CN114859249B - Method and device for detecting battery pack capacity - Google Patents
Method and device for detecting battery pack capacity Download PDFInfo
- Publication number
- CN114859249B CN114859249B CN202210785324.5A CN202210785324A CN114859249B CN 114859249 B CN114859249 B CN 114859249B CN 202210785324 A CN202210785324 A CN 202210785324A CN 114859249 B CN114859249 B CN 114859249B
- Authority
- CN
- China
- Prior art keywords
- battery pack
- algorithm model
- lstm algorithm
- capacity
- preset
- 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.)
- Active
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]
- G01R31/385—Arrangements for measuring battery or accumulator variables
- G01R31/387—Determining ampere-hour charge capacity or SoC
-
- 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
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
Abstract
The embodiment of the invention discloses a method and a device for detecting the capacity of a battery pack, wherein the method comprises the steps of obtaining the capacity detection data of a target battery pack according to an ampere-hour integral method; importing the capacity detection data into a preset LSTM algorithm model group, and determining a target LSTM algorithm model; and detecting the capacities of a plurality of battery packs to be detected based on the target LSTM algorithm model. According to the method and the device, the preset LSTM algorithm model group protecting the segmented LSTM algorithm model is established, the capacity detection data of the target battery pack is matched with the LSTM algorithm model in the preset LSTM algorithm model group, and the capacity of the battery pack is detected for a batch of battery packs to be detected by using the matched target LSTM algorithm model, so that the technical problems that in the prior art, the time consumption is long and the detection cost is high when the battery pack capacity is detected are solved, the detection efficiency and the detection precision of the battery pack are improved, and the technical effect of reducing the detection cost of the battery pack is achieved.
Description
Technical Field
The embodiment of the invention relates to the technical field of battery capacity detection, in particular to a method and a device for detecting the capacity of a battery pack.
Background
During the production process of the battery pack, a capacity test needs to be carried out. The capacity parameter of the battery is an indirectly measured parameter, and the measurement is generally completed by performing full-process discharge on the battery pack by adopting an ampere-hour integration method. In the actual production process, enterprises adopt an original method and use charging and discharging equipment to carry out ampere-hour integration to carry out capacity detection.
The battery pack needs to be fully inspected according to relevant industry standard requirements. If the capacity of a production line is 10GPH (10 battery packs are produced per hour), according to the existing test scheme, the charging and discharging test channels generally need 10x (2.5 to 3) h, namely the number of the channels from 25 to 30, so that the full test requirement of the production line capacity can be ensured. This results in a huge amount of charging and discharging equipment investment, and the detection cost remains high. Some manufacturers may choose a sampling inspection mode to perform quality management, so that the validity of parameters cannot be guaranteed for undetected products.
Disclosure of Invention
The embodiment of the invention provides a method and a device for detecting the capacity of a battery pack, which solve the technical problems of long time consumption and high detection cost in the detection of the capacity of the battery pack in the prior art.
The embodiment of the invention provides a method for detecting the capacity of a battery pack, which comprises the following steps:
acquiring capacity detection data of a target battery pack according to an ampere-hour integration method, wherein the target battery pack is any one of a plurality of battery packs to be detected, and the capacity detection data at least comprises capacity data and voltage data;
importing the capacity detection data into a preset LSTM algorithm model group to determine a target LSTM algorithm model, wherein the preset LSTM algorithm model group comprises a plurality of LSTM algorithm models, and the target LSTM algorithm model is a model which can enable the detected capacity of the target battery pack to reach preset accuracy in the preset LSTM algorithm model group;
and detecting the capacity of the battery packs to be detected based on the target LSTM algorithm model.
Further, the preset LSTM algorithm model group is obtained by the following method:
sampling basic parameters of a preset number of sample battery packs at a preset sampling frequency, wherein the basic parameters at least comprise initial open-circuit voltage, termination open-circuit voltage, initial current, termination current, discharge starting time and discharge ending time of the sample battery packs;
and segmenting the basic parameters according to a time sequence, and inputting the segmented basic parameters into an LSTM neural network as input quantity to obtain the preset LSTM algorithm model group.
Further, the dividing the basic parameters according to the time sequence, and inputting the divided basic parameters into the LSTM neural network as input quantities to obtain a preset LSTM algorithm model group includes:
dividing the sampling time of the basic parameter according to the time for every 1% reduction of the charge amount;
inputting the segmented basic parameters into an LSTM neural network as input quantity, and performing segmented training at preset electric charge intervals to obtain a plurality of LSTM algorithm models, wherein the plurality of LSTM algorithm models form a preset LSTM algorithm model group, and the preset electric charge intervals are larger than 1%.
Further, the dividing the basic parameters according to the time sequence, and inputting the divided basic parameters into the LSTM neural network as input quantities to obtain a preset LSTM algorithm model group includes:
dividing the basic parameters according to a time sequence, and inputting the divided basic parameters into a combined network model of an LSTM neural network and other neural networks as input quantities to obtain a preset LSTM algorithm model group, wherein the other neural networks at least comprise one of the following components: convolutional neural networks, cyclic neural networks, gated cyclic networks.
Further, the segmented basic parameters are used as input quantities to be input into an LSTM neural network, segmented training is carried out at preset electric quantity intervals, and a plurality of LSTM algorithm models are obtained and comprise:
inputting the segmented basic parameters serving as input quantity into an LSTM neural network, and performing segmented training at the interval of 5% of electric charge quantity to obtain S 1 =SOC(5%-0%)、S 2 =SOC(10%-0%)、……、S 19 =SOC(100%-0%)、S 20 =SOC(100%-5%)、S 21 = SOC (95% -5%), … …, sm = (55% -35%), … …, sn = SOC (100% -95%), where S represents the LSTM algorithm model, n represents the number of LSTM algorithm models, and n > m > 1.
