BBU category identification model construction, life prediction method, device and equipment

Abstract

The present invention relates to the field of computer technology and discloses a method, apparatus, and device for constructing a BBU category identification model and predicting lifespan. The method comprises: obtaining a capacity change value and a voltage change value generated by each of a plurality of battery backup units after operating under a plurality of different preset conditions; inputting the capacity change value and voltage change value generated by a first type of battery backup unit after operating under each set of preset conditions into a pre-constructed initial category identification model to obtain a plurality of sets of preset model parameter values; and constructing a category identification model corresponding to the first type of battery backup unit based on the plurality of sets of preset model parameter values, the capacity change parameters, the voltage change parameters, and the preset model parameters. This method can be used to obtain a category identification model corresponding to each of the plurality of battery backup units. These category identification models can then be used to identify the type of battery backup unit.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to BBU category identification model construction, life prediction method, device and equipment. Background Art

[0002] Current storage systems typically utilize a PC power supply unit (PSU) and a battery backup unit (BBU) for redundant power supply. When the mains power in the data center goes out, the storage system detects the PSU power anomaly in real time and seamlessly switches to the BBU. Existing technology requires predicting the remaining lifespan of the BBU to prevent BBU performance degradation or failures from seriously impacting the reliability and security of the storage system.

[0003] Currently, methods used to predict the remaining life of a BBU are typically data-driven. These methods can only train degradation data and predict the service life of a single BBU type. Existing technologies are unable to identify different BBU types and, therefore, are unable to extract corresponding test data to predict the BBU lifespan. Summary of the Invention

[0004] In view of this, the present invention provides a BBU category identification model construction, life prediction method, device and equipment to solve the problem in the prior art that different types of BBU categories cannot be identified and the corresponding types of test data cannot be extracted to predict the life of the BBU.

[0005] In a first aspect, the present invention provides a method for constructing a BBU category identification model, the method comprising:

[0006] Obtaining a capacity change value and a voltage change value generated by each of the multiple battery backup units after the battery backup unit operates under multiple sets of different preset conditions;

[0007] Inputting the capacity change value and the voltage change value generated by the first battery backup unit operating under each set of preset conditions into a pre-built initial category identification model to obtain multiple sets of preset model parameter values, wherein the initial category identification model includes the capacity change parameter, the voltage change parameter, and the preset model parameter, and the first battery backup unit is any one of the multiple battery backup units;

[0008] A category recognition model corresponding to the first type of battery backup unit is constructed based on multiple sets of preset model parameter values, capacity change parameters, voltage change parameters, and preset model parameters.

[0009] The present invention provides a method for constructing a BBU category identification model, which has the following advantages:

[0010] Obtain the capacity change value and voltage change value generated by each of the multiple battery backup units after working under multiple sets of different preset conditions. After inputting the capacity change value and voltage change value generated by the first battery backup unit after working under each set of preset conditions into the pre-built category recognition model, multiple sets of preset model parameter values ​​can be obtained. According to the multiple sets of preset model parameter values, obtain the parameter values ​​corresponding to the final preset model parameters. Input the parameter values ​​corresponding to the final preset model parameters into the pre-built category recognition model to obtain a category recognition model in which only the capacity change parameter and the voltage change parameter are unknown. This model happens to be the category recognition model corresponding to the first battery backup unit. Similarly, the category recognition model corresponding to each of the multiple battery backup units can be obtained respectively through the above method. These category recognition models are used to identify the type of battery backup unit in the future.

[0011] In an optional embodiment, obtaining a capacity change value and a voltage change value generated by each of the multiple battery backup units after operating under multiple sets of different preset conditions includes:

[0012] The capacity change value and voltage change value are obtained after each battery backup unit is discharged at a constant current according to a preset current value for a preset time period under different preset charge states.

[0013] Specifically, by inputting the capacity change values ​​and voltage change values ​​obtained under different states of charge into a pre-built category recognition model, the preset model parameters under different states of charge can be fitted, and then the parameter values ​​of the final preset model parameters can be obtained based on the parameter values ​​of the preset model parameters under different states of charge, so that the parameter values ​​of the preset model parameters finally obtained are more representative, thereby making the category recognition model finally obtained more accurate in identifying the category of the battery backup unit.

[0014] In an optional embodiment, each of the multiple sets of preset model parameter values ​​includes a parameter value of a first preset model parameter and a parameter value of a second preset model parameter. A category recognition model corresponding to the first type of battery backup unit is constructed based on the multiple sets of preset model parameter values, the capacity change parameter, the voltage change parameter, and the preset model parameters, including:

[0015] Obtaining a first average parameter value corresponding to the first preset model parameter based on the plurality of sets of parameter values ​​of the first preset model parameter; and obtaining a second average parameter value corresponding to the second preset model parameter based on the plurality of sets of parameter values ​​of the second preset model parameter;

[0016] A category identification model corresponding to the first type of battery backup unit is constructed according to the first average parameter value, the second average parameter value, the capacity variation parameter, and the voltage variation parameter.

[0017] Specifically, the average value can better reflect the universality and versatility of a parameter. Therefore, constructing a category recognition model corresponding to the first type of battery backup unit based on the first average parameter value, the second average parameter value, the capacity change parameter, and the voltage change parameter will be more accurate when performing type recognition.

[0018] In an optional embodiment, the initial category recognition model is expressed using the following formula:

[0019] ΔQ=a1·jexp(a2)

[0020] Wherein, ΔQ is a capacity change parameter, j is a voltage change parameter, a1 is a first preset model parameter, and a2 is a second preset model parameter. In a second aspect, the present invention provides a BBU remaining life prediction method, the method comprising:

[0021] Obtaining an actual capacity change value and an actual voltage change value generated by the battery backup unit to be identified under preset conditions;

[0022] Inputting the actual capacity change value into different category identification models constructed by the BBU category identification model construction method according to any one of claims 1 to 4, respectively, to obtain a reference voltage change value corresponding to each category identification model;

[0023] Determine the type of the battery backup unit to be identified based on the actual voltage change value and the reference voltage change value corresponding to each category identification model;

[0024] Acquiring test data corresponding to the type of the battery backup unit to be identified;

[0025] The test data is input into the target remaining life prediction model, and after multiple predictions are made using the target remaining life prediction model, the remaining life prediction interval of the battery backup unit is obtained.

[0026] The present invention provides a method for predicting the remaining life of a BBU, which has the following advantages:

[0027] The actual capacity change and voltage change values ​​generated by the battery backup unit to be identified under preset conditions are obtained. The actual capacity change values ​​are then input into different classification identification models constructed using the method of any embodiment of the first aspect, thereby obtaining reference voltage values ​​corresponding to each classification identification model. Based on a comparison of the actual voltage change and the reference voltage values, one of the multiple classification identification models is selected as a target classification identification model, and the type of battery backup unit corresponding to the target classification model is determined to be the type of the battery backup unit to be identified. Test data corresponding to the type of battery backup unit to be identified is then extracted and input into a target remaining life prediction model. Multiple predictions are performed using the target remaining life prediction model to obtain a predicted remaining life interval for the battery backup unit. Compared to a single battery remaining life value, a predicted remaining life interval provides greater flexibility and a wider range for battery maintenance or replacement. Users can replace and maintain batteries within this predicted remaining life interval, minimizing delays in replacement or maintenance and ensuring the safety of the battery backup unit during use. This also minimizes battery waste caused by premature replacement, reducing maintenance costs and the likelihood of fatal failures.

[0028] In an optional embodiment, determining the type of the battery backup unit to be identified according to the actual voltage change value and the reference voltage change value corresponding to each category identification model includes:

[0029] Compare the actual voltage change value with the reference voltage change value generated by each category recognition model respectively;

[0030] When it is determined that the difference between the actual voltage change value and the reference voltage change value generated by the first type identification model is the smallest, the type of the battery backup unit corresponding to the first type identification model is determined to be the type corresponding to the battery backup unit to be identified.

[0031] Specifically, the actual voltage change value is compared with the reference voltage change value generated by each category recognition model, and the category recognition model corresponding to the reference voltage change value closest to the actual voltage change value is selected as the target recognition model. The selection method will be closer to the type of battery backup unit to be identified.

[0032] In an optional embodiment, the test data is input into a target remaining life prediction model, and after multiple predictions are performed using the target remaining life prediction model, a remaining life prediction interval of the battery backup unit is obtained, including:

[0033] In the current iteration period, the initial remaining life prediction model is iteratively trained, and when the trained remaining life prediction model meets the preset training stop condition, the target remaining life prediction model corresponding to the current iteration period is obtained;

[0034] Input the test data into the target remaining life prediction model corresponding to the current iteration cycle, obtain the prediction results, and then enter the next iteration cycle;

[0035] When the number of iterations reaches the preset threshold, the iteration is stopped;

[0036] After obtaining a preset number of prediction results, perform big data statistics on the preset number of prediction results and output the remaining life prediction interval.

[0037] Specifically, the initial remaining life prediction model is trained in each iteration cycle. When the trained remaining life prediction model meets the preset training stop condition, the target remaining life prediction model corresponding to the current iteration cycle is obtained. The target remaining life prediction model corresponding to the current iteration cycle is then used to predict the test data to obtain a prediction result. Then, the initial remaining life prediction model is iteratively trained in the next iteration cycle. Because the training of the remaining life prediction model itself is random, the remaining life prediction model will be trained again after each prediction, and when the trained remaining life prediction model meets the training stop condition, the target remaining life prediction model corresponding to the next iteration cycle is obtained. After repeating the above operation for a preset number of times, a preset number of prediction results can be obtained. Then, big data statistics are performed on the preset number of prediction results, and finally the remaining life prediction interval can be output.

[0038] In an optional embodiment, the method further includes: outputting a confidence level corresponding to the remaining life prediction interval.

[0039] In an optional embodiment, the test data is capacity degradation data of the battery backup unit.

[0040] In an optional embodiment, the preset condition includes: performing constant current discharge on the battery backup unit to be identified according to a preset current value for a preset time period under the current state of charge of the battery backup unit to be identified.

[0041] In a third aspect, the present invention provides a device for constructing a BBU category identification model, the device comprising:

[0042] an acquisition module, configured to acquire a capacity change value and a voltage change value generated by each of the multiple battery backup units after the battery backup unit operates under multiple sets of different preset conditions;

[0043] a processing module, configured to input a capacity change value and a voltage change value generated by the first battery backup unit after operating under each set of preset conditions into a pre-built initial category identification model to obtain multiple sets of preset model parameter values, wherein the initial category identification model includes a capacity change parameter, a voltage change parameter, and a preset model parameter, and the first battery backup unit is any one of the multiple types of battery backup units;

[0044] The construction module is used to construct a category recognition model corresponding to the first type of battery backup unit according to multiple groups of preset model parameter values, capacity change parameters, voltage change parameters, and preset model parameters.

[0045] The present invention provides a BBU category identification model construction device with the following advantages:

[0046] Obtain the capacity change value and voltage change value generated by each of the multiple battery backup units after working under multiple sets of different preset conditions. After inputting the capacity change value and voltage change value generated by the first battery backup unit after working under each set of preset conditions into the pre-built category recognition model, multiple sets of preset model parameter values ​​can be obtained. According to the multiple sets of preset model parameter values, obtain the parameter values ​​corresponding to the final preset model parameters. Input the parameter values ​​corresponding to the final preset model parameters into the pre-built category recognition model to obtain a category recognition model in which only the capacity change parameter and the voltage change parameter are unknown. This model happens to be the category recognition model corresponding to the first battery backup unit. Similarly, the category recognition model corresponding to each of the multiple battery backup units can be obtained respectively through the above method. These category recognition models are used to identify the type of battery backup unit in the future.

[0047] In a fourth aspect, the present invention provides a BBU remaining life prediction device, the device comprising:

[0048] The first acquisition module is used to obtain an actual capacity change value and an actual voltage change value of the battery backup unit to be identified under preset conditions;

[0049] a processing module, configured to input the actual capacity change value into different category identification models constructed by the BBU category identification model construction method according to any one of claims 1 to 4, respectively, to obtain a reference voltage change value corresponding to each category identification model;

[0050] a determination module, configured to determine the type of the battery backup unit to be identified based on the actual voltage change value and the reference voltage change value corresponding to each category identification model;

[0051] A second acquisition module is used to acquire test data corresponding to the type of the battery backup unit to be identified;

[0052] The prediction module is used to input the test data into the target remaining life prediction model, and obtain the remaining life prediction interval of the battery backup unit after multiple predictions using the target remaining life prediction model.

[0053] The present invention provides a BBU remaining life prediction device with the following advantages:

[0054] The actual capacity change and voltage change values ​​generated by the battery backup unit to be identified under preset conditions are obtained. The actual capacity change values ​​are then input into different classification identification models constructed using the method of any embodiment of the first aspect, thereby obtaining reference voltage values ​​corresponding to each classification identification model. Based on a comparison of the actual voltage change and the reference voltage values, one of the multiple classification identification models is selected as a target classification identification model, and the type of battery backup unit corresponding to the target classification model is determined to be the type of the battery backup unit to be identified. Test data corresponding to the type of battery backup unit to be identified is then extracted and input into a target remaining life prediction model. Multiple predictions are performed using the target remaining life prediction model to obtain a predicted remaining life interval for the battery backup unit. Compared to a single remaining life value, a predicted remaining life interval provides greater flexibility and a wider range for battery maintenance or replacement. Users can replace and maintain batteries within this predicted remaining life interval, minimizing delays in replacement or maintenance and ensuring the safety of the battery backup unit during use. This also minimizes battery waste caused by premature replacement, reducing maintenance costs and the likelihood of fatal failures.

[0055] In a fifth aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to thereby execute the BBU category identification model construction method of the above-mentioned first aspect or any corresponding embodiment thereof; or, execute the BBU remaining life prediction method of the above-mentioned second aspect or any corresponding embodiment thereof.

[0056] In a sixth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to cause a computer to execute the BBU category identification model construction method of the above-mentioned first aspect or any corresponding embodiment thereof; or, to execute the BBU remaining life prediction method of the above-mentioned second aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0058] Figure 1 This is a flow chart of a method for constructing a BBU category identification model provided by an embodiment of the present invention;

[0059] Figure 2 This is a flowchart of another method for constructing a BBU category identification model provided by an embodiment of the present invention;

[0060] Figure 3 This is a flow chart of a BBU remaining life prediction method provided by an embodiment of the present invention;

[0061] Figure 4 1 is a flow chart of another BBU remaining life prediction method provided by an embodiment of the present invention;

[0062] Figure 5 (a) is a schematic diagram of the network structure of the AT-IndRNN provided by the present invention;

[0063] Figure 5 (b) is a schematic diagram of the network structure of the traditional RNN provided by the present invention;

[0064] Figure 6 This is a schematic diagram of the structure formed by expanding the AT-IndRNN according to time provided by the present invention;

[0065] Figure 7 This is a schematic diagram of the overall framework for obtaining a BBU remaining life prediction interval provided by an embodiment of the present invention;

[0066] Figure 8 Schematic diagram of the mapping relationship between the actual capacity prediction value and RUL provided by an embodiment of the present invention;

[0067] Figure 9 This is a schematic diagram of the overall implementation architecture including category recognition model selection and battery life prediction provided by an embodiment of the present invention;

[0068] Figure 10 This is a schematic diagram of the structure of a BBU category identification model building device provided by an embodiment of the present invention;

[0069] Figure 11 This is a schematic structural diagram of a BBU remaining life prediction device provided by an embodiment of the present invention;

[0070] Figure 12 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0071] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0072] Current storage systems in the industry typically use a PSU + BBU redundant power supply system. If the mains power in the data center fails, the storage system detects the abnormal PSU power supply and seamlessly switches to the BBU. The BBU provides continuous power, ensuring that data in the storage system controller's write cache is fully and securely written to non-volatile media such as HDDs and SSDs, preventing data loss.

[0073] The voltage of a single lithium-ion battery is relatively low, so in actual use, several cells must be connected in series to form a BBU battery pack that meets the corresponding voltage and power requirements. Due to differences in lithium-ion battery manufacturing processes and the influence of the operating environment, inconsistencies can occur during battery cycling. These inconsistencies primarily manifest as differences in characteristics such as terminal voltage, capacity, state of charge (SOC), internal resistance, self-discharge rate, and charge / discharge capacity among the individual cells within the pack. After several charge and discharge cycles, a lithium-ion battery pack will experience charge imbalance between individual cells. This unbalanced charge and discharge state ultimately leads to reduced capacity and service life. Furthermore, as the number of charge and discharge cycles increases, the performance of the lithium-ion battery, such as its available capacity, gradually decreases. When the capacity drops below the failure threshold, the battery fails. This severely impacts the lifespan and safety of the lithium-ion battery, which in turn leads to performance degradation or failure of the BBU, severely impacting the reliability and safety of the storage system.

[0074] Remaining Useful Life (RUL) prediction leverages known operating status information to predict battery capacity degradation trends in advance, thereby estimating the number of charge and discharge cycles required for the battery to reach its failure threshold. During battery cyclical use, improper charging and discharging methods or environmental influences can lead to corrosion of internal electrode materials, continuous electrolyte loss, and reduced internal activity. This reduces the battery's available capacity and gradually degrades battery performance, severely impacting battery life and safety. This can lead to performance degradation or even failure of the storage BBU. Accurate RUL prediction enables proactive maintenance or replacement of the storage BBU, reducing maintenance costs and minimizing the risk of fatal failures.

[0075] Because storage systems have varying power consumption and backup requirements, they require different types of BBUs, such as 4S3P (4 serial and 3 parallel), 4S2P (4 serial and 2 parallel), and 4S1P (4 serial and 1 parallel). To predict the lifespan of various BBUs, it's necessary to first identify the different types. However, currently, no effective method exists for identifying BBU types. Consequently, it's impossible to extract sample data and test data of the corresponding types to predict the remaining lifespan of a BBU.

[0076] To address the above-mentioned issues, embodiments of the present invention provide multiple embodiments for predicting the remaining life of a BBU. It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system (computer device) including, for example, a set of computer-executable instructions. Moreover, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown.

[0077] In this embodiment, a method for constructing a BBU category identification model is provided, which can be used for the above-mentioned terminal devices, such as mobile phones, tablet computers, etc. Figure 1 FIG. 1 is a flow chart of a method for constructing a BBU category identification model provided by an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0078] Step S101 : obtaining a capacity change value and a voltage change value generated by each of the multiple types of battery backup units after each type of battery backup unit operates under multiple sets of different preset conditions.

[0079] In an optional embodiment, obtaining a capacity change value and a voltage change value generated by each of the multiple battery backup units after operating under multiple sets of different preset conditions includes:

[0080] The capacity change value and voltage change value are obtained after each battery backup unit is discharged at a constant current according to a preset current value for a preset time period under different preset charge states.

[0081] For example, multiple sets of different preset conditions may include, for example, discharging at a constant current of 3A for 20 minutes at a battery state of charge (SOC) of 30%, 40%, 50%, 60%, 70%, 80%, 90%, and 100%. After different types of BBUs operate under these multiple sets of different preset conditions, the capacity change and voltage change values ​​generated under each set of preset conditions can be recorded. For example, discharging at a constant current of 3A for 20 minutes at an SOC of 30%. Discharging at a constant current of 3A for 20 minutes at an SOC of 40%, and so on. Discharging at a constant current of 3A for 20 minutes at an SOC of 100%. Each type of BBU needs to operate under any of the aforementioned sets of preset conditions and record the capacity change and voltage change values ​​generated.

[0082] Specifically, the discharge current and discharge time in the preset conditions are not clearly defined and can be determined based on actual conditions. However, the product of the discharge current and discharge time cannot exceed a certain percentage of the rated discharge capacity of the battery cell. For example, when the discharge capacity of a battery cell under normal operating conditions is between 30% and 80%, the product of the discharge current and discharge time cannot exceed 30% of the rated discharge capacity of the battery cell.

[0083] The specific reason is that in order to extend the service life of the battery cell, the battery cell power will not be discharged to 0. Therefore, the battery cell can be set to stop discharging after it reaches 30% of the rated discharge capacity, and then the battery cell can be charged to prevent over-discharge. Then, the capacity change value and voltage change value obtained at different states of charge are input into the pre-built initial category recognition model, and the preset model parameters at different states of charge can be fitted. Then, based on the parameter values ​​of the preset model parameters at different states of charge, the final parameter values ​​of the preset model parameters can be obtained, so that the parameter values ​​of the preset model parameters finally obtained are more representative, thereby making the final category recognition model more accurate in identifying the category of the battery backup unit.

[0084] In step S102, the capacity change value and voltage change value generated by the first type of battery backup unit operating under each set of preset conditions are input into a pre-built initial category recognition model to obtain multiple sets of preset model parameter values.

[0085] Step S103 , constructing a category recognition model corresponding to the first type of battery backup unit according to multiple sets of preset model parameter values, capacity change parameters, voltage change parameters, and preset model parameters.

[0086] Specifically, in an optional embodiment, the initial category identification model includes a capacity change parameter, a voltage change parameter, and a preset model parameter, and the first type of battery backup unit is any one of a plurality of battery backup units.

[0087] Therefore, by replacing each set of capacity change values ​​and voltage change values ​​with the corresponding capacity change parameters and voltage change parameters in the initial class recognition model, the parameter values ​​of the preset model parameters can be determined. In a similar manner, multiple sets of preset model parameter values ​​can be obtained.

[0088] Finally, a category identification model corresponding to the first type of battery backup unit is constructed based on multiple sets of preset model parameter values, capacity variation parameters, voltage variation parameters, and preset model parameters. In this category identification model, the values ​​of the capacity variation parameters and the voltage variation parameters can be obtained based on actual conditions. The values ​​of the preset model parameters can be ultimately determined based on the multiple sets of preset model parameter values.

[0089] In an optional manner, for example, the average value of multiple sets of preset model parameter values ​​may be selected as the final preset model parameter value, or the average value may be selected as the final preset model parameter value after removing the maximum and minimum values. The specific manner of obtaining the preset model parameter value is not limited here.

[0090] This embodiment provides a method for constructing a BBU category identification model. The method obtains capacity change and voltage change values ​​generated by each of multiple types of battery backup units (BBUs) operating under multiple sets of preset conditions. The capacity change and voltage change values ​​generated by the first type of BBU operating under each set of preset conditions are input into a pre-constructed category identification model to obtain multiple sets of preset model parameter values. Based on these multiple sets of preset model parameter values, parameter values ​​corresponding to the final preset model parameters are obtained. The parameter values ​​corresponding to the final preset model parameters are input into the pre-constructed category identification model to obtain a category identification model in which only the capacity change and voltage change parameters are unknown. This model is the category identification model corresponding to the first type of BBU. Similarly, the above method can be used to obtain category identification models corresponding to each of the multiple types of BBUs. These category identification models can then be used to identify the type of BBU.

[0091] In this embodiment, a method for constructing a BBU category identification model is provided, which can be used in the above-mentioned mobile terminals, such as mobile phones, tablet computers, etc. Figure 2 FIG. 1 is a flow chart of another method for constructing a BBU category identification model provided by an embodiment of the present invention. Figure 2As shown, the process includes the following steps:

[0092] Step S201 : obtaining a capacity change value and a voltage change value generated by each of the multiple types of battery backup units after each type of battery backup unit operates under multiple sets of different preset conditions.

[0093] The specific implementation process of step S201 has been introduced in the previous text, so it will not be repeated here.

[0094] In step S202, the capacity change value and voltage change value generated by the first type of battery backup unit operating under each set of preset conditions are input into a pre-built initial category recognition model to obtain multiple sets of preset model parameter values.

[0095] In an optional embodiment, the initial category recognition model is expressed using the following formula:

[0096] ΔQ = a1·jexp(a2) (Formula 1)

[0097] Wherein, ΔQ is a capacity change parameter, j is a voltage change parameter, a1 is a first preset model parameter, and a2 is a second preset model parameter.

[0098] By substituting the capacity and voltage changes generated by the first type of battery backup unit operating under each set of preset conditions as known parameters into Formula 1, we can obtain multiple sets of relationship equations for a1 and a2. These two relationship equations can then be used to determine the actual values ​​of a1 and a2, respectively. By randomly combining different relationship equations, we can obtain multiple sets of a1 and a2 values.

[0099] Step S203 , constructing a category recognition model corresponding to the first type of battery backup unit according to multiple sets of preset model parameter values, capacity change parameters, voltage change parameters, and preset model parameters.

[0100] In an optional embodiment, each of the multiple sets of preset model parameter values ​​includes a parameter value of the first preset model parameter and a parameter value of the second preset model parameter. The specific implementation process of step S203 can be implemented as follows, see the following method steps:

[0101] Step S2031 : acquiring a first average parameter value corresponding to the first preset model parameter according to the parameter values ​​of the plurality of groups of first preset model parameters.

[0102] Step S2032: Obtain a second average parameter value corresponding to the second preset model parameter according to the parameter values ​​of the plurality of groups of second preset model parameters.

[0103] Specifically, in an optional embodiment, the first average parameter value may be the average of all parameter values ​​in multiple sets of first preset model parameters, or the average value obtained after removing the maximum and minimum values. Similarly, the second average parameter value may be the average of all parameter values ​​in multiple sets of second preset model parameters, or the average value obtained after removing the maximum and minimum values.

[0104] Step S2033 , constructing a category recognition model corresponding to the first type of battery backup unit according to the first average parameter value, the second average parameter value, the capacity variation parameter, and the voltage variation parameter.

[0105] The average value can better reflect the universality and versatility of a parameter. Therefore, based on multiple sets of preset model parameter values, a first average parameter value corresponding to the first preset model parameter and a second average parameter value corresponding to the second preset model parameter are obtained. Then, based on the first and second average parameter values, the capacity variation parameter, and the voltage variation parameter, a category recognition model corresponding to the first type of battery backup unit is constructed. This category recognition model will also provide more accurate type recognition.

[0106] In an optional embodiment, in order to facilitate the subsequent use of the corresponding type of BBU category identification model, after constructing the category identification model corresponding to each type of BBU respectively, the method may also include configuring identification information for each category identification model to facilitate the subsequent selection of the corresponding category identification model for use based on the identification information.

[0107] In this embodiment, a BBU remaining life prediction method is provided, which can be used in the above-mentioned mobile terminals, such as mobile phones, tablet computers, etc. Figure 3 FIG. 1 is a flow chart of a BBU remaining life prediction method provided by an embodiment of the present invention, such as Figure 3 As shown, the process includes the following steps:

[0108] Step S301 : obtaining an actual capacity change value and an actual voltage change value generated by a battery backup unit to be identified under preset conditions.

[0109] The specific implementation process is described in the previous article. It is the same or similar to the acquisition method in the previous article, so I will not go into details here.

[0110] In an optional embodiment, the preset condition here is slightly different from the preset condition in the aforementioned embodiment. The preset condition in this embodiment may be: performing a constant current discharge on the BBU to be identified at a preset current value for a preset period of time under the current state of charge of the BBU to be identified. For example, a constant current of 3A may be discharged for 20 minutes under the current state of charge of the BBU to be identified.

[0111] In step S302 , the actual capacity change value is input into different category identification models constructed by the BBU category identification model construction method described in any of the above embodiments, and a reference voltage change value corresponding to each category identification model is obtained.

[0112] Specifically, the class identification model has been introduced in any of the aforementioned embodiments. For example, the class identification model is represented by ΔQ = a1·jexp(a2) in Formula 1. Here, a1 and a2 are already determined values. ΔQ is the data calculated in step S301. Then, using Formula 1, j, the reference voltage change value, can be derived.

[0113] Step S303 : determining the type of the battery backup unit to be identified according to the actual voltage change value and the reference voltage change value corresponding to each type identification model.

[0114] Specifically, after obtaining the actual voltage change value, it can be compared with the reference voltage value corresponding to each category identification model respectively, and then the closest reference voltage change value can be determined, and the type of battery backup unit corresponding to the closest reference voltage change value can be determined as the type of battery backup unit corresponding to the actual voltage change value.

[0115] Step S304: Acquire sample test data corresponding to the type of the battery backup unit to be identified.

[0116] Specifically, test data corresponding to the type of the battery backup unit to be identified is extracted from a test database.

[0117] Step S305 : inputting the test data into the target remaining life prediction model, performing multiple predictions using the target remaining life prediction model, and obtaining a remaining life prediction interval of the battery backup unit.

[0118] Specifically, the target remaining life prediction model is used to predict the same set of test data. Because each prediction carries a certain degree of randomness, multiple prediction results can be obtained after multiple predictions. Ultimately, the remaining life prediction interval of the battery backup unit can be determined based on these multiple prediction results.

[0119] An embodiment of the present invention provides a BBU remaining life prediction method. The method obtains the actual capacity change and voltage change values ​​generated by a battery backup unit (BBU) to be identified under preset conditions. The actual capacity change values ​​are then input into different category recognition models constructed using the method of any embodiment of the first aspect, respectively, to obtain reference voltage values ​​corresponding to each category recognition model. Based on a comparison between the actual voltage change and the reference voltage values, one of the multiple category recognition models is selected as a target category recognition model, and the BBU type corresponding to the target recognition model is determined to be the type of the BBU to be identified. Test data corresponding to the BBU type to be identified is then extracted and input into a target remaining life prediction model. Multiple predictions are performed using the target remaining life prediction model to obtain a predicted remaining life interval for the BBU. Compared to a single remaining life value, a predicted remaining life interval provides greater flexibility and a wider range of coverage for battery maintenance or replacement. Users can replace or maintain batteries within this predicted remaining life interval, minimizing delays in replacement or maintenance and ensuring the safety of the BBU during use. It can also avoid as much as possible the waste of batteries caused by early replacement, reduce maintenance costs and reduce the chance of fatal failures.

[0120] In this embodiment, a BBU remaining life prediction method is provided, which can be used in the above-mentioned mobile terminals, such as mobile phones, tablet computers, etc. Figure 4 FIG. 1 is a flow chart of another BBU remaining life prediction method provided by an embodiment of the present invention. Figure 4 As shown, the process includes the following steps:

[0121] Step S401 : obtaining an actual capacity change value and an actual voltage change value generated by a battery backup unit to be identified under preset conditions.

[0122] In step S402 , the actual capacity change value is input into different category identification models constructed by the BBU category identification model construction method described in any of the above embodiments, and a reference voltage change value corresponding to each category identification model is obtained.

[0123] The specific implementation processes of step S401 and step S402 have been described in detail in the previous embodiment, specifically referring to step S301 and step S302 , and therefore will not be described again here.

[0124] Step S403 : determining the type of the battery backup unit to be identified according to the actual voltage change value and the reference voltage change value corresponding to each type identification model.

[0125] In an optional implementation, step S403 may be implemented as follows:

[0126] Step S4031 : Compare the actual voltage change value with the reference voltage change value generated by each category recognition model.

[0127] Step S4032: When it is determined that the difference between the actual voltage change value and the reference voltage change value generated by the first type identification model is the smallest, the type of the battery backup unit identified by the first type identification model is determined to be the type corresponding to the battery backup unit to be identified.

[0128] The actual voltage change value is compared with the reference voltage change value generated by each classification recognition model, and the classification recognition model corresponding to the reference voltage change value that is closest to the actual voltage change value is selected as the target recognition model to identify the type of the battery backup unit to be identified. This method determines the type of the backup unit to be identified, which is closer to the actual type of the battery backup unit to be identified.

[0129] In an optional embodiment, in addition to determining the difference between the actual voltage change value and the reference voltage change value generated by the first category recognition model, the similarity between the actual voltage change value and the reference voltage change value can also be determined. If the actual voltage change value has the highest similarity to reference voltage value A, this also indicates that the type of battery backup unit identified by the first category recognition model corresponding to reference voltage value A is closer to the type of battery backup unit to be identified.

[0130] Step S404: Acquire test data corresponding to the type of the battery backup unit to be identified.

[0131] Step S405 : inputting the test data into the target remaining life prediction model, performing multiple predictions using the target remaining life prediction model, and obtaining a remaining life prediction interval of the battery backup unit.

[0132] In an optional implementation, step S405 may be implemented as follows:

[0133] Step a1: within the current iteration period, iteratively train the initial remaining life prediction model. When the trained remaining life prediction model meets the preset training stop condition, obtain the target remaining life prediction model corresponding to the current iteration period.

[0134] Specifically, the data-driven method is a commonly used method for predicting battery life. The data-driven method does not consider the electrochemical mechanism inside the lithium-ion battery, but directly mines the implicit battery health status information and its evolution law from the battery capacity degradation data or status monitoring data to achieve battery RUL prediction. It has high dynamic accuracy and good universality. For example, the Long Short-term Memory Recurrent Neural Network (LSTM RNN) captures the long-term dependency of battery capacity attenuation, trains the network based on online capacity data, and recursively obtains the long-term capacity degradation trend of the battery. The network can also be trained with offline data to achieve early RUL prediction of the battery.

[0135] However, RNNs generally suffer from the vanishing and exploding gradient problems, making it difficult to maintain long-term learning patterns. While LSTM RNNs address these issues to some extent, they utilize tanh or sigmoid saturation activation functions, and the lack of independence between neurons in traditional RNNs or LSTM RNNs. When stacked in multiple layers, inter-layer gradients decay significantly, resulting in insufficient long-term learning performance.

[0136] Therefore, in the embodiment of the present invention, the remaining life prediction model adopts an adaptive independent recurrent neural network (Self-adaptiveIndependently RNN, abbreviated as AT-IndRNN) model.

[0137] (1) The structure of AT-IndRNN is introduced below.

[0138] The network structure of AT-IndRNN and traditional RNN is as follows Figure 5 (a) and 5(b). The hidden layer neurons of the traditional RNN are interconnected. Each neuron at the current moment is connected to the output of all neurons at the previous moment. In other words, these neurons are not independent. For details, see Figure 5 (a). Compared with traditional RNN, AT-IndRNN learns long-term memory of time series in a new way. During training, each neuron in the hidden layer only receives input information and its own hidden state. The neurons in each layer are independent of each other, which makes it easy to explain the behavior of neurons in each layer of AT-IndRNN. For details, see Figure 5 (b) shown.

[0139] The structure of AT-IndRNN after expanding by time is as follows Figure 6As shown in the figure. "W" represents the input processing of each step with Relu as the activation function, and "Recurrent+ReLU" represents the cyclic process with Relu as the activation function. Compared with the sigmoid or hyperbolic tangent function-based structures of RNN and LSTMRNN, Relu is a non-saturated activation function with a gradient of 0 or 1, which avoids the gradient vanishing problem on the layer. The expression of the Relu activation function is as follows:

[0140]

[0141] In addition, Batch Normalization (BN) is used before and after the activation function to fix the mean and variance of the input signal at each layer. This normalization effect of BN increases the response or gradient of deeper layers in the network model, thereby solving the "vanishing gradient" problem that is common in deep network training.

[0142] By superimposing this basic structure, a deep AT-IndRNN network can be constructed. The AT-IndRNN network can be defined as follows:

[0143] h t =σ(Wx t +L⊙h t-1 +b) (Formula 3)

[0144] in and h t ∈R η are the input and hidden layer states of AT-IndRNN at time t, h t-1 ∈R η is the input of AT-IndRNN at time t-1, R represents the matrix, express Column-row matrix, R η represents an η-column-row matrix, express A matrix of rows and columns, and and L∈R η Then they are the weights in the corresponding states, b∈R η is the bias of the neuron, ⊙ represents the Hadamard product, σ is the activation function of the neuron, and η are the length of the input and the number of neurons in the current layer, respectively. Since the neurons in each layer of the AT-IndRNN are independent of each other, its recurrent input is a vector rather than a matrix. Specifically, for the nth neuron at time t, its hidden state can be obtained as follows:

[0145]

[0146] Among them L n and h n,t are the weight and hidden state of the nth neuron in this layer respectively.

[0147] (2) Training of AT-IndRNN

[0148] AT-IndRNN uses time-dependent gradient backpropagation in each layer. Since neurons in each layer are independent of each other, the gradient of each neuron in AT-IndRNN can be calculated independently. k Time objective function J k Minimize, where the objective function J k As follows:

[0149]

[0150] in, For the battery at t k The remaining lifespan predicted at any moment, and t k Then, for the nth neuron, propagate backward to time step t k-l The gradient is:

[0151]

[0152] in, is the derivative of the activation function. As can be seen from the above formula, the gradient of AT-IndRNN only involves an easily adjustable scalar value L n The exponential term, and the gradient of the activation function are bounded within a certain range, compared to the gradient of RNN:

[0153]

[0154] in, is the Jacobian matrix of the activation function. The gradient of AT-IndRNN mainly depends on the loop weight L n The value of . The gradient of RNN mainly depends on the matrix product. Therefore, the training of AT-IndRNN is more robust than that of traditional RNN.

[0155] In order to solve the problem of gradient explosion and disappearance over time, for AT-IndRNN, it is necessary to adjust the exponential term In an appropriate range. The activation function of AT-IndRNN uses the ReLU function, whose gradient is 0 or 1, so the gradient product operation will not cause the gradient to explode or disappear. In order to maintain the long-term memory of AT-IndRNN, in particular, the time step tk-l can still effectively affect the time step t after a long time interval.k state, that is, in order to maintain t k -t k-l time steps of memory, then for the nth neuron, the recurrent weight L n The range can be set to:

[0156]

[0157] Where ε represents the minimum effective gradient. In other words, in order to avoid the gradient vanishing of the neuron, the above constraints should be met. To avoid the gradient explosion problem, it is necessary to further limit the range to:

[0158]

[0159] Among them, γ is the maximum gradient value when gradient explosion does not occur.

[0160] After performing the above operations, when the trained remaining life prediction model meets the constraints of formula 8 or 9, it can be determined that the trained remaining life prediction model meets the preset training stop conditions, and the target remaining life prediction model corresponding to the current iteration cycle is obtained.

[0161] Step a2: input the test data into the target remaining life prediction model corresponding to the current iteration cycle, obtain the prediction results, and enter the next iteration cycle.

[0162] Step a3: When the number of iterations reaches a preset threshold, the iteration is stopped.

[0163] Specifically, after each training of the AT-IndRNN model, the target RLU prediction model corresponding to the current iteration cycle is obtained. The test data is then input into the target RLU prediction model corresponding to the current iteration cycle to obtain the prediction results. Then, the next iteration cycle begins, and the initial RLU prediction model corresponding to the next iteration cycle is iteratively trained again. When the iterative training meets the preset training stop conditions, the target RLU prediction model corresponding to the next iteration cycle is obtained. The prediction data is then input into the target RLU prediction model corresponding to the next iteration cycle to obtain the prediction results.

[0164] After repeating the above steps a preset number of times, a preset number of preset results can be obtained. For example, repeat the above steps 30 times to obtain 30 prediction results, and then stop the iteration operation.

[0165] In one optional embodiment, if the current iteration cycle is the first iteration cycle, the initial RLS prediction model is the original RLS prediction model initially constructed. If the current iteration cycle is not the first iteration cycle, the initial RLS prediction model may also be the target RLS prediction model corresponding to the previous iteration cycle. This is equivalent to performing further iterative optimization based on the optimized RLS prediction model obtained in each iterative training. This ensures that each target RLS prediction model obtained produces a more accurate prediction result than the previous one.

[0166] Step a4: After obtaining a preset number of prediction results, perform big data statistics on the preset number of prediction results and output the remaining life prediction interval.

[0167] The predicted battery lifespan can be output through big data statistics. In practice, probabilistic statistics can be used to obtain the final predicted battery lifespan. This range is typically relatively narrow. For example, the remaining lifespan prediction range could be between 80 and 82 charge and discharge cycles, or between 80 and 90 cycles.

[0168] By obtaining a prediction interval, compared to a single numerical value, battery maintenance or replacement can be more flexible and calibrated. By replacing or maintaining batteries within this predicted remaining lifespan, users can minimize delays in replacement or maintenance, ensuring the safety of the battery backup unit during use. This also minimizes battery waste caused by premature replacement, reduces maintenance costs, and reduces the chance of fatal failures.

[0169] For example, if the remaining life of the battery is 85 charge and discharge cycles, and the actual value obtained by using the existing technology is 80 times. The remaining life prediction interval obtained by using the method of the embodiment of the present invention is 80-85 times. Assume that the user will choose a compromise data, for example, choose to replace the battery at the 83rd time. Compared with the existing technology, the probability of battery waste caused by early replacement of the battery may be reduced as much as possible. Or, for example, if the remaining life of the battery is 80 charge and discharge cycles, and the actual value obtained by using the existing technology is 100 times. The user maintains the battery at the 82nd time, compared with the existing technology, the implementation method of the present invention can also reduce the hysteresis of replacing or maintaining the battery as much as possible.

[0170] In an optional embodiment, to increase the credibility of the remaining life prediction interval obtained in the embodiments of the present invention, the present invention further includes outputting a confidence level corresponding to the remaining life prediction interval. In a specific example, the confidence level can be 95% or 97.5%. That is, the battery life prediction results within this confidence level are more reliable.

[0171] In an optional embodiment, the test data is capacity degradation data of the battery backup unit.

[0172] Specifically, before executing the above operations, it is also possible to perform a degradation and aging test on various types of BBUs in advance to obtain a BBU capacity degradation data set, and pre-process the data to eliminate abnormal capacity data.

[0173] Then, based on the BBU category recognition model, the BBU category to be predicted and the capacity degradation training data and test data of the category are screened out.

[0174] The training data is used to train the AT-IndRNN model. The test data, the same set of test data, is fed into the AT-IndRNN model after each training session to generate multiple predictions. Ultimately, the remaining life prediction interval and the corresponding confidence level are obtained.

[0175] For details, see Figure 7 shown. Figure 7 The overall framework diagram for obtaining the remaining life prediction interval is shown. Figure 7 The input is a capacity degradation dataset corresponding to a specific type of BBU. After data screening and preprocessing, training and test datasets are generated. These datasets are normalized, and the AT-IndRNN network model is subsequently trained and tested. Future capacity predictions are obtained. The predicted capacity values ​​are denormalized to obtain the actual capacity predictions. Finally, the corresponding RUL predictions are derived based on the actual capacity predictions. The RUL prediction interval and the corresponding confidence level are then determined based on these multiple predictions.

[0176] In an optional embodiment, see Figure 8 As shown, Figure 8 A schematic diagram illustrating the mapping relationship between the actual capacity prediction value and the RUL is shown in FIG. Figure 8 The vertical axis on the left is the actual capacity forecast value. Figure 8The horizontal axis in the middle represents the RUL prediction value. Therefore, a similar mapping relationship can be used to determine the RUL prediction result corresponding to the actual capacity prediction value. Finally, the RUL prediction interval and the corresponding confidence level are obtained based on the 30 prediction results. Figure 8 It also includes information such as degradation state identification and performance prediction, but other contents are not important to explain in the embodiment of the present invention, so they will not be introduced here in detail.

[0177] Figure 9 Schematic diagram of the overall implementation architecture corresponding to the aforementioned embodiment. Among them, the judgment factor is the actual voltage change value introduced above. The BBU analog recognition model is determined by the actual voltage change value and the reference voltage change value. Then the BBU data set to be predicted is selected. Figure 9 The diagram shows three types: 4S3P (4 series and 3 parallel), 4S2P (4 series and 2 parallel), and 4S1P (4 series and 1 parallel).

[0178] The dataset (training data and test data) is then fed into the AT-IndRNN model for prediction, ultimately yielding the predicted space for the BBU's remaining lifetime and the corresponding confidence level. The specific implementation process has been detailed previously, so I won't go into further detail here.

[0179] This embodiment also provides a BBU category identification model construction device and a BBU remaining life prediction device. These devices are used to implement the above-mentioned embodiments and preferred implementations, and details already described are omitted. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.

[0180] This embodiment provides a BBU category identification model construction device, such as Figure 10 As shown, it includes: an acquisition module 1001, a processing module 1002 and a construction module 1003.

[0181] An acquisition module 1001 is configured to acquire a capacity change value and a voltage change value generated by each of the multiple types of battery backup units after each type of battery backup unit operates under multiple sets of different preset conditions;

[0182] a processing module 1002 configured to input a capacity change value and a voltage change value generated by the first type of battery backup unit after operating under each set of preset conditions into a pre-built initial category identification model to obtain multiple sets of preset model parameter values, wherein the initial category identification model includes a capacity change parameter, a voltage change parameter, and a preset model parameter, and the first type of battery backup unit is any one of the multiple types of battery backup units;

[0183] The construction module 1003 is configured to construct a category recognition model corresponding to the first type of battery backup unit according to multiple sets of preset model parameter values, capacity change parameters, voltage change parameters, and preset model parameters.

[0184] In an optional embodiment, the acquisition module 1001 is specifically configured to obtain a capacity change value and a voltage change value after performing constant current discharge on each battery backup unit at a preset current value for a preset period of time under different preset states of charge. In an optional embodiment, each of the multiple sets of preset model parameter values ​​includes a parameter value of a first preset model parameter and a parameter value of a second preset model parameter. The acquisition module 1001 is specifically configured to obtain a first average parameter value corresponding to the first preset model parameter based on the multiple sets of parameter values ​​of the first preset model parameter; and obtain a second average parameter value corresponding to the second preset model parameter based on the multiple sets of parameter values ​​of the second preset model parameter.

[0185] A category identification model corresponding to the first type of battery backup unit is constructed according to the first average parameter value, the second average parameter value, the capacity variation parameter, and the voltage variation parameter.

[0186] In an optional embodiment, the initial category recognition model is expressed using the following formula:

[0187] ΔQ=a1·jexp(a2)

[0188] Wherein, ΔQ is a capacity change parameter, j is a voltage change parameter, a1 is a first preset model parameter, and a2 is a second preset model parameter.

[0189] The BBU remaining life prediction device in this embodiment is presented in the form of a functional module, where the module refers to an application-specific integrated circuit (ASIC), a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0190] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0191] An embodiment of the present invention provides a BBU category identification model construction device that obtains the capacity change value and voltage change value generated by each of multiple battery backup units after operating under multiple sets of different preset conditions. After inputting the capacity change value and voltage change value generated by the first type of battery backup unit operating under each set of preset conditions into a pre-constructed category identification model, multiple sets of preset model parameter values ​​can be obtained. Based on the multiple sets of preset model parameter values, parameter values ​​corresponding to the final preset model parameters are obtained. The parameter values ​​corresponding to the final preset model parameters are input into the pre-constructed category identification model to obtain a category identification model in which only the capacity change parameter and the voltage change parameter are unknown. This model is exactly the category identification model corresponding to the first type of battery backup unit. Similarly, the above method can be used to obtain a category identification model corresponding to each of the multiple types of battery backup units. These category identification models can then be used to identify the type of battery backup unit.

[0192] The embodiment of the present invention provides a schematic diagram of the structure of a BBU remaining life prediction device, such as Figure 11 As shown, it includes: a first acquisition module 1101 , a processing module 1102 , a determination module 1103 , a second acquisition module 1104 , and a prediction module 1105 .

[0193] The first acquisition module 1101 is configured to acquire an actual capacity change value and an actual voltage change value generated by a battery backup unit to be identified under preset conditions;

[0194] A processing module 1102 is configured to input the actual capacity change value into different category identification models constructed by the BBU category identification model construction method according to any one of claims 1 to 4, and obtain a reference voltage change value corresponding to each category identification model;

[0195] a determination module 1103 for determining the type of the battery backup unit to be identified based on the actual voltage change value and the reference voltage change value corresponding to each category identification model;

[0196] A second acquisition module 1104 is configured to acquire test data corresponding to the type of the battery backup unit to be identified;

[0197] The prediction module 1105 is configured to input the test data into the target remaining life prediction model, perform multiple predictions using the target remaining life prediction model, and obtain a predicted remaining life interval of the battery backup unit.

[0198] In an optional embodiment, the determination module 1103 is specifically configured to compare the actual voltage change value with the reference voltage change value generated by each category recognition model;

[0199] When it is determined that the difference between the actual voltage change value and the reference voltage change value generated by the first category identification model is the smallest, it is determined that the type of the battery backup unit identified by the first category identification model is the type corresponding to the battery backup unit to be identified.

[0200] In an optional embodiment, the prediction module 1105 is specifically configured to iteratively train the initial remaining life prediction model within a current iteration period, and when the trained remaining life prediction model meets a preset training stop condition, obtain a target remaining life prediction model corresponding to the current iteration period;

[0201] Input the test data into the target remaining life prediction model corresponding to the current iteration cycle, obtain the prediction results, and then enter the next iteration cycle;

[0202] When the number of iterations reaches the preset threshold, the iteration is stopped;

[0203] After obtaining a preset number of prediction results, perform big data statistics on the preset number of prediction results and output the remaining life prediction interval.

[0204] In an optional embodiment, the prediction module 1105 is further configured to output a confidence level corresponding to the remaining life prediction interval.

[0205] In an optional embodiment, the test data is capacity degradation data of the battery backup unit.

[0206] In an optional embodiment, the preset conditions include:

[0207] Under the current state of charge of the battery backup unit to be identified, the battery backup unit to be identified is subjected to constant current discharge for a preset time period according to a preset current value.

[0208] The BBU remaining life prediction device in this embodiment is presented in the form of a functional module, where the module refers to an application-specific integrated circuit (ASIC), a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0209] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0210] An embodiment of the present invention provides a BBU remaining life prediction device that obtains the actual capacity change and actual voltage change values ​​generated by a battery backup unit (BBU) to be identified under preset conditions. The actual capacity change values ​​are then input into different category recognition models constructed using the method of any embodiment of the first aspect, respectively, to obtain reference voltage values ​​corresponding to each category recognition model. Based on a comparison of the actual voltage change and the reference voltage values, one of the multiple category recognition models is selected as a target category recognition model, and the BBU type corresponding to the target recognition model is determined to be the type of the BBU to be identified. Test data corresponding to the BBU type to be identified is then extracted and input into a target remaining life prediction model. Multiple predictions are performed using the target remaining life prediction model to obtain a predicted remaining life interval for the BBU. Compared to a single remaining life value, a predicted remaining life interval provides greater flexibility and a wider range of coverage for battery maintenance or replacement. Users can replace or maintain batteries within this predicted remaining life interval, minimizing delays in replacement or maintenance and ensuring the safety of the BBU during use. It can also avoid as much as possible the waste of batteries caused by early replacement, reduce maintenance costs and reduce the chance of fatal failures.

[0211] The embodiment of the present invention also provides a computer device having the above Figure 10 The BBU category identification model building device shown, and Figure 11 The BBU remaining life prediction device shown.

[0212] See also Figure 12 , Figure 12 is a structural diagram of a computer device provided by an optional embodiment of the present invention, such as Figure 12 As shown, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in the memory or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 12 A processor 10 is taken as an example.

[0213] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0214] The memory 20 stores instructions that can be executed by at least one processor 10, so as to enable at least one processor 10 to execute the method shown in the above embodiment.

[0215] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created based on the use of a computer device for displaying a small program landing page, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0216] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0217] The computer device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 12 The bus connection is taken as an example.

[0218] The input device 30 can receive input digital or character information and generate key signal input related to user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a trackpad, a touch pad, an indicator stick, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 can include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor). The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display, and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0219] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0220] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for constructing a BBU category identification model, characterized in that: The method comprises: Obtaining a capacity change value and a voltage change value generated by each of the multiple battery backup units after the battery backup unit operates under multiple sets of different preset conditions; Inputting the capacity change value and the voltage change value generated after the first type of battery backup unit operates under each set of preset conditions into a pre-built initial category identification model to obtain multiple sets of preset model parameter values, wherein the initial category identification model includes a capacity change parameter, a voltage change parameter, and a preset model parameter, and the first type of battery backup unit is any one of the multiple types of battery backup units; Constructing a category identification model corresponding to the first type of battery backup unit according to the multiple sets of preset model parameter values, the capacity change parameter, the voltage change parameter, and the preset model parameter; The step of obtaining a capacity change value and a voltage change value generated by each of the multiple battery backup units after the battery backup units respectively operate under multiple sets of different preset conditions includes: The capacity change value and voltage change value are obtained after each of the battery backup units is discharged at a constant current according to a preset current value for a preset period of time under different preset states of charge.

2. The method according to claim 1, characterized in that Each of the multiple sets of preset model parameter values ​​includes a parameter value of a first preset model parameter and a parameter value of a second preset model parameter. Constructing a category identification model corresponding to the first type of battery backup unit based on the multiple sets of preset model parameter values, the capacity change parameter, the voltage change parameter, and the preset model parameters includes: Obtaining, based on the parameter values ​​of the plurality of sets of first preset model parameters, a first average parameter value corresponding to the first preset model parameter; and obtaining, based on the parameter values ​​of the plurality of sets of second preset model parameters, a second average parameter value corresponding to the second preset model parameter; A category identification model corresponding to the first type of battery backup unit is constructed according to the first average parameter value, the second average parameter value, the capacity change parameter, and the voltage change parameter.

3. The method according to claim 2, characterized in that The initial category recognition model is expressed by the following formula: in, is the capacity change parameter, is the voltage variation parameter, is the first preset model parameter, is the second preset model parameter.

4. A BBU remaining life prediction method, characterized in that: The method comprises: Obtaining an actual capacity change value and an actual voltage change value generated by the battery backup unit to be identified under preset conditions, wherein the preset conditions include: performing constant current discharge on the battery backup unit to be identified at a preset current value for a preset time period under a current state of charge of the battery backup unit to be identified; Inputting the actual capacity change value into different category identification models constructed by the BBU category identification model construction method according to any one of claims 1 to 3, respectively, to obtain a reference voltage change value corresponding to each category identification model; Determining the type of the battery backup unit to be identified according to the actual voltage change value and the reference voltage change value corresponding to each category identification model; Acquiring test data corresponding to the type of the battery backup unit to be identified; The test data is input into a target remaining life prediction model, and after multiple predictions are performed using the target remaining life prediction model, a remaining life prediction interval of the battery backup unit is obtained.

5. The method according to claim 4, characterized in that The determining the type of the battery backup unit to be identified according to the actual voltage change value and the reference voltage change value corresponding to each category identification model includes: comparing the actual voltage change value with a reference voltage change value generated by each category recognition model respectively; When it is determined that the difference between the actual voltage change value and the reference voltage change value generated by the first category identification model is the smallest, it is determined that the type of the battery backup unit identified by the first category identification model is the type corresponding to the battery backup unit to be identified.

6. The method according to claim 4 or 5, characterized in that The step of inputting the test data into a target remaining life prediction model and performing multiple predictions using the target remaining life prediction model to obtain a remaining life prediction interval for the battery backup unit includes: Iteratively training the initial remaining life prediction model within the current iteration period, and when the trained remaining life prediction model meets the preset training stop condition, obtaining the target remaining life prediction model corresponding to the current iteration period; Inputting the test data into the target remaining life prediction model corresponding to the current iteration cycle, obtaining the prediction results, and entering the next iteration cycle; When the number of iterations reaches the preset threshold, the iteration is stopped; After obtaining a preset number of prediction results, big data statistics are performed on the preset number of prediction results to output the remaining life prediction interval.

7. The method according to claim 6, characterized in that The method further comprises: Output the confidence level corresponding to the remaining life prediction interval.

8. A BBU category identification model construction device, characterized in that: The device comprises: an acquisition module, configured to acquire a capacity change value and a voltage change value generated by each of the multiple battery backup units after the battery backup unit operates under multiple sets of different preset conditions; a processing module, configured to input the capacity change value and the voltage change value generated by the first type of battery backup unit after operating under each set of preset conditions into a pre-built initial category identification model to obtain multiple sets of preset model parameter values, wherein the initial category identification model includes a capacity change parameter, a voltage change parameter, and a preset model parameter, and the first type of battery backup unit is any one of the multiple types of battery backup units; a construction module, configured to construct a category identification model corresponding to the first type of battery backup unit based on the multiple sets of preset model parameter values, the capacity change parameter, the voltage change parameter, and the preset model parameter; The acquisition module is specifically configured to acquire the capacity change value and the voltage change value after performing constant current discharge on each of the battery backup units according to a preset current value for a preset period of time under different preset states of charge.

9. A BBU remaining life prediction device, characterized in that: The device comprises: A first acquisition module is configured to obtain an actual capacity change value and an actual voltage change value of the battery backup unit to be identified under preset conditions, wherein the preset conditions include: performing constant current discharge on the battery backup unit to be identified according to a preset current value for a preset time period under a current state of charge of the battery backup unit to be identified; a processing module, configured to input the actual capacity change value into different category identification models constructed by the BBU category identification model construction method according to any one of claims 1 to 3, respectively, to obtain a reference voltage change value corresponding to each category identification model; a determination module, configured to determine the type of the battery backup unit to be identified based on the actual voltage change value and the reference voltage change value corresponding to each category identification model; A second acquisition module is used to acquire test data corresponding to the type of the battery backup unit to be identified; The prediction module is used to input the test data into a target remaining life prediction model, and obtain a remaining life prediction interval of the battery backup unit after performing multiple predictions using the target remaining life prediction model.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the BBU category identification model construction method according to any one of claims 1 to 3; or executes the BBU remaining life prediction method according to any one of claims 4 to 7.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, which are used to enable a computer to execute the BBU category identification model construction method described in any one of claims 1 to 3; or to execute the BBU remaining life prediction method described in any one of claims 4 to 7.

Images