A method for online detection of charging performance of smart wearable devices

Abstract

The present application relates to the field of charging detection technology, and specifically to a method for online detection of charging performance for smart wearable devices. The method comprises: analyzing the voltage difference, current difference, and temperature difference between different time periods under each charging performance, determining the voltage deviation value, current deviation value, and temperature deviation value of the battery in each time period under each charging performance, and forming a charging performance vector with the temperature change value and the power loss characteristic value. The charging performance vector and its preset label are formed into a data unit; based on all data units, an ensemble learning algorithm is used to iteratively optimize the weak classifier to detect the charging performance of the battery in the smart wearable device. The present application solves the problems of reduced sensitivity and missed detection caused by fixed current threshold charging performance detection, and improves the sensitivity and reliability of charging performance detection of smart wearable devices.

Description

Technical Field

[0001] The present application relates to the field of charging detection technology, and in particular to an online detection method for charging performance of a smart wearable device. Background Art

[0002] Smart wearable devices are electronic devices that can be worn on the human body and have multiple functions, such as smart watches, smart bracelets, and smart glasses. The compact design of smart wearable devices means their battery capacity is limited, making efficient charging technology and reliable charging management particularly important. Online monitoring of charging performance not only facilitates adjustment of charging strategies but also effectively prevents potential failures and extends device lifespan.

[0003] Due to the size and battery capacity limitations of smart wearable devices, abnormal charging performance signals are often subtle, making detection difficult. Existing technologies detect charging performance by setting current thresholds. However, when applied to smart wearable devices, this method struggles to accurately capture weak abnormal signals during the charging process, resulting in insufficient detection sensitivity. Furthermore, fixed threshold settings reduce the adaptability of detection across different charging modes, increasing the missed detection rate during online testing of smart wearable device charging performance and reducing the sensitivity and reliability of smart wearable device charging performance testing. Summary of the Invention

[0004] In order to solve the above technical problems, the present application provides a method for online detection of charging performance of a smart wearable device to solve the existing problems.

[0005] The present invention provides a method for online detection of charging performance of a smart wearable device using the following technical solutions:

[0006] Use smart sensors to collect real-time voltage, current and temperature data of batteries in smart wearable devices under each charging performance;

[0007] Analyze the battery's energy loss during each preset time period under each charging performance, and determine the characteristic value of the battery's energy loss during each time period under each charging performance; combine each time period and a preset number of time periods before it into a characteristic time period, and determine the temperature change value under each charging performance by analyzing the difference between the average distribution of the backward difference values ​​of all temperature data under each charging performance during the characteristic time period and the maximum temperature data, and combining the difference between the temperature data collected at the beginning and end of the characteristic time period;

[0008] Analyze all voltage differences, all current differences, and all temperature data average distribution differences between each time period under each charging performance and the remaining time periods, respectively, determine the voltage deviation value, current deviation value, and temperature deviation value of the battery in each time period under each charging performance, and combine the temperature change value and the power loss characteristic value to form a charging performance vector of the battery in each time period under each charging performance, and form the charging performance vector and its preset label into a data unit;

[0009] Based on all data units, the ensemble learning algorithm is used to iteratively optimize the weak classifiers, specifically:

[0010] Based on the missed detection rate of each weak classifier, the characteristic stretch coefficient of each weak classifier is determined. In combination with the charging performance vector in the data unit corresponding to each weak classifier in each iteration, the charging performance characteristic value of each weak classifier in the next iteration is adjusted to obtain the optimized weak classifier;

[0011] Based on the optimized weak classifier, the charging performance of the battery in the smart wearable device at the current moment is detected.

[0012] Preferably, the categories of charging performance include: normal battery charging, poor battery charging contact, over-temperature battery charging, and battery aging.