Further, when the target LSTM algorithm model is SOC (x% -y%), the detecting the capacity of the plurality of battery packs to be tested based on the target LSTM algorithm model includes:
discharging the battery pack to be tested from the initial charge amount to the charge amount of 0%, and recording the discharge time t1;
after the battery pack to be tested is placed still for first standing time, the battery pack to be tested is charged until the electric charge amount is x% by using an ampere-hour integration method, and charging time t2 is recorded;
after the battery pack to be tested is placed still for a second standing time, discharging the battery pack to be tested until the charge amount is an initial charge amount, and recording the discharging time t3, wherein the initial charge amount is less than or equal to y%;
and intercepting the battery pack parameters of the battery pack to be detected with the electric charge amount within the interval [ x% -y% ], and detecting the capacity of the battery pack to be detected based on the target LSTM algorithm model.
Further, importing the capacity detection data into a preset LSTM algorithm model group, and determining a target LSTM algorithm model comprises:
importing the capacity detection data of the target battery pack into the preset LSTM algorithm model group, and detecting the capacity of the target battery pack according to each LSTM algorithm model in the preset LSTM algorithm model group respectively to obtain a plurality of detection capacity value curves of the target battery pack;
judging whether the matching degree between the detection capacity value curve and the prediction curve of the LSTM algorithm model reaches the preset accuracy degree or not;
if so, determining the LSTM algorithm model reaching the preset accuracy as the target LSTM algorithm model;
if not, acquiring the capacity detection data of the target battery pack again, or reselecting the preset LSTM algorithm model group, and determining the target LSTM algorithm model based on the reselected preset LSTM algorithm model group.
Further, under the condition of presetting constant current, acquiring the capacity detection data of the target battery pack according to an ampere-hour integration method comprises the following steps:
charging the target battery pack from the initial charge amount to the charge amount of 100%, and recording the charging time t4;
after the target battery pack is placed still for a third standing time, discharging the target battery pack until the charge amount is 0%, recording the discharging time t5, and meanwhile calculating the discharging capacity value;
after the target battery pack is placed still for a fourth standing time, charging the target battery pack to the initial charge amount, and recording the charging time t6;
and recording and storing time data, voltage data and capacity data of the target battery pack at each charging and discharging stage as capacity detection data of the target battery pack.
The embodiment of the invention also provides a device for detecting the capacity of the battery pack, which comprises:
the device comprises a data acquisition unit, a data processing unit and a data processing unit, wherein the data acquisition unit is used for acquiring capacity detection data of a target battery pack according to an ampere-hour integration method, the target battery pack is any one of a plurality of battery packs to be detected, and the capacity detection data at least comprises capacity data and voltage data;
the model determining unit is used for importing the capacity detection data into a preset LSTM algorithm model group and determining a target LSTM algorithm model, wherein the preset LSTM algorithm model group comprises a plurality of LSTM algorithm models, and the target LSTM algorithm model is a model which can enable the detected capacity of the target battery pack to reach preset accuracy in the preset LSTM algorithm model group;
and the capacity detection unit is used for detecting the capacity of the plurality of battery packs to be detected based on the target LSTM algorithm model.
Further, the device also comprises single-channel charging and discharging equipment and a temperature control environment cabin;
the single-channel charging and discharging equipment is used for providing electric energy for the battery pack to be tested;
the temperature control environment bin is used for providing a preset temperature environment for the battery pack to be tested.
The embodiment of the invention discloses a method and a device for detecting the capacity of a battery pack, wherein the method comprises the steps of obtaining the capacity detection data of a target battery pack according to an ampere-hour integral method; importing the capacity detection data into a preset LSTM algorithm model group to determine a target LSTM algorithm model; and detecting the capacities of a plurality of battery packs to be detected based on the target LSTM algorithm model. According to the method and the device, the preset LSTM algorithm model group protecting the segmented LSTM algorithm model is established, the capacity detection data of the target battery pack is matched with the LSTM algorithm model in the preset LSTM algorithm model group, and the capacity of the battery pack is detected for a batch of battery packs to be detected by using the matched target LSTM algorithm model, so that the technical problems that in the prior art, the time consumption is long and the detection cost is high when the battery pack capacity is detected are solved, the detection efficiency and the detection precision of the battery pack are improved, and the technical effect of reducing the detection cost of the battery pack is achieved.
Drawings
Fig. 1 is a flowchart of a method for detecting capacity of a battery pack according to an embodiment of the present invention;
FIG. 2 is a cell structure diagram of an LSTM provided in an embodiment of the present invention;
fig. 3 is a structural diagram of a device for detecting the capacity of a battery pack according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
It should be noted that the terms "first", "second", and the like in the description and claims of the present invention and the accompanying drawings are used for distinguishing different objects, and are not used for limiting a specific order. The following embodiments of the present invention may be implemented individually, or in combination with each other, and the embodiments of the present invention are not limited in this respect.
Fig. 1 is a flowchart of a method for detecting capacity of a battery pack according to an embodiment of the present invention. As shown in fig. 1, the method for detecting the capacity of the battery pack specifically includes the following steps:
s101, capacity detection data of a target battery pack is obtained according to an ampere-hour integration method, wherein the target battery pack is any one of a plurality of battery packs to be detected, and the capacity detection data at least comprises capacity data and voltage data.
