Equipment health prediction method and computer readable storage medium

By building a health prediction model, predicting the health status of terminal equipment batteries, the problem of the existing technology being unable to be warned in advance is solved, and the prevention of equipment failures and the reduction of operation and maintenance costs are achieved.

CN118962498APending Publication Date: 2024-11-15SHENZHEN ZOLON TECH CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202410918445.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-09
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing technology cannot effectively predict the health status of terminal equipment batteries, resulting in equipment performance decay and failure, and cannot be warned in advance, resulting in economic losses and increased operation and maintenance costs.

Method used

By determining the current battery health data of the target device, building a health prediction model, predicting future battery health status, providing early warning and maintenance suggestions, and reducing the risk of failure.

Benefits of technology

It realizes accurate prediction of the health status of terminal equipment batteries, early warning, avoid equipment failures, reduce operation and maintenance costs, and improve work efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118962498A_ABST
    Figure CN118962498A_ABST
Patent Text Reader

Abstract

The invention discloses a device health prediction method and a computer readable storage medium, and the method comprises the steps: determining first health data of a target device, the first health data comprising battery health data of the target device in a current time period; and determining battery health prediction data of the target equipment in a future preset period according to the first health data. According to the scheme, the battery health condition of the target equipment in the future preset period can be efficiently and accurately predicted. Therefore, early warning can be further carried out before the equipment possibly breaks down, health management of the equipment can be effectively achieved, equipment faults are prevented, economic losses caused by the equipment faults are reduced, the operation and maintenance cost can be saved, and the working efficiency is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the technical field of equipment health prediction, and in particular, relates to an equipment health prediction method and a computer-readable storage medium. Background Art

[0002] With the development of manufacturing industry and information technology, terminal devices are constantly developing in the direction of intelligence, efficiency and diversified scenarios. Terminal devices can assist people in completing various processing tasks and facilitate people's lives. The good use status of terminal devices is a necessary condition to ensure the smooth completion of processing tasks. If the terminal devices are not in good use status, it will be difficult to complete the processing tasks, and even cause damage to the interests of users.

[0003] Terminal devices are usually powered by batteries with limited capacity, and the health of the battery directly affects the health of the device. Taking smart payment devices including point of sale terminals as an example, as the number of charge and discharge times of the battery in the smart payment device increases, the performance of the battery gradually decays, resulting in the decay of the overall performance of the device. If the performance of the device decays to a certain extent and is still used, it is very likely that the device will fail during payment, resulting in a series of serious problems such as transaction interruption, sales loss, customer dissatisfaction, data loss, etc., which will bring inevitable economic losses to merchants.

[0004] In the prior art, some solutions focus on the impact of battery performance on the health of terminal devices, but usually only involve a qualitative evaluation of the current health status of the battery in the terminal device, and do not actually involve an advance prediction of the health status of the terminal device. Therefore, these solutions still cannot effectively solve the problem of equipment failure caused by terminal device performance degradation. Summary of the invention

[0005] The embodiments of the present application provide a device health prediction method, apparatus, terminal device and computer-readable storage medium, which can efficiently and accurately predict the battery health status of a target device within a preset period in the future, so that an early warning can be issued before the device may fail. Therefore, the health management of the device can be effectively achieved, the occurrence of failures can be prevented, and the economic losses caused by equipment failures can be reduced, thereby saving operation and maintenance costs and improving work efficiency.

[0006] A first aspect of an embodiment of the present application provides a device health prediction method, including:

[0007] Determine first health data of the target device, wherein the first health data includes battery health data of the target device in a current period;

[0008] Determine battery health prediction data of the target device within a preset future period according to the first health data.

[0009] A second aspect of an embodiment of the present application provides a device health prediction apparatus, including:

[0010] A determination module, configured to determine first health data of a target device, wherein the first health data includes battery health data of the target device in a current period;

[0011] The prediction module is used to determine the battery health data of the target device within a preset period in the future according to the first health data.

[0012] A third aspect of an embodiment of the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the steps of the above-mentioned device health prediction method when executing the computer program.

[0013] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned device health prediction method are implemented.

[0014] The first aspect of the embodiment of the present application provides a device health prediction method, which first determines the first health data including the battery health data of the target device in the current time period, and then determines the battery health prediction data of the target device in a future preset period based on the first health data. This scheme can efficiently and accurately predict the battery health status of the target device in a future preset period. Therefore, it is possible to further provide an early warning before the device may fail, thereby effectively achieving health management of the device, preventing the occurrence of equipment failures, reducing economic losses caused by equipment failures, and thus saving operation and maintenance costs and improving work efficiency.

[0015] It can be understood that the beneficial effects of the second to fourth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 is a schematic flow chart of a device health prediction method provided by an embodiment of the present application;

[0018] Figure 2a is a partial schematic diagram of a data list in an initial health data set provided by an embodiment of the present application;

[0019] Figure 2b Yes Figure 2a A schematic diagram of a data list obtained by preliminary conversion of a data list;

[0020] Figure 2c is a visualization diagram of battery health data of the first device after time series conversion;

[0021] Figure 2d It is a partial schematic diagram of a data list of a device health data set provided by an embodiment of the present application;

[0022] Figure 3a is a schematic diagram of battery health value changing with device age provided by an embodiment of the present application;

[0023] Figure 3b This is a comparative schematic diagram of battery health values ​​of devices with different usage frequencies as the age of the devices provided by an embodiment of the present application;

[0024] Figure 3c This is a comparative schematic diagram of battery health values ​​of devices of different network types as they age, provided by an embodiment of the present application;

[0025] Figure 3d This is a comparative diagram of battery health values ​​of devices in different usage areas as they age, provided by an embodiment of the present application;

[0026] Figure 4a It is a schematic diagram of the division of the training data set and the verification data set provided by an embodiment of the present application;

[0027] Figure 4b is a schematic diagram of a prediction result of a device health prediction model provided by an embodiment of the present application;

[0028] Figure 5 is a schematic diagram of a report setting interface provided by an embodiment of the present application;

[0029] Figure 6 It is a structural schematic diagram of a device health prediction apparatus provided by an embodiment of the present application;

[0030] Figure 7 It is a structural diagram of a terminal device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0031] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0032] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0033] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0034] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0035] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0036] As mentioned above, terminal devices are usually powered by batteries with limited capacity. As the number of battery charge and discharge times increases, the performance of the battery will gradually decay, causing the performance of the terminal device to be greatly reduced. Therefore, the health of the battery in the terminal device is closely related to the health of the device. For example, normally for a lithium battery in a terminal device, when the maximum capacity of the lithium battery decays to 80% of the rated capacity, it is considered that the battery has reached the failure threshold. Therefore, battery health prediction is the key to terminal device health prediction. However, the battery health prediction scheme in the prior art actually only involves a qualitative evaluation of the current health status of the battery in the terminal device, and does not really involve an advance prediction of the health status of the terminal device. Therefore, it cannot effectively solve the problem of equipment failure caused by the performance degradation of the terminal device.