[0013] Preferably, the method for determining the characteristic value of the battery's power loss in each time period under each charging performance is:

[0014] Calculate the input power of the battery in each time period under each charging performance based on the current and voltage of the battery in each time period under each charging performance; obtain the remaining capacity of the battery at the beginning and end of each time period under each charging performance, and use the difference between the remaining capacity of the battery at the end and the beginning of the time period as the stored power of the battery in each time period under each charging performance;

[0015] The difference between the input power and the stored power of the battery in each period under each charging performance is divided by the stored power, which is taken as the power loss characteristic value of the battery in each period under each charging performance.

[0016] Preferably, the method for determining the temperature change value under each charging performance is:

[0017] Calculate the result of dividing the maximum temperature data under each charging performance in the characteristic period by the average of the backward difference values ​​of all temperature data, and record it as the temperature difference value. Multiply the difference between the temperature data between the first and last collection moments of the characteristic period by the temperature difference value as the temperature change value under each charging performance.

[0018] Preferably, the determining of the voltage deviation value, current deviation value and temperature deviation value of the battery in each time period under each charging performance includes:

[0019] Calculate all voltage differences between each time period and the remaining time periods under each charging performance, and take the minimum difference as the voltage deviation value of the battery in each time period under each charging performance;

[0020] With respect to the current and temperature data, the current deviation value and the temperature deviation value of the battery in each time period under each charging performance are obtained according to the method for obtaining the voltage deviation value.

[0021] Preferably, the method for determining the preset label is:

[0022] The labels of the battery charging performance vectors in each period under normal battery charging in all types of charging performance are set to 1, and the labels of the battery charging performance vectors in each period under all other types of charging performance are set to -1.

[0023] Preferably, the expression of the characteristic stretching coefficient of each weak classifier is: γ i =γ1+FNR i ×(γ2-γ1); where γ i Represents the feature stretching coefficient of weak classifier i; FNR i represents the missed detection rate of weak classifier i; γ1 and γ2 represent the preset first feature stretching coefficient and the preset second feature stretching coefficient respectively, wherein the preset first feature stretching coefficient is smaller than the preset second feature stretching coefficient.

[0024] Preferably, the adjusting the charging performance characteristic value of each weak classifier in the next iteration includes:

[0025] Record all elements in the charging performance vector in the data unit as charging performance feature values, wherein each weak classifier includes a preset number of charging performance feature values ​​in the charging performance vector;

[0026] The nth charging performance characteristic value of weak classifier i at the k+1th iteration The expression is: Where, represents the nth charging performance characteristic value of weak classifier i at the kth iteration, where the initial value of the nth charging performance characteristic value is the nth charging performance characteristic value in the charging performance vector in the data unit corresponding to weak classifier i; norm() represents the normalized value; γ i Represents the feature stretching coefficient of weak classifier i.

[0027] Preferably, the optimized weak classifier includes:

[0028] During the weak classifier optimization iteration process, the charging performance characteristic value of the weak classifier is adjusted. When the accuracy of the weak classifier no longer increases or the iteration reaches the preset maximum number of iterations, the obtained weak classifier is used as the optimized weak classifier.

[0029] Preferably, the detecting of the charging performance of the battery in the smart wearable device at the current moment includes:

[0030] According to the method for obtaining the battery charging performance vectors for each time period under each charging performance, the charging performance vectors for all charging performances in the time period before the current moment, which is the same length as each time period, are obtained and recorded as the charging performance vector at the current moment. All the charging performance vectors at the current moment are respectively input into the optimized weak classifiers to obtain the predicted labels output by all the optimized weak classifiers;

[0031] The expression of the charging performance prediction label L of the battery in the smart wearable device at the current moment is: Where FL i Indicates the predicted label output by the optimized weak classifier i at the current moment; FNR i 、FPR i They represent the missed detection rate and false detection rate of the optimized weak classifier i respectively; I represents the number of all optimized classifiers; ln[] represents the logarithmic function with a natural constant as the base; sgn{} represents the sign function;

[0032] If the value of the predicted label L is 0 or 1, the battery charging performance of the smart wearable device is normal at the current moment; otherwise, the battery charging performance is abnormal.