Specifically, the battery capacity detection generally refers to a discharge capacity (SOH) test of the battery pack, that is, a total amount of electricity discharged from the battery pack from SOC =100% to SOC =0%, wherein the battery capacity is one of important performance indexes for measuring the battery performance, and it indicates the amount of electricity discharged from the battery under certain conditions (discharge rate, temperature, end voltage, etc.), that is, the capacity of the battery, and is generally expressed in units of ampere-hour (abbreviated as a · H, 1A · H = 3600C); SOC is the State of Charge (State of Charge) of the battery, which reflects the remaining capacity of the battery, and is numerically defined as the ratio of the remaining capacity to the battery capacity, expressed as a percentage. The value range is 0~1, which indicates that the battery is completely discharged when SOC =0 and indicates that the battery is completely charged when SOC = 1. The battery pack capacity detection means that when the battery pack leaves a factory, the actual capacity of the battery is detected, namely: SOC =100% to SOC =0%, the electrical energy is discharged.
In order to ensure that the temperature data of the battery pack to be tested in the detection process are consistent, the temperature control environment cabin is adopted to ensure the detection environment temperature of the battery pack to be tested, and meanwhile, single-channel charging and discharging equipment is selected to provide electric energy for the battery pack to be tested. In order to reduce the detection time, firstly, one of the battery packs to be detected in the same batch of battery packs to be detected with the capacity to be detected is randomly selected as a target battery pack, and the capacity is detected by adopting a traditional ampere-hour integration method to obtain capacity detection data. Generally, the capacity detection data at least includes the above-mentioned capacity data and voltage data of the battery pack to be detected, and may also include other data obtained when the capacity of the battery pack is detected by using a conventional ampere-hour integration method.
Optionally, S101, obtaining capacity detection data of the target battery pack according to the ampere-hour integration method includes: under the condition of a preset constant current, charging the target battery pack from an initial charge amount to a charge amount of 100%, and recording the charging time t4; after the target battery pack is placed still for a third standing time, discharging the target battery pack until the electric charge amount is 0%, recording the discharging time t5, and meanwhile, calculating the discharging capacity value; after the target battery pack is kept still for the fourth standing time, the target battery pack is charged to the initial charge amount, and the charging time t6 is recorded; and recording and storing time data, voltage data and capacity data of the target battery pack at each charging and discharging stage as capacity detection data of the target battery pack.
Specifically, according to general production process requirements, the capacity of the battery pack when the battery pack leaves a factory is about 10% to 20%, exemplarily, the capacity of the battery pack under production is 20%, the 1C charging rate is used as an example, the initial charge amount of the battery pack is usually 20%, then the target battery pack is charged from the initial charge amount to the charge amount of 100%, that is, the target battery pack is charged from the initial SOC =20% to the SOC =100%, and the recording takes t4=0.8 hours; standing for 0.25 hours (i.e., the third standing time); discharging the target battery pack until the SOC =0% and the time t5=1 hour, and simultaneously calculating a discharge capacity value; standing for 0.25 hour again (i.e. the fourth standing time); charging the target battery pack to SOC =20%, and recording the time consumption t6=0.2 hours; the process is a process of carrying out capacity detection on the battery pack by using a traditional ampere-hour integration method, and the whole process takes about 2.5 hours.
The ampere-hour integration method can be used for obtaining capacity detection data of the target battery pack, including capacity data, voltage data, time data and the like in the charging and discharging processes.
And S102, importing the capacity detection data into a preset LSTM algorithm model group, and determining a target LSTM algorithm model, wherein the preset LSTM algorithm model group comprises a plurality of LSTM algorithm models, and the target LSTM algorithm model is a model which can enable the capacity of the target battery pack obtained through detection to reach preset accuracy in the preset LSTM algorithm model group.
Optionally, S102, importing the capacity detection data into a preset LSTM algorithm model group, and determining the target LSTM algorithm model includes: importing the capacity detection data of the target battery pack into a preset LSTM algorithm model group, and detecting the capacity of the target battery pack according to each LSTM algorithm model in the preset LSTM algorithm model group respectively to obtain a plurality of detection capacity value curves of the target battery pack; judging whether the matching degree between the detection capacity value curve and the prediction curve of the LSTM algorithm model reaches preset accuracy or not; if so, determining the LSTM algorithm model reaching the preset accuracy as a target LSTM algorithm model; and if not, re-acquiring the capacity detection data of the target battery pack, or re-selecting the preset LSTM algorithm model group, and determining the target LSTM algorithm model based on the re-selected preset LSTM algorithm model group.
Specifically, LSTM (Long Short-Term Memory network) is a time-cycle neural network used for result prediction of time-series data. Inputting the capacity data, voltage data, time data and the like of the target battery pack at different charging and discharging stages into a preset LSTM algorithm model group, calculating the capacity of the target battery pack by respectively adopting LSTM algorithm models at different charge quantity stages in the preset LSTM algorithm model group, matching the obtained detection capacity value curve with a prediction curve of the LSTM algorithm model, and performing capacity detection on a large batch of battery packs to be detected by taking the matched optimal result as the target LSTM algorithm model. For example, if the best matching result is that the prediction accuracy of the LSTM algorithm model of the SOC (50%, 30%) interval is 99%, it may be determined that the LSTM algorithm model of the interval is the target LSTM algorithm model.
If the matching degree between the detected capacity value curve and the prediction curve of the LSTM algorithm model is judged not to reach the preset accuracy, it is indicated that the capacity prediction of each LSTM algorithm model in the current preset LSTM algorithm model set on the battery pack to be tested cannot reach the prediction accuracy, and at the moment, the following methods can be adopted to solve the problem: (1) The capacity detection data of the target battery pack is obtained again, and because the problem that the acquired data are deviated due to environmental influence or equipment influence exists in the process of obtaining the capacity detection data of the target battery pack, the obtained detection capacity value curve may be inaccurate, and therefore a target LSTM algorithm model which accords with the prediction accuracy cannot be found; (2) And reselecting the preset LSTM algorithm model group, and determining the target LSTM algorithm model based on the reselected preset LSTM algorithm model group. The preset LSTM algorithm model group used currently may be obtained based on LSTM neural network training only, and if the target LSTM algorithm model meeting the prediction accuracy cannot be matched, other preset LSTM algorithm model groups obtained through training of the LSTM neural network and other neural networks can be selected from the database, so that the target LSTM algorithm model meeting the prediction accuracy is determined.
S103, detecting the capacity of the multiple battery packs to be detected based on the target LSTM algorithm model.
Specifically, after the target LSTM algorithm model is determined, the capacity of the battery packs to be tested is detected by using the target LSTM algorithm model.
Optionally, when the target LSTM algorithm model is SOC (x% -y%), S103, detecting the capacities of the multiple battery packs to be tested based on the target LSTM algorithm model includes: under the condition of a preset constant current, discharging the battery pack to be tested from the initial charge amount to the charge amount of 0%, and recording the discharge time t1; after the battery pack to be tested is placed still for the first standing time, the battery pack to be tested is charged until the charge amount is x% by using an ampere-hour integration method, and the charging time t2 is recorded; after the battery pack to be tested is placed still for the second standing time, discharging the battery pack to be tested until the charge amount is the initial charge amount, and recording the discharging time t3, wherein the initial charge amount is less than or equal to y%; and intercepting battery pack parameters of the battery pack to be detected within the interval [ x% -y% ] of the charge amount of the battery pack to be detected, and detecting the capacity of the battery pack to be detected based on a target LSTM algorithm model.
Specifically, assuming that the matched optimal LSTM algorithm model, that is, the target LSTM algorithm model is that the charge amount is SOC (50% -30%), discharging the battery pack to be tested from the initial SOC =20% to SOC =0%, and recording the time consumption t1=0.2 hours; standing for 0.25 hours (namely the first standing time); charging the battery pack to be tested to SOC =50% and consuming time t2=0.5 hours; standing for 0.25 hours again (namely the second standing time); discharging the battery pack to be tested until the SOC =20%, and recording the time consumption t3=0.3 hour; the total time consumed in the charging and discharging process is 1.5 hours, the battery pack parameters of which the electric charge amount is in the interval of (50% -30%) are intercepted, and then the capacity (SOH) of the battery pack to be detected is detected based on a target LSTM algorithm model.
For example, assuming that the batch of battery packs to be tested has 8 packs, the ampere-hour integration method needs to charge the battery to full charge and then discharge the battery to the lowest level, and the actual capacity of the battery pack is calculated in the discharging step, and the consumed electric energy is generally the nominal capacity of the battery, for example: the 100-degree battery pack needs to consume 100-degree electricity. The battery content detection by using the ampere-hour integration method takes 8 × 2.5=20 hours; and the time consumption of 1 × 2.5+7 × 1.5=13 hours for battery pack capacity detection by using the preset LSTM algorithm model group provided in the scheme is reduced by 35% and the energy consumption is reduced by about 50% compared with the traditional ampere-hour integration method.
According to the method and the device, the preset LSTM algorithm model group protecting the segmented LSTM algorithm model is established, the capacity detection data of the target battery pack is matched with the LSTM algorithm model in the preset LSTM algorithm model group, and the capacity of the battery pack is detected for a batch of battery packs to be detected by using the matched target LSTM algorithm model, so that the technical problems that in the prior art, the time consumption is long and the detection cost is high when the battery pack capacity is detected are solved, the detection efficiency and the detection precision of the battery pack are improved, and the technical effect of reducing the detection cost of the battery pack is achieved.
Optionally, the preset LSTM algorithm model set is obtained by: sampling basic parameters of a preset number of sample battery packs at a preset sampling frequency, wherein the basic parameters at least comprise initial open-circuit voltage, termination open-circuit voltage, initial current, termination current, discharge starting time and discharge finishing time of the sample battery packs; and (3) dividing the basic parameters according to the time sequence, and inputting the divided basic parameters into an LSTM neural network as input quantity to obtain a preset LSTM algorithm model group.
Specifically, in the process of detecting the SOH of a preset number of sample battery packs by using an ampere-hour integration method, sampling basic parameters at a preset sampling frequency, wherein the basic parameters at least comprise initial open-circuit voltage, termination open-circuit voltage, initial current, termination current, discharge starting time and discharge finishing time of the sample battery packs; and segmenting the basic parameters by using a time sequence, inputting the segmented basic parameters into an LSTM neural network as input quantity for model training, and finally obtaining a preset LSTM algorithm model group.
In the embodiment of the present invention, the LSTM is characterized in that valve nodes of each layer are added outside an RNN (Recurrent Neural Network) structure. The valves are of type 3: forgetting the valve (forget gate), the input valve (input gate) and the output valve (output gate). These valves can be opened or closed to add a determination of whether the memory state of the model network (the state of the previous network) at the layer output reaches a threshold value to the current layer calculation.
FIG. 2 is a cell structure diagram of the LSTM provided in the examples of the present invention. As shown in FIG. 2, the LSTM has two transmission states, one beingc t (cell status), one is h t (hidden state). The transfer formula between the states is as follows:
；
；
；
(ii) a Wherein,
after the splicing vector is multiplied by the weight matrix, the splicing vector is converted into a numerical value between 0 and 1 through a sigmoid activation function to serve as a gating state. And z is the value between-1 and 1 to which the result is converted by a tanh activation function. Wherein,
，
，
(ii) a w is a weight parameter of the perceptual model,
threshold structure weight, x, for output gate sigmoid t As the current input, y t Is the output of the last moment in time,
the function is activated for sigmoid.
