CN116068444A - Battery health degree estimation method and device - Google Patents
Battery health degree estimation method and device Download PDFInfo
- Publication number
- CN116068444A CN116068444A CN202111294820.2A CN202111294820A CN116068444A CN 116068444 A CN116068444 A CN 116068444A CN 202111294820 A CN202111294820 A CN 202111294820A CN 116068444 A CN116068444 A CN 116068444A
- Authority
- CN
- China
- Prior art keywords
- charge
- health
- battery
- historical
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/70—Energy storage systems for electromobility, e.g. batteries
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
- Tests Of Electric Status Of Batteries (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention provides a battery health degree estimation method, which comprises the following steps: receiving operation data of a vehicle; judging whether the operation data can be used for estimating the health degree of the battery; and if so, calculating a value for each of the plurality of health features from the operational data and providing the calculated value to a trained neural network model of the vehicle to estimate the battery health, wherein the trained neural network model is used to represent a relationship between the plurality of health features and the battery health. The scheme of the invention considers the factors of the use process and the environment of the battery, so that the obtained battery SOH is more accurate. When the method is executed online, namely, the operation data transmitted by the vehicle-mounted networking terminal is utilized to estimate the SOH of the battery at the cloud server of the host factory, the host factory can dynamically grasp the battery health condition of the sold vehicle and dynamically monitor the battery service condition of the user, so that after-sale and maintenance service is provided, and the user experience is improved.
Description
Technical Field
The present invention relates generally to the field of battery technology. More particularly, the present invention relates to a battery health estimation method, apparatus, computing device, computer readable storage medium and computer program product.
Background
With the development of technology, pure electric vehicles and hybrid vehicles are increasingly used. To supply driving energy, a power battery (or referred to as a battery pack) formed by connecting a plurality of battery cells in series is generally mounted on a vehicle body. The power battery is inevitably aged in the long-term use process, so that the aging condition of the power battery needs to be estimated. The aging condition of the power battery is estimated typically by calculating the State of Health (SOH).
SOH of a power battery is currently calculated mainly by means of off-line measurement, including off-line measurement of battery capacity and off-line measurement of impedance. Although there may be some way to calculate SOH online, the establishment of its computational model depends mainly on experimental data and results. Alternatively, SOH may be calculated in a battery management system (Battery Management System, BMS) of the vehicle, and the calculated SOH may be collected by an on-board networking terminal (e.g., T-BOX) of the vehicle.
However, SOH of a power battery is affected by many factors such as battery materials, manufacturing, use, and environment, etc., and existing SOH calculation methods do not consider some important factors, especially actual use process and environmental factors of a battery having a great influence on SOH, which makes SOH calculated through off-line measurement or through a model built based on experimental data and results inaccurate. As for the way of calculating SOH in BMS, which is mainly dependent on data in BMS, there is also a problem that insufficient data causes SOH to be inaccurate.
Furthermore, the existing SOH calculation model has a narrow application range, because it is applicable to only certain types of batteries, but not others. Meanwhile, optimization and improvement of the calculation model require a large number of experiments or recalculations, and thus are costly to maintain.
Disclosure of Invention
As described above, the existing SOH calculation method does not consider some important factors affecting SOH, particularly the actual use process and environmental factors of the battery having a great influence on SOH, which makes SOH calculated by off-line measurement or by a model built based on experimental data and results inaccurate. The manner of calculating SOH in the BMS also has a problem in that data is insufficient, resulting in insufficient accuracy of SOH. In addition, the existing SOH calculation model has a narrow application range and high maintenance cost.
In view of the foregoing technical problem, a first aspect of the present invention provides a method for estimating a battery health, including: receiving operational data of the vehicle; judging whether the operation data can be used for estimating the health degree of the battery; and if so, calculating a value for each of a plurality of health features based on the operational data and providing the calculated value to a trained neural network model of the vehicle to estimate the battery health, wherein the trained neural network model is used to represent a relationship between the plurality of health features and the battery health.
According to the method, the values of the predefined plurality of health features can be extracted from the running data of the vehicle, and the battery SOH can be estimated by utilizing the values of the health features and the pre-trained neural network model. The battery SOH obtained by the method is more accurate because of considering the factors of the battery use process and the environment. When the method is executed online, namely, the operation data transmitted by the vehicle-mounted networking terminal is utilized to estimate the SOH of the battery at the cloud server of the host factory, the host factory can dynamically grasp the battery health condition of the sold vehicle and dynamically monitor the battery service condition of the user, so that after-sale and maintenance service is provided, and the user experience is improved. In addition, the estimation method is not affected by the service time of the vehicle, the type of the vehicle and the type of the battery, and is applicable to any electric or hybrid vehicle.
According to some alternative embodiments, determining whether the operational data is available for estimating battery health further comprises: judging whether the operation data meet a charge-discharge cycle or not; if so, judging whether the charge start charge state in the operation data is smaller than a first threshold value and the charge end charge state is larger than a second threshold value, wherein when the charge start charge state is smaller than the first threshold value and the charge end charge state is larger than the second threshold value, the operation data can be used for estimating the battery health.
According to some alternative embodiments, the neural network model is a Long Short Term Memory (LSTM) neural network model.
According to some optional embodiments, before calculating the value of each of the plurality of health features based on the operational data, the method further comprises: and data cleaning is carried out on the operation data, and the operation data are divided according to charge and discharge cycles.
According to some alternative embodiments, the method further comprises: obtaining historical operation data of the vehicle in a preset historical period, wherein the preset historical period comprises a plurality of charge and discharge cycles; carrying out data cleaning on the historical operation data and dividing the historical operation data according to charge and discharge cycles; and training the neural network model based on the partitioned historical operating data.