[0033] An embodiment of the present application provides a method for online detection of charging performance of a smart wearable device, the method comprising the following steps:

[0034] This application has at least the following beneficial effects:

[0035] This application uses smart sensors to collect status data of smart wearable devices during charging. By calculating the power loss value, the energy loss during the charging process can be quantified, which helps to identify abnormal charging performance caused by problems such as battery aging or poor contact; further, by calculating the temperature change, the detection of short-term abnormal temperature rise can be enhanced, thereby improving the ability to identify overheating risks during the charging process of smart wearable devices; further, by analyzing the differences in voltage, current and temperature data between different time periods, voltage deviation values, current deviation values ​​and temperature deviation values ​​are constructed, which helps to identify small fluctuations in the charging process and improve the sensitivity of charging performance detection of smart wearable devices. and reliability; further, the energy loss characteristic value, temperature change value, and voltage, current and temperature deviation values ​​are combined into a charging performance image, which can comprehensively reflect the state of the charging performance of the smart wearable device, and the charging performance vector and its label are combined into a data unit for subsequent classifier training and optimization, which helps to improve the ability to distinguish different charging performances of smart wearable devices, thereby improving the sensitivity and reliability of the performance detection of smart wearable devices; further, an integrated learning algorithm is adopted, and combined with feature stretching technology, the weak classifier is iteratively optimized to amplify weak abnormal features, reduce the missed detection rate, and improve the sensitivity and reliability of the detection of smart wearable devices. This application extracts the charging features during the charging process of the smart wearable device, adopts an integrated learning algorithm based on the charging features, iteratively optimizes the weak classifier, reduces the missed detection rate, and improves the sensitivity and reliability of the charging performance detection of the smart wearable device. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. 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 any creative work.

[0037] Figure 1 A flowchart of a method for online detection of charging performance of a smart wearable device provided in one embodiment of the present application;

[0038] Figure 2 A schematic diagram of a data unit acquisition process provided for one embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to further illustrate the technical means and effects adopted by this application to achieve the predetermined invention purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation method, structure, features and effects of an online detection method for charging performance of a smart wearable device proposed in this application. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics of one or more embodiments may be combined in any suitable form.

[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0041] The following describes in detail a specific solution of a method for online detection of charging performance of a smart wearable device provided by the present application with reference to the accompanying drawings.

[0042] An embodiment of the present application provides a method for online detection of charging performance of a smart wearable device. Specifically, the following method is provided for online detection of charging performance of a smart wearable device. Figure 1 , the method comprises the following steps:

[0043] Step S1: Using smart sensors to collect voltage, current and temperature data of the battery in the smart wearable device at each charging performance in real time.

[0044] Multiple smart sensors are installed on the smart wearable device to measure the voltage data, current data and temperature data of the smart wearable device in real time. The data acquisition frequency is set to f. In this embodiment, the value of the data acquisition frequency f is manually set. In this embodiment, the value of the data acquisition frequency f is 10 MHz. In actual application, as other implementation methods, the implementer can also set it by himself according to the specific situation. This embodiment does not impose any special restrictions.

[0045] Furthermore, low-power Bluetooth is used to transmit the data collected by smart sensors to the cloud in real time.

[0046] Step S2: Analyze the voltage distribution, current distribution, and temperature distribution under each charging performance to obtain data units.

[0047] The current, voltage, and temperature data in the charging circuit of smart wearable devices all contain a large number of sampling points. Directly using this high-dimensional data as input for charging performance monitoring models can lead to excessive computational burden, overfitting, and the potential introduction of unnecessary noise. Furthermore, the abnormal charging performance characteristics of smart wearable devices are often very subtle. Feature extraction helps capture the underlying patterns of data changes during the charging process, amplifying subtle abnormalities and improving the sensitivity of online charging performance monitoring.

[0048] Energy loss during charging is one of the important indicators reflecting charging performance. Abnormal charging performance is often accompanied by large energy loss. Therefore, the greater the energy loss, the greater the probability of abnormal charging performance. Therefore, this embodiment analyzes the energy loss of the battery in each preset time period under each charging performance, and determines the energy loss characteristic value of the battery in each time period under each charging performance, specifically:

[0049] As an implementation method, in this embodiment, the input power of the battery in each time period under each charging performance is calculated based on the current and voltage of the battery in each time period under each charging performance. In this embodiment, the remaining capacity of the battery at the beginning and end of each time period under each charging performance is read by the BMS chip, and the difference between the remaining capacity of the battery at the end and the beginning of the time period is used as the stored power of the battery in each time period under each charging performance.

[0050] The difference between the input power and the stored power of the battery in each period under each charging performance is divided by the stored power, which is taken as the power loss characteristic value of the battery in each period under each charging performance.

[0051] It should be understood that the method for constructing the preset time periods is: all moments within the preset time length before each moment are composed into the preset time periods, where the value of the preset time length is set manually. In this embodiment, the value of the preset time length is 200ms. In actual application, as other implementation methods, the implementer can also set the value of the preset time length or construct the method of presetting each time period based on the specific situation. This embodiment does not impose any special restrictions.

[0052] It should be noted that the value of the preset number is set manually. In this embodiment, the value of the preset number is 49. In actual application, as other implementation methods, the implementer can also set it by himself based on the specific situation. This embodiment does not impose any special restrictions.

[0053] Among them, the process of calculating the input power of the battery in each time period under each charging performance based on the current and voltage of the battery in each time period, and obtaining the remaining battery capacity using the BMS chip are all well-known technologies, and the method of calculating the input power and the process of obtaining the remaining battery capacity will not be described in detail again.

[0054] Furthermore, since temperature rise is another important indicator of abnormal charging performance, charging circuit failure or battery aging can evenly cause excessive battery temperature rise. Therefore, this embodiment combines each time period and a preset number of time periods before it into a characteristic time period. By analyzing the difference between the average distribution of the backward difference values ​​of all temperature data under each charging performance within the characteristic time period and the maximum temperature data, and combining the difference in temperature data between the first and last collection moments of the characteristic time period, the temperature change value under each charging performance is determined to improve the identification of overheating risks during the charging process. Specifically,

[0055] Calculate the result of dividing the maximum temperature data under each charging performance in the characteristic period by the average of the backward difference values ​​of all temperature data, and record it as the temperature difference value. Take the product of the difference in temperature data between the first and last collection moments of the characteristic period and the temperature difference value as the temperature change value under each charging performance; among them, the difference in temperature data between the end moment and the initial moment of the characteristic period reflects the overall temperature rise degree in the characteristic period. Use the temperature difference value as the weight to enhance the detection of short-term abnormal temperature rise. The greater the temperature rise during the charging process, the greater the temperature change value of the corresponding period. The greater the obtained temperature change value, the greater the possibility of abnormal charging performance.

[0056] It should be noted that there are many methods for measuring the difference between data. In this embodiment, the absolute value of the difference between the temperature data at the beginning and end of the characteristic period is used as the difference between the temperature data at the beginning and end of the characteristic period. In actual application, as other implementation methods, the implementer may also use other methods for measuring the difference between data, such as the square or ratio of the difference, based on the specific circumstances. This embodiment does not impose any special restrictions on the selection of the method for measuring the difference between data.

[0057] It is additionally noted that, in this embodiment, all contents involving calculation of differences between data adopt the method of taking the absolute value of the difference.

[0058] The calculation method of the backward difference value is a well-known technology, and its specific calculation process will not be repeated here.

[0059] Furthermore, this embodiment determines the voltage deviation value, current deviation value, and temperature deviation value of the battery in each time period under each charging performance by analyzing the differences in all voltages, all currents, and all temperature data average distributions between each time period under each charging performance and the remaining time periods, specifically:

[0060] As an implementation manner, in this embodiment, by calculating all voltage differences between each time period under each charging performance and the remaining time periods, the minimum difference is used as the voltage deviation value of the battery in each time period under each charging performance.

[0061] With respect to the current and temperature data, the current deviation value and the temperature deviation value of the battery in each time period under each charging performance are obtained according to the method for obtaining the voltage deviation value.

[0062] It should be noted that there are many methods for measuring the differences between data groups. In this embodiment, the DTW distance of all voltages between each time period and the remaining time periods under each charging performance is used as the difference between all voltages between each time period and the remaining time periods under each charging performance. In actual application, as other implementation methods, implementers may also adopt other methods for measuring the differences between data groups, such as Euclidean distance or Manhattan distance, based on specific circumstances. This embodiment does not impose any special restrictions on the selection of methods for measuring the differences between data groups.

[0063] The calculation method of the DTW distance is a well-known technology, and its specific calculation process is not repeated here.

[0064] Furthermore, in this embodiment, the voltage deviation value, current deviation value, temperature deviation value, temperature change value and power loss characteristic value of the battery in each time period under each charging performance constitute the charging performance vector of the battery in each time period under each charging performance.

[0065] Furthermore, in this embodiment, the charging performance vector of the battery in each time period under each charging performance and its corresponding label are formed into a data unit, wherein the method for obtaining the preset label is: setting the label of the charging performance vector of the battery in each time period under normal battery charging in all types of charging performance to 1, and setting the label of the charging performance vector of the battery in each time period under all other types of charging performance to -1.

[0066] Preferably, the data unit acquisition process diagram provided in this embodiment is as follows Figure 2 shown.

[0067] So far, this embodiment has analyzed the voltage, current and temperature distribution during the charging process of the smart wearable device, extracted the energy loss characteristic values, temperature change values, and voltage, current and temperature deviation values, and constructed a charging performance vector. These charging performance vectors and their corresponding labels constitute a data unit for charging performance detection, which can effectively identify abnormal conditions during the charging process and improve the sensitivity and reliability of charging performance detection for smart wearable devices.

[0068] Step S3: Based on all data units, an ensemble learning algorithm is used to iteratively optimize the weak classifier.

[0069] The battery capacity of smart wearable devices is small, and the range of variation of battery voltage and current is usually narrow. The temperature rise caused by minor faults is usually very small, resulting in charging abnormalities with characteristics similar to those in normal conditions, resulting in a high missed detection rate for abnormal charging performance detection.

[0070] Therefore, this embodiment adopts the Boosting method in the ensemble learning algorithm and performs stretching processing on the characteristic values ​​of different charging performances according to the missed detection rate in the iterative optimization process of its weak classifier, thereby enhancing the differences in the changes in characteristic values ​​between different charging performances. This enables the weak classifier to better identify abnormal characteristics of the charging performance of smart wearable devices, improves the model's ability to distinguish different charging performances, and reduces the missed detection rate of the charging performance detection model. Specifically:

[0071] This embodiment determines the characteristic stretching coefficient of each weak classifier based on the missed detection rate of each weak classifier, and adjusts the charging performance characteristic value of each weak classifier in the next iteration in combination with the charging performance vector in the data unit corresponding to each weak classifier in each iteration to obtain an optimized weak classifier, specifically:

[0072] In this embodiment, the ratio of the number of data units under four charging performance conditions, namely, normal battery charging, poor battery charging contact, over-temperature battery charging, and battery aging, is set to 5:2:2:1. The number of data units collected and obtained is 50,000, constituting a data set, which is divided into a training set and a validation set in a ratio of 8:2.

[0073] Furthermore, the five elements of the voltage deviation value, current deviation value, temperature deviation value, temperature change value and power loss characteristic value in the charging performance vector are collectively referred to as charging performance characteristic values. Each time, a preset number of non-repeated charging performance characteristic values ​​are selected from the five charging performance characteristic values ​​as classification features of the weak classifier. In this embodiment, the preset number is 3, that is, each time 3 charging performance characteristic values ​​are selected from the five charging performance characteristic values ​​as classification features of the weak classifier. According to the selection method of the charging performance characteristic values, a total of That is, there are 10 weak classifiers. The construction process of the weak classifier is: the training set and three charging performance feature values ​​​​selected from the charging performance vector are used as the input of the CART algorithm, and the resulting decision tree is a weak classifier.

[0074] Among them, the Boosting method and CART algorithm in the ensemble learning algorithm are both well-known technologies, and their specific principles are not described in detail here.

[0075] Furthermore, the charging performance characteristic values ​​are stretched to amplify weak abnormal features, thereby improving the classification accuracy of the model while reducing the training time of the model. That is, based on the missed detection rate of each weak classifier, the characteristic stretching coefficient of each weak classifier is determined, and combined with the charging performance vector in the data unit corresponding to each weak classifier in each iteration, the charging performance characteristic value of each weak classifier in the next iteration is adjusted, specifically:

[0076] As an implementation method, in this embodiment, the feature stretching coefficient γ of the weak classifier i is i The expression of γ is:i =γ1+FNR i ×(γ2-γ1); where γ i Represents the feature stretching coefficient of weak classifier i; FNR i represents the missed detection rate of weak classifier i; γ1 and γ2 represent the preset first feature stretching coefficient and the preset second feature stretching coefficient respectively, wherein the preset first feature stretching coefficient is smaller than the preset second feature stretching coefficient.

[0077] It should be noted that, in this embodiment, the preset value of the first characteristic stretching coefficient is 1, and the preset value of the second characteristic stretching coefficient is 2. In actual application, as other implementation methods, the implementer can also set them according to the specific situation. This embodiment does not impose any special restrictions.

[0078] The calculation method of the missed detection rate of the weak classifier is a well-known technology, and the specific calculation process is not repeated here.

[0079] Furthermore, considering that the larger the charging performance characteristic value, the greater the probability that the smart wearable device has charging performance abnormalities, for weak anomalies in the charging process, that is, when the difference between the charging performance characteristic values ​​of the normal data unit and the abnormal data unit in the training set is small, resulting in a high missed detection rate of a single weak classifier, a larger stretching coefficient is set at this time to improve the weak classifier's ability to recognize weak anomalies, specifically:

[0080] As an implementation method, in this embodiment, the nth charging performance characteristic value of the weak classifier i at the k+1th iteration is The expression is: Where, represents the nth charging performance characteristic value of weak classifier i at the kth iteration, where the initial value of the nth charging performance characteristic value is the nth charging performance characteristic value in the charging performance vector in the data unit corresponding to weak classifier i; norm() represents the normalized value; γ i Represents the feature stretching coefficient of weak classifier i.

[0081] Furthermore, during the weak classifier optimization iteration process, the charging performance characteristic value of the weak classifier is adjusted. When the accuracy of the weak classifier no longer increases or the iteration reaches a preset maximum number of iterations, the obtained weak classifier is used as the optimized weak classifier.

[0082] It should be noted that the maximum number of iterations is set to 20 in this embodiment, and the implementer can also set it according to the specific situation. This embodiment does not impose any special restrictions.

[0083] The calculation method of the accuracy of the weak classifier is a well-known technology, and the specific calculation process will not be described in detail.

[0084] So far, this embodiment adopts the Boosting algorithm in ensemble learning, combined with feature stretching technology, to iteratively optimize the weak classifier to enhance the ability to identify abnormal features of charging performance of smart wearable devices, adjust the feature stretching coefficient according to the missed detection rate, amplify weak anomalies, improve the model classification accuracy, reduce training time, and reduce the missed detection rate, thereby improving the sensitivity and reliability of charging performance detection of smart wearable devices.

[0085] Step S4: Detecting the charging performance of the battery in the smart wearable device at the current moment based on the optimized weak classifier.

[0086] Based on the optimized weak classifier obtained in step S3, further, according to the method for obtaining the charging performance vector of the battery in each time period under each charging performance, the charging performance vectors under all kinds of charging performance in the time period with the same length as each time period before the current moment are obtained, recorded as the charging performance vector at the current moment, and all the charging performance vectors are respectively input into the optimized weak classifier to obtain the predicted labels output by all optimized weak classifiers.

[0087] As an implementation manner, in this embodiment, the expression of the charging performance prediction label L of the battery in the smart wearable device at the current moment is: Where FL i Indicates the predicted label output by the optimized weak classifier i at the current moment; FNR i 、FPR i They represent the missed detection rate and false detection rate of the optimized weak classifier i respectively; I represents the number of all optimized classifiers; ln[] represents the logarithmic function with a natural constant as the base; sgn{} represents the sign function.

[0088] If the value of the predicted tag L is 0 or 1, the battery charging performance of the smart wearable device at the current moment is normal. Conversely, if the value of the predicted tag L is -1, the battery charging performance of the smart wearable device at the current moment is abnormal.

[0089] The calculation method of the false positive rate of the weak classifier is a well-known technique, and the specific calculation process will not be described in detail.

[0090] At this point, this embodiment collects voltage, current and temperature data in real time, and extracts power loss, temperature changes, and voltage, current and temperature deviation values ​​to construct a charging performance vector. It uses the Boosting algorithm in ensemble learning and combines iterative optimization of weak classifiers with feature stretching technology, thereby enhancing the ability to recognize weak abnormal features, improving the classification accuracy and detection sensitivity of the model, and reducing the missed detection rate, thereby effectively improving the sensitivity and reliability of charging performance detection of smart wearable devices.

[0091] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0092] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0093] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them. Modifications to the technical solutions described in the aforementioned embodiments, or equivalent replacements of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for online detection of charging performance of a smart wearable device, characterized in that: The method comprises the following steps: Use smart sensors to collect real-time voltage, current and temperature data of batteries in smart wearable devices under each charging performance; Analyze the battery's energy loss during each preset time period under each charging performance, and determine the characteristic value of the battery's energy loss during each time period under each charging performance; combine each time period and a preset number of time periods before it into a characteristic time period, and determine the temperature change value under each charging performance by analyzing the difference between the average distribution of the backward difference values ​​of all temperature data under each charging performance during the characteristic time period and the maximum temperature data, and combining the difference between the temperature data collected at the beginning and end of the characteristic time period; Analyze all voltage differences, all current differences, and all temperature data average distribution differences between each time period under each charging performance and the remaining time periods, respectively, determine the voltage deviation value, current deviation value, and temperature deviation value of the battery in each time period under each charging performance, and combine the temperature change value and the power loss characteristic value to form a charging performance vector of the battery in each time period under each charging performance, and form the charging performance vector and its preset label into a data unit; Based on all data units, the ensemble learning algorithm is used to iteratively optimize the weak classifiers, specifically: Based on the missed detection rate of each weak classifier, the characteristic stretch coefficient of each weak classifier is determined. In combination with the charging performance vector in the data unit corresponding to each weak classifier in each iteration, the charging performance characteristic value of each weak classifier in the next iteration is adjusted to obtain the optimized weak classifier; Based on the optimized weak classifier, the charging performance of the battery in the smart wearable device at the current moment is detected.

2. The method for online detection of charging performance of a smart wearable device according to claim 1, wherein: The categories of charging performance include: normal battery charging, poor battery charging contact, over-temperature battery charging and battery aging.

3. The method for online detection of charging performance of a smart wearable device according to claim 1, wherein: The method for determining the characteristic value of the battery's power loss in each time period under each charging performance is as follows: Calculate the input power of the battery in each time period under each charging performance based on the current and voltage of the battery in each time period under each charging performance; obtain the remaining capacity of the battery at the beginning and end of each time period under each charging performance, and use the difference between the remaining capacity of the battery at the end and the beginning of the time period as the stored power of the battery in each time period under each charging performance; The difference between the input power and the stored power of the battery in each period under each charging performance is divided by the stored power, which is taken as the power loss characteristic value of the battery in each period under each charging performance.

4. The method for online detection of charging performance of a smart wearable device according to claim 1, wherein: The method for determining the temperature change value under each charging performance is: Calculate the result of dividing the maximum temperature data under each charging performance in the characteristic period by the average of the backward difference values ​​of all temperature data, and record it as the temperature difference value. Multiply the difference between the temperature data between the first and last collection moments of the characteristic period by the temperature difference value as the temperature change value under each charging performance.

5. The method for online detection of charging performance of a smart wearable device according to claim 1, wherein: The step of respectively determining the voltage deviation value, current deviation value, and temperature deviation value of the battery in each time period under each charging performance includes: Calculate all voltage differences between each time period and the remaining time periods under each charging performance, and take the minimum difference as the voltage deviation value of the battery in each time period under each charging performance; With respect to the current and temperature data, the current deviation value and the temperature deviation value of the battery in each time period under each charging performance are obtained according to the method for obtaining the voltage deviation value.

6. The method for online detection of charging performance of a smart wearable device according to claim 2, wherein: The method for determining the preset label is: The labels of the battery charging performance vectors in each period under normal battery charging in all types of charging performance are set to 1, and the labels of the battery charging performance vectors in each period under all other types of charging performance are set to -1.

7. The method for online detection of charging performance of a smart wearable device according to claim 1, wherein: The expression of the characteristic stretching coefficient of each weak classifier is: γ i =γ1+FNR i ×(γ2-γ1); where γ i Represents the feature stretching coefficient of weak classifier i; FNR i represents the missed detection rate of weak classifier i; γ1 and γ2 represent the preset first feature stretching coefficient and the preset second feature stretching coefficient respectively, wherein the preset first feature stretching coefficient is smaller than the preset second feature stretching coefficient.

8. The method for online detection of charging performance of a smart wearable device according to claim 1, wherein: The adjusting of the charging performance characteristic values ​​of each weak classifier in the next iteration includes: Record all elements in the charging performance vector in the data unit as charging performance feature values, wherein each weak classifier includes a preset number of charging performance feature values ​​in the charging performance vector; The nth charging performance characteristic value of weak classifier i at the k+1th iteration The expression is: Where, represents the nth charging performance characteristic value of weak classifier i at the kth iteration, where the initial value of the nth charging performance characteristic value is the nth charging performance characteristic value in the charging performance vector in the data unit corresponding to weak classifier i; norm() represents the normalized value; γ i Represents the feature stretching coefficient of weak classifier i.

9. The method for online detection of charging performance of a smart wearable device according to claim 1, wherein: The optimized weak classifier includes: During the weak classifier optimization iteration process, the charging performance characteristic value of the weak classifier is adjusted. When the accuracy of the weak classifier no longer increases or the iteration reaches the preset maximum number of iterations, the obtained weak classifier is used as the optimized weak classifier.

10. The method for online detection of charging performance of a smart wearable device according to claim 1, wherein: The detecting of the charging performance of the battery in the smart wearable device at the current moment includes: According to the method for obtaining the battery charging performance vectors for each time period under each charging performance, the charging performance vectors for all charging performances in the time period before the current moment, which is the same length as each time period, are obtained and recorded as the charging performance vector at the current moment. All the charging performance vectors at the current moment are respectively input into the optimized weak classifiers to obtain the predicted labels output by all the optimized weak classifiers; The expression of the charging performance prediction label L of the battery in the smart wearable device at the current moment is: Where FL i Indicates the predicted label output by the optimized weak classifier i at the current moment; FNR i 、FPR i They represent the missed detection rate and false detection rate of the optimized weak classifier i respectively; I represents the number of all optimized classifiers; ln[] represents the logarithmic function with a natural constant as the base; sgn{} represents the sign function; If the value of the predicted label L is 0 or 1, the battery charging performance of the smart wearable device is normal at the current moment; otherwise, the battery charging performance is abnormal.

Images