Optionally, the dividing the basic parameters according to the time sequence, and inputting the divided basic parameters into the LSTM neural network as input quantities to obtain the preset LSTM algorithm model group includes: dividing the sampling time of the basic parameter according to the time for each 1% reduction of the charge amount; and inputting the segmented basic parameters serving as input quantities into an LSTM neural network, and performing segmented training at preset electric charge intervals to obtain a plurality of LSTM algorithm models, wherein the plurality of LSTM algorithm models form a preset LSTM algorithm model group, and the preset electric charge intervals are larger than 1%.
Specifically, the basic parameter is divided in a cyclic recursion manner at intervals of every 1% of SOC, that is, the sampling time of the basic parameter is divided according to the time taken for every 1% of charge amount to decrease, the divided basic parameter is input to the LSTM neural network as an input quantity, and the segmented training is performed at preset charge amount intervals, wherein the preset charge amount intervals can be selected from 5%, 10% and the like as required, and no specific limitation is made herein, and after the LSTM algorithm models of different charge amount intervals are obtained, the LSTM algorithm models of different charge amount intervals form a preset LSTM algorithm model group.
Preferably, the segmented basic parameters are input into the LSTM neural network as input quantities, and segmented training is performed at intervals of a preset charge quantity, and obtaining a plurality of LSTM algorithm models includes: inputting the divided basic parameters into an LSTM neural network as input quantity, and performing segmentation training at the interval of 5% of electric charge quantity to obtain S 1 =SOC(5%-0%)、S 2 =SOC(10%-0%)、……、S 19 =SOC(100%-0%)、S 20 =SOC(100%-5%)、S 21 = SOC (95% -5%), … …, sm = (55% -35%), … …, sn = SOC (100% -95%), where S represents the LSTM algorithm model, n represents the number of LSTM algorithm models, and n > m > 1.
Specifically, it is preferable that the preset charge amount interval is 5%, and S can be obtained by dividing the preset charge amount interval in a cyclic recursive manner 1 =SOC(5%-0%)、S 2 =SOC(10%-0%)、……、S 19 =SOC(100%-0%)、S 20 =SOC(100%-5%)、S 21 = SOC (95% -5%), … …, sm = (55% -35%), … …, sn = SOC (100% -95%) segmented LSTM algorithm model, where S is a linear transformation model 19 The data generated in the interval of (1) is the complete ampere-hour integration method.
Preferably, the dividing the basic parameters according to the time sequence, and inputting the divided basic parameters into the LSTM neural network as input quantities to obtain the preset LSTM algorithm model group includes: dividing the basic parameters according to the time sequence, and inputting the divided basic parameters serving as input quantities into a combined network model of the LSTM neural network and other neural networks to obtain a preset LSTM algorithm model group, wherein the other neural networks at least comprise one of the following parts: convolutional neural networks, cyclic neural networks, gated cyclic networks.
In particular, there are many variations of LSTM, such as: RNN-LSTM, BI-LSTM (bidirectional long-short term memory network), and the like, besides LSTM variant neural networks, other networks such as convolutional neural networks, recurrent neural networks, gated recurrent networks, and the like, in order to make the trained model more accurate, model training can be performed by using a combined network model of LSTM and one or more of the other neural networks to obtain a preset LSTM algorithm model group. In the subsequent detection stage of the battery pack to be detected, the capacity value of the battery pack to be detected can be accurately obtained only by importing the capacity detection data of the target battery pack into the preset LSTM algorithm model group and finding the optimal target LSTM algorithm model (which may be obtained by training the LSTM in combination with other neural networks).
Fig. 3 is a structural diagram of a device for detecting the capacity of a battery pack according to an embodiment of the present invention, the device mainly includes: a data acquisition unit 31, a model determination unit 32, a capacity detection unit 33, wherein:
the data acquisition unit 31 is configured to acquire capacity detection data of a target battery pack according to an ampere-hour integration method, where the target battery pack is any one of a plurality of battery packs to be detected, and the capacity detection data at least includes capacity data and voltage data;
the model determining unit 32 is configured to import the capacity detection data into a preset LSTM algorithm model group, and determine a target LSTM algorithm model, where the preset LSTM algorithm model group includes a plurality of LSTM algorithm models, and the target LSTM algorithm model is a model that can enable the detected capacity of the target battery pack to reach a preset accuracy in the preset LSTM algorithm model group;
and a capacity detection unit 33, configured to detect capacities of the multiple battery packs to be tested based on the target LSTM algorithm model.
Optionally, the device for detecting the capacity of the battery pack further comprises a single-channel charging and discharging device and a temperature control environment bin;
the single-channel charging and discharging equipment is used for providing electric energy for the battery pack to be tested;
the temperature control environment bin is used for providing a preset temperature environment for the battery pack to be tested.
Optionally, when the target LSTM algorithm model is SOC (x% -y%), the capacity detecting unit 33 is specifically configured to:
under the condition of a preset constant current, discharging the battery pack to be tested from the initial charge amount to the charge amount of 0%, and recording the discharge time t1;
after the battery pack to be tested is placed still for the first standing time, the battery pack to be tested is charged until the charge amount is x% by using an ampere-hour integration method, and the charging time t2 is recorded;
after the battery pack to be tested is placed still for the second standing time, discharging the battery pack to be tested until the charge amount is the initial charge amount, and recording the discharging time t3, wherein the initial charge amount is less than or equal to y%;
and intercepting battery pack parameters of the battery pack to be detected within the interval [ x% -y% ] of the charge amount of the battery pack to be detected, and detecting the capacity of the battery pack to be detected based on a target LSTM algorithm model.
Optionally, the model determining unit 32 is specifically configured to:
importing the capacity detection data of the target battery pack into a preset LSTM algorithm model group, and detecting the capacity of the target battery pack according to each LSTM algorithm model in the preset LSTM algorithm model group respectively to obtain a plurality of detection capacity value curves of the target battery pack;
judging whether the matching degree between the detection capacity value curve and the prediction curve of the LSTM algorithm model reaches preset accuracy or not;
if so, determining the LSTM algorithm model reaching the preset accuracy as a target LSTM algorithm model;
and if not, re-acquiring the capacity detection data of the target battery pack, or re-selecting the preset LSTM algorithm model group, and determining the target LSTM algorithm model based on the re-selected preset LSTM algorithm model group.
Optionally, the data obtaining unit 31 is specifically configured to:
under the condition of a preset constant current, charging the target battery pack from an initial charge amount to a charge amount of 100%, and recording the charging time t4;
after the target battery pack is kept still for the third standing time, discharging the target battery pack until the electric charge amount is 0%, recording the discharging time t5, and meanwhile, calculating the discharging capacity value;
after the target battery pack is placed still for the fourth standing time, charging the target battery pack to the initial charge amount, and recording the charging time t6;
and recording and storing time data, voltage data and capacity data of the target battery pack at each charging and discharging stage as capacity detection data of the target battery pack.
The device provided by the embodiment of the present invention has the same implementation principle and technical effect as the method embodiments, and for the sake of brief description, reference may be made to the corresponding contents in the method embodiments without reference to the device embodiments.
The device for detecting the capacity of the battery pack provided by the embodiment of the invention has the same technical characteristics as the method for detecting the capacity of the battery pack provided by the embodiment, so that the same technical problems can be solved, and the same technical effects are achieved.
In the description of the embodiments of the present invention, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
Finally, it should be noted that the above-mentioned embodiments are only preferred embodiments of the present invention and the technical principles applied. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (7)
1. A method for detecting the capacity of a battery pack, the method comprising:
acquiring capacity detection data of a target battery pack according to an ampere-hour integration method, wherein the target battery pack is any one of a plurality of battery packs to be detected, and the capacity detection data at least comprises capacity data and voltage data;
importing the capacity detection data into a preset LSTM algorithm model group to determine a target LSTM algorithm model, wherein the preset LSTM algorithm model group comprises a plurality of LSTM algorithm models, the plurality of LSTM algorithm models are LSTM algorithm models in different charge quantity stages, and the target LSTM algorithm model is a model which can enable the detected capacity of the target battery pack to reach preset accuracy in the preset LSTM algorithm model group;
detecting the capacity of a plurality of battery packs to be detected based on the target LSTM algorithm model;
the preset LSTM algorithm model group is obtained by the following method:
sampling basic parameters of a preset number of sample battery packs at a preset sampling frequency, wherein the basic parameters at least comprise initial open-circuit voltage, termination open-circuit voltage, initial current, termination current, discharge starting time and discharge ending time of the sample battery packs;
dividing the basic parameters according to a time sequence, and inputting the divided basic parameters serving as input quantities into an LSTM neural network to obtain the preset LSTM algorithm model group;
inputting the divided basic parameters into an LSTM neural network as input quantity, and performing segmentation training at 5% charge quantity intervals to obtain S1= SOC (5% -0%), S2= SOC (10% -0%), … …, S19= SOC (100% -0%), S20= SOC (100% -5%), S21= SOC (95% -5%), … …, sm = (55% -35%), … … and Sn = SOC (100% -95%), wherein S represents an LSTM algorithm model, n represents the number of LSTM algorithm models, and n > m > 1;
when the target LSTM algorithm model is SOC (x% -y%), the detecting the capacity of the plurality of battery packs to be tested based on the target LSTM algorithm model includes:
under the condition of a preset constant current, discharging the battery pack to be tested from the initial charge amount to the charge amount of 0%, and recording the discharge time t1;
after the battery pack to be tested is placed still for first standing time, the battery pack to be tested is charged until the charge amount is x% by using an ampere-hour integration method, and charging time t2 is recorded;
after the battery pack to be tested is placed still for a second standing time, discharging the battery pack to be tested until the charge amount is an initial charge amount, and recording the discharge time t3, wherein the initial charge amount is less than or equal to y%;
and intercepting battery pack parameters of the battery pack to be detected with the electric charge amount within an interval [ x% -y% ] and detecting the capacity of the battery pack to be detected based on the target LSTM algorithm model.
2. The method for detecting battery pack capacity according to claim 1, wherein the step of dividing the basic parameters according to a time sequence and inputting the divided basic parameters into an LSTM neural network as input quantities to obtain a preset LSTM algorithm model set comprises:
dividing the sampling time of the basic parameter according to the time for every 1% reduction of the charge amount;
inputting the segmented basic parameters into an LSTM neural network as input quantity, and performing segmented training at preset electric charge intervals to obtain a plurality of LSTM algorithm models, wherein the plurality of LSTM algorithm models form a preset LSTM algorithm model group, and the preset electric charge intervals are larger than 1%.
3. The method for detecting battery pack capacity according to claim 1, wherein the step of dividing the basic parameters according to a time sequence and inputting the divided basic parameters into an LSTM neural network as input quantities to obtain a preset LSTM algorithm model set comprises:
dividing the basic parameters according to a time sequence, and inputting the divided basic parameters into a combined network model of an LSTM neural network and other neural networks as input quantities to obtain a preset LSTM algorithm model group, wherein the other neural networks comprise: convolutional neural networks, cyclic neural networks, gated cyclic networks.