[0037] The embodiment of the present application provides a device health prediction method, which can be used to predict the health status of various terminal devices, and can predict the health status of the terminal devices after a period of time, so as to easily further build a predictive maintenance strategy for the terminal devices, and effectively avoid the problem of equipment failure caused by the performance degradation of the terminal devices. The terminal devices to which the device health prediction method is applicable include but are not limited to: smart phones, POS machines, tablet computers, laptops, smart watches, smart home devices (such as smart door locks, smart monitoring devices, smart speakers), etc. For simplicity, the following text takes the terminal device as an example of a POS machine to expand the specific implementation method of the device health prediction method.

[0038] like Figure 1 As shown, the device health prediction method provided in the embodiment of the present application includes the following steps S120 and S140:

[0039] Step S120 , determining first health data of the target device, wherein the first health data includes battery health data of the target device in a current period.

[0040] In an embodiment of the present application, the target device is a terminal device for which health prediction is desired. The number of target devices may be one or more. In other words, the health status of a terminal device may be predicted and managed, and batch prediction management may be performed on multiple terminal devices or even a batch of terminal devices at the same time. In the case where the number of target devices is multiple, the first health data of each target device may be determined separately in this step.

[0041] In an embodiment of the present application, the first health data at least includes the battery health data of the target device in the current time period. In addition, in one example, the first health data may also include the battery health data of the target device in at least one time period before the current time period. In another example, the first health data may also include associated feature data related to the battery health of the target device. The associated feature data may specifically be feature data of at least one feature dimension of the target device. These feature dimensions can be arbitrarily set according to actual needs, including but not limited to intrinsic features of the device and extrinsic features of the device. Intrinsic features of the device may include features related to the performance of the device itself, such as battery property features such as battery rated capacity and battery materials. External features of the device may include device usage features such as the time the device has been in use, the frequency of use of the device by the user, the area of ​​use of the device, the network type of the device, the charging frequency, and the charging method.

[0042] The time period can be a predefined unit time period that is close to the prediction time. The unit time period can be 1 hour, 1 day, 1 week, etc. Taking the unit time period as an example, the current time period can be the day of the prediction, or it can be the previous day. For the case where the current time period is the prediction day, the first health data can at least include the battery health data of the target device on the prediction day. Alternatively, it can also include the battery health data of the target device one week or one month before the prediction day. Of course, it can also include feature data of one or more feature dimensions of the target device mentioned above.

[0043] Battery health data may include one or more physical quantities that can characterize the health or health of the battery. For example, battery health data may include battery health value or battery health index (BHI), current capacity, original rated capacity, cycle count (the total number of charge and discharge cycles experienced by the battery), battery voltage level in different states, battery internal resistance, charge and discharge rate, charge and discharge curve, etc.

[0044] In one embodiment, the battery health data of the target device includes a battery health value of the target device. The battery health value may specifically be a ratio of the relative capacity of the battery to its original rated capacity, reflecting the performance state and aging degree of the battery. The battery health value may be calculated using any suitable method. Step S120 determines the first health data of the target device, including step S121.

[0045] Step S121, determining the battery health value of the target device in the current period according to charging parameters of the target device in at least one charging cycle in the current period, wherein the charging parameters include charging current, charging time and battery rated capacity.

[0046] Taking the unit time period as one day as an example, in this step, the battery health value of the target device in the current time period can be determined according to the charging parameters of the target device in at least one charging cycle of the day.

[0047] For example, the target device can be a mobile phone with an Android system. In this step, the real-time charging current, charging time, battery rated capacity and other underlying parameters of the target device in multiple charging cycles on the same day can be obtained through the native Android API, and the value of the milliampere-hour charged into the mobile phone each time (the actual milliampere-hour change) can be calculated. Then, the actual milliampere-hour change during each charge is calculated and divided by the battery power increase percentage to obtain the current actual capacity (estimated capacity) of the mobile phone. After that, the ratio of the actual capacity to the battery rated capacity is calculated and multiplied by 100% to obtain the battery health value of the mobile phone at the current charge. For example, if the mobile phone is charged once on the same day, the battery health value of the day can be obtained based on the calculation result of the charge. For the case where the mobile phone is charged multiple times on the same day, there may be a certain degree of randomness in the single calculation result. Therefore, after multiple calculations, the median of the most recent several calculation values ​​can be used as the final battery health value to realize the monitoring and evaluation of the device battery life. For example, if the mobile phone is charged 5 times on the same day, the battery health value corresponding to each charge can be calculated using the above method, that is, 5 battery health values ​​are obtained, and then the median of the 5 battery health values ​​can be used as the battery health value of the day.

[0048] Similarly, for a method in which the first health data also includes the battery health data of the target device in other time periods before the current time period, the above calculation method can also be used to obtain the battery health values ​​of other time periods. For example, the battery health value of the mobile phone to be predicted accumulated every day in a month is calculated as the first health data of the mobile phone to be predicted. Alternatively, the battery health value of the mobile phone to be predicted accumulated every day in a month together with the feature data of at least one feature dimension of the mobile phone can be used as the first health data of the mobile phone.

[0049] Step S140, determining battery health prediction data of the target device within a preset future period according to the first health data.

[0050] The battery health prediction data of the target device in the future preset period determined in this step includes the battery health prediction data of the target device at least one time point in the future preset period. The future preset period can be arbitrarily set according to actual needs, as long as it is a time that has not yet arrived after the current period. For example, the future preset period can be the next week, the next month, the next 2 months, the next 3 months, the next half year or the next year, etc. Taking the future preset period as the next 2 months as an example, the battery health prediction data determined in this step can be the battery health prediction data of any suitable time period within the next 2 months. Taking the unit time period as one day and the battery health data including the battery health value as an example, the battery health prediction data may include the battery health value of the target device on any one or more days in the next 2 months. For example, in this step, the battery health value of the target device on each day in the next 2 months can be obtained.

[0051] In this step, a variety of suitable methods can be used to determine the battery health prediction data of the target device in the future preset period according to the first health data. Exemplarily and not restrictively, a health prediction model can be pre-built. For example, the target device is a mobile phone with an Android system installed. A large amount of battery health data of this type of mobile phone can be collected in advance, including the current battery health value of each mobile phone, the battery health value in the past period of time, and associated feature data (daily usage time, application usage, game and video playback, etc., charging frequency, charging time, whether fast charging is used, etc.). Then, the downward trend of the battery health value of each mobile phone can be analyzed based on historical data. And the impact of these associated feature data on the battery health value can be analyzed. Afterwards, a health prediction model that can predict the battery health data of the target device in the future preset period can be constructed or screened and trained based on the data analysis results. In one example, the health prediction model can be a traditional statistical model. For example, the health prediction model can be an auto-regression moving average model (Self-Regression Moving Average Model, referred to as SMA) or an auto-regression integrated moving average model (Auto Regressive Integrated Moving Average Model, referred to as ARIMA) that can model a one-dimensional time series. The SMA model uses the average of the observed values ​​of the past n time points as the predicted value of the next point, which can effectively reflect the change of trends and has a small amount of calculation. The ARIMA model combines three parts: autoregression (AR), differencing (I, i.e. Integrated) and moving average (MA). It is particularly suitable for time series data with non-stationarity. It converts non-stationary time series into stationary time series for prediction through differentiation. In another example, the health prediction model can also be a machine learning model, such as a deep learning model. In this step, the first health data of the target device can be input into the constructed or trained health prediction model to obtain the battery health prediction data of the target device predicted by the model in the future preset period.

[0052] It can be understood that since what is obtained in this step is the battery health prediction data of the target device in the future preset period. Therefore, the accurate prediction of the health of the target device is truly achieved. In this way, an effective warning can be given before the battery of the target device fails, giving the user enough time to maintain or update the device. For example, it can help users or equipment managers to formulate battery maintenance and replacement plans to avoid unexpected downtime; the battery replacement time can also be predicted based on the prediction results, so the budget can be reasonably arranged to reduce emergency procurement costs; the use and charging strategies of the equipment can also be optimized based on the prediction results to extend the battery life; effective measures can be taken in advance to prevent data loss or equipment damage; planning based on the prediction results can ensure that mission-critical equipment can operate reliably when needed and improve system reliability; and measures can be taken to save or transfer key data before the battery is exhausted to improve the security of data stored in the target device; a battery health management plan can also be formulated based on the prediction results, thereby extending the service life of the entire device.

[0053] The device health prediction method provided in the embodiment of the present application first determines the first health data including the battery health data of the target device in the current time period, and then determines the battery health prediction data of the target device in a future preset period based on the first health data. This scheme can efficiently and accurately predict the battery health status of the target device in a future preset period. Therefore, it is possible to further provide an early warning before the device may fail, thereby effectively achieving health management of the device, preventing the occurrence of equipment failures, reducing economic losses caused by equipment failures, and thus saving operation and maintenance costs and improving work efficiency.

[0054] In one implementation, step S140 determines battery health prediction data of the target device within a future preset period based on the first health data, including step S141.

[0055] Step S141, inputting the first health data into a health prediction model to obtain battery health prediction data of the target device in a future preset period. The health prediction model is a deep learning model for time series prediction.

[0056] The health prediction model can be a variety of suitable deep learning models that can realize time series prediction. In one example, the health prediction model can be a long short-term memory network (LSTM model for short). The model solves the gradient vanishing problem of the recurrent neural network to a certain extent by introducing input gate, forget gate, and output gate structures, and can realize the fitting and prediction of the time series data of each target device separately. In another example, the health prediction model can be a DeepAR model. This model belongs to a probabilistic prediction model based on the autoregressive recurrent neural network architecture, which can realize the global learning of time series relationship and distribution characteristics from the monitoring data. The DeepAR model can easily incorporate other features of the prediction object into the calculation at the same time. For devices with different feature data (such as devices in different regions, devices of different models, and devices of different factory batches), the device health prediction results output by the DeepAR model can be different. In other words, the DeepAR model can use feature data of different feature dimensions to predict the battery health of the device. The prediction target of the DeepAR model is the probability distribution of the sequence at each time step. In addition, the DeepAR model can also use cold start prediction, that is, it can perform new time series predictions through similar time series in the absence of almost any historical data. For example, after the DeepAR model is trained, you can only input the battery health data of the target device in the current period (without entering other historical data), and the DeepAR model can predict the battery health data in the future. In addition, the DeepAR model can establish a unified prediction model for multiple time series, which is suitable for massive data scenarios. For example, the battery health prediction of a large number of devices can still have high computational efficiency and more accurate prediction results.

[0057] In the above scheme, the use of deep learning models designed for time series prediction can improve the accuracy of equipment health prediction. Deep learning models can capture complex nonlinear relationships and long-term dependencies, providing high-precision prediction results. In addition, deep learning models can automatically learn and extract features from raw data, reducing the need for manual feature engineering. And properly trained deep learning models have good generalization capabilities and can adapt to different data distributions and prediction tasks. In addition, deep learning models can handle large-scale data sets, and the performance of the model usually improves as the amount of data increases. And deep learning models can be designed for real-time prediction, respond quickly to newly collected data, and update predictions in real time. In summary, the above scheme can improve the accuracy and efficiency of health predictions for target devices.

[0058] In one embodiment, the battery health data of the target device includes a battery health value of the target device. Step S141 inputs the first health data into a health prediction model to obtain battery health prediction data of the target device in a future preset period, including: step S141a and step S141b.

[0059] Step S141a, input the first health data into the health prediction model, and output the probability distribution of the predicted value of the battery health value of the target device at each time point in the future preset period. Step S141b, determine the estimated value of the battery health value of the target device at the time point according to the probability distribution of the predicted value of the battery health value at each time point.

[0060] Take the case where the health prediction model is a DeepAR model, the target device is a mobile phone, and the current time period is the predicted day. In step S141a, at least the battery health value of the mobile phone on the predicted day can be input into the DeepAR model, and the model can output the probability distribution of the predicted value of the battery health value of the mobile phone every day (every week or every month) in the next two months. In some preferred examples, for the case of prediction of different target devices, at least one characteristic data such as the used time of the mobile phone, the rated capacity of the battery, the frequency of use of the user's device, the use area, the network type, the charging frequency, the charging method, etc. and the battery health value of the mobile phone on the predicted day can also be input into the DeepAR model together, and the probability distribution of the predicted value of the battery health value of the mobile phone every day in the next two months can be accurately obtained. In step S141b, the estimated value of the battery health value of the day can be determined according to the probability distribution (such as the probability density function) of the predicted value of the battery health value of the mobile phone output by the model. For example, the predicted value corresponding to the maximum probability density value can be determined as the estimated value of the battery health value of the day. Alternatively, the predicted value corresponding to the median can also be determined as the estimated value of the battery health value of the day.

[0061] In this solution, more comprehensive prediction information can be given by providing a probability distribution of the battery health value at each time point instead of a single estimate. The probability distribution can quantify the uncertainty of the prediction and help users understand the reliability of the prediction. The probability distribution can be used for risk assessment, for example, by identifying high-risk battery health value decline intervals to take measures in advance; equipment maintenance, replacement and use strategies can provide data-driven decision support. In addition, the model can be continuously updated based on new data to adapt to changes in battery health status.

[0062] It can be understood that in specific scenarios (such as predictive maintenance of a large number of target devices), probability distribution is more meaningful than single-point prediction. For example, in equipment health prediction, if the probability of failure of the target device in the future period is known (the probability that the predicted value of the equipment health value is less than or equal to the predicted value threshold corresponding to the equipment failure), the operation optimization method can be used to make the best maintenance decision. For example, the number of target devices expected to be predicted and managed is 1,000, the predicted value threshold can be 80%, and a suitable confidence threshold (such as 90%) can be pre-set. The future preset period is, for example, the next 2 months. Exemplarily, according to the model output results, the number of target devices with a confidence level higher than 90% whose battery health value is lower than 80% in the next 2 months can be found, and the equipment inventory quantity or operation and maintenance work for the next period of time can be planned according to the statistical quantity. In other words, according to the probability distribution, the target devices that are more likely to have equipment failures (the width of the confidence interval is smaller and more concentrated) can be processed first, which can reduce maintenance costs and improve maintenance efficiency.

[0063] In one embodiment, before inputting the first health data into the health prediction model in step S141, the method further includes steps S111 and S112.

[0064] Step S111, determining a device health data set, wherein the device health data set at least includes a time series of battery health data of each of a plurality of first devices within a first historical period.

[0065] The first historical time can be set arbitrarily according to actual needs. For example, it can be a continuous time period before the current time period, such as 1 year, 2 years, etc. The device health data set includes at least the battery health data of each of the multiple first devices in the first historical time period and the timestamp corresponding to each battery health data. The multiple time periods can be multiple consecutive time periods in the first historical time. Taking the number of first devices as 900, the first historical time is 165 days before the predicted time, and each time period is 1 day as an example, the device health data set determined in this step can include at least 900 first devices (for example, 900 mobile phones installed with Android system) The battery health value of each day in the 165 days and the timestamp corresponding to each battery health value (for example, absolute date or relative date). For example, the 900 mobile phones can be used as test devices in advance, and the underlying parameters such as real-time charging current, charging time, and battery rated capacity of each mobile phone in multiple charging cycles per day are obtained through the native Android API. The method for calculating the battery health value in the aforementioned example is used to calculate the battery health value of each mobile phone on a daily basis. Multiple first devices can be different devices with the same intrinsic and external characteristics of the device. Alternatively, the multiple first devices may also be devices with different intrinsic device characteristics and / or extrinsic device characteristics, so that the battery health data of different first devices with different characteristics can be distinguished, and the scope of application is wider. In this example, the device health data set may also include the battery attributes, network type, usage time, usage frequency, network type, and usage region of each first device.

[0066] In one implementation, step S111 determines the device health data set, which may specifically include step S111a and step S111b.

[0067] Step S111a, obtaining an initial health data set, wherein the initial health data set at least includes battery health data of multiple second devices at multiple time points within a second historical time, the multiple second devices include multiple first devices, and the second historical time includes the first historical time.

[0068] The number of test devices included in the initial health data set may be greater than or equal to the number of devices in the device health data set. For example, the initially collected health data set contains battery health data sets of 3017 test devices (second devices), while the device health data set that finally participates in the training only includes health data sets of 886 test devices (first devices). Similarly, the time span corresponding to the battery health data in the initial health data set may be greater than or equal to the time span corresponding to the battery health data in the device health data set participating in the training. For example, the second historical time is 200 consecutive days before the prediction time, and the first historical time is 165 consecutive days before the prediction time; or, the second historical time and the first historical time may be the same. For example, both are 165 consecutive days from July 28, 2022 to January 6, 2023.

[0069] Taking the second historical time as 165 consecutive days from July 28, 2022 to January 6, 2023 as an example, the battery health data of each of the 3017 test devices can be counted in the second historical time, and these battery health data can be uploaded to the battery activity data set in the device management platform. This data set can be used to record the daily battery consumption and battery health of the device. For example, in the pre-operation and maintenance work of the smart payment device, this data set can be used to let users understand the activity of the device, monitor and evaluate the frequency of use of the device, the time when the user mainly uses the device, the type of network mainly used, and other related feature data, and analyze the battery health change trend of the device. The initial health data set obtained after the data collection is completed can include at least the battery health data of the 3017 test devices in 165 days and the timestamps corresponding to each battery health data. Alternatively, the initial health data set can also include feature data of each feature dimension of each test device. These feature data can be feature data recorded every day, or feature data that can reflect the overall situation of the device in 165 days after processing the feature data recorded for 165 days.

[0070] Step S111b, performing a preprocessing operation on the initial health data set to obtain a device health data set consisting of multiple health data sequences. Each health data sequence corresponds to a first device, and each health data sequence includes a time series of battery health data of the corresponding first device within a first historical time. The preprocessing operation includes at least one of the following operations: abnormal data elimination operation, missing data filling operation, data format conversion operation, and data standardization operation.

[0071] like Figure 2aAs shown, after completing the collection of battery health data of 3017 test devices for 165 days, 8 columns and 120860 rows of data can be obtained. Each row includes not only the battery health value and time (timestamp corresponding to the data) of a second device on a certain day, but also the device serial number, frequency of use, device age (i.e., the time the device has been used, specifically in months), country (i.e., the area where the device is used), model, and network type. Characteristic data. These data can be regarded as one-dimensional data. First, these one-dimensional data can be converted into two-dimensional data for easy preprocessing. The data of each device can be placed in one column, and the battery health value of each device can be arranged in chronological order to facilitate subsequent processing.

[0072] Then, abnormal data removal operations and missing data filling operations can be performed. Specifically, abnormal battery health values ​​can be removed according to the abnormal value removal rules. For example, for some devices that are not actively used, they may not be charged on some of the 165 days, so the battery health value cannot be obtained. In other words, the battery health values ​​of these devices at certain time points are null values. In an embodiment of the present application, all battery health data of devices with more than half of them being null values ​​can be removed to ensure the objectivity of the training process. Alternatively, for some unreasonable battery health values, for example, in the case where the battery health value of a certain device within 165 days has a jump value that is significantly different from the battery health value of the previous day and the next day, such jump values ​​can be deleted. This can avoid the adverse effects of such obviously abnormal battery health values ​​on training.

[0073] After the abnormal data removal operation is completed, the missing data filling operation can also be performed. For example, the empty values ​​of the battery health values ​​of each remaining device after the data removal can be filled. For example, if the battery health value of a certain device on a certain day is empty, any one of the battery health values ​​of the day before and the day after that day or the average value can be used as the battery health value of that day. Finally, if Figure 2b As shown, there are 886 columns of data left, that is, there are 886 valid health data of the test devices (first devices) left, and each column of data also contains associated characteristic data of the test device.

[0074] In order to facilitate training, the processed health data after elimination and filling can be converted into data format. Figure 2b The 886 columns of data in the .time series format can also be visualized using third-party plotting libraries such as the matplotlib plug-in to obtain Figure 2c , such as Figure 2c As shown, for each first device, the battery health value of the device also fluctuates over time.

[0075] In order to more accurately predict the battery health of the device, the health data of each device can also be standardized. For example, if the device health data set includes daily battery health values ​​and device-related feature data, a variety of appropriate standardization processing methods can be used to standardize the battery health values ​​and device-related feature data as feature parameters for model training. Figure 2d As shown in the figure, the device health data set finally obtained includes the standardized battery health value of each first device on a daily basis (the value range is between 0 and 1), as well as the standardized values ​​of the feature data of the five feature dimensions of each device, namely, country (usage area), model, network type, usage frequency, and device age (month). The standardized values ​​of these feature data constitute the feature standardized array of each device. Figure 2d As shown, each training data in the device health data set (for a health data sequence of each first device) includes the start period of the first historical time (2022-07-25), the target value (i.e., the battery health value), and a feature normalization array.

[0076] The above solution can ensure the quality and consistency of model training data, enhance the generalization ability of the model, improve model performance, and reduce computing resource consumption to ensure the reliability and efficiency of model training.

[0077] Step S112, training the health prediction model using the device health data set until a preset end condition is met.

[0078] In this step, a variety of suitable training methods can be used to train the health prediction model. During the training process, adjust the model parameters and optimize the hyperparameters. Consider using different evaluation indicators, such as accuracy, recall, F1 score, ROC-AUC, etc. The preset end conditions can include model performance reaching a certain threshold, training time limit, training round number limit, etc. By continuously iterating the training until the preset end conditions are met, the model can be continuously optimized until the best performance is achieved.

[0079] It can be understood that time series data can capture the trend of device battery health status changing over time. Using historical data to train the model can help the model learn the law of device health status changing over time, thereby improving the reliability of the prediction. Setting preset end conditions can effectively control the model training process, avoid resource waste, and ensure that the model stops training after reaching the expected performance. Model evaluation and adjustment during training can help improve the generalization ability of the model and ensure that the model performs well on new and unseen data. In addition, this solution can be applied to different types of equipment and different health indicators, has good scalability, and can be used in a variety of scenarios. Therefore, it can provide scientific decision support for equipment operation and maintenance, reduce operational risks, and improve equipment utilization efficiency and economic benefits.

[0080] In one embodiment, step S120 determines the first health data of the target device, including: determining the battery health data of the target device at least in the current time period and characteristic data of at least one characteristic dimension of the target device as the first health data; wherein the characteristic dimension includes one or more of the following: battery attributes, network type, usage time, usage frequency, network type, and usage area.

[0081] For example, before predicting the device health of a POS terminal, the battery health value of the POS terminal on the current day and in the previous period can be calculated, and the characteristic data of the battery rated capacity of the POS terminal, the main network type (data network 4G, 5G or WiFi, etc.) of the user using the POS terminal, the usage time (e.g., the cumulative usage months, the POS terminal has been used for a total of 12 months), and the usage frequency (e.g., the number of times charged per week) can be obtained. These characteristic data and the battery health value can be standardized using a method similar to step S111b to obtain the first health data.

[0082] It can be understood that in this scheme, not only the current battery health data of the device is considered, but also the characteristic data of some characteristic dimensions that affect the battery health during the use of the device. Comprehensive consideration of data from multiple dimensions can provide a more comprehensive perspective and help to more accurately predict the health of the battery. Characteristic data such as device age, frequency of use, etc. are closely related to the battery health status and can significantly improve the accuracy of the prediction model. Different usage habits and environmental conditions have different effects on battery health. Considering these factors can achieve more personalized predictions. In addition, the analysis of multi-dimensional features can reveal the interactions and influences between different factors, providing more information for a deeper understanding of battery health. In general, comprehensive consideration of the current battery health data of the device and other related features can make the prediction results more accurate and provide strong support for device maintenance and user decision-making.

[0083] By analyzing the battery health data curves of devices with different feature data, it is found that the battery health data curves of different devices perform differently. For example, devices with high usage frequency decay faster than devices with low usage frequency; devices using mobile networks decay faster than devices using wifi. In order to determine the feature dimensions that have a greater impact on battery health, a data-driven approach can be used to screen out at least one feature dimension (for example, called a key feature dimension) from a variety of possible feature dimensions, and then the feature data of the key feature dimensions of the test device are also used as training data for the model. The device battery can be regarded as a black box, and a complex mapping relationship between the feature data of the key feature dimensions of these devices and the battery health can be established through statistical and machine learning methods. The data-driven approach does not require an in-depth understanding of the battery working mechanism and aging mechanism, and does not require the establishment of a specific physical model.

[0084] In one embodiment, the device health data set also includes feature data of at least one feature dimension of each first device. The method further includes steps S101 to S103.

[0085] Step S101, obtain the battery health data of multiple third devices at multiple time points within a third historical time and the feature data of multiple feature dimensions of multiple third devices, wherein the feature data of each feature dimension of the multiple third devices are not exactly the same. The third device may be the above-mentioned second device, or it may be another device different from the second device. The third historical time may be the second historical time, or it may be different from the second historical time. The screening of feature dimensions using other pre-collected data may be performed before obtaining the initial health data set, or after obtaining the initial health data set, the data in the initial health data set may be analyzed and screened. The feature dimensions involved in the statistics may include at least four types: network type, area of ​​use, battery rated capacity, and time used, and may also include other feature dimensions. For example, a total of 20 feature dimensions are selected for correlation statistical analysis.

[0086] Step S102, analyzing the feature data of each feature dimension of the plurality of third devices and the battery health data of the plurality of third devices at a plurality of time points within a third historical time, to obtain a correlation between each feature dimension and the battery health data.

[0087] For example, the battery health data of 3,000 test devices within one year are obtained in advance. The feature data of the 20 feature dimensions of these 3,000 test devices are not exactly the same. Then, the feature data of each feature dimension can be analyzed, and the correlation coefficient between the feature data of each feature dimension and the battery health value can be calculated separately, such as the Pearson correlation coefficient, Spearman rank correlation or Kendall rank correlation. And hypothesis testing can be performed to determine whether the correlation is statistically significant. According to the size and significance of the correlation coefficient, the correlation between the feature data of each feature dimension and the battery health value is explained. You can also use visualization tools such as scatter plots and heat maps to show the relationship between variables.

[0088] For the usage time (e.g., called device age), the trend of battery health value changing with device age can be analyzed for one device or multiple devices. Figure 3a As shown in the figure, the battery health values ​​of multiple devices of the same model decrease as the age of the devices increases. These devices experienced battery degradation at the 20th month, and by the 31st month, the median battery health had dropped to 91%. The correlation coefficient between the device health value and the device age can also be obtained through data correlation statistical analysis.