According to some optional embodiments, training the neural network model based on the partitioned historical operating data further comprises: selecting a charge-discharge cycle from the plurality of charge-discharge cycles having a charge-start state of charge less than a first threshold and a charge-end state of charge greater than a second threshold; calculating, for each selected charge-discharge cycle, a value for each health feature in its historical battery health and a preset health feature set based on the historical operating data; and training the neural network model using the calculated values of the set of health features and the historical battery health.
According to some alternative embodiments, the historical battery health is calculated by the following formula:
wherein eta is the coulombic efficiency of the battery，t 1 For stopping the charging start time, t 2 For the stop charge end time, I is the current during the stop charge, SOC start To charge the initial state of charge, SOC end To end charge state, Q t To the full charge capacity of the battery, Q 0 SOH is the battery health, which is the rated capacity of the battery.
According to some alternative embodiments, before training the neural network model, the method further comprises: respectively calculating the correlation between each health feature in the health feature set and the historical battery health degree; and selecting a preset number of health features from the health feature set as the plurality of health features in the order of high-to-low correlation.
According to some alternative embodiments, the correlation between each health feature and the historical battery health is calculated by the following formula:
wherein n is the number of charge-discharge cycles selected, X i For each health feature at the value of the ith charge and discharge cycle,
mean value of each health feature in n charge-discharge cycles, Y i Historical battery health for the ith charge and discharge cycle, +.>
For an average historical battery health, ρ, over n charge-discharge cycles xy For each of the health features a correlation with the historical battery health.
According to some optional embodiments, training the neural network model using the calculated values of the set of health features and the historical battery health further comprises: the neural network model is trained by inputting values of the plurality of health features for each selected charge-discharge cycle as samples and outputting the historical battery health for each selected charge-discharge cycle as samples.
A second aspect of the present invention proposes a battery health estimation apparatus comprising: a data receiving unit configured to receive operation data of a vehicle; a data determination unit configured to determine whether the operation data is usable for estimating a battery health; and a health estimation unit configured to calculate a value of each of a plurality of health features based on the operation data and provide the calculated value to a trained neural network model of the vehicle to estimate the battery health, when the data determination unit determines that the operation data is usable to estimate the battery health, wherein the trained neural network model is used to represent a relationship between the plurality of health features and the battery health.
According to the device, the values of the predefined plurality of health features can be extracted from the running data of the vehicle, and the battery SOH is estimated by utilizing the values of the health features and the pre-trained neural network model. The battery SOH obtained by the method is more accurate because of considering the factors of the battery use process and the environment. When the method is executed online, namely, the operation data transmitted by the vehicle-mounted networking terminal is utilized to estimate the SOH of the battery at the cloud server of the host factory, the host factory can dynamically grasp the battery health condition of the sold vehicle and dynamically monitor the battery service condition of the user, so that after-sale and maintenance service is provided, and the user experience is improved. Furthermore, such an estimation device is not affected by the vehicle service time, the vehicle type and the battery type, and is suitable for any electric or hybrid vehicle.
According to some optional embodiments, the data determination unit is further configured to: judging whether the operation data meet a charge-discharge cycle or not; if so, judging whether the charge start charge state in the operation data is smaller than a first threshold value and the charge end charge state is larger than a second threshold value, wherein when the charge start charge state is smaller than the first threshold value and the charge end charge state is larger than the second threshold value, the operation data can be used for estimating the battery health.
According to some alternative embodiments, the neural network model is a Long Short Term Memory (LSTM) neural network model.
According to some alternative embodiments, the apparatus further comprises: and a first preprocessing unit configured to perform data cleansing on the operation data and divide the operation data by a charge-discharge cycle before the health degree estimation unit calculates a value of each of a plurality of health features based on the operation data.
According to some alternative embodiments, the apparatus further comprises: a data obtaining unit configured to obtain historical operation data of the vehicle within a preset history period including a plurality of charge-discharge cycles; the second preprocessing unit is configured to perform data cleaning on the historical operation data and divide the historical operation data according to charge-discharge cycles; and a model training unit configured to train the neural network model based on the divided historical operating data.
According to some optional embodiments, the model training unit is further configured to: selecting a charge-discharge cycle from the plurality of charge-discharge cycles having a charge-start state of charge less than a first threshold and a charge-end state of charge greater than a second threshold; calculating, for each selected charge-discharge cycle, a value for each health feature in its historical battery health and a preset health feature set based on the historical operating data; and training the neural network model using the calculated values of the set of health features and the historical battery health.
According to some alternative embodiments, the historical battery health is calculated by the following formula:
wherein eta is the coulombic efficiency of the battery, t 1 For stopping the charging start time, t 2 For the stop charge end time, I is the current during the stop charge, SOC start To charge the initial state of charge, SOC end To end charge state, Q t To the full charge capacity of the battery, Q 0 SOH is the battery health, which is the rated capacity of the battery.
According to some alternative embodiments, the apparatus further comprises: a health feature selection unit configured to: respectively calculating the correlation between each health feature in the health feature set and the historical battery health degree; and selecting the plurality of health features from the set of health features in a high-to-low order of relevance.
According to some alternative embodiments, the correlation between each health feature and the historical battery health is calculated by the following formula:
wherein n is the number of charge-discharge cycles selected, X i For each health feature at the value of the ith charge and discharge cycle,
mean value of each health feature in n charge-discharge cycles, Y i Historical battery health for the ith charge and discharge cycle, +.>
For an average historical battery health, ρ, over n charge-discharge cycles xy For each of the health features a correlation with the historical battery health.
According to some optional embodiments, the model training unit is further configured to: the neural network model is trained by inputting values of the plurality of health features for each selected charge-discharge cycle as samples and outputting the historical battery health for each selected charge-discharge cycle as samples.
A third aspect of the present invention proposes a computing device comprising: a processor; and a memory for storing computer-executable instructions that, when executed, cause the processor to perform a method according to any of the above embodiments.
A fourth aspect of the invention proposes a computer readable storage medium having stored thereon computer executable instructions for performing a method according to any of the above embodiments.
A fifth aspect of the invention proposes a computer program product stored on a computer readable storage medium and comprising computer executable instructions which, when executed, cause at least one processor to perform a method according to any of the above-described embodiments.
Drawings
Features, advantages and other aspects of embodiments of the present invention will become more apparent upon reference to the following detailed description, taken in conjunction with the accompanying drawings, wherein like reference numerals designate the same or similar parts throughout the several embodiments thereof, which are shown by way of illustration and not limitation.
FIG. 1 is a general flow chart of a battery health estimation method according to the present invention;
FIG. 2 illustrates a flow chart of a method of training the neural network model of FIG. 1;
fig. 3 shows several sub-steps of step 23 in fig. 2;
FIG. 4 shows a flowchart of a battery health estimation method according to one embodiment of the invention;
fig. 5 illustrates a battery health estimation apparatus according to an embodiment of the present invention; and
FIG. 6 illustrates a computing device for estimating battery health according to one embodiment of the invention.
Detailed Description
Various exemplary embodiments of the present invention are described in detail below with reference to the drawings. While the exemplary methods, apparatus described below include software and/or firmware executed on hardware among other components, it should be noted that these examples are merely illustrative and should not be considered as limiting. For example, it is contemplated that any or all of the hardware, software, and firmware components could be embodied exclusively in hardware, exclusively in software, or in any combination of hardware and software. Thus, while exemplary methods and apparatus have been described below, those skilled in the art will readily appreciate that the examples provided are not intended to limit the manner in which such methods and apparatus may be implemented.
Furthermore, the flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of methods and systems according to various embodiments of the present invention. It should be noted that the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by special purpose hardware-based systems which perform the specified functions or operations, or combinations of special purpose hardware and computer instructions.
For convenience of description, some terms appearing in the present invention are explained below, and it should be understood that terms used in the present application should be interpreted as having meanings consistent with their meanings in the context of the present application specification and the relevant field. The terms "comprising," including, "and similar terms in the present invention should be construed as open-ended terms, i.e., including, but not limited to," mean that other elements may also be included.
In embodiments of the present invention, the term "based on" is based at least in part on.
In an embodiment of the present invention, the term "one embodiment" means "at least one embodiment".
In an embodiment of the present invention, the term "another embodiment" means "at least one other embodiment" and so forth.
In an embodiment of the invention, the term "vehicle" may be an automobile, truck, SUV, van, bus, or any other rolling platform, which may be a pure electric vehicle or a hybrid vehicle.
In embodiments of the present invention, the term "battery" or "power cell" may include a single cell, a battery of multiple cells, or a battery pack.
Currently, the existing SOH calculation method does not consider some important factors affecting SOH, especially the actual use process and environmental factors of the battery with a great influence on SOH, which makes the SOH calculated by offline measurement or by a model established based on experimental data and results inaccurate. The manner of calculating SOH in the BMS also has a problem in that data is insufficient, resulting in insufficient accuracy of SOH. In addition, the existing SOH calculation model has a narrow application range and high maintenance cost.
To this end, an embodiment of the present invention proposes a battery health estimation method capable of extracting values of a predefined plurality of health features from operation data and estimating a battery SOH using the values of the health features and a pre-trained neural network model. The battery SOH obtained by the method is more accurate because of considering the factors of the battery use process and the environment. When the method is executed online, namely, the operation data transmitted by the vehicle-mounted networking terminal is utilized to estimate the SOH of the battery at the cloud server of the host factory, the host factory can dynamically grasp the battery health condition of the sold vehicle and dynamically monitor the battery service condition of the user, so that after-sale and maintenance service is provided, and the user experience is improved. In addition, the estimation method is not affected by the service time of the vehicle, the type of the vehicle and the type of the battery, and is applicable to any electric or hybrid vehicle.
The following describes the invention in terms of several embodiments. Fig. 1 shows an overall flowchart of a battery health estimation method according to the present invention. The method of fig. 1 may be performed by a cloud server of a host factory or by a computing device on a vehicle. Referring to fig. 1, method 100 begins at step 11. In step 11, operational data of a vehicle is received. As is well known in the art, an in-vehicle networking device such as a T-BOX may communicate with a cloud server of a host factory via a mobile network (e.g., 4G, 5G network) and send operational data of the vehicle to the cloud server at certain times (e.g., when the vehicle is powered up and when it is parked for charging), including, but not limited to, operational status of the vehicle and battery status, such as vehicle speed, mileage, gear, voltage current of the battery, state of Charge (SOC), motor data, and the like. Thus, the cloud server may receive operational data of the vehicle from the in-vehicle networking device. Further, the operational data may also be received directly by a computing device on the vehicle.
Next, in step 12, it is determined whether the operation data is available for estimating the battery health. In some embodiments, step 12 further comprises: judging whether the operation data meets a charge-discharge cycle or not; if so, judging whether the charge start SOC in the operation data is smaller than a first threshold value and the charge end SOC is larger than a second threshold value, wherein when the charge start SOC is smaller than the first threshold value and the charge end SOC is larger than the second threshold value, the operation data can be used for estimating the health of the battery. The process between two stopping charges of the vehicle is called a charge-discharge cycle. That is, one charge-discharge cycle includes a parking charge phase and a vehicle travel phase. First, it is determined whether the operation data satisfies a charge determination cycle. In the present invention, a neural network model is employed to estimate the battery health, and in order to make the model more accurate when training the neural network model, historical operation data in a charge-discharge cycle in which the charge start SOC is smaller than a first threshold value and the charge end SOC is larger than a second threshold value is selected for model training (to be described later in detail). Therefore, when using the trained model, if the operation data satisfies one charge determination cycle, it is determined whether the charge start SOC in the operation data is smaller than the first threshold value and the charge end SOC is larger than the second threshold value. The values of the first threshold and the second threshold may be set as needed, for example, the first threshold is set to 40% and the second threshold is set to 60%.
If it is determined that the operational data is available for estimating battery health, a value for each of a plurality of health features is calculated based on the operational data and the calculated values are provided to a trained neural network model of the vehicle to estimate battery health, step 13. The trained neural network model is used to represent a relationship between a plurality of health features and battery health. The health characteristics may be characteristics that affect the health of the battery, such as the accumulated use time of the battery, the accumulated driving mileage of the vehicle, the current, voltage, and power of the battery when charged, the temperature of the battery when driving, the discharge power, etc., and their values may be calculated from the operation data. In some embodiments, the neural network model is a Long Short Term Memory (LSTM) neural network model. In other embodiments, other forms of neural network models may be employed.
According to the method, the values of the predefined plurality of health features can be extracted from the running data of the vehicle, and the battery SOH can be estimated by utilizing the values of the health features and the pre-trained neural network model. The battery SOH obtained by the method is more accurate because of considering the factors of the battery use process and the environment. When the method is executed online, namely, the operation data transmitted by the vehicle-mounted networking terminal is utilized to estimate the SOH of the battery at the cloud server of the host factory, the host factory can dynamically grasp the battery health condition of the sold vehicle and dynamically monitor the battery service condition of the user, so that after-sale and maintenance service is provided, and the user experience is improved. In addition, the estimation method is not affected by the service time of the vehicle, the type of the vehicle and the type of the battery, and is applicable to any electric or hybrid vehicle.
In some embodiments, the method 100 further comprises: before calculating the value of each of the plurality of health features based on the operational data, the operational data is data-washed and divided by charge-discharge cycles. The data cleaning may include sorting the operational data according to a time sequence and an accumulated mileage sequence, removing erroneous values and outliers from the data, and filling missing values from the data by means of linear interpolation. And then dividing the operation data according to the flag bit and the charge and discharge cycle.
In some embodiments, the method 100 further includes building and training a neural network model. Fig. 2 shows a flowchart of a training method of the neural network model in fig. 1. It will be appreciated by those skilled in the art that the training of the model should be performed prior to estimating battery health using the model, for example prior to step 13, but may be performed sequentially or in parallel with steps 11 and 12. The frequency and time of model training may be set as desired. For example, when the vehicle service time is short, the model may be trained repeatedly at a higher frequency because of less historical operating data. As the vehicle service time increases, the frequency of model training may decrease. Referring to fig. 2, in step 21, historical operation data of the vehicle is obtained for a preset history period including a plurality of charge-discharge cycles. The history period may be set as needed, for example, 6 months, 1 year, or the like.
In step 22, the historical operating data is data-cleaned and divided according to charge-discharge cycles. Similarly, the data cleaning comprises the steps of sorting the historical operation data according to the time sequence and the accumulated mileage sequence, removing error values and abnormal values in the data, and filling missing values in the data in a linear interpolation mode. And then, dividing the historical operation data according to the charge and discharge cycles according to the flag bits. As previously described, one charge-discharge cycle includes a park charge phase and a vehicle travel phase.
Thereafter, in step 23, a neural network model is trained based on the partitioned historical operating data. As described above, in some embodiments, the neural network model may be an LSTM neural network model. In other embodiments, other types of neural network models may be used. The trained neural network model has a high accuracy because the data used to train the neural network model is sufficient and correlated to the actual use of the battery.
Fig. 3 shows several sub-steps of step 23 in fig. 2. As shown in fig. 3, in step 231, a charge-discharge cycle in which the charge start SOC is smaller than the first threshold value and the charge end SOC is larger than the second threshold value is selected from the plurality of charge-discharge cycles. The values of the first threshold and the second threshold may be set as needed, for example, the first threshold is set to 40% and the second threshold is set to 60%. By selecting a satisfactory charge-discharge cycle, the accuracy of the trained neural network model can be further improved. In step 232, for each selected charge-discharge cycle, a value for each of its historical battery health and a set of preset health features is calculated based on the historical operating data. That is, historical operating data in each charge and discharge cycle is used to calculate a historical battery health for that charge and discharge cycle and a value for each health feature in a set of preset health features.
In the present embodiment, the historical battery health is calculated by the following formulas (1) - (3):
wherein eta is the coulombic efficiency of the battery, t 1 Charging step for parkingStart time of parking charge of section t 2 The stop charge end time in the stop charge phase is the current, SOC, during the stop charge phase start To charge the initial state of charge, SOC end To end charge state, Q t To the full charge capacity of the battery, Q 0 SOH is the battery health, which is the rated capacity of the battery.
In the present embodiment, 26 health features are preset as a health feature set. These 26 health features are shown in table 1 below.
TABLE 1
It should be noted that the health feature set shown in table 1 is for illustration purposes only. In other embodiments, other numbers and/or defined health features may be set. The values of the health features described above may be extracted or calculated from historical operating data.
Next, in step 233, a correlation between each health feature in the health feature set and the historical battery health is calculated, respectively. In the present embodiment, the correlation between each health feature and the historical battery health is calculated by the following formula (4):
wherein n is the number of charge-discharge cycles selected, X i For each health feature at the value of the ith charge and discharge cycle,
mean value of each health feature in n charge-discharge cycles, Y i Historical battery for ith charge-discharge cycleHealth degree, and/or->
For an average historical battery health, ρ, over n charge-discharge cycles xy For each of the health features a correlation with the historical battery health. The closer the value calculated by equation (4) is to 1, the higher the correlation between the health characteristic and the historical battery health, and vice versa. />
In step 234, a predetermined number of health features are selected from the set of health features as a plurality of health features in a high-to-low correlation order. The number of selected health features may be set as desired, for example 10. In step 235, the neural network model is trained by taking as sample inputs the values of the plurality of health features for each selected charge-discharge cycle and taking as sample outputs the historical battery health for each selected charge-discharge cycle. That is, in training the neural network model, only the values of the plurality of health features and the corresponding historical battery health for each charge-discharge cycle selected are used as the sample set. In this embodiment, the model is trained by dividing the sample set into a training set and a test set at a ratio of 8:2.
It should be understood by those skilled in the art that, in this embodiment, a part of health features with higher correlation is selected from the health feature set according to the correlation with the historical battery health, so as to improve the accuracy of the model. However, in other embodiments, the model may also be trained without regard to correlation with historical battery health, but rather with the values of all of the health features in the set of health features as sample inputs to the model.
Fig. 4 shows a flowchart of a battery health estimation method according to an embodiment of the present invention. The battery health estimation method 400 shown in fig. 4 is performed by a cloud server in communication with an in-vehicle networking device. In method 400, steps 41-42 are used to train the model and are therefore performed offline. Steps 43-47 are used to estimate battery health online and are therefore performed online. In step 41, historical operating data of the vehicle is obtained. In step 42, the LSTM model is trained and model parameters are determined. The process of training the model has been described in detail above and will not be described in detail here. In this embodiment, steps 41-42 are performed before steps 43-46. However, in other embodiments, steps 41-42 may also be performed in parallel with steps 42-46. After the model is trained, the cloud server stores the trained model in memory for subsequent use. One skilled in the art will appreciate that a battery health estimation model may be trained for each vehicle.
When it is desired to estimate battery health online, first, in step 43, operational data of the vehicle is received. Step 44 then includes determining whether the received operational data is available for estimating battery health. If so, the method 400 proceeds to step 45, otherwise it returns to step 43 to continue receiving operational data. In step 45, data cleansing and charge-discharge cycle division are performed. Next, step 46 includes calculating values of a plurality of health features, and in step 47, estimating battery health based on the trained battery health estimation model. The estimated battery health may be used to provide after-market and maintenance services. For example, by analyzing the correlation between each health characteristic and battery health, a better vehicle usage habit may be recommended or suggested to the user, extending battery life. For another example, when the battery health is below a certain threshold, a warning may be issued to alert the user to replace the battery in time. For another example, battery health may be used as a reference factor in evaluating the value of a second hand vehicle.
Fig. 5 illustrates a battery health estimation apparatus according to an embodiment of the present invention. The units in fig. 5 may be implemented in software, hardware (e.g., integrated circuits, FPGAs, etc.), or a combination of software and hardware. Referring to fig. 5, the apparatus 500 includes a data receiving unit 501, a data judging unit 502, and a health estimating unit 503. The data receiving unit 501 is configured to receive operation data of the vehicle. The data determination unit 502 is configured to determine whether the operation data is available for estimating the battery health. The health estimation unit 503 is configured to calculate a value of each of the plurality of health features based on the operation data and provide the calculated value to a trained neural network model of the vehicle to estimate the battery health when the data determination unit 502 determines that the operation data is available to estimate the battery health. The trained neural network model is used to represent a relationship between a plurality of health features and battery health.
In some embodiments, the data determination unit 502 is further configured to: judging whether the operation data meets a charge-discharge cycle or not; if so, judging whether the charge start charge state in the operation data is smaller than a first threshold value and the charge end charge state is larger than a second threshold value, wherein when the charge start charge state is smaller than the first threshold value and the charge end charge state is larger than the second threshold value, the operation data can be used for estimating the health degree of the battery.
In some embodiments, the neural network model is a Long Short Term Memory (LSTM) neural network model.
In some embodiments, the apparatus 500 further comprises a first preprocessing unit (not shown in fig. 5). The first preprocessing unit is configured to perform data cleansing on the operation data and divide by a charge-discharge cycle before the health estimation unit 503 calculates a value of each of the plurality of health features based on the operation data.
In some embodiments, the apparatus 500 further comprises a data obtaining unit, a second preprocessing unit, and a model training unit (not shown in fig. 5). The data obtaining unit is configured to obtain historical operation data of the vehicle within a preset history period including a plurality of charge-discharge cycles. The second preprocessing unit is configured to perform data cleaning on the historical operation data and divide the historical operation data according to charge-discharge cycles. The model training unit is configured to train a neural network model based on the partitioned historical operating data.
In some embodiments, the model training unit is further configured to: selecting a charge-discharge cycle from a plurality of charge-discharge cycles having a charge-start state of charge less than a first threshold and a charge-end state of charge greater than a second threshold; calculating, for each selected charge-discharge cycle, a value for each health feature in its historical battery health and a preset health feature set based on the historical operating data; and training a neural network model using the calculated values of the set of health features and the historical battery health.
In some embodiments, the historical battery health is calculated by the following formula:
wherein eta is the coulombic efficiency of the battery, t 1 For stopping the charging start time, t 2 For the stop charge end time, I is the current during the stop charge, SOC start To charge the initial state of charge, SOC end To end charge state, Q t To the full charge capacity of the battery, Q 0 SOH is the battery health, which is the rated capacity of the battery.
In some embodiments, the apparatus 500 further comprises a health feature selection unit (not shown in fig. 5). The health feature selection unit is configured to: respectively calculating the correlation between each health feature in the health feature set and the historical battery health degree; and selecting a preset number of health features from the health feature set as a plurality of health features in the order of high-to-low correlation.
In some embodiments, the correlation between each health feature and historical battery health is calculated by the following formula:
wherein n is the number of charge-discharge cycles selected, X i Charging at ith for each health featureThe value of the discharge cycle is such that,
mean value of each health feature in n charge-discharge cycles, Y i Historical battery health for the ith charge and discharge cycle, +.>
For an average historical battery health, ρ, over n charge-discharge cycles xy For each of the health features a correlation with the historical battery health.
In some embodiments, the model training unit is further configured to: the neural network model is trained by inputting values of a plurality of health features of each selected charge-discharge cycle as samples and outputting historical battery health of each selected charge-discharge cycle as samples.
FIG. 6 illustrates a computing device for estimating battery health according to one embodiment of the invention. It should be appreciated that the computing device 600 may be implemented to implement the functionality of the battery health estimation method 100 of fig. 1. As can be seen in fig. 6, the computing device 600 includes a Central Processing Unit (CPU) 601 (e.g., a processor) that can perform various suitable actions and processes in accordance with computer program instructions stored in a Read Only Memory (ROM) 602 or loaded from a storage unit 608 into a Random Access Memory (RAM) 603. In RAM 603, various programs and data required for the operation of the computing device 600 may also be stored. The CPU 601, ROM 602, and RAM 603 are connected to each other through a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.
Various components in computing device 600 are connected to I/O interface 605, including: an input unit 606, an output unit 607, a storage unit 608, and a communication unit 609. The communication unit 609 allows the computing device 600 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks.
Various methods described above, such as the battery health estimation method 100, may be performed by the CPU 601. For example, in some embodiments, the battery health estimation method 100 may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 608. In some embodiments, some or all of the computer program may be loaded and/or installed onto computing device 600 via ROM 602 and/or communication unit 609. When the computer program is loaded into RAM 603 and executed by processor CPU 601, one or more actions or steps of the battery health estimation method 100 described above may be performed.
Accordingly, in another embodiment, the present invention is directed to a computer-readable storage medium having computer-executable instructions stored thereon for performing the battery health estimation method of the various embodiments of the present invention.
In another embodiment, the invention is directed to a computer program product tangibly stored on a computer-readable storage medium and comprising computer-executable instructions that, when executed, cause at least one processor to perform the method of battery health estimation in various embodiments of the invention.
According to the above embodiment, it is possible to extract values of a predefined plurality of health features from the operation data of the vehicle, and estimate the battery SOH using the values of the health features and the neural network model trained in advance. The battery SOH obtained by the method is more accurate because of considering the factors of the battery use process and the environment. When the method is executed online, namely, the operation data transmitted by the vehicle-mounted networking terminal is utilized to estimate the SOH of the battery at the cloud server of the host factory, the host factory can dynamically grasp the battery health condition of the sold vehicle and dynamically monitor the battery service condition of the user, so that after-sale and maintenance service is provided, and the user experience is improved. In addition, the estimation method is not affected by the service time of the vehicle, the type of the vehicle and the type of the battery, and is applicable to any electric or hybrid vehicle.
Computer readable program instructions or computer program products for executing aspects of the present invention can also be stored in the cloud end, and when a call is required, a user can access the computer readable program instructions stored on the cloud end for executing aspects of the present invention through the mobile internet, the fixed network or other networks, thereby implementing the technical solutions disclosed in the aspects of the present invention.
In general, the various example embodiments of the invention may be implemented in hardware or special purpose circuits, software, firmware, logic or any combination thereof. Some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device. While aspects of the embodiments of the invention are illustrated or described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that blocks, apparatus, systems, techniques or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.
While various example embodiments of the invention are described above as being implemented in hardware or dedicated circuitry, the computing device described above may be implemented in either hardware or software, as: in the 90 s of the 20 th century, a technical improvement could easily be made whether the improvement was a hardware improvement (e.g., a improvement in the circuit structure of diodes, transistors, switches, etc.) or a software improvement (e.g., a improvement in the process flow). However, with the continuous development of technology, many improvements of the method flow today can be almost all achieved by programming the improved method flow into a hardware circuit, in other words, by programming a different program for the hardware circuit to obtain a corresponding hardware circuit structure, i.e. to achieve a change of the hardware circuit structure, so that such improvements of the method flow can also be regarded as direct improvements of the hardware circuit structure. Therefore, an improvement of a method flow cannot be said to be realized by a hardware entity module. For example, a programmable logic device (Programmable Logic Device: PLD) (e.g., field programmable gate array (Field Programmable Gate Array: FPGA)) is an integrated circuit whose logic function is determined by the programming of the device by a user. A digital system is "integrated" on a programmable logic device by the designer's own programming without requiring the chip manufacturer to design and fabricate application specific integrated circuit chips. Moreover, instead of manually fabricating integrated circuit chips, such programming is now mostly implemented with "logic compiler" software, which is similar to the software compiler used in program development and writing, and the original code before being compiled is also written in a specific programming language, which is called hardware description language (Hardware Description Language: HDL), but HDL is not just one kind, but a plurality of kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), confluence, CUPL (Cornell University Programming Language), HDCal, JML (Java Hardware Description Language), lava, lola, myHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., and VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog are currently most commonly used. It will also be apparent to those skilled in the art that a hardware circuit implementing the logic method flow can be readily obtained by merely slightly programming the method flow into an integrated circuit using several of the hardware description languages described above.
Although embodiments of the present invention have been described with reference to a number of specific embodiments, it should be understood that embodiments of the present invention are not limited to the specific embodiments disclosed. The embodiments of the invention are intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims. The scope of the claims is to be accorded the broadest interpretation so as to encompass all such modifications and equivalent structures and functions.
Claims (23)
1. A battery health estimation method, comprising:
receiving operation data of a vehicle;
judging whether the operation data can be used for estimating the health degree of the battery; and
if so, calculating a value for each of a plurality of health features based on the operational data and providing the calculated values to a trained neural network model of the vehicle to estimate the battery health, wherein,
the trained neural network model is used to represent a relationship between the plurality of health features and the battery health.
2. The method of claim 1, wherein determining whether the operational data is available for estimating battery health further comprises:
judging whether the operation data meet a charge-discharge cycle or not;
If so, judging whether the charge start charge state in the operation data is smaller than a first threshold value and whether the charge end charge state is larger than a second threshold value, wherein,
the operational data may be used to estimate the battery health when the charge start state of charge is less than the first threshold and the charge end state of charge is greater than the second threshold.
3. The method of claim 1, wherein the neural network model is a long-short-term memory (LSTM) neural network model.
4. The method of claim 1, wherein prior to calculating the value for each of a plurality of health features based on the operational data, the method further comprises:
and data cleaning is carried out on the operation data, and the operation data are divided according to charge and discharge cycles.
5. The method of claim 1, wherein prior to providing the calculated values to the trained neural network model of the vehicle, the method further comprises:
obtaining historical operation data of the vehicle in a preset historical period, wherein the preset historical period comprises a plurality of charge and discharge cycles;
carrying out data cleaning on the historical operation data and dividing the historical operation data according to charge and discharge cycles; and
The neural network model is trained based on the partitioned historical operating data.
6. The method of claim 5, wherein training the neural network model based on the partitioned historical operating data further comprises:
selecting a charge-discharge cycle from the plurality of charge-discharge cycles having a charge-start state of charge less than a first threshold and a charge-end state of charge greater than a second threshold;
calculating, for each selected charge-discharge cycle, a value for each health feature in its historical battery health and a preset health feature set based on the historical operating data; and
training the neural network model using the calculated values of the set of health features and the historical battery health.
7. The method of claim 6, wherein the historical battery health is calculated by the formula:
wherein eta is the coulombic efficiency of the battery, t 1 For stopping the charging start time, t 2 For the stop charge end time, I is the current during the stop charge, SOC start To charge the initial state of charge, SOC end To end charge state, Q t To the full charge capacity of the battery, Q 0 SOH is the battery health, which is the rated capacity of the battery.
8. The method of claim 6, wherein prior to training the neural network model, the method further comprises:
respectively calculating the correlation between each health feature in the health feature set and the historical battery health degree; and
a preset number of health features are selected from the health feature set as the plurality of health features in a sequence of high-to-low correlation.
9. The method of claim 8, wherein the correlation between each health feature and the historical battery health is calculated by the following formula:
wherein n is the number of charge-discharge cycles selected, X i For each health feature at the value of the ith charge and discharge cycle,
mean value of each health feature in n charge-discharge cycles, Y i Historical battery health for the ith charge and discharge cycle, +.>
For an average historical battery health, ρ, over n charge-discharge cycles xy For each of the health features a correlation with the historical battery health.
10. The method of claim 8, wherein training the neural network model with the calculated values of the set of health features and the historical battery health further comprises:
The neural network model is trained by inputting values of the plurality of health features for each selected charge-discharge cycle as samples and outputting the historical battery health for each selected charge-discharge cycle as samples.
11. A battery health estimation device, comprising:
a data receiving unit configured to receive operation data of a vehicle;
a data determination unit configured to determine whether the operation data is usable for estimating a battery health; and
a health estimation unit configured to calculate a value of each of a plurality of health features based on the operation data and to provide the calculated value to a trained neural network model of the vehicle to estimate the battery health when the data determination unit determines that the operation data is usable to estimate the battery health, wherein,
the trained neural network model is used to represent a relationship between the plurality of health features and the battery health.
12. The apparatus of claim 11, wherein the data determination unit is further configured to:
Judging whether the operation data meet a charge-discharge cycle or not;
if so, judging whether the charge start charge state in the operation data is smaller than a first threshold value and whether the charge end charge state is larger than a second threshold value, wherein,
the operational data may be used to estimate the battery health when the charge start state of charge is less than the first threshold and the charge end state of charge is greater than the second threshold.
13. The apparatus of claim 11, wherein the neural network model is a long-short-term memory (LSTM) neural network model.
14. The apparatus of claim 11, further comprising: a first preprocessing unit configured to:
the operation data is subjected to data cleaning and divided according to charge-discharge cycles before the health degree estimation unit calculates the value of each of the plurality of health features based on the operation data.
15. The apparatus of claim 11, further comprising:
a data obtaining unit configured to obtain historical operation data of the vehicle within a preset history period including a plurality of charge-discharge cycles;
The second preprocessing unit is configured to perform data cleaning on the historical operation data and divide the historical operation data according to charge-discharge cycles; and
a model training unit configured to train the neural network model based on the partitioned historical operating data.
16. The apparatus of claim 15, wherein the model training unit is further configured to:
selecting a charge-discharge cycle from the plurality of charge-discharge cycles having a charge-start state of charge less than a first threshold and a charge-end state of charge greater than a second threshold;
calculating, for each selected charge-discharge cycle, a value for each health feature in its historical battery health and a preset health feature set based on the historical operating data; and
training the neural network model using the calculated values of the set of health features and the historical battery health.
