parameter vector is then input into the frost prediction model deployed in the main control board memory of the target device. Based on the complex nonlinear mapping relationships learned during the training phase, the frost prediction model performs a series of mathematical operations on the input features (e.g., matrix multiplication and activation function transformation in a neural network, or path determination in a tree model). After internal processing by the frost prediction model, a specific numerical value, the current frost index, is finally generated at the output layer.

[0091] In this embodiment, by lightweighting and embedding the frosting prediction model into the main control system of the target device, and combining it with the target device's existing sensor network, real-time non-invasive prediction of the evaporator's frosting level can be achieved without relying on expensive peripherals such as cameras. This enables low-cost, high-precision intelligent frosting control to be implemented in mass-produced target devices, thereby providing accurate data support for reducing energy consumption and improving performance of the target devices.

[0092] In one feasible implementation, refer to Figure 4 After inputting the current multidimensional operating parameters into the frosting prediction model and outputting the current frosting index of the evaporator of the target equipment through the frosting prediction model, the frosting prediction method may also include steps S401~S404, as follows: S401: When the current frost index is greater than the index threshold, control the target device to perform defrosting and acquire defrosting data.

[0093] In this embodiment, the index threshold is a preset critical value for the degree of frost. When the predicted current frost index exceeds this index threshold, it will be determined that defrosting is required for the target equipment. Specifically, a command can be sent to the execution unit (defrosting heater, fan, etc.) of the target equipment to start the defrosting program, thereby achieving automated defrosting of the target equipment.

[0094] In this embodiment, defrosting data refers to the result data recorded after defrosting the target equipment. This defrosting data is used to evaluate the actual effect of the defrosting process. Specific defrosting data may include defrosting duration, evaporator temperature recovery curve after defrosting, defrosting energy consumption, etc.

[0095] In this embodiment, defrosting data can reflect the actual degree of frost formation on the target device. For example, when the defrosting data is the defrosting duration, the longer the defrosting duration, the greater the actual degree of frost formation on the target device, and the longer the defrosting process takes.

[0096] S402: Generate a frost prediction log based on the current multidimensional operating parameters, the current frost index, and defrosting data.

[0097] In this embodiment, the frost prediction log refers to a set of structured data packaged and uploaded for model optimization purposes. The frost prediction log contains at least a complete "input-output-result" information chain that triggers the defrosting decision, namely: the current multidimensional operating parameters at the trigger time (input), the current frost index predicted by the model (output / decision basis), and the defrosting data collected subsequently (result / verification information).

[0098] S403: Upload the frost prediction log to the cloud and receive the model update parameters returned by the cloud.

[0099] In this embodiment, the model update parameters are determined by training the cloud model based on the frost prediction logs to obtain the updated cloud model parameters, and then determining the updated cloud model parameters.

[0100] In this embodiment, the cloud refers to a server or server cluster deployed on the Internet. The cloud has powerful data storage and computing capabilities, which are used to centrally process the frost prediction logs uploaded from a large number of mass-produced devices and to retrain the cloud model.

[0101] In practice, the target device can upload frost prediction logs to the cloud via its network modules (such as cellular networks, Wi-Fi, etc.). The cloud can then aggregate frost prediction logs from the target device and a large number of other devices, and retrain or fine-tune the cloud model based on the aggregated frost prediction logs, thereby generating model update parameters that are more adapted to real-world, complex usage environments.

[0102] In this embodiment, the model update parameter can represent the change in model parameters, which can be the difference between the model parameters of the cloud model and the model parameters of the frost prediction model. That is, instead of sending the complete model parameters of the cloud model to the target device, the cloud only sends the change in model parameters to the target device, so that the target device updates the frost prediction model according to the change in model parameters. This significantly reduces the amount of data transmitted for each collaborative learning session, and compared to full parameter synchronization, this mechanism effectively reduces network bandwidth usage.

[0103] S404: Update the model parameters of the frost prediction model based on the model update parameters.

[0104] In this embodiment, after receiving the model update parameters returned by the cloud, the target device can update the parameters of the frost prediction model deployed on the target device, thereby achieving parameter synchronization between the frost prediction model and the cloud model.

[0105] In this embodiment, the model update parameters are the parameters used to update the model parameters of the frost prediction model. For example, when the model update parameters use the change in model parameters, the target device can update the latest model parameters of the frost prediction model by summing the change in model parameters and the model parameters of the frost prediction model.

[0106] In this embodiment, by adopting an architecture that combines edge computing and cloud-based federated learning, and by continuously iterating and optimizing the frost prediction model using massive frost prediction logs from real-world scenarios, the frost prediction model can achieve real-time inference and continuous optimization. This effectively adapts to different regional climates, diverse user habits, and drift caused by device aging, thereby maintaining high prediction accuracy and intelligent control over the long term, and improving the long-term reliability and energy efficiency of the product.

[0107] In one feasible implementation, the step of training a cloud-based model based on frost prediction logs to obtain updated model parameters may specifically include: parsing the frost prediction logs to obtain current multidimensional operating parameters, current frost index, and defrosting data; determining frost prediction evaluation information based on the current frost index and defrosting data; generating a retraining dataset based on the current multidimensional operating parameters and current frost index when the frost prediction evaluation information meets the evaluation criteria; and training the cloud-based model based on the retraining dataset to obtain updated model parameters.

[0108] In this embodiment, the cloud can decode and extract data from the frost prediction logs of various target devices to obtain current multi-dimensional operating parameters, the current frost index, and defrost data. Since defrost data reflects the actual degree of frost on the target device, frost prediction evaluation information is determined based on the current frost index and defrost data. This frost prediction evaluation information serves as a quantitative representation of the accuracy of the current frost index.

[0109] In this embodiment, the step of determining the frost prediction assessment information based on the current frost index and defrosting data may specifically include: determining the actual defrosting index based on the correspondence between defrosting data and preset defrosting data and the actual defrosting index; and determining the absolute value of the difference between the actual defrosting index and the current frost index as the frost prediction assessment information.

[0110] In this embodiment, the actual defrost index is an estimate of the actual degree of frost formation on the evaporator before defrost, calculated back from defrost data. The principle is that there is a strong correlation between the energy or time required for defrosting and the amount of frost to be melted (i.e., the degree of frost formation). For example, when the defrost data is defrost time, the defrost time can be converted to the actual defrost index using a pre-defined correspondence between defrost time and the actual defrost index. Alternatively, when the defrost data is defrost energy consumption, the defrost energy consumption can be converted to the actual defrost index using a pre-defined correspondence between defrost energy consumption and the actual defrost index.

[0111] In this embodiment, the evaluation metric is a threshold condition used to determine whether the prediction accuracy of the frost prediction model is acceptable. The evaluation metric can be: the frost prediction evaluation information of a single prediction is less than a first error threshold and / or the average value of N consecutive frost prediction evaluation information is less than a second error threshold. Here, N is a positive integer greater than or equal to 2.

[0112] For example, the first error threshold can be set to 15%, meaning that the frost prediction evaluation information meets the evaluation criteria when the absolute value of the difference between the current frost index output by the frost prediction model and the actual frost index is less than 15%. The second error threshold can be set to 10%, and N can be set to 3, meaning that the frost prediction evaluation information meets the evaluation criteria when the average value of the absolute values ​​of the differences between the current frost index output by the frost prediction model and the actual frost index obtained from three consecutive calculations is less than 10%.

[0113] In this implementation, the retraining dataset refers to the dataset constructed to optimize the cloud model. If the frost prediction evaluation information meets the evaluation criteria, it indicates that the frost prediction model has relatively accurate prediction accuracy. At this time, the current multidimensional running parameters and the current frost index in the frost prediction log are paired to form a high-quality retraining sample, which is then added to the retraining dataset.

[0114] In this implementation, the cloud can use a retraining dataset containing multiple high-quality retraining samples to train or fine-tune the cloud model. The retraining process aims to minimize the prediction loss of the cloud model on these new samples, i.e., adjust the model parameters so that the output of the cloud model is closer to the more realistic frost index. After training is complete, the updated model parameters of the cloud model are saved, and the updated model parameters to be deployed are calculated based on the updated model parameters.

[0115] In this embodiment, the actual defrosting index of the evaporator frost level is derived by using defrosting data, overcoming the limitation that mass-produced target equipment cannot directly measure the frost layer. The monitoring signal required for optimization is obtained indirectly through physical processes, enabling the cloud model to continuously learn from the actual operating effects of a large number of target equipment, accurately correct prediction deviations, and achieve the ability of the cloud model and the frost prediction model to continuously optimize and maintain long-term accuracy without increasing any hardware costs, thereby significantly improving the reliability and adaptability of intelligent defrosting control.

[0116] To facilitate better implementation of the frosting prediction method of this application, this application also provides a frosting prediction device based on the above-described frosting prediction method. The meanings of the terms used are the same as in the frosting prediction method described above, and specific implementation details can be found in the descriptions of the method embodiments.

[0117] Based on the same inventive concept, and referring to Figure 5 This application provides a frosting prediction device 500, which includes: The sample acquisition module 501 is used to acquire the multidimensional operating parameters of the sample device and the evaporator image of the sample device. The index determination module 502 is used to determine the frosting index of the sample based on the evaporator image; The data generation module 503 is used to generate target training data based on the multidimensional operating parameters of the samples and the frost index of the samples; The model training module 504 is used to train the initial model based on the target training data to obtain the frost prediction model; the frost prediction model is used to predict frost on the evaporator.

[0118] In one embodiment, the index determination module 502 includes: The image recognition submodule is used to perform image recognition on evaporator images to obtain frost coverage and frost thickness. The index determination submodule is used to determine the frost index of a sample based on frost coverage and frost thickness.

[0119] In one embodiment, the multidimensional sample operating parameters include first sample operating parameters; the data generation module includes: The feature extraction submodule is used to extract features from the running parameters of the first sample to obtain the time series features of the sample; The data generation submodule is used to generate target training data by using the multidimensional operating parameters and time series features of the samples as training samples and the frost index of the samples as sample labels.

[0120] In one embodiment, the multidimensional sample operating parameters further include a second sample operating parameter; the first sample operating parameter includes one or more of the following: sample ambient temperature, sample ambient humidity, sample evaporator temperature, sample refrigerator temperature, sample freezer temperature, and cumulative running time of the sample compressor; the second sample operating parameter includes one or more of the following: number of times the sample door is opened, total sample door opening time, and sample set temperature.

[0121] In one embodiment, the frost prediction device 500 further includes: The parameter acquisition module is used to acquire the current multi-dimensional operating parameters of the target device; The frost prediction module is used to input the current multi-dimensional operating parameters into the frost prediction model, and output the current frost index of the evaporator of the target device through the frost prediction model.

[0122] In one embodiment, the frost prediction device 500 further includes: The defrosting module is used to control the target device to perform defrosting when the current frost index is greater than the index threshold, and to acquire defrosting data. The log generation module is used to generate frost prediction logs based on the current multidimensional operating parameters, the current frost index, and defrosting data. The data transceiver module is used to upload frost prediction logs to the cloud and receive model update parameters returned by the cloud. The model update parameters are determined by the cloud training the cloud model based on the frost prediction logs to obtain the updated cloud model parameters. The parameter update module is used to update the model parameters of the frost prediction model based on the model update parameters.

[0123] In one embodiment, the cloud model is trained based on frost prediction logs to obtain updated model parameters, including: The frost prediction log is analyzed to obtain the current multidimensional operating parameters, the current frost index, and defrosting data; Based on the current frost index and defrosting data, determine the frost prediction and assessment information; When the frost prediction evaluation information meets the evaluation criteria, a retraining dataset is generated based on the current multidimensional operating parameters and the current frost index. The cloud model is trained based on the retraining dataset to obtain the updated model parameters.

[0124] The technical solution of this application, by utilizing image recognition technology to obtain high-precision sample frost indices during the R&D phase and combining them with synchronously acquired multi-dimensional operating parameters of the samples to train an initial model, successfully transfers the direct visual detection capability of frost conditions to mass-production equipment that relies solely on conventional sensors. This method avoids installing costly and complex image acquisition devices on each mass-production device, significantly reducing hardware costs and installation difficulty. Simultaneously, compared to traditional timed or single-parameter threshold methods, it more accurately reflects the actual frost dynamics, enabling on-demand defrosting, effectively reducing unnecessary defrosting cycles or avoiding defrosting delays, ultimately achieving the beneficial effects of reducing equipment energy consumption and improving cooling efficiency and stability.

[0125] For specific limitations regarding the frosting prediction device 500, please refer to the limitations of the frosting prediction method above, which will not be repeated here. Each module in the aforementioned frosting prediction device 500 can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0126] In addition, this application also provides an electronic device, such as Figure 6 As shown, it illustrates the structural diagram of the electronic device involved in this application, specifically: The electronic device may include components such as a processor 601 with one or more processing cores and a memory 602 with one or more computer-readable storage media. Those skilled in the art will understand that... Figure 6 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein: The processor 601 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines, and performs various functions and processes data by running or executing software programs and / or modules stored in the memory 602, and by calling data stored in the memory 602, thereby providing overall monitoring of the electronic device. Optionally, the processor 601 may include one or more processing cores; preferably, the processor 601 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 601.

[0127] The memory 602 can be used to store software programs and modules. The processor 601 executes various functional applications and data processing by running the software programs and modules stored in the memory 602. The memory 602 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 602 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 602 may also include a memory controller to provide the processor 601 with access to the memory 602.

[0128] In one feasible implementation, the electronic device further includes a power supply 603 that supplies power to the various components. Preferably, the power supply 603 can be logically connected to the processor 601 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 603 may also include one or more DC or AC power supplies, recharging systems, power equipment debugging circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0129] In one feasible implementation, the electronic device may further include an input unit 604, which can be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0130] Although not shown, the electronic device may also include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 601 in the electronic device loads the executable files corresponding to the processes of one or more applications into the memory 602 according to the following instructions, and the processor 601 runs the applications stored in the memory 602, thereby implementing the steps in any of the frosting prediction methods provided in the embodiments of this application.

[0131] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0132] In one feasible implementation, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the methods described in any embodiment of this application.

[0133] In one feasible implementation, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the methods described in any embodiment of this application.

[0134] In one feasible implementation, a computer program product is also proposed, comprising a computer program or instructions that, when executed by a processor, implement the methods described in any embodiment of this application.

[0135] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0136] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0137] Therefore, this application provides a computer-readable storage medium storing a computer program that can be loaded by a processor to perform the steps in any of the frost prediction methods provided in this application.

[0138] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0139] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0140] Since the instructions stored in the computer-readable storage medium can execute the steps of any of the frost prediction methods provided in this application, the beneficial effects that any of the frost prediction methods provided in this application can achieve can be realized, as detailed in the preceding embodiments, and will not be repeated here.

[0141] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0142] The above provides a detailed description of the frost prediction method, apparatus, electronic device, and computer-readable storage medium provided in this application. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, those skilled in the art will recognize that there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for predicting frost formation, characterized in that, The method includes: Acquire multidimensional operating parameters of the sample device and an image of the evaporator of the sample device; Based on the evaporator image, the frost index of the sample is determined; Based on the multidimensional operating parameters of the sample and the frost index of the sample, target training data is generated; The initial model is trained based on the target training data to obtain a frosting prediction model; the frosting prediction model is used to predict frosting on the evaporator.

2. The frost prediction method according to claim 1, characterized in that, The determination of the sample frost index based on the evaporator image includes: Image recognition is performed on the evaporator image to obtain the frost coverage and frost thickness; The frost index of the sample is determined based on the frost coverage and the frost thickness.

3. The frost prediction method according to claim 1, characterized in that, The multidimensional operating parameters of the samples include first sample operating parameters; the generation of target training data based on the multidimensional operating parameters of the samples and the sample frost index includes: Feature extraction is performed on the operating parameters of the first sample to obtain the sample time series features; The target training data is generated by using the multidimensional operating parameters of the sample and the time series features of the sample as training samples, and using the frost index of the sample as the sample label.

4. The frost prediction method according to claim 3, characterized in that, The multidimensional operating parameters of the sample also include a second sample operating parameter; the first sample operating parameter includes one or more of the following: sample ambient temperature, sample ambient humidity, sample evaporator temperature, sample refrigerator temperature, sample freezer temperature, and cumulative running time of the sample compressor; the second sample operating parameter includes one or more of the following: number of times the sample door is opened, total sample door opening time, and sample set temperature.

5. The frost prediction method according to claim 1, characterized in that, After training the initial model based on the target training data to obtain the frost prediction model, the method further includes: Obtain the current multidimensional operating parameters of the target device; The current multidimensional operating parameters are input into the frost prediction model, and the frost prediction model outputs the current frost index of the evaporator of the target device.

6. The frost prediction method according to claim 5, characterized in that, After inputting the current multidimensional operating parameters into the frosting prediction model and outputting the current frosting index of the evaporator of the target device through the frosting prediction model, the method further includes: When the current frost index is greater than the index threshold, the target device is controlled to perform defrosting and defrosting data is acquired. Based on the current multidimensional operating parameters, the current frost index, and the defrosting data, a frost prediction log is generated; The frost prediction log is uploaded to the cloud, and the model update parameters returned by the cloud are received. The model update parameters are determined by the cloud training the cloud model based on the frost prediction log to obtain the updated cloud model parameters. The model parameters of the frost prediction model are updated based on the model update parameters.

7. The frost prediction method according to claim 6, characterized in that, The process of training the cloud model based on the frost prediction log to obtain updated model parameters includes: The frost prediction log is parsed to obtain the current multidimensional operating parameters, the current frost index, and the defrosting data; Based on the current frost index and the defrosting data, frost prediction and evaluation information is determined; When the frost prediction evaluation information meets the evaluation criteria, a retraining dataset is generated based on the current multidimensional operating parameters and the current frost index. Based on the retraining dataset, the cloud model is trained to obtain the updated model parameters of the cloud model.

8. A frosting prediction device, characterized in that, The device includes: The sample acquisition module is used to acquire multidimensional operating parameters of the sample device and an image of the evaporator of the sample device. An index determination module is used to determine the frosting index of a sample based on the evaporator image; The data generation module is used to generate target training data based on the multidimensional operating parameters of the sample and the frost index of the sample; The model training module is used to train the initial model based on the target training data to obtain the frost prediction model; the frost prediction model is used to predict frost formation on the evaporator.

9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps in the frost prediction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the frost prediction method as described in any one of claims 1 to 7.

Images