[0089] For other feature dimensions, the control variable method can be used to analyze the correlation between each feature dimension and battery health data. Figure 3b As shown in the figure, the characteristic data of all test devices involved in the analysis are the same except for the frequency of device use. There are significant differences in the battery health values ​​of devices with different usage frequencies: devices charged 1-2 times a week will not experience battery degradation until the 22nd month, and the battery health is still 94% in the 31st month. However, devices charged more than 3 times a week will experience battery degradation as early as the 10th month, and the battery health is only 86% in the 31st month. Figure 3c As shown in the figure, the characteristic data of all test devices involved in the analysis are the same except for the network type. There are also significant differences in the battery health values ​​of devices with different network types: devices using 4G will experience battery degradation in the 11th month, and the battery health is 87% in the 31st month. Devices using WiFi will experience battery degradation in the 19th month, and the battery health is 86% in the 31st month. Figure 3d As shown in the figure, the characteristic data of all the test devices involved in the analysis are the same except for the region of use. There are also significant differences in the battery health values ​​of devices in different regions of use: the battery of the device from China showed battery degradation in the 17th month, and the battery health was still 86% in the 31st month. However, due to the very cold climate, the median battery health of the device from Russia was only 80% in the 10th month, and only 71% in the 31st month.

[0090] Step S103: Screen out at least one characteristic dimension from the multiple characteristic dimensions according to the correlation relationship, wherein the correlation between each of the at least one characteristic dimension and the battery health data is greater than or equal to a correlation threshold.

[0091] The correlation threshold can be set arbitrarily according to actual needs. For example, the correlation between the feature data of each feature dimension and the battery health value can be represented by a correlation coefficient. In step S103, the correlation threshold can be a coefficient threshold of 0.7. After obtaining the correlation coefficient between the feature data of each feature dimension and the battery health value in step S102, in this step, the feature dimension with a correlation coefficient greater than or equal to 0.7 can be determined as the above-mentioned key feature dimension.

[0092] It can be understood that by selecting features that are highly correlated with battery health data, the accuracy of the prediction model can be improved. Moreover, screening out the most relevant features can reduce unnecessary features and simplify the model structure. And features with strong correlation are more likely to maintain their predictive ability on new data, thereby improving the generalization of the model. By eliminating features that are not strongly correlated with battery health data, the risk of model overfitting can be reduced, and the computational cost of model training and prediction can also be reduced. The screened features are easier to interpret and help understand the decision-making process of the model. And the screened features may be more adaptable to different equipment and environmental conditions. And correlation analysis can also help to discover the key factors affecting battery health and provide direction for further research. By identifying features that are highly correlated with battery health, risks in equipment operation can be better managed and predicted. In general, this scheme can further improve the accuracy of equipment health prediction and provide a scientific basis for equipment maintenance and health management.

[0093] In one implementation, step S112 uses the device health data set to train the health prediction model, including the following steps:

[0094] Step S112a, divide the device health data set into a training data set and a verification data set, wherein the training data set includes battery health data of multiple first devices in a first period in a first historical time, and the verification data set includes battery health data of multiple first devices in a second period in the first historical time, and the first period is earlier than the second period.

[0095] like Figure 4a As shown in FIG. 1 , the battery health data of 886 valid test devices after preprocessing for 165 days can be divided into a training data set and a validation data set. The first period is, for example, the first 105 days, and the second period can be the last 60 days. That is, the data of the first 105 days can be used as the training data set ( Figure 4a The data represented by the green line in the middle), and the data of the next 60 days are used as the validation data set ( Figure 4a This makes it easier to verify the data based on these 60 days and use the data from these 60 days to predict the battery health in the future.

[0096] Step S112b, training the health prediction model using the training data set, the validation data set, and the first loss function until a preset end condition is met. The value of the first loss function is positively correlated with the difference between the predicted value of the battery health data of each first device at each time point in the second period output by the health prediction model and the actual value of the battery health data of the device at each time point in the validation data set.

[0097] For example, the health prediction model is a DeepAR model. The preprocessed and segmented training data set and validation data set (for example, a time series of battery health data of 886 valid test devices and feature data of key feature dimensions) can be input into the DeepAR model for training, and finally a battery health prediction model that can be applied to devices in various scenarios (with different features) is obtained.

[0098] In a specific training example, the batch size of the training data can be set to 20, the learning rate starts at 0.0005, and 200 epochs are trained each time. The data of the first 60 days of the battery health data time series of each device can be used as the training context. The training will stop when no progress is made in 20 epochs, and the model with the lowest loss will be returned as the final model. In order to evaluate the accuracy of the model, the root mean square error (RMSE) can be used as an evaluation indicator, that is, as the first loss function for training the health prediction model. The root mean square error (RMSE) formula is:

[0099]

[0100] In the formula, and z i Respectively represent the predicted value of the battery health data of each first device at each time point in the second period (for example, the last 60 days) output by the health prediction model and the true value of the battery health data of the device at each time point in the verification data set, and n represents the number of samples, that is, the number of first devices. It can be understood that RMSE is suitable for evaluating situations where the estimation error is not obvious. By minimizing the above-mentioned first loss function, the model reaches the convergence condition, and then the training can be terminated. For example, after training, the final RMSE result obtained by the model is 0.06, which shows that the selected experimental data can more accurately reflect the attenuation degree of the device battery, and can adapt to the differences caused by the inconsistency of device characteristics.

[0101] For example, after the training is completed, the battery health of the target device in the next 60 days can be predicted based on the battery health data of the target device (for example, each first device is used as the target device) and the existing characteristic parameters of the device, including country, model, network type, usage frequency, and device age (month). Figure 4bAs shown, the blue line in the figure represents the true value in the validation data set, the red line represents the predicted value (the predicted value corresponding to the median of the probability distribution), and the light red area represents the 90% confidence interval. There are two parts of the prediction results in the figure. The first part is from November 2022 to January 2023. It can be seen that the prediction results are compared with the validation data. The 90% confidence interval of the predicted value covers the true value at most times, which solves the problem of the estimation model's lack of extreme value fitting ability. The second part is the predicted data of the battery health values ​​of each first device from February 2023 to March 2023, which successfully predicts the battery health decay trend in the next two months.

[0102] This approach can effectively improve the performance and reliability of health prediction models by rationally utilizing historical data, setting up appropriate training and validation processes, and adopting a loss function that is positively correlated with the error.

[0103] In one embodiment, the battery health prediction data includes the predicted values ​​of the battery health values ​​of the target device at various time points in a preset future period. The method further includes the following steps:

[0104] Step S151, determining whether a predicted value of a battery health value of a target device at at least one time point in a preset future period is less than a health value threshold;

[0105] Step S152: If yes, then send an early warning message and / or corresponding equipment maintenance suggestion message to the target equipment before a preset period in the future.

[0106] The health value threshold can be set arbitrarily according to actual needs. For example, based on historical experience, devices with battery health below 80% will greatly affect daily use, and users are generally advised to replace batteries or replace devices, so the health value threshold can be set to 80%. Of course, multiple health value thresholds corresponding to different risk levels can also be set. The future preset period is, for example, the next two months. In a specific example, the battery health value of each device in the multiple target devices managed in the next month (for example, the 30th day from the prediction time) and the battery health value in the next two months can be predicted respectively. And it can be determined whether the battery health value is less than 80%. For each device, when it is identified that the predicted value of the battery health value of the device in the next month or the next two months is less than 80%, an early warning prompt can be issued one month in advance. For example, a pop-up window prompt can be sent to the APP installed on the device or through an email or an in-station letter to remind the user to pay attention to maintenance or replacement of the device. Equipment maintenance suggestions can also be made, such as suggesting to close unnecessary applications, pay attention to turning off the device during idle time, and so on.

[0107] In this solution, by predicting the battery health value and setting thresholds, actions can be taken before the battery performance drops to an unacceptable level to achieve preventive maintenance. Timely warning and maintenance can reduce the degradation rate of the battery, extend the battery life, and reduce the frequency and cost of replacement. In addition, by predicting battery health and taking corresponding measures, it is possible to ensure that the equipment remains operational in critical tasks and improve the reliability and stability of the equipment. The early warning system can alert users before the battery performance drops to the point where it affects the operation of the equipment, reducing unexpected downtime and production interruptions caused by battery problems. In addition, by predicting battery health, companies can better plan maintenance resources, such as manpower, spare parts, etc., to achieve optimal allocation of resources. It can also effectively reduce equipment failures caused by battery problems, which can improve user satisfaction and loyalty.

[0108] In one embodiment, the method further comprises the following steps:

[0109] Step S153, providing a user interface, and displaying health prediction information of the target device in a preset future period on the user interface. The health prediction information includes battery health prediction data and / or a health risk level corresponding to the battery health prediction data. Step S154, updating the health prediction information at a preset frequency.

[0110] In one example, the battery health value list of each target device and / or the health risk level of each target device in the next month and / or the next two months can be regularly pushed to the device management interface of the operation and maintenance personnel, such as a low risk if the health value is higher than 85%, a medium risk if the health value is between 80% and 85%, and a high risk if the health value is lower than 80%. This makes it easy for users to count the number of devices with health risks in the future and to update the device inventory in a timely manner. Alternatively, the number of medium-risk devices and high-risk devices can also be displayed in real time on the interface. Of course, other information that affects the availability of the target device, such as the age of the device and the remaining memory of the device, as well as the risk level corresponding to this information, can also be displayed. For example, the age of the device is greater than 24 months, which is a high risk; the age of the device is 1-2 years, which is a medium risk; the age of the device is less than 1 year, which is a low risk; the remaining memory of the device is less than 3G, which is a high risk; the remaining memory of the device is between 3G-8G, which is a medium risk; the remaining memory of the device is greater than 8G, which is a low risk. In another example, the above user interface may also correspond to the display interface of the APP installed on the target device.

[0111] The preset frequency can be set arbitrarily according to actual needs. For example, it can be updated once a day, once a week, or once a month. In this way, the battery health prediction information of the target device can be informed to the user in a timely manner, so that the user can pay attention to or maintain it in advance.

[0112] By combining the user interface with regularly updated forecast information, users can obtain accurate device health information in a timely manner, making it easier to better manage and maintain the health of the device battery.

[0113] like Figure 5 As shown, a report setting interface can also be provided, and the push setting information for health information of various devices set by the user can be received through the report setting interface. The push setting information may include: the notification method of the message, the calculation frequency of the health data (corresponding to the update frequency of the health prediction information in step S154), and may also include attention screening conditions for different health information. For example, if the device age is greater than 24 months, it is a high concern level, if the device memory remaining is less than 3GB, it is a high concern level, if the battery health value in the next month is less than 85%, it is a medium concern level, and if the battery health value in the next two months is less than 85%, it is a medium concern level, etc.

[0114] The embodiment of the present application also provides a device health prediction apparatus for executing the method steps in the above method embodiment. The apparatus may be a virtual appliance in a terminal device, which is run by a processor of the terminal device, or the terminal device itself.

[0115] like Figure 6 As shown, the device health prediction device 600 provided in the embodiment of the present application includes:

[0116] A determination module 610 is configured to determine first health data of a target device, wherein the first health data includes battery health data of the target device in a current period;

[0117] The prediction module 620 is used to determine the battery health data of the target device within a preset period in the future according to the first health data.

[0118] In one implementation, the prediction module 620 is specifically configured to:

[0119] The first health data is input into a health prediction model to obtain battery health prediction data of the target device in a future preset period; wherein the health prediction model is a deep learning model for time series prediction.

[0120] In one implementation, the battery health data of the target device includes a battery health value of the target device, and the prediction module 620 includes:

[0121] An input-output unit, configured to input the first health data into a health prediction model, and output a probability distribution of predicted values ​​of the battery health value of the target device at various time points in a future preset period;

[0122] The first determination unit is used to determine an estimated value of the battery health value of the target device at each time point according to the probability distribution of the predicted value of the battery health value at the time point.

[0123] In one embodiment, the device health prediction apparatus 600 further includes:

[0124] A first data collection module, configured to determine a device health data set, wherein the device health data set comprises at least a time series of battery health data of each of a plurality of first devices within a first historical time;

[0125] The training module is used to train the health prediction model using the equipment health data set until a preset end condition is met.

[0126] In one embodiment, the training module includes:

[0127] a dividing unit, configured to divide a device health data set into a training data set and a verification data set, wherein the training data set includes battery health data of a plurality of first devices in a first period of a first historical time, and the verification data set includes battery health data of a plurality of first devices in a second period of the first historical time, the first period being earlier than the second period;

[0128] A training unit is used to train a health prediction model using a training data set, a validation data set, and a first loss function until a preset end condition is met; wherein the value of the first loss function is positively correlated with the difference between the predicted value of the battery health data of each first device at each time point in the second period output by the health prediction model and the actual value of the battery health data of the device at each time point in the validation data set.

[0129] In one embodiment, the first data collection module includes:

[0130] a data collection unit, configured to determine an initial health data set, wherein the initial health data set at least includes battery health data of a plurality of second devices at a plurality of time points within a second historical time, the plurality of second devices including a plurality of first devices, and the second historical time including the first historical time;

[0131] A preprocessing unit is used to perform a preprocessing operation on an initial health data set to obtain a device health data set consisting of multiple health data sequences; wherein each health data sequence corresponds to a first device, and each health data sequence includes a time series of battery health data of the corresponding first device within a first historical time; the preprocessing operation includes at least one of the following operations: an abnormal data elimination operation, a missing data filling operation, a data format conversion operation, and a data standardization operation.

[0132] In one implementation, the determination module 610 includes:

[0133] The second determination unit is used to determine the battery health data of the target device at least in the current time period and the characteristic data of at least one characteristic dimension of the target device as the first health data; wherein the characteristic dimension includes one or more of the following: battery attributes, network type, usage time, usage frequency, network type and usage area.

[0134] In one implementation, the device health data set also includes feature data of at least one feature dimension of each first device, and the device health prediction device 600 also includes:

[0135] a second data collection module, used to determine battery health data of a plurality of third devices at a plurality of time points within a third historical time and feature data of a plurality of feature dimensions of the plurality of third devices, wherein the feature data of each feature dimension of the plurality of third devices are not completely the same;

[0136] an analysis module, configured to analyze feature data of each feature dimension of the plurality of third devices and battery health data of the plurality of third devices at a plurality of time points within a third historical time, to obtain a correlation between each feature dimension and the battery health data;

[0137] The screening module is used to screen out at least one feature dimension from multiple feature dimensions according to the correlation relationship, wherein the correlation between each of the at least one feature dimension and the battery health data is greater than or equal to a correlation threshold.

[0138] In one embodiment, the device health prediction apparatus 600 further includes:

[0139] Providing a module for providing a user interface and displaying health prediction information of the target device in a preset future period on the user interface;

[0140] An update module is used to update health prediction information at a preset frequency; wherein the health prediction information includes battery health prediction data and / or a health risk level corresponding to the battery health prediction data.

[0141] In one embodiment, the battery health prediction data includes predicted values ​​of battery health values ​​of the target device at various time points in a future preset period, and the device health prediction apparatus 600 further includes:

[0142] A determination module, used to determine whether a predicted value of a battery health value of a target device at at least one time point in a preset future period is less than a health value threshold;

[0143] The sending module is used to send early warning information and / or corresponding equipment maintenance suggestion information to the target equipment before a preset period in the future.

[0144] like Figure 7As shown, the embodiment of the present application further provides a terminal device 700, including: at least one processor 710 ( Figure 7 Only one processor is shown in the figure), a memory 720, and a computer program 730 stored in the memory 720 and executable on at least one processor 710. When the processor 710 executes the computer program 730, the steps of the above-mentioned device health prediction method are implemented.

[0145] The terminal device may include but is not limited to a processor and a memory. Figure 5 It is only an example of a terminal device and does not constitute a limitation on the terminal device. It may include more or fewer components than shown in the figure, or combine certain components, or different components. The processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0146] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / modules are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0147] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example for illustration. In actual applications, the above-mentioned function allocation can be completed by different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The functional modules in the embodiment can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software functional modules. In addition, the specific names of the functional modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0148] An embodiment of the present application also provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned device health prediction method can be implemented.

[0149] An embodiment of the present application provides a computer program product. When the computer program product is executed on a terminal device, the terminal device can implement the steps in the above-mentioned device health prediction method.

[0150] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A device health prediction method, characterized in that: include: Determine first health data of a target device, wherein the first health data comprises battery health data of the target device in a current period; Determine battery health prediction data of the target device within a preset future period according to the first health data.

2. The device health prediction method according to claim 1, characterized in that: The step of determining the battery health prediction data of the target device within a preset future period according to the first health data includes: The first health data is input into a health prediction model to obtain battery health prediction data of the target device in a future preset period; wherein the health prediction model is a deep learning model for time series prediction.

3. The device health prediction method according to claim 2, characterized in that: The battery health data of the target device includes a battery health value of the target device, and the inputting the first health data into a health prediction model to obtain battery health prediction data of the target device within a preset future period includes: Inputting the first health data into the health prediction model, and outputting a probability distribution of predicted values ​​of the battery health value of the target device at each time point in the future preset period; According to the probability distribution of the predicted value of the battery health value at each time point, the estimated value of the battery health value of the target device at that time point is determined.

4. The device health prediction method according to claim 2, characterized in that: Before inputting the first health data into the health prediction model, the method further includes: determining a device health data set, wherein the device health data set includes at least a time series of battery health data of each of a plurality of first devices over a first historical time; The health prediction model is trained using the device health data set until a preset end condition is met.

5. The device health prediction method according to claim 4, characterized in that: The using the device health data set to train the health prediction model includes: dividing the device health data set into a training data set and a validation data set, wherein the training data set includes battery health data of the plurality of first devices in a first period in the first historical time, and the validation data set includes battery health data of the plurality of first devices in a second period in the first historical time, the first period being earlier than the second period; Training the health prediction model using the training data set, the validation data set, and the first loss function until a preset end condition is met; Among them, the value of the first loss function is positively correlated with the difference between the predicted value of the battery health data of each first device at each time point in the second period output by the health prediction model and the true value of the battery health data of the device at each time point in the verification data set.

6. The device health prediction method according to claim 4, characterized in that: The determining of the equipment health data set comprises: Determine an initial health data set, wherein the initial health data set at least includes battery health data of a plurality of second devices at a plurality of time points within a second historical time, the plurality of second devices include the plurality of first devices, and the second historical time includes the first historical time; Performing a preprocessing operation on the initial health data set to obtain the device health data set consisting of a plurality of health data sequences; Among them, each health data sequence corresponds to a first device, and each health data sequence includes a time series of battery health data of the corresponding first device within the first historical time; the preprocessing operation includes at least one of the following operations: abnormal data elimination operation, missing data filling operation, data format conversion operation, and data standardization operation.

7. The device health prediction method according to any one of claims 1 to 6, characterized in that: The determining the first health data of the target device includes: Determine battery health data of the target device at least in a current period and feature data of at least one feature dimension of the target device as the first health data; The characteristic dimensions include one or more of the following: battery attributes, network type, usage time, usage frequency, network type and usage area.

8. The device health prediction method according to claim 7 as cited in claim 4, characterized in that: The device health data set also includes feature data of the at least one feature dimension of each first device, and the method further includes: Determine battery health data of a plurality of third devices at a plurality of time points within a third historical time and feature data of a plurality of feature dimensions of the plurality of third devices, wherein the feature data of each feature dimension of the plurality of third devices are not completely the same; Analyze the feature data of each feature dimension of the plurality of third devices and the battery health data of the plurality of third devices at a plurality of time points in the third historical time to obtain a correlation between each feature dimension and the battery health data; According to the correlation, the at least one characteristic dimension is screened out from the multiple characteristic dimensions, wherein the correlation between each of the at least one characteristic dimension and the battery health data is greater than or equal to a correlation threshold.

9. The device health prediction method according to any one of claims 1 to 6, characterized in that: The battery health prediction data includes predicted values ​​of the battery health value of the target device at various time points in a future preset period, and the method further includes: Determine whether a predicted value of a battery health value of the target device at at least one time point within the future preset period is less than a health value threshold; If so, an early warning message and / or corresponding equipment maintenance suggestion message is sent to the target device before the future preset period.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the device health prediction method according to any one of claims 1 to 9 are implemented.

Citation Information

Cited By

  • Equipment management method and device, computer equipment and storage medium

    CN119648196A

  • Device health prediction method and computer-readable storage medium

    WO2026011944A1