17. The apparatus of claim 16, wherein the historical battery health is calculated by the formula:
wherein eta is the coulombic efficiency of the battery, t 1 For stopping the charging start time, t 2 For the stop charge end time, I is the current during the stop charge, SOC start To charge the initial state of charge, SOC end To end charge state, Q t To the full charge capacity of the battery, Q 0 SOH is the battery health, which is the rated capacity of the battery.
18. The apparatus of claim 16, further comprising a health feature selection unit configured to:
respectively calculating the correlation between each health feature in the health feature set and the historical battery health degree; and
and selecting a preset number of health features from the health feature set to serve as the health features according to the sequence of the correlation from high to low.
19. The apparatus of claim 18, wherein the correlation between each health feature and the historical battery health is calculated by the formula:
wherein n is the number of charge-discharge cycles selected, X i For each health feature at the value of the ith charge and discharge cycle,
mean value of each health feature in n charge-discharge cycles, Y i Historical battery health for the ith charge and discharge cycle, +.>
For an average historical battery health, ρ, over n charge-discharge cycles xy For each of the health features a correlation with the historical battery health.
20. The apparatus of claim 18, wherein the model training unit is further configured to:
the neural network model is trained by inputting values of the plurality of health features for each selected charge-discharge cycle as samples and outputting the historical battery health for each selected charge-discharge cycle as samples.
21. A computing device, comprising:
a processor; and
a memory for storing computer-executable instructions that, when executed, cause the processor to perform the method of any of claims 1-10.
22. A computer readable storage medium having stored thereon computer executable instructions for performing the method according to any of claims 1-10.
23. A computer program product stored on a computer readable storage medium and comprising computer executable instructions that, when executed, cause at least one processor to perform the method of any one of claims 1-10.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111294820.2A CN116068444A (en) | 2021-11-03 | 2021-11-03 | Battery health degree estimation method and device |
| PCT/EP2022/078969 WO2023078673A1 (en) | 2021-11-03 | 2022-10-18 | Method and apparatus for estimating state of health of battery |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111294820.2A CN116068444A (en) | 2021-11-03 | 2021-11-03 | Battery health degree estimation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116068444A true CN116068444A (en) | 2023-05-05 |
Family
ID=84331327
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111294820.2A Pending CN116068444A (en) | 2021-11-03 | 2021-11-03 | Battery health degree estimation method and device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN116068444A (en) |
| WO (1) | WO2023078673A1 (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116298931B (en) * | 2023-05-12 | 2023-09-01 | 四川新能源汽车创新中心有限公司 | Cloud data-based lithium ion battery capacity estimation method |
| CN117269805A (en) * | 2023-11-23 | 2023-12-22 | 中国人民解放军国防科技大学 | Vehicle-mounted lithium battery pack health state evaluation model training and predicting method and device |
| CN117782647B (en) * | 2024-02-23 | 2024-06-04 | 杭州中港地铁装备维护有限公司 | Subway running part bearing health state monitoring system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20180084358A (en) * | 2017-01-17 | 2018-07-25 | 삼성전자주식회사 | Method and apparatus for estimating state of battery |
| KR20190032780A (en) * | 2017-09-20 | 2019-03-28 | 삼성전자주식회사 | Device and method to estimate state of battery |
-
2021
- 2021-11-03 CN CN202111294820.2A patent/CN116068444A/en active Pending
-
2022
- 2022-10-18 WO PCT/EP2022/078969 patent/WO2023078673A1/en active Application Filing
Also Published As
| Publication number | Publication date |
|---|---|
| WO2023078673A1 (en) | 2023-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116068444A (en) | Battery health degree estimation method and device | |
| US9533597B2 (en) | Parameter identification offloading using cloud computing resources | |
| US9446678B2 (en) | Battery model with robustness to cloud-specific communication issues | |
| EP3605127A1 (en) | Device and method for estimating battery resistance | |
| CN102455411A (en) | Adaptive slowly-varying current detection | |
| EP3551494B1 (en) | A method of estimating a charge state for a battery cell | |
| CN105128853A (en) | System used for calculating driving mileage of extended-range hybrid power automobile | |
| CN113219336B (en) | Battery degradation determination system, method, and non-transitory storage medium storing program | |
| CN111376790A (en) | Method, apparatus, device and medium for determining a mileage of a vehicle | |
| CN110542866B (en) | Method for estimating residual electric quantity parameter of battery | |
| CN112526378B (en) | Battery inconsistency fault early warning method and device | |
| CN112703518A (en) | Remaining mileage prediction method and battery remote service system | |
| CN111591152A (en) | Battery pack power change decision method, device and system in charge and change separation mode | |
| KR20190120794A (en) | Method and placement for balancing battery packs | |
| CN113665431A (en) | Method and device for estimating remaining endurance mileage of electric automobile | |
| CN115877243A (en) | Storage battery degradation estimation device, system, method, and storage medium | |
| CN112630661A (en) | Battery state of charge (SOC) estimation method and device | |
| WO2016200319A1 (en) | A method and a monitoring unit for monitoring a battery system | |
| CN112673320A (en) | Method for estimating operating parameters of a battery cell in a vehicle | |
| CN109986997B (en) | Power battery SOC prediction device, automobile and method | |
| CN107038493B (en) | Method for rapidly predicting battery replacement load of battery replacement station of electric vehicle | |
| WO2022024848A1 (en) | Battery management system, calculation system, battery degradation prediction method, and battery degradation prediction program | |
| Enthaler et al. | Significance of internal battery resistance on the remaining range estimation of electric vehicles | |
| CN115598531A (en) | Method for estimating state of health of vehicle battery and training method of estimation model thereof | |
| CN116298945A (en) | Method and device for determining remaining available energy of battery and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |