Method for predicting state of health of battery and related apparatus

Abstract

The present disclosure relates to a method for predicting the state of health of a battery and a related apparatus. The method comprises: acquiring source domain data comprising voltage-time curves of reference batteries in full life cycles and target domain data comprising a voltage-time curve of a target battery early in its life; identifying feature voltage change ranges; constructing source domain samples, of which sample inputs comprise the feature values and the states of health of each reference battery in first multiple feature voltage change ranges previous to each feature voltage change range, and sample outputs comprise the states of health of the reference battery in second multiple feature voltage change ranges subsequent to the feature voltage change range; on the basis of the source domain samples, constructing a metric function for measuring sample input similarities; constructing target domain samples in a manner similar to that of constructing the source domain samples; and using the similarities between sample inputs of the target domain samples and the sample inputs of the source domain samples to weight the sample outputs of the source domain samples, so as to synthesize predicted values of the state of health of the target battery.

Description

Method and related apparatus for predicting battery health status

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application is based on the application with CN application number 202310699711.1 and application date June 13, 2023, and claims its priority. The disclosed content of the CN application is hereby introduced as a whole into this application. Technical Field

[0003] The present disclosure generally relates to the field of battery operation and maintenance technology, and more particularly to a method and related apparatus for predicting the health status of a battery. Background Art

[0004] The change pattern of the battery's state of health (SOH) over time, that is, the battery's aging pattern, can be used to assess the battery's remaining life. Therefore, battery health prediction is one of the core issues in the field of battery operation and maintenance technology.

[0005] Summary of the Invention

[0006] According to a first aspect of the present disclosure, a method for predicting the health state of a battery is provided, comprising: acquiring source domain battery data and target domain battery data, the source domain battery data comprising a voltage variation curve over time of each reference battery in a plurality of reference batteries under a corresponding one of a plurality of reference conditions over the entire life cycle, and the target domain battery data comprising a voltage variation curve over time of the target battery in an early stage of its life under actual conditions; identifying a common characteristic voltage variation interval of the plurality of reference batteries based on the voltage variation curves over time of the plurality of reference batteries, the voltage variation pattern of each reference battery in the characteristic voltage variation interval appearing periodically in the voltage variation curve of the reference battery over time; constructing a source domain sample set, comprising constructing a source domain sample for each characteristic voltage variation interval appearing in the voltage variation curve of each reference battery, the sample input of the source domain sample comprising the voltage variation pattern of the reference battery in the characteristic voltage variation interval The method comprises the following steps: constructing a target domain sample for each characteristic voltage change interval appearing in the voltage-time change curve of the target battery, constructing a target domain sample for each characteristic voltage change interval appearing in the voltage-time change curve of the target battery, and constructing a target domain sample for each characteristic voltage change interval appearing in the voltage-time change curve of the target battery, and constructing a target domain sample for each characteristic voltage change interval appearing in the voltage-time change curve of the target battery, and constructing a target domain sample for each characteristic voltage change interval appearing in the characteristic voltage change interval appearing in the characteristic voltage change interval appearing in the target domain sample, and constructing a target domain sample for each ... sample, and constructing a target domain sample for each characteristic voltage change interval appearing in the target domain sample sample sample, and constructing a target domain sample for each characteristic voltage change interval appearing in the target domain sample sample sample sample, and constructing a target domain sample for each characteristic voltage change interval appearing in the target domain sample sample sample sample, and constructing a target domain sample for each characteristic voltage change interval appearing in the target domain sample sample sample sample, and constructing a target domain sample for each characteristic voltage change interval appearing in the target domain sample sample sample sample, and constructing a target domain sample for each characteristic voltage change interval appearing

[0007] According to a second aspect of the present disclosure, there is provided an apparatus for predicting the health state of a battery, comprising: a battery data acquisition module configured to acquire source domain battery data and target domain battery data, wherein the source domain battery data comprises a voltage variation curve over time of each reference battery in a corresponding one of a plurality of reference conditions over the entire life cycle, and the target domain battery data comprises a voltage variation curve over time of the target battery in an early stage of its life under actual conditions; a characteristic interval identification module configured to identify a common characteristic voltage variation interval of the plurality of reference batteries based on the voltage variation curves over time of the plurality of reference batteries, wherein the voltage variation pattern of each reference battery in the characteristic voltage variation interval appears periodically in the voltage variation curve over time of the reference battery; and a source domain sample construction module configured to construct a source domain sample set, comprising constructing a source domain sample for each characteristic voltage variation interval appearing in the voltage variation curve over time of each reference battery, wherein the sample input of the source domain sample comprises the voltage variation interval of the reference battery in the characteristic voltage variation interval. The characteristic values ​​and health status of the first multiple characteristic voltage change intervals before, the characteristic values ​​are related to the health status, and the sample output of the source domain sample includes the health status of the reference battery in the second multiple characteristic voltage change intervals after the characteristic voltage change interval; a metric function construction module, configured to construct a metric function based on the source domain sample set, and the metric function is used to measure the similarity between sample inputs; a target domain sample construction module, configured to construct a target domain sample for each characteristic voltage change interval appearing in the voltage change curve of the target battery over time, and the sample input of the target domain sample includes the characteristic values ​​and health status of the target battery in the first multiple characteristic voltage change intervals before the characteristic voltage change interval; and a health status prediction module, configured to use the similarity between the sample input of the target domain sample determined based on the metric function and the sample input of the source domain sample in the source domain sample set as a weight to weight the sample output of the source domain sample in the source domain sample set, thereby synthesizing the sample output of the target domain sample as the predicted value of the health status of the target battery.

[0008] According to a third aspect of the present disclosure, a computing device for predicting the health state of a battery is provided, comprising: one or more processors; and a memory storing computer-executable instructions, wherein the computer-executable instructions, when executed by the one or more processors, enable the one or more processors to execute the method for predicting the health state of a battery according to any embodiment of the first aspect of the present disclosure.

[0009] According to a fourth aspect of the present disclosure, a non-volatile storage medium having computer-executable instructions stored thereon is provided, wherein when the computer-executable instructions are executed by a computer, the computer executes the method for predicting the health status of a battery according to any embodiment of the first aspect of the present disclosure.

[0010] Other features and advantages of the present disclosure will become more apparent from the following detailed description of exemplary embodiments of the present disclosure with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The foregoing and other features and advantages of the present disclosure will become apparent from the following description of the embodiments of the present disclosure taken in conjunction with the accompanying drawings, which are incorporated herein and form a part of the specification and serve to further explain the principles of the present disclosure and enable those skilled in the art to make and use the present disclosure.

[0012] Figure 1 schematically depicts the battery SOH variation curves over time under different usage conditions;

[0013] FIG2 schematically depicts a battery SOH variation curve over time over the entire life cycle in a laboratory test scenario and a battery SOH variation curve over time in the early life cycle in a real prediction scenario;

[0014] FIG3 shows a flowchart of a method for predicting the health status of a battery according to some embodiments of the present disclosure;

[0015] FIG4 shows a flowchart of a method for predicting a battery's state of health according to other embodiments of the present disclosure;

[0016] FIG5 schematically depicts the battery voltage variation curves over time under different usage conditions;

[0017] FIG6 shows a schematic block diagram of an apparatus for predicting a battery state of health according to some embodiments of the present disclosure;

[0018] FIG7 shows a schematic block diagram of a computing device for predicting the health status of a battery according to some embodiments of the present disclosure.

[0019] Note that in the embodiments described below, the same reference numerals are sometimes used in common across different drawings to denote the same parts or parts having the same functions, and their repeated descriptions are omitted. In some cases, similar reference numerals and letters are used to denote similar items, so once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0020] For ease of understanding, the positions, sizes, and ranges of various structures shown in the drawings and the like may not represent actual positions, sizes, and ranges, etc. Therefore, the present disclosure is not limited to the positions, sizes, and ranges disclosed in the drawings and the like. DETAILED DESCRIPTION

[0021] Various exemplary embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. It should be noted that unless otherwise specifically stated, the relative arrangements of components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present disclosure.

[0022] The following description of at least one exemplary embodiment is merely illustrative and is not intended to limit the present disclosure, its application, or use. In other words, the structures and methods herein are presented in an exemplary manner to illustrate various embodiments of the structures and methods of the present disclosure. However, those skilled in the art will appreciate that these are merely exemplary of the disclosure that may be implemented, and are not exhaustive. Furthermore, the drawings are not necessarily drawn to scale, and some features may be exaggerated to illustrate details of specific components.

[0023] In addition, technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0024] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0025] The aging pattern of a battery is closely related to the battery's usage (i.e., how the battery is used, including charging mode, discharging mode, or workload mode, etc.). Even the same battery will experience different aging patterns under different usage conditions, as shown in Figure 1. Therefore, a battery health status prediction model trained from battery data under a given usage condition (e.g., a full charge and discharge test in a laboratory) may be difficult to adapt to predicting how the battery's health status changes over time under new usage conditions. In addition, when it is actually necessary to predict the battery's health status, there is often only a small amount of early-life data for the battery, unlike when testing the battery in the laboratory where full life cycle data can be obtained, as shown in Figure 2. Therefore, in real prediction scenarios, the scarcity of battery data samples may make it difficult to accurately predict the battery's health status.

[0026] The present disclosure provides a method for predicting the health status of a battery. The method can accurately predict the long-term changes of the health status of the battery over time in the future by searching for batteries in a source domain that have similar early aging patterns to those of batteries in a target domain, across various usage conditions, using a small amount of early-life data of the battery in the target domain, thereby obtaining the full life cycle aging pattern of the battery and realizing the prediction of the calendar life of the battery, which is more intuitive and simple than the prediction of the cycle life.

[0027] FIG3 illustrates a method 100 for predicting the state of health of a battery according to some embodiments of the present disclosure. Method 100 includes, at step S102, obtaining source domain battery data and target domain battery data. The source domain battery data includes a voltage-over-time curve for each of a plurality of reference batteries under a corresponding reference condition from among multiple reference use cases over the entire life cycle. The target domain battery data includes a voltage-over-time curve for the target battery at an early stage of its life under actual use cases. In this context, a curve is not limited to data in the form of a graph; it may also be data in other formats, such as text. For example, it may be text-converted into an array, matrix, or set, as long as the data can be presented as a curve. Both the reference battery and the target battery may be, for example, rechargeable batteries such as lithium batteries. The multiple reference use cases may include various known use cases, such as backup batteries in data centers that require a full discharge and recharge at regular intervals, and power batteries in new energy electric vehicles that require incomplete charging every few days based on the owner's driving habits. The actual use case of the target battery does not necessarily need to be one of the aforementioned known use cases; it may also be a new use case. The target battery may have been used for a period of time, so the battery data (such as the change of voltage over time, the change of SOH over time, etc.) in the early stage of its life is known.

[0028] Method 100 includes: at step S104, identifying a common characteristic voltage variation interval for the multiple reference batteries based on the voltage variation curves over time for the multiple reference batteries, wherein the voltage variation pattern of each reference battery within the characteristic voltage variation interval periodically appears in the voltage variation curve of the reference battery over time. For example, referring to FIG5 , the reference battery under use case 1 undergoes a complete charge and discharge at regular intervals, while the reference battery under use case 2 undergoes an incomplete charge and discharge at regular intervals. It can be noted that although the voltage variation curves for use case 1 and use case 2 appear very different, they each vary from voltage v1 to voltage v2 at regular intervals, and therefore share a common, periodically occurring characteristic voltage variation interval [v1, v2]. Although the duration and / or occurrence period of the characteristic voltage variation interval [v1, v2] may not be the same for use case 1 and use case 2, given that the characteristic voltage variation interval [v1, v2] occurs periodically under all use cases, the aging pattern of the battery over time can be directly determined by determining the aging pattern of the battery as a function of the number of occurrences of the characteristic voltage variation interval [v1, v2]. In some embodiments, the voltage variation pattern of each reference battery within the characteristic voltage variation interval may be monotonic. In this article, a "monotonic" voltage variation pattern may refer to a voltage that continues to increase over time, or a voltage that continues to decrease over time. When the voltage increases over time, the voltage variation pattern may be considered to be in a charging stage. When the voltage decreases over time, the voltage variation pattern may be considered to be in a discharging stage. In some embodiments, the voltage variation pattern of each reference battery within the characteristic voltage variation interval may be in a charging stage. In some embodiments, the voltage variation pattern of each reference battery within the characteristic voltage variation interval may be in a discharging stage. In some examples, the common characteristic voltage variation interval of the multiple reference batteries may be identified by the following operations: let B be the set of the multiple reference batteries, b be the reference battery in B, V b (t) is the voltage variation curve of reference battery b over time, and the characteristic voltage variation range [v1, v2] is: v1, So that for any b∈B and a non-negative integer n, there exists time t1, time t2, period T, satisfying: t1<t2 and V b (t1+nT)=v1,V b (t2+nT)=v2 and for time t′1, t′2∈[t1+nT, t2+nT], t′1<t′2, there is V b (t′1)<V b(t′2). In this way, the characteristic voltage variation interval can be selected so that the voltage variation pattern of each reference battery within the characteristic voltage variation interval is monotonic and in the charging stage. This may be preferable in some cases because a monotonic voltage variation pattern facilitates finding the periodic law in the battery voltage variation curve over time, and the charging stage is often more stable and controllable than the discharging stage. Assume T b N is the period of the characteristic voltage change interval in the voltage change curve of reference battery b over time, b is the number of times the characteristic voltage variation interval appears in the voltage variation curve of reference battery b over time. As mentioned above, for different reference batteries b, T b 、N b Can be different.

[0029] Next, method 100 includes: at step S106, constructing a source domain sample set. A source domain sample can be constructed for each characteristic voltage variation interval that appears in the voltage-over-time variation curve of each reference battery. The sample input of the source domain sample includes the characteristic values ​​and health status of the reference battery for the first multiple characteristic voltage variation intervals before the characteristic voltage variation interval, and the sample output of the source domain sample includes the health status of the reference battery for the second multiple characteristic voltage variation intervals after the characteristic voltage variation interval. Feature values ​​related to the health status can be extracted from the characteristic voltage variation interval using any suitable method currently known or later developed. The feature value can be any suitable variable that can indicate a change in the health status, such as, but not limited to, a change in charge value per unit voltage variation range. The battery health status can be characterized using any suitable method currently known or later developed, such as, but not limited to, the ratio of battery capacity to rated capacity, the ratio of battery internal resistance to rated internal resistance, etc.

[0030] In some embodiments, let B be the set of the plurality of reference batteries, b be a reference battery in B, and the characteristic voltage variation interval i of the reference battery b has a characteristic value and health status Then construct the source domain sample set The method comprises: constructing a source domain sample for each characteristic voltage change interval appearing in the voltage-time change curve of each reference battery b, wherein the sample input of the source domain sample comprises the characteristic value and health status of the reference battery in the r characteristic voltage change intervals before the characteristic voltage change interval, and the sample output of the source domain sample comprises the health status of the reference battery in the (R×s) characteristic voltage change intervals after the characteristic voltage change interval, so that the source domain sample of the kth characteristic voltage change interval of the reference battery b comprises:

[0031] Sample input and

[0032] Sample Output Wherein r, R, s, and k are positive integers.

[0033] In some examples, r can be determined based on the number of occurrences of a characteristic voltage variation interval in the voltage-over-time curve of the target battery, and (R×s) can be determined based on a preset time-domain scaling ratio of the sample output relative to the sample input. For example, for a lithium battery with a full cycle life of 1000-3000 cycles, r=100, R=100, and s=10 can be set. In this way, the characteristic values ​​and health status of 100 consecutive characteristic voltage variation intervals in the early life can be used to predict the long-term health status changes of the next 1000 characteristic voltage variation intervals.

[0034] In some embodiments, a source domain sample set is constructed It may also include: performing time domain scaling on the source domain samples according to different period scaling ratios to obtain source domain samples enhanced according to different period scaling ratios, wherein the source domain samples enhanced according to the period scaling ratio l of the kth characteristic voltage variation interval of the reference battery b include

[0035] Sample input and

[0036] Sample Output Where l is a positive integer. For example, for a lithium battery with a full cycle life of 1000-3000 cycles, l can be set to 1, 2, or 3. In some examples, the cycle scaling ratio l can be determined based on the ratio of the period of the characteristic voltage change interval in the voltage change curve of the target battery to the period of the characteristic voltage change interval in the voltage change curve of the reference battery. In some examples, the cycle scaling ratio l can be determined based on the ratio of the aging rate of the target battery to the aging rate of the reference battery. By reasonably setting the value of l, it is suitable to transfer the knowledge of the source domain battery data to the target battery with different aging rates.

[0037] 3, the method 100 further includes: at step S108, constructing a metric function based on the source domain sample set, wherein the metric function is used to measure the similarity between sample inputs. In some embodiments, the source domain sample set can be used to construct a metric function based on the source domain sample set. Construct the metric function f θ , making

[0038] in, Represents sample input The similarities between Indicates that when the sample input is When the sample input is The probability of Indicates that when the sample input is When the sample input is The probability of In some embodiments, it is possible to further calculate the expected value by using a source domain sample set. Create a training dataset to train the neural network model to solve the metric function f θ The parameter θ, the loss function L of the neural network model θ For example, it can be constructed as

[0039] in, is the source domain sample set In the randomly sampled batch data, MSE is used to find the mean square error, and the metric function f θ The parameter θ is determined so that the loss function L θ Minimize the parameter θ. For example, the loss function L θ The metric function f θ The parameter θ is determined as Although such a metric function f θ Although trained using data from a source domain sample set, it can be used to measure the similarity of sample inputs across domains (e.g., between the source domain and the target domain). The identification of characteristic voltage variation intervals in step S104 ensures comparability between sample inputs across domains. The construction of source domain samples in various ways in step S106 increases the diversity of source domain sample inputs, thereby improving the generalization ability of the metric function.

[0040] Still referring to FIG3 , method 100 further includes: at step S110, constructing a target domain sample. Similar to constructing a source domain sample, a target domain sample can be constructed for each characteristic voltage variation interval that appears in the voltage variation curve of the target battery over time. The sample input of the target domain sample includes the characteristic values ​​and health status of the target battery in the first multiple characteristic voltage variation intervals before the characteristic voltage variation interval. In some embodiments, let c be the target battery (T c is the period of the characteristic voltage change interval in the voltage variation curve of the target battery c over time, N c is the number of times the characteristic voltage variation interval appears in the voltage variation curve of the target battery c over time), and the characteristic voltage variation interval j of the target battery c has a characteristic value and health status Then constructing the target domain sample may include: constructing a target domain sample for each characteristic voltage change interval that appears in the voltage change curve of the target battery c over time, wherein the sample input of the target domain sample includes the characteristic value and health status of the target battery in the r characteristic voltage change intervals before the characteristic voltage change interval, so that the target domain sample of the g-th characteristic voltage change interval of the target battery c includes the sample input Where g is a positive integer.

[0041] The method 100 further includes: at step S112, using the similarity between the sample input of the target domain sample determined based on the metric function and the sample input of the source domain sample in the source domain sample set as a weight to weight the sample output of the source domain sample in the source domain sample set, thereby synthesizing the sample output of the target domain sample as a predicted value of the health state of the target battery. In some examples, the future health state of the target battery can be determined as in are respectively the predicted health status values ​​of the target battery when the characteristic voltage variation interval occurs for the (k+s), (k+2s), ..., (k+R×s)th time. In some embodiments, the method 100 further includes: at step S120, the predicted health status value of the target battery (e.g. ) to interpolate, thereby predicting the health status of the target battery over time under actual usage conditions throughout its life cycle.

[0042] FIG4 illustrates a method 100′ for predicting the state of health of a battery according to some other embodiments. Method 100′ differs from method 100 in that step S112 is replaced by steps S114, S116, and S118. Specifically, method 100′ includes: at step S114, performing time domain scaling on target domain samples according to different period scaling ratios to obtain target domain samples enhanced according to the different period scaling ratios; at step S116, determining the degree of match between the target domain samples enhanced according to the different period scaling ratios and source domain samples in a source domain sample set based on a metric function, and determining the target domain sample with the highest degree of match as the target domain sample enhanced according to the optimal period scaling ratio; and at step S118, using the similarity between the sample input of the target domain sample enhanced according to the optimal period scaling ratio and the sample input of the source domain samples in the source domain sample set determined based on the metric function as a weight to weight the sample outputs of the source domain samples in the source domain sample set, thereby synthesizing the sample output of the target domain sample enhanced according to the optimal period scaling ratio as the predicted value of the state of health of the target battery.

[0043] Specifically, in some embodiments, the target domain sample of the target battery c after the g-th characteristic voltage variation interval is enhanced by the period scaling ratio q includes the sample input Where q is a positive integer. The cycle scaling ratio q can be set independently of the cycle scaling ratio l. In some embodiments, only the cycle scaling ratio q can be set without setting the cycle scaling ratio l. In some embodiments, only the cycle scaling ratio l can be set without setting the cycle scaling ratio q. In some embodiments, both the cycle scaling ratio q and the cycle scaling ratio l can be set. By coordinating the cycle scaling ratio q with the cycle scaling ratio l, it is possible to facilitate the migration of source domain battery data knowledge to the target domain when the aging rate of the reference battery is a non-integer multiple of the aging rate of the target battery.

[0044] In some embodiments, based on the metric function f θ Determine the target domain samples and source domain sample sets after enhancement by different period scaling ratios The matching degree between the source domain samples in may include: quantifying the matching degree as The optimal cycle scaling ratio can be determined as The sample input of the target domain sample after enhancement by the optimal period scaling ratio can be expressed as In some embodiments, the future state of health of the target battery may be determined as:

[0045] The target battery is The predicted value of the health state when the characteristic voltage variation interval occurs for the first time. Further, the method 100' may also include: at step S120, the predicted value of the health state of the target battery (for example, ) to interpolate the target battery's state of health over time under actual usage and throughout its lifecycle. This interpolated state of health over time curve can be used to conveniently assess the target battery's remaining calendar life. For example, the time until its state of health drops to a certain value (e.g., 80% of the rated value) can be determined. This calendar life provides a more intuitive concept of life compared to cycle life.

[0046] With reference to FIG6 , the present disclosure further provides an apparatus 200 for predicting the state of health of a battery. Apparatus 200 includes a battery data acquisition module 202, a characteristic interval identification module 204, a source domain sample construction module 206, a metric function construction module 208, a target domain sample construction module 210, and a state of health prediction module 212. Battery data acquisition module 202 is configured to acquire source domain battery data and target domain battery data. The source domain battery data includes a voltage variation curve over time for each of a plurality of reference batteries under a corresponding one of a plurality of reference usage conditions over the entire life cycle. The target domain battery data includes a voltage variation curve over time for the target battery under actual usage conditions at an early stage of its life. Characteristic interval identification module 204 is configured to identify a common characteristic voltage variation interval for the plurality of reference batteries based on the voltage variation curves over time for the plurality of reference batteries. The voltage variation pattern of each reference battery within the characteristic voltage variation interval periodically appears in the voltage variation curve of the reference battery. The source domain sample construction module 206 is configured to construct a source domain sample set, including constructing a source domain sample for each characteristic voltage variation interval that appears in the voltage variation curve of each reference battery over time. The sample input of the source domain sample includes the characteristic value and health status of the reference battery in the first multiple characteristic voltage variation intervals before the characteristic voltage variation interval, the characteristic value is related to the health status, and the sample output of the source domain sample includes the health status of the reference battery in the second multiple characteristic voltage variation intervals after the characteristic voltage variation interval. The metric function construction module 208 is configured to construct a metric function based on the source domain sample set, the metric function being used to measure the similarity between the sample inputs. The target domain sample construction module 210 is configured to construct a target domain sample for each characteristic voltage variation interval that appears in the voltage variation curve of the target battery over time. The sample input of the target domain sample includes the characteristic value and health status of the target battery in the first multiple characteristic voltage variation intervals before the characteristic voltage variation interval. The health status prediction module 212 is configured to weight the sample outputs of the source domain samples in the source domain sample set using the similarity between the sample inputs of the target domain samples determined based on the metric function and the sample inputs of the source domain samples in the source domain sample set as a weight, thereby synthesizing the sample outputs of the target domain samples as the predicted value of the health status of the target battery.

[0047] In some embodiments, the target domain sample construction module 210 can also be configured to perform time domain scaling on the target domain samples according to different period scaling ratios to obtain target domain samples enhanced according to different period scaling ratios. The health status prediction module 212 can also be configured to: determine the degree of matching between the target domain samples enhanced according to different period scaling ratios and the source domain samples in the source domain sample set based on a metric function, and determine the one with the highest matching degree as the target domain sample enhanced according to the optimal period scaling ratio; and use the similarity between the sample input of the target domain sample enhanced according to the optimal period scaling ratio determined based on the metric function and the sample input of the source domain sample in the source domain sample set as a weight to weight the sample output of the source domain sample in the source domain sample set, thereby synthesizing the sample output of the target domain sample enhanced according to the optimal period scaling ratio as the predicted value of the health status of the target battery.

[0048] The embodiment of the device 200 is basically similar to the aforementioned method embodiment, so it will not be described again here. For relevant details, please refer to the description of the aforementioned method embodiment.

[0049] The present disclosure also provides a computing device for predicting the state of health of a battery, which may include one or more processors and a memory storing computer-executable instructions. When executed by the one or more processors, the computer-executable instructions cause the one or more processors to perform the method for predicting the state of health of a battery as described in any of the aforementioned embodiments of the present disclosure. As shown in FIG7 , computing device 300 may include one or more processors 302 and a memory 304 storing computer-executable instructions. When executed by the one or more processors 302, the computer-executable instructions cause the one or more processors 302 to perform the method for predicting the state of health of a battery as described in any of the aforementioned embodiments of the present disclosure. The one or more processors 302 may, for example, be a central processing unit (CPU) of computing device 300. The one or more processors 302 may be any type of general-purpose processor, or may be a processor specifically designed for predicting the state of health of a battery, such as an application-specific integrated circuit ("ASIC"). The memory 304 may include various computer-readable media accessible by the one or more processors 302. In various embodiments, the memory 304 described herein may include volatile and non-volatile media, and removable and non-removable media. For example, the memory 304 may include any combination of the following: random access memory ("RAM"), dynamic RAM ("DRAM"), static RAM ("SRAM"), read-only memory ("ROM"), flash memory, cache memory, and / or any other type of non-transitory computer-readable medium. The memory 304 may store instructions that, when executed by the processor 302, cause the processor 302 to perform the method for predicting the state of health of a battery according to any of the aforementioned embodiments of the present disclosure.

[0050] The present disclosure also provides a non-transitory storage medium having computer-executable instructions stored thereon, which, when executed by a computer, causes the computer to perform the method for predicting the health status of a battery according to any of the aforementioned embodiments of the present disclosure.

[0051] The present disclosure also provides a computer program product, which may include instructions that, when executed by a processor, may implement the method for predicting the state of health of a battery according to any of the aforementioned embodiments of the present disclosure. The instructions may be any set of instructions to be executed directly by one or more processors, such as machine code, or any set of instructions to be executed indirectly, such as a script. The instructions may be stored in an object code format for direct processing by one or more processors, or in any other computer language, including a script or collection of independent source code modules that are interpreted on demand or compiled in advance.

[0052] The words "left," "right," "front," "back," "top," "bottom," "up," "down," "high," "low," and the like, if any, in the specification and claims, are used for descriptive purposes and are not necessarily intended to describe invariant relative positions. It should be understood that the words so used are interchangeable under appropriate circumstances such that the embodiments of the present disclosure described herein, for example, are capable of operation in orientations other than those illustrated or otherwise described herein. For example, features previously described as "above" other features could be described as "below" the other features when the apparatus in the accompanying drawings is turned over. The apparatus can also be otherwise oriented (rotated 90 degrees or in other orientations) and relative spatial relationships would be interpreted accordingly.

[0053] In the specification and claims, when an element is referred to as being "on," "attached" to, "connected" to, "coupled" to, "coupled to," or "in contact with," the element may be directly on, directly attached to, directly connected to, directly coupled to, directly coupled to, or directly in contact with, another element, or one or more intervening elements may be present. In contrast, when an element is referred to as being "directly" "on," "directly attached" to, "directly connected" to, "directly coupled" to, "directly coupled to," or "in direct contact with," another element, there may be no intervening elements. In the specification and claims, when a feature is arranged "adjacent" to another feature, it may mean that the feature has a portion that overlaps with the adjacent feature or a portion that is located above or below the adjacent feature.

[0054] As used herein, the word "exemplary" means "serving as an example, instance, or illustration," rather than as a "model" to be precisely copied. Any implementation described as exemplary is not necessarily to be construed as preferred or advantageous over other implementations. Furthermore, this disclosure is not to be bound by any expressed or implied theory presented in the technical field, background, summary, or detailed description.

[0055] As used herein, the term "substantially" is intended to encompass any minor variations due to design or manufacturing imperfections, device or component tolerances, environmental influences, and / or other factors. The term "substantially" also allows for deviations from a perfect or ideal condition due to parasitic effects, noise, and other practical considerations that may be present in actual implementations.

[0056] Additionally, terms such as "first," "second," and the like may also be used herein for reference purposes only and are not intended to be limiting. For example, the terms "first," "second," and other numerical terms referring to structures or elements do not imply a sequence or order unless the context clearly indicates otherwise.

[0057] It should also be understood that when the term “include / comprises” is used in this document, it indicates the presence of the specified features, integers, steps, operations, units and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, units and / or components and / or their combinations.

[0058] In this disclosure, the term "provide" is used in a broad sense to cover all ways of obtaining an object, and thus "providing an object" includes but is not limited to "purchasing", "preparing / manufacturing", "arranging / setting up", "installing / assembling", and / or "ordering" an object, etc.

[0059] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the present disclosure. As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

[0060] Those skilled in the art will appreciate that the boundaries between the above-mentioned operations are merely illustrative. Multiple operations can be combined into a single operation, a single operation can be distributed among additional operations, and operations can be performed at least partially overlapping in time. Moreover, alternative embodiments can include multiple instances of a particular operation, and the order of operations can be changed in various other embodiments. However, other modifications, variations, and replacements are also possible. Aspects and elements of all embodiments disclosed above can be combined in any manner and / or in combination with aspects or elements of other embodiments to provide multiple additional embodiments. Therefore, this specification and the accompanying drawings should be considered illustrative, not restrictive.

[0061] Although some specific embodiments of the present disclosure have been described in detail by way of examples, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present disclosure. The various embodiments disclosed herein may be combined in any manner without departing from the spirit and scope of the present disclosure. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A method for predicting the health status of a battery, include: Acquire source domain battery data and target domain battery data, wherein the source domain battery data includes a voltage variation curve over time of each reference battery in a plurality of reference batteries under a corresponding one of a plurality of reference conditions over the entire life cycle, and the target domain battery data includes a voltage variation curve over time of the target battery in an early life under an actual condition; Identifying a common characteristic voltage variation interval of the plurality of reference batteries based on the voltage variation curves of the plurality of reference batteries over time, wherein a voltage variation pattern of each reference battery within the characteristic voltage variation interval periodically appears in the voltage variation curve of the reference battery over time; Constructing a source domain sample set, including constructing a source domain sample for each characteristic voltage change interval appearing in the voltage-time change curve of each reference battery, the sample input of the source domain sample including the characteristic value and health status of the reference battery in the first multiple characteristic voltage change intervals before the characteristic voltage change interval, the characteristic value is related to the health status, and the sample output of the source domain sample including the health status of the reference battery in the second multiple characteristic voltage change intervals after the characteristic voltage change interval; Constructing a metric function based on a source domain sample set, wherein the metric function is used to measure the similarity between sample inputs; For each characteristic voltage variation interval that appears in the voltage variation curve of the target battery over time, a target domain sample is constructed, wherein the sample input of the target domain sample includes the characteristic values ​​and health status of the target battery in the first plurality of characteristic voltage variation intervals before the characteristic voltage variation interval; as well as The similarity between the sample input of the target domain sample determined based on the metric function and the sample input of the source domain sample in the source domain sample set is used as a weight to weight the sample output of the source domain sample in the source domain sample set, thereby synthesizing the sample output of the target domain sample as a predicted value of the health state of the target battery.

2. The method according to claim 1, further comprising: include: The target domain samples are respectively time-scaled according to different period scaling ratios to obtain target domain samples enhanced according to different period scaling ratios; Determine the matching degree between the target domain samples enhanced by different periodic scaling ratios and the source domain samples in the source domain sample set based on the metric function, and determine the one with the highest matching degree as the target domain sample enhanced by the optimal periodic scaling ratio; as well as The similarity between the sample input of the target domain sample enhanced by the optimal period scaling ratio determined by the metric function and the sample input of the source domain sample in the source domain sample set is used as a weight to evaluate the source domain samples in the source domain sample set. The sample outputs are weighted to synthesize the sample outputs of the target domain samples enhanced by the optimal cycle scaling ratio as the predicted value of the health state of the target battery.

3. The method according to claim 1 or 2, in, The voltage variation pattern of each reference battery within the characteristic voltage variation interval is monotonic and is in the charging stage.

4. The method according to claim 3, in, The common characteristic voltage variation interval of the plurality of reference batteries is identified by the following operations: Let B be the set of the plurality of reference batteries, b be the reference battery in B, V b (t) is the voltage variation curve of reference battery b over time, then the characteristic voltage variation interval [v 1 ,v 2 ]for: So that for any b∈B and a non-negative integer n, Existence time t 1 , time t 2 , period T, satisfies: t 1 <t 2 and V b (t 1 +nT) = v 1 ,V b (t 2 +nT) = v 2 and For time t′ 1 ,t′ 2 ∈[t 1 +nT,t 2 +nT],t′ 1 <t′ 2 , with V b (t′ 1 ) <V b (t′ 2 ).

5. The method according to claim 1 or 2, in, The characteristic value is the change in charge per unit voltage change range.

6. The method according to claim 2, in, Let B be the set of the plurality of reference batteries, b be a reference battery in B, and the characteristic voltage variation interval i of reference battery b has a characteristic value and health status Then construct the source domain sample set include: For each characteristic voltage variation interval that appears in the voltage variation curve of each reference battery b over time, a source domain sample is constructed, the sample input of the source domain sample includes the characteristic value and health status of the reference battery in the r characteristic voltage variation intervals before the characteristic voltage variation interval, and the sample output of the source domain sample includes the health status of the reference battery in the (R×s) characteristic voltage variation intervals after the characteristic voltage variation interval, so that the source domain sample of the kth characteristic voltage variation interval of the reference battery b includes Sample Input and Sample Output Wherein r, R, s, and k are positive integers.

7. The method according to claim 6, in, r is determined based on the number of times the characteristic voltage variation interval appears in the voltage variation curve of the target battery over time, and (R×s) is determined based on a preset time domain scaling ratio of the sample output relative to the sample input.

8. The method according to claim 6, in, Construct source domain sample set The invention also includes: performing time domain scaling on the source domain samples according to different period scaling ratios to obtain source domain samples enhanced according to different period scaling ratios, wherein the source domain samples enhanced according to the period scaling ratio l of the kth characteristic voltage variation interval of the reference battery b include Sample Input and Sample Output Wherein l is a positive integer.

9. The method according to claim 8, in, Determine the cycle scaling ratio l based on the ratio of the period in which the characteristic voltage variation interval appears in the voltage variation time curve of the target battery to the period in which the characteristic voltage variation interval appears in the voltage variation time curve of the reference battery; or The cycle scaling ratio l is determined based on the ratio of the aging rate of the target battery relative to the aging rate of the reference battery.

10. The method according to claim 8, in, Based on the source domain sample collection Construct the metric function f θ , so that in, Represents sample input The similarities between It means that when the sample input is When the sample input is The probability of It means that when the sample input is When the sample input is The probability of Used to seek expectations.

11. The method according to claim 10, in, Based on the source domain sample collection Create a training dataset to train the neural network model to solve the metric function f θ The parameter θ, the loss function L of the neural network model θ is constructed as in, is the source domain sample set The batch data randomly sampled in, MSE is used to find the mean square error, And wherein, by minimizing the loss function L θ The metric function f θ The parameter θ is determined as 12. The method according to claim 10, in, Let c be the target battery. The characteristic voltage variation interval j of the target battery c has the characteristic value and health status Then constructing the target domain sample includes: For each characteristic voltage change interval that appears in the voltage change curve of the target battery c over time, a target domain sample is constructed. The sample input of the target domain sample includes the characteristic value and health status of the target battery in the r characteristic voltage change intervals before the characteristic voltage change interval, so that the target domain sample of the g-th characteristic voltage change interval of the target battery c includes Sample Input And wherein, the target domain samples enhanced by the period scaling ratio q in the g-th characteristic voltage variation interval of the target battery c include Sample Input Wherein g and q are positive integers.

13. The method according to claim 12, in, The cycle scaling factor q is set independently of the cycle scaling factor l.

14. The method according to claim 12, in, Based on the metric function f θ Determine the target domain samples and source domain sample sets enhanced by different periodic scaling ratios The matching degree between the source domain samples in includes: The matching degree is quantified as And where the optimal cycle scaling ratio is determined as The sample input of the target domain sample after enhancement by the optimal cycle scaling ratio is 15. The method according to claim 14, in, The target battery's future state of health is determined as in The target battery is The predicted value of the health status when the characteristic voltage change interval occurs for the first time.

16. The method according to claim 1, further comprising: include: The predicted value of the health state of the target battery is interpolated to predict the health state change curve of the target battery over the entire life cycle under actual usage conditions.

17. A device for predicting the health status of a battery, include: The battery data acquisition module is configured as follows: Acquire source domain battery data and target domain battery data, wherein the source domain battery data includes a voltage variation curve over time of each reference battery in a plurality of reference batteries under a corresponding one of a plurality of reference conditions over the entire life cycle, and the target domain battery data includes a voltage variation curve over time of the target battery in an early life under an actual condition; The feature interval recognition module is configured as follows: Identifying a common characteristic voltage variation interval of the plurality of reference batteries based on the voltage variation curves of the plurality of reference batteries over time, wherein a voltage variation pattern of each reference battery within the characteristic voltage variation interval periodically appears in the voltage variation curve of the reference battery over time; The source domain sample construction module is configured as follows: Constructing a source domain sample set, including constructing a source domain sample for each characteristic voltage change interval appearing in the voltage-time change curve of each reference battery, the sample input of the source domain sample including the characteristic value and health status of the reference battery in the first multiple characteristic voltage change intervals before the characteristic voltage change interval, the characteristic value is related to the health status, and the sample output of the source domain sample including the health status of the reference battery in the second multiple characteristic voltage change intervals after the characteristic voltage change interval; The metric function construction module is configured as follows: Constructing a metric function based on a source domain sample set, wherein the metric function is used to measure the similarity between sample inputs; The target domain sample construction module is configured as follows: For each characteristic voltage variation interval that appears in the voltage variation curve of the target battery over time, a target domain sample is constructed, wherein the sample input of the target domain sample includes the characteristic values ​​and health status of the target battery in the first plurality of characteristic voltage variation intervals before the characteristic voltage variation interval; as well as The health status prediction module is configured as follows: The similarity between the sample input of the target domain sample determined based on the metric function and the sample input of the source domain sample in the source domain sample set is used as a weight to weight the sample output of the source domain sample in the source domain sample set, thereby synthesizing the sample output of the target domain sample as a predicted value of the health state of the target battery.

18. A computing device for predicting the state of health of a battery, include: one or more processors; as well as A memory storing computer executable instructions which, when executed by the one or more processors, cause the one or more processors to perform the method for predicting the health state of a battery according to any one of claims 1 to 16.

19. A non-transitory storage medium having computer executable instructions stored thereon, the computer executable instructions, when executed by a computer, causing the computer to perform the method for predicting the health state of a battery according to any one of claims 1 to 16.

20. A computer program product comprising instructions, which when executed by a processor implement the method for predicting the state of health of a battery according to any one of claims 1 to 16.

Images