4. The method for detecting the capacity of the battery pack according to claim 1, wherein the importing the capacity detection data into a preset LSTM algorithm model group, and the determining the target LSTM algorithm model comprises:
importing the capacity detection data of the target battery pack into the preset LSTM algorithm model group, and detecting the capacity of the target battery pack according to each LSTM algorithm model in the preset LSTM algorithm model group respectively to obtain a plurality of detection capacity value curves of the target battery pack;
judging whether the matching degree between the detection capacity value curve and the prediction curve of the LSTM algorithm model reaches the preset accuracy degree or not;
if so, determining the LSTM algorithm model reaching the preset accuracy as the target LSTM algorithm model;
if not, acquiring the capacity detection data of the target battery pack again, or reselecting the preset LSTM algorithm model group, and determining the target LSTM algorithm model based on the reselected preset LSTM algorithm model group.
5. The method for detecting the capacity of the battery pack according to claim 1, wherein the step of obtaining the capacity detection data of the target battery pack according to the ampere-hour integration method comprises the steps of:
under the condition of a preset constant current, charging the target battery pack from an initial charge amount to a charge amount of 100%, and recording charging time t4;
after the target battery pack is placed still for a third standing time, discharging the target battery pack until the charge amount is 0%, recording the discharging time t5, and meanwhile calculating the discharging capacity value;
after the target battery pack is placed still for a fourth standing time, charging the target battery pack to the initial charge amount, and recording the charging time t6;
and recording and storing time data, voltage data and capacity data of the target battery pack at each charging and discharging stage as capacity detection data of the target battery pack.
6. An apparatus for detecting the capacity of a battery pack, the apparatus comprising:
the device comprises a data acquisition unit, a data processing unit and a data processing unit, wherein the data acquisition unit is used for acquiring capacity detection data of a target battery pack according to an ampere-hour integration method, the target battery pack is any one of a plurality of battery packs to be detected, and the capacity detection data at least comprises capacity data and voltage data;
the model determining unit is used for importing the capacity detection data into a preset LSTM algorithm model group and determining a target LSTM algorithm model, wherein the preset LSTM algorithm model group comprises a plurality of LSTM algorithm models, the LSTM algorithm models are LSTM algorithm models in different charge quantity stages, and the target LSTM algorithm model is a model which can enable the capacity of the target battery pack obtained through detection to reach preset accuracy in the preset LSTM algorithm model group;
the capacity detection unit is used for detecting the capacity of the battery packs to be detected based on the target LSTM algorithm model;
the preset LSTM algorithm model group is obtained by the following method:
sampling basic parameters of a preset number of sample battery packs at a preset sampling frequency, wherein the basic parameters at least comprise initial open-circuit voltage, final open-circuit voltage, initial current, final current, discharge starting time and discharge ending time of the sample battery packs;
dividing the basic parameters according to a time sequence, and inputting the divided basic parameters into an LSTM neural network as input quantity to obtain the preset LSTM algorithm model group;
inputting the divided basic parameters into an LSTM neural network as input quantity, and performing segmentation training at 5% charge quantity intervals to obtain S1= SOC (5% -0%), S2= SOC (10% -0%), … …, S19= SOC (100% -0%), S20= SOC (100% -5%), S21= SOC (95% -5%), … …, sm = (55% -35%), … … and Sn = SOC (100% -95%), wherein S represents an LSTM algorithm model, n represents the number of LSTM algorithm models, and n > m > 1;
when the target LSTM algorithm model is SOC (x% -y%), the capacity detection unit is specifically configured to:
under the condition of a preset constant current, discharging the battery pack to be tested from the initial charge amount to the charge amount of 0%, and recording the discharge time t1;
after the battery pack to be tested is placed still for the first standing time, the battery pack to be tested is charged until the charge amount is x% by using an ampere-hour integration method, and the charging time t2 is recorded;
after the battery pack to be tested is placed still for the second standing time, discharging the battery pack to be tested until the charge amount is the initial charge amount, and recording the discharging time t3, wherein the initial charge amount is less than or equal to y%;
and intercepting battery pack parameters of the battery pack to be detected within the interval [ x% -y% ] of the charge amount of the battery pack to be detected, and detecting the capacity of the battery pack to be detected based on a target LSTM algorithm model.
7. The device for detecting the capacity of the battery pack according to claim 6, further comprising a single-channel charging and discharging device and a temperature control environment bin;
the single-channel charging and discharging equipment is used for providing electric energy for the battery pack to be tested;
the temperature control environment bin is used for providing a preset temperature environment for the battery pack to be tested.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210785324.5A CN114859249B (en) | 2022-07-06 | 2022-07-06 | Method and device for detecting battery pack capacity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210785324.5A CN114859249B (en) | 2022-07-06 | 2022-07-06 | Method and device for detecting battery pack capacity |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114859249A CN114859249A (en) | 2022-08-05 |
| CN114859249B true CN114859249B (en) | 2022-10-18 |
Family
ID=82627046
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210785324.5A Active CN114859249B (en) | 2022-07-06 | 2022-07-06 | Method and device for detecting battery pack capacity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114859249B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117054892B (en) * | 2023-10-11 | 2024-02-27 | 特变电工西安电气科技有限公司 | Evaluation method, device and management method for battery state of energy storage power station |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110824364A (en) * | 2019-10-24 | 2020-02-21 | 重庆邮电大学 | Lithium battery SOH estimation and RUL prediction method based on AST-LSTM neural network |
| CN111220921A (en) * | 2020-01-08 | 2020-06-02 | 重庆邮电大学 | Lithium battery capacity estimation method based on improved convolution-long short-term memory neural network |
| CN111948563A (en) * | 2020-06-19 | 2020-11-17 | 浙江大学 | A method for predicting the remaining life of electric forklift lithium battery based on multi-neural network coupling |
| CN112051511A (en) * | 2020-08-26 | 2020-12-08 | 华中科技大学 | Method and system for estimating state of health of power battery based on multi-channel technology |
| CN112098848A (en) * | 2020-09-14 | 2020-12-18 | 北京大学深圳研究生院 | Battery discharge capacity prediction method and system and readable storage medium |
| CN112241608A (en) * | 2020-10-13 | 2021-01-19 | 国网湖北省电力有限公司电力科学研究院 | A Lithium Battery Life Prediction Method Based on LSTM Network and Transfer Learning |
| CN112986830A (en) * | 2021-04-22 | 2021-06-18 | 湖北工业大学 | Lithium battery capacity estimation method based on convolution time memory neural network |
| CN113253116A (en) * | 2021-05-18 | 2021-08-13 | 齐鲁工业大学 | Lithium ion battery state of charge estimation method and storage medium |
| CN114545277A (en) * | 2022-02-18 | 2022-05-27 | 北京理工新源信息科技有限公司 | Power battery retirement prediction method based on capacity attenuation |
-
2022
- 2022-07-06 CN CN202210785324.5A patent/CN114859249B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110824364A (en) * | 2019-10-24 | 2020-02-21 | 重庆邮电大学 | Lithium battery SOH estimation and RUL prediction method based on AST-LSTM neural network |
| CN111220921A (en) * | 2020-01-08 | 2020-06-02 | 重庆邮电大学 | Lithium battery capacity estimation method based on improved convolution-long short-term memory neural network |
| CN111948563A (en) * | 2020-06-19 | 2020-11-17 | 浙江大学 | A method for predicting the remaining life of electric forklift lithium battery based on multi-neural network coupling |
| CN112051511A (en) * | 2020-08-26 | 2020-12-08 | 华中科技大学 | Method and system for estimating state of health of power battery based on multi-channel technology |
| CN112098848A (en) * | 2020-09-14 | 2020-12-18 | 北京大学深圳研究生院 | Battery discharge capacity prediction method and system and readable storage medium |
| CN112241608A (en) * | 2020-10-13 | 2021-01-19 | 国网湖北省电力有限公司电力科学研究院 | A Lithium Battery Life Prediction Method Based on LSTM Network and Transfer Learning |
| CN112986830A (en) * | 2021-04-22 | 2021-06-18 | 湖北工业大学 | Lithium battery capacity estimation method based on convolution time memory neural network |
| CN113253116A (en) * | 2021-05-18 | 2021-08-13 | 齐鲁工业大学 | Lithium ion battery state of charge estimation method and storage medium |
| CN114545277A (en) * | 2022-02-18 | 2022-05-27 | 北京理工新源信息科技有限公司 | Power battery retirement prediction method based on capacity attenuation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114859249A (en) | 2022-08-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Battery life estimation based on cloud data for electric vehicles | |
| Ling et al. | State-of-charge and state-of-health estimation for lithium-ion batteries based on dual fractional-order extended Kalman filter and online parameter identification | |
| Pan et al. | State of charge estimation of lithium-ion batteries using a grey extended Kalman filter and a novel open-circuit voltage model | |
| Nuroldayeva et al. | State of health estimation methods for lithium‐ion batteries | |
| KR102106775B1 (en) | Diagnosis method of battery using Deep Learning | |
| CN114280479B (en) | Retired battery rapid sorting method based on electrochemical impedance spectroscopy | |
| Takyi-Aninakwa et al. | An optimized long short-term memory-weighted fading extended Kalman filtering model with wide temperature adaptation for the state of charge estimation of lithium-ion batteries | |
| CN112051511A (en) | Method and system for estimating state of health of power battery based on multi-channel technology | |
| Li et al. | The lithium-ion battery state-of-charge estimation using random forest regression | |
| CN117054892B (en) | Evaluation method, device and management method for battery state of energy storage power station | |
| Zhou et al. | Research on online estimation of available capacity of lithium batteries based on daily charging data | |
| CN115158076A (en) | Metering error evaluation method, device and computer readable storage medium | |
| CN113687251B (en) | Double-model-based lithium ion battery pack voltage abnormality fault diagnosis method | |
| CN117805649A (en) | Method for identifying abnormal battery cells based on SOH quantized battery capacity attenuation | |
| CN107037375B (en) | Method and device for measuring direct current internal resistance of battery | |
| CN110412472A (en) | A battery state of charge estimation method based on normal gamma filter | |
| Jiang et al. | A battery capacity estimation method based on the equivalent circuit model and quantile regression using vehicle real-world operation data | |
| CN114859249B (en) | Method and device for detecting battery pack capacity | |
| CN114720899A (en) | Retired battery echelon utilization and sorting method and system, electronic equipment and storage medium | |
| Ye et al. | Enhanced robust capacity estimation of lithium-ion batteries with unlabeled dataset and semi-supervised machine learning | |
| Xiong et al. | A novel method for state of charge estimation of lithium-ion batteries at low-temperatures | |
| CN113884934A (en) | A lithium battery self-discharge screening method, device, electronic device and storage medium | |
| CN111257758B (en) | SOH estimation method for emergency lead-acid storage battery of power station | |
| Mullapudi et al. | Unified Approach for Estimation of State of Charge and Remaining Useful Life of Li-Ion batteries using Deep Learning | |
| CN117347862A (en) | Method and device for predicting discharge capacity based on lithium battery voltage |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |