Method, device, equipment and medium for predicting service life of lithium-ion power battery

Abstract

The application relates to a service life prediction method, device, equipment and medium of a lithium ion power battery, and belongs to the technical field of batteries. The application can acquire charging data; wherein the charging data at least includes SOC values corresponding to each sampling period; target charging segments are divided according to the difference between the SOC values corresponding to adjacent sampling periods; statistical information corresponding to each target charging segment is determined; a common SOC interval for each target charging segment is selected according to the statistical information and a preset SOC interval determination condition; capacity information corresponding to the common SOC interval is calculated according to a preset ampere-hour integral calculation formula; the capacity information is used as training data to train a to-be-trained model, and a trained sequence prediction model is obtained; the remaining service life of the lithium ion power battery is predicted through the trained sequence prediction model, and a prediction result is obtained. Through the application, the service life prediction accuracy of the lithium ion power battery can be improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of batteries, and particularly relates to a service life prediction method, device, equipment and medium of a lithium-ion power battery. BACKGROUND

[0002] At present, the lithium-ion power battery has been widely applied to the field of new energy vehicles and serves as one of the main power sources of new energy vehicles.

[0003] In practice, it is found that the lithium-ion power battery will have a performance degradation phenomenon with the normal use time. In this regard, the service life of the lithium-ion power battery needs to be predicted so as to better use the lithium-ion power battery based on the service life as a reference. However, the existing service life prediction method of the lithium-ion power battery usually obtains the capacity according to the full charging segment, and the matching degree with the actual use habit is not high, thereby resulting in an inaccurate prediction result. Moreover, the existing service life prediction method is only applicable to the short-term capacity decline prediction problem, and has the problem of low accuracy for the long-term capacity decline prediction.

[0004] In summary, the existing service life prediction method of the lithium-ion power battery has the problem of low prediction accuracy. SUMMARY

[0005] Therefore, the application provides a service life prediction method, device, equipment and medium of a lithium-ion power battery, which helps to improve the service life prediction accuracy of the lithium-ion power battery.

[0006] To achieve the above object, the application adopts the following technical solutions:

[0007] In a first aspect, the application provides a service life prediction method of a lithium-ion power battery, which comprises the following steps:

[0008] obtaining charging data; wherein the charging data at least includes an SOC value corresponding to each sampling period; and dividing to obtain each target charging segment according to the difference between the SOC values corresponding to adjacent sampling periods;

[0009] determining statistical information corresponding to each target charging segment;

[0010] selecting a common SOC interval for each target charging segment according to the statistical information and a preset SOC interval determination condition;

[0011] calculating capacity information corresponding to the common SOC interval according to a preset ampere-hour integral calculation formula;

[0012] using the capacity information as training data to train a to-be-trained model to obtain a trained sequence prediction model;

[0013] The sequence prediction model is trained to predict the remaining service life of the lithium-ion power battery, and a prediction result is obtained.

[0014] Further, the difference between the SOC values corresponding to adjacent sampling periods is divided to obtain each target charging segment, including:

[0015] For each sampling period, the difference between the SOC value of the sampling period and the SOC value of the last sampling period is calculated;

[0016] If the difference between the SOC value of the sampling period and the SOC value of the last sampling period is equal to 0 or 1, it is determined that the sampling period and the last sampling period belong to the same charging segment;

[0017] If the difference between the SOC value of the sampling period and the SOC value of the last sampling period is not equal to 0 and not equal to 1, it is determined that the sampling period and the last sampling period do not belong to the same charging segment;

[0018] According to the sampling periods belonging to the same charging segment, each initial charging segment is generated;

[0019] The initial charging segments are selected to obtain the target charging segments.

[0020] Further, the initial charging segments are selected to obtain the target charging segments, including:

[0021] Obtain the SOC value distribution data corresponding to each initial charging segment;

[0022] According to the SOC value distribution data, a target SOC threshold is determined; wherein the target SOC threshold includes at least one of the following: a maximum initial SOC value corresponding to a preset percentile, a minimum cutoff SOC value, and a minimum charging SOC change value;

[0023] According to the target SOC threshold, the initial charging segments are selected to obtain the target charging segments.

[0024] Further, the statistical information includes statistical information of original data features in the target charging segments, and the original data features include at least one of the following: time, temperature, current, voltage, SOC value, and charging frequency; and

[0025] The preset SOC interval determination condition includes:

[0026] The interval range of the SOC interval is greater than a preset range threshold; and

[0027] The SOC interval corresponds to the same charging current; and

[0028] The number of target charging segments containing the SOC interval is greater than a preset number threshold.

[0029] Further, capacity information corresponding to the common SOC interval is calculated according to a preset ampere-hour integral calculation formula, including:

[0030] A charging initial state battery state of charge amount, a charging final state battery state of charge amount, a charging current, and a charging time corresponding to the common SOC interval are determined;

[0031] The charging initial state battery state of charge amount, the charging final state battery state of charge amount, the charging current, and the charging time are substituted into the preset ampere-hour integral calculation formula to obtain an initial capacity sequence corresponding to the common SOC interval;

[0032] The initial capacity sequence is subjected to capacity value correction processing and outlier elimination processing to obtain a target capacity sequence;

[0033] The target capacity sequence is subjected to normalization processing to obtain a normalized capacity sequence, and the normalized capacity sequence is determined as the capacity information.

[0034] Further, the trained sequence prediction model includes an LSTM encoder, an LSTM decoder, and an attention mechanism.

[0035] Further, the capacity information is used as training data to train a to-be-trained model to obtain a trained sequence prediction model, including:

[0036] A first capacity sequence of a first time step length and a second capacity sequence of a second time step length are selected from the capacity information;

[0037] The first capacity sequence and a charging frequency sequence of the first time step length are used as input data of the to-be-trained model, and the second capacity sequence is used as output data of the to-be-trained model;

[0038] The input data is input into the to-be-trained model to obtain prediction data of the to-be-trained model;

[0039] The to-be-trained model is iteratively trained based on the prediction data and the output data to obtain the trained sequence prediction model.

[0040] In a second aspect, the application provides a service life prediction device for a lithium ion power battery, including:

[0041] The data acquisition unit is configured to acquire charging data, wherein the charging data at least includes SOC values corresponding to respective sampling periods;

[0042] The segment division unit is configured to divide the respective target charging segments according to differences between the SOC values corresponding to adjacent sampling periods.

[0043] The statistical unit is configured to determine statistical information corresponding to the respective target charging segments.

[0044] The interval determination unit is configured to select a common SOC interval for the respective target charging segments according to the statistical information and a preset SOC interval determination condition.

[0045] The capacity calculation unit is configured to calculate capacity information corresponding to the common SOC interval according to a preset ampere-hour integral calculation formula.

[0046] The model training unit is configured to train a to-be-trained model by taking the capacity information as training data, to obtain a trained sequence prediction model.

[0047] The prediction unit is configured to predict the remaining service life of the lithium-ion power battery by using the trained sequence prediction model, to obtain a prediction result.

[0048] Further, the segment division unit is specifically configured to:

[0049] For each sampling period, calculate a difference between the SOC value of the sampling period and the SOC value of the previous sampling period.

[0050] If the difference between the SOC value of the sampling period and the SOC value of the previous sampling period is equal to 0 or 1, it is determined that the sampling period and the previous sampling period belong to the same charging segment.

[0051] If the difference between the SOC value of the sampling period and the SOC value of the previous sampling period is not equal to 0 and not equal to 1, it is determined that the sampling period and the previous sampling period do not belong to the same charging segment.

[0052] According to the sampling periods belonging to the same charging segment, the respective initial charging segments are generated.

[0053] The respective initial charging segments are selected, to obtain the respective target charging segments.

[0054] Further, the segment division unit is specifically configured to:

[0055] Obtain SOC value distribution data corresponding to the respective initial charging segments.

[0056] determining a target SOC threshold according to the SOC value distribution data, wherein the target SOC threshold comprises at least one of: a maximum initial SOC value corresponding to a preset percentile, a minimum cut-off SOC value, and a minimum charging SOC change value;

[0057] performing selection on the initial charging segments according to the target SOC threshold, to obtain target charging segments.

[0058] Further, the statistical information comprises statistical information of original data features in the target charging segments, and the original data features comprise at least one of: time, temperature, current, voltage, SOC value, and charging frequency.

[0059] The preset SOC interval determination condition comprises:

[0060] an interval range of the SOC interval is greater than a preset range threshold; and

[0061] the SOC interval corresponds to the same charging current; and

[0062] a number of target charging segments containing the SOC interval is greater than a preset number threshold.

[0063] Further, the capacity calculation unit is specifically configured to:

[0064] determine a charging initial state battery charge state amount, a charging final state battery charge state amount, a charging current, and a charging time corresponding to the common SOC interval;

[0065] substitute the charging initial state battery charge state amount, the charging final state battery charge state amount, the charging current, and the charging time into the preset ampere-hour integral calculation formula, to obtain an initial capacity sequence corresponding to the common SOC interval;

[0066] perform capacity value correction processing and outlier elimination processing on the initial capacity sequence, to obtain a target capacity sequence;

[0067] perform normalization processing on the target capacity sequence, to obtain a normalized capacity sequence, and determine the normalized capacity sequence as the capacity information.

[0068] Further, the trained sequence prediction model comprises an LSTM encoder, an LSTM decoder, and an attention mechanism.

[0069] Further, the model training unit is specifically configured to:

[0070] select a first capacity sequence of a first time step length and a second capacity sequence of a second time step length from the capacity information;

[0071] input data of the to-be-trained model, and the second capacity sequence is output data of the to-be-trained model;

[0072] input the input data into the to-be-trained model to obtain prediction data of the to-be-trained model;

[0073] based on the prediction data and the output data, iteratively train the to-be-trained model to obtain the trained sequence prediction model.

[0074] In a third aspect, the present application provides an electronic device, comprising:

[0075] at least one processor; and

[0076] a memory in communication connection with the at least one processor; wherein

[0077] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method of the first aspect.

[0078] In a fourth aspect, the present application provides a non-transitory computer readable storage medium storing computer instructions, characterized in that the computer instructions are used to enable the computer to execute the method of the first aspect.

[0079] In a fifth aspect, the present application provides a computer program product comprising a computer program, which, when executed by a processor, implements the method of the first aspect.

[0080] The present application has at least the following beneficial effects:

[0081] Through the present application, the plurality of charging segments can be divided according to the difference between the SOC values corresponding to each adjacent sampling period, and the common SOC interval is selected according to the charging condition statistical information of each charging segment and in combination with the preset SOC interval determination condition. The capacity information auxiliary sequence prediction model is trained based on the common SOC interval. Since the capacity information calculation is more accurate and the sequence prediction method has better prediction effect in long-term prediction, the use of the trained sequence prediction model in the present application for remaining useful life prediction can improve the prediction accuracy of the useful life of the lithium ion power battery.

[0082] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0083] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced. Obviously, the accompanying drawings in the following description only only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0084] Figure 1 is a flow chart of a lithium-ion power battery life prediction method according to an exemplary embodiment;

[0085] Figure 2 is a flow chart of a lithium-ion power battery life prediction method according to another exemplary embodiment;

[0086] Figure 3 is a block diagram structure schematic diagram of a lithium-ion power battery life prediction device according to an exemplary embodiment;

[0087] Figure 4 is a block diagram structure schematic diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0088] In order to make the purpose, technical solutions and advantages of the present application more clear, the technical solutions of the present application will be described in detail below. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of the present application.

[0089] Please refer to Figure 1 , Figure 1 is a flow chart of a lithium-ion power battery life prediction method according to an exemplary embodiment, the lithium-ion power battery life prediction method comprising the following steps:

[0090] Step S101, obtaining charging data; wherein the charging data at least includes the SOC value corresponding to each sampling period.

[0091] In this embodiment, the execution subject can be a terminal device, a server or other electronic device.

[0092] The execution subject can set a sampling period in advance, and obtain the related data of the lithium-ion power battery in the new energy vehicle according to the sampling period. The execution subject can judge the parking charging state of the new energy vehicle. If the new energy vehicle is in the parking charging state, the related data of the lithium-ion power battery collected in each sampling period at this time is determined as the charging data. For example, the sampling period can be every ten seconds, and the SOC value corresponding to each sampling period can be the SOC value every ten seconds in the charging process.

[0093] The related data of the lithium-ion power battery can include but is not limited to time, temperature, current, voltage, SOC (state-of-charge), and charging frequency.

[0094] The SOC value can be the ratio of the charging capacity to the rated capacity, expressed in percentage, and the value range is 0-1. When SOC=0, it indicates that the battery is completely discharged, and when SOC=1, it indicates that the battery is completely charged.

[0095] It can be understood that the SOC value corresponding to each sampling period in the charging data presents a non-decreasing change trend with the increase of time.

[0096] In step S102, the SOC values corresponding to adjacent sampling periods are divided into target charging segments according to the difference between the SOC values.

[0097] In this embodiment, the execution subject can calculate the difference between the SOC value of each sampling period and the SOC value of the previous sampling period, and then judge whether the two sampling periods belong to the same charging process according to the difference, so as to divide each target charging segment of the new energy vehicle based on the difference.

[0098] Each target charging segment is associated with the related data of the lithium-ion power battery corresponding to the target charging segment.

[0099] In step S103, the statistical information corresponding to each target charging segment is determined.

[0100] In this embodiment, the execution subject can perform data statistics according to the related data of the lithium-ion power battery associated with each target charging segment, and obtain statistical information. The statistical information can reflect the charging habit information of the new energy vehicle, and can include but is not limited to the overall charging frequency habit, the charging depth habit, the charging current information, the charging environment, the charging temperature change, etc., which are not limited in this embodiment.

[0101] Specifically, the statistical information can include, but is not limited to, statistical data between initial and final SOC points and charging frequency distribution, statistical data between charging depth and charging frequency, statistical data between current and SOC distribution, and statistical data between temperature and charging frequency.

[0102] In step S104, a common SOC interval for the target charging segments is selected according to the statistical information and a preset SOC interval determination condition.

[0103] In this embodiment, after obtaining the statistical information, the execution subject can obtain the distribution, timing change, overall charging habit, charging habit, and temperature change information of the related data of the lithium ion power battery in each target charging segment based on analysis of the statistical information.

[0104] Then, the execution subject can select a common SOC interval for each target charging segment in combination with the preset SOC interval determination condition and the above statistical information.

[0105] The preset SOC interval determination condition can be used to eliminate the influence of different charging depths and different current sizes in each charging phase on the charging capacity in irregular charging segments. Specifically, the preset SOC interval determination condition can include: the SOC interval range is as large as possible, the same charging current size and strategy when calculating the capacity are ensured, and the number of charging segment samples containing the common SOC interval is as large as possible. When the selected SOC interval range is as large as possible, a longer charging process can be obtained, and more capacity change data can be obtained. Selecting the common SOC charging interval of the charging segment ensures the same charging current size and strategy when calculating the capacity, which can reduce the error caused by different current sizes and strategies. The number of charging segment samples containing the common SOC interval is as large as possible, which can obtain a more accurate SOC interval and meet the charging habits of different types of new energy vehicles.

[0106] Specifically, the execution subject can analyze the above statistical information to determine an SOC interval that meets the preset SOC interval determination condition, and take it as the common SOC interval for each target charging segment.

[0107] In step S105, the capacity information corresponding to the common SOC interval is calculated according to a preset ampere-hour integral calculation formula.

[0108] In this embodiment, the preset ampere-hour integral calculation formula is as follows:

[0109]

[0110] wherein C is the capacity, SOC tSOC0 is the battery charge state of the end-of-charge state of the common SOC interval, SOC0 is the battery charge state of the start-of-charge state of the common SOC interval, η is the charging and discharging efficiency coefficient, approximately 1, I(t) is the charging current, and t is the charging time.

[0111] Wherein, after obtaining the common SOC interval, the execution subject can first determine the battery charge state of the end-of-charge state of the common SOC interval, the battery charge state of the start-of-charge state, the charging current and the charging time, and then substitute these parameters into the preset ampere-hour integral calculation formula to calculate the capacity and obtain the capacity sequence. The execution subject can further obtain the degradation and aging relationship between the charge number sequence and the capacity sequence based on the capacity sequence and the charge number sequence corresponding to the common SOC interval. Wherein, the charge number sequence and the capacity sequence have a non-linear downward trend relationship, and this degradation and aging relationship is used for subsequent auxiliary prediction of capacity aging through the time sequence prediction model.

[0112] Optionally, after calculating the capacity through the preset ampere-hour integral calculation formula to obtain the capacity sequence, the capacity sequence can also be post-processed to obtain a normalized capacity sequence, and then based on the normalized capacity sequence and the charge number sequence corresponding to the common SOC interval, the degradation and aging relationship between the charge number sequence and the capacity sequence is obtained. The post-processing operation here can include but is not limited to using temperature correction coefficient to correct capacity value, using DBSCAN clustering method to identify and eliminate abnormal values, and using normalization method to obtain normalized capacity.

[0113] Step S106, training the to-be-trained model with the capacity information as training data to obtain a trained sequence prediction model.

[0114] In this embodiment, the to-be-trained model can be a sequence-to-sequence model with attention mechanism, which is used for time series prediction. The network structure of the to-be-trained model can include an LSTM encoder and an LSTM decoder, and the LSTM encoder and the LSTM decoder contain an attention mechanism.

[0115] Wherein, when training the to-be-trained model, the LSTM encoder can read the capacity sequence in the capacity information and convert the capacity sequence into an intermediate state vector. Then, the attention mechanism can calculate the attention distribution of the intermediate state vector, and then calculate the weighted average information corresponding to the capacity sequence according to the attention distribution, and summarize the weighted average information to obtain an input sequence. At this time, the LSTM encoder can focus on part of the content in the input sequence to obtain a hidden state, and superimpose the hidden state with the hidden state of the LSTM decoder to obtain the output of this time step through the LSTM decoder, and start to generate the output sequence step by step.

[0116] And the execution subject can set the time step length of the input sequence and the output sequence in advance, and train the to-be-trained model according to the time step length. Compared with single-step forward prediction, the model training method with the self-defined time step length can reduce the number of iterations, thereby reducing the cumulative error of iterations and improving the accuracy of model training. Moreover, the introduction of the attention mechanism in the time prediction model can capture important information in the input sequence and increase the training weight, so that the time prediction model has better extraction ability for important information in the input sequence, and the accuracy of the single sequence prediction result is improved.

[0117] Optionally, after each iteration prediction, the output sequence of this time can be integrated into the input data set to construct a new input sequence, thereby updating the to-be-trained model and further improving the prediction reliability.

[0118] Step S107, predicting the remaining useful life of the lithium-ion power battery by the trained sequence prediction model to obtain a prediction result.

[0119] In the embodiment, the execution subject can train the to-be-trained model by using the data before the prediction starting point to obtain a trained sequence prediction model. Then, the execution subject can predict the remaining useful life of the lithium-ion power battery by the trained sequence prediction model to obtain a prediction result.

[0120] The prediction of the remaining useful life of the lithium-ion power battery by the trained sequence prediction model to obtain a prediction result can include: obtaining a current capacity sequence of the lithium-ion power battery, inputting the current capacity sequence into the trained sequence prediction model, if the current capacity sequence indicates that the capacity value has dropped to a preset capacity threshold, outputting an RUL prediction result based on the sequence prediction model. Wherein, RUL refers to the SOH (capacity) under the current charge and discharge cycle (at the prediction starting point), and the remaining available charge and discharge cycle number when the decline drops to the set capacity threshold. The RUL prediction result can be obtained when the capacity drops to the threshold. Moreover, if the current capacity sequence indicates that the capacity value has not dropped to the preset capacity threshold, the current capacity sequence can be used as the input sequence for subsequent prediction to update the sequence prediction model.

[0121] According to the difference between the SOC values corresponding to each adjacent sampling period, a plurality of charging segments can be divided, and according to the charging condition statistical information of each charging segment, a common SOC interval is selected in combination with the preset SOC interval determination condition, and a capacity information auxiliary sequence prediction model is trained based on the common SOC interval. Since the capacity information calculation is more accurate and the sequence prediction method has better prediction effect in long-term prediction, the trained sequence prediction model in the present application is used for remaining useful life prediction, which can improve the prediction accuracy of the useful life of the lithium ion power battery.

[0122] Referring to Figure 2 , Figure 2 is a flow chart of a method for predicting the useful life of a lithium ion power battery according to another exemplary embodiment, the method for predicting the useful life of the lithium ion power battery comprising the following steps:

[0123] Step S201, obtaining charging data; wherein the charging data at least includes SOC values corresponding to each sampling period.

[0124] In the present embodiment, the execution subject can be a terminal device, a server or the like.

[0125] The execution subject can set the sampling period in advance and obtain the related data of the lithium ion power battery in the new energy vehicle according to the sampling period. The execution subject can judge the parking charging state of the new energy vehicle, and if the new energy vehicle is in the parking charging state, the related data of the lithium ion power battery collected in each sampling period at this time is determined as the above charging data. For example, the sampling period can be every ten seconds, and the SOC value corresponding to each sampling period can be the SOC value every ten seconds in the charging process.

[0126] The related data of the lithium ion power battery can include but is not limited to time, temperature, current, voltage, SOC (state-of-charge), and charging times.

[0127] The SOC value can be the ratio of the charging capacity to the rated capacity, expressed in percentage, and the value range is 0-1. When SOC=0, it means that the battery is completely discharged, and when SOC=1, it means that the battery is completely charged.

[0128] It can be understood that the SOC value corresponding to each sampling period in the charging data increases with time and presents a non-decreasing trend.

[0129] Step S202, dividing each target charging segment according to the difference between the SOC values corresponding to adjacent sampling periods.

[0130] In the embodiment, the execution subject can calculate the difference between the SOC value of each sampling period and the SOC value of the last sampling period, and determine whether the two sampling periods belong to the same charging process according to the difference, so as to divide the target charging segments of the new energy vehicle based on the difference.

[0131] Each target charging segment is associated with relevant data of the lithium ion power battery corresponding to the target charging segment.

[0132] As an optional implementation, the dividing of the target charging segments according to the difference between the SOC values corresponding to adjacent sampling periods comprises: calculating the difference between the SOC value of each sampling period and the SOC value of the last sampling period; if the difference between the SOC value of the sampling period and the SOC value of the last sampling period is equal to 0 or 1, it is determined that the sampling period and the last sampling period belong to the same charging segment; if the difference between the SOC value of the sampling period and the SOC value of the last sampling period is not equal to 0 and not equal to 1, it is determined that the sampling period and the last sampling period do not belong to the same charging segment; and generating each initial charging segment according to the sampling periods belonging to the same charging segment, and selecting the initial charging segments to obtain the target charging segments.

[0133] In the embodiment, the difference between the SOC value of each sampling period and the SOC value of the last sampling period can be calculated from the second sampling period, and if the difference between the SOC value of the sampling period and the SOC value of the last sampling period is equal to 0 or 1, it is determined that the sampling period and the last sampling period belong to the same charging segment. Otherwise, it is determined that the sampling period and the last sampling period do not belong to the same charging segment, and the sampling time of the sampling period can be taken as the starting time of the next charging segment and the ending time of the current charging segment. Then, the execution subject can collect the sampling times corresponding to the sampling periods belonging to the same charging segment to generate each initial charging segment. For example, if the sampling period is every 10 seconds, for the second sampling period, if it is determined that the second sampling period and the last sampling period do not belong to the same charging segment, the 0th second to the 10th second can be determined as an initial charging segment, and the 10th second can be determined as the starting time of another initial charging segment. Then, the execution subject can select the initial charging segments based on the SOC information of each initial charging segment to obtain the target charging segments.

[0134] As another optional implementation, the selecting the target charging segments from the initial charging segments comprises: obtaining SOC value distribution data corresponding to each of the initial charging segments; determining a target SOC threshold according to the SOC value distribution data; wherein the target SOC threshold comprises at least one of a maximum initial SOC value corresponding to a preset percentile, a minimum cut-off SOC value, and a minimum charging SOC change value; and selecting the target charging segments from the initial charging segments according to the target SOC threshold.

[0135] In the embodiment, the execution subject can calculate distribution statistics of initial SOC and cut-off SOC in all initial charging segments respectively, obtain a maximum initial SOC value of a preset percentile (e.g., 75 percentile), a minimum cut-off SOC value within a reasonable value, and a minimum charging SOC change value as a screening judgment condition of a charging segment, and select the target charging segments from the initial charging segments.

[0136] In step S203, statistical information corresponding to each of the target charging segments is determined.

[0137] In the embodiment, the execution subject can perform data statistics according to the related data of the lithium-ion power battery associated with each of the target charging segments to obtain statistical information. The statistical information can reflect the charging habit information of the new energy vehicle, and can specifically include but is not limited to the overall charging frequency habit, the charging depth habit, the charging current information, the charging environment, and the charging temperature change, etc., which are not limited in the embodiment.

[0138] Specifically, the statistical information can include but is not limited to statistical data between initial and final SOC points and charging frequency distribution, statistical data between charging depth and charging frequency, statistical data between current and SOC distribution, and statistical data between temperature and charging frequency.

[0139] In step S204, a common SOC interval for the target charging segments is selected according to the statistical information and a preset SOC interval determination condition.

[0140] In the embodiment, after obtaining the statistical information, the execution subject can analyze the statistical information to obtain the distribution, the timing change, the overall charging habit, the charging habit, and the temperature change information of the related data of the lithium-ion power battery in each of the target charging segments.

[0141] Then, the execution subject can select a common SOC interval for each of the target charging segments in combination with the preset SOC interval determination condition and the statistical information.

[0142] The preset SOC interval determination condition can be used to eliminate the influence of different charging depths in irregular charging segments and different current sizes in each charging stage on the charging capacity. Specifically, the preset SOC interval determination condition can include: the SOC interval range is as large as possible, the same charging current size and strategy when calculating the capacity, and the number of charging segment samples containing the common SOC interval is as large as possible. Wherein, the selected SOC interval range is as large as possible, so that the charging process is as long as possible, and more capacity change data is obtained. Selecting the common SOC charging interval of the charging segment ensures the same charging current size and strategy when calculating the capacity, which can reduce the error caused by different current sizes and strategies. The number of charging segment samples containing the common SOC interval is as large as possible, so that a more accurate SOC interval can be obtained to meet the charging habits of different types of new energy vehicles.

[0143] Specifically, the execution subject can analyze the above statistical information, determine the SOC interval meeting the preset SOC interval determination condition, and take it as the common SOC interval for each target charging segment.

[0144] As an optional implementation, the statistical information includes statistical information of original data features in the target charging segments, and the original data features include at least one of the following: time, temperature, current, voltage, SOC value, and charging times.

[0145] The preset SOC interval determination condition includes:

[0146] The interval range of the SOC interval is greater than a preset range threshold; and

[0147] The SOC interval corresponds to the same charging current; and

[0148] The number of target charging segments containing the SOC interval is greater than a preset number threshold.

[0149] In this embodiment, the preset SOC interval determination condition can include that the interval range is greater than a preset range threshold, so as to select a longer charging process as the basis for subsequent capacity calculation. In addition, the preset SOC interval determination condition can also include that the charging current corresponding to the SOC interval is the same, and the number of target charging segments containing the SOC interval is greater than a preset number threshold. Through these SOC interval determination conditions, the same charging current size and strategy when calculating the capacity can be ensured, and the charging habits of different new energy vehicles can be met.

[0150] In step S205, the charging initial state battery charge state amount, the charging end state battery charge state amount, the charging current, and the charging time corresponding to the common SOC interval are determined.

[0151] Step S206, the initial state of charge battery charge state quantity, the end of charging state battery charge state quantity, the charging current and the charging time, into the preset ampere-hour integral calculation formula, the initial capacity sequence corresponding to the common SOC interval is obtained.

[0152] In this embodiment, the preset ampere-hour integral calculation formula is as follows:

[0153]

[0154] Wherein, C is the capacity, SOC t is the end of charging state battery charge state quantity of the common SOC interval, SOC0 is the initial state of charge battery charge quantity of the common SOC interval, η is the charging and discharging efficiency coefficient, about 1, I(t) is the charging current, t is the charging time.

[0155] Wherein, after obtaining the common SOC interval, the subject can first determine the end of charging state battery charge state quantity, the initial state of charge battery charge quantity, the charging current and the charging time of the common SOC interval, and then put these parameters into the preset ampere-hour integral calculation formula to calculate the capacity, and obtain the capacity sequence. The subject can further obtain the degradation aging relationship between the charge frequency sequence and the capacity sequence based on the capacity sequence and the charge frequency sequence corresponding to the common SOC interval. Wherein, the charge frequency sequence and the capacity sequence have a non-linear downward trend relationship, and such degradation aging relationship is used for subsequent auxiliary prediction of capacity aging through the time sequence prediction model.

[0156] Step S207, the initial capacity sequence is subjected to capacity value correction processing and outlier elimination processing to obtain a target capacity sequence.

[0157] In this embodiment, the subject can use the temperature correction coefficient to perform capacity value correction processing, and use the DBSCAN clustering method to perform outlier elimination processing.

[0158] Step S208, the target capacity sequence is subjected to normalization processing to obtain a normalized capacity sequence, and the normalized capacity sequence is determined as the capacity information.

[0159] After calculating the capacity through the preset ampere-hour integral calculation formula to obtain the capacity sequence, the capacity sequence can also be subjected to post-processing operation to obtain the normalized capacity sequence, and then based on the normalized capacity sequence and the charge frequency sequence corresponding to the common SOC interval, the degradation aging relationship between the charge frequency sequence and the capacity sequence is obtained. The post-processing operation here can include but is not limited to using temperature correction coefficient to correct capacity value, using DBSCAN clustering method to identify and eliminate outliers, using normalization method to obtain normalized capacity.

[0160] Step S209, selecting a first capacity sequence of a first time step length and a second capacity sequence of a second time step length from the capacity information.

[0161] In the embodiment, the first time step length is preferably 50, and the second time step length is preferably 20. The input time step length and the output time step length can be used as hyperparameters of the network, and the optimal combination of (50, 20) is selected by the grid method.

[0162] Step S210, taking the first capacity sequence and the charging frequency sequence of the first time step length as input data of the to-be-trained model, and taking the second capacity sequence as output data of the to-be-trained model.

[0163] In the embodiment, the execution subject uses the charging frequency sequence and the capacity sequence of the time step length of 50 as input data, and uses the capacity sequence of the next time step length of 20 as output data to train the model.

[0164] Step S211, inputting the input data into the to-be-trained model to obtain prediction data of the to-be-trained model.

[0165] Step S212, iteratively training the to-be-trained model based on the prediction data and the output data to obtain the trained sequence prediction model.

[0166] In the embodiment, the execution subject iteratively trains to make the prediction data approach the output data.

[0167] The trained sequence prediction model includes an LSTM encoder, an LSTM decoder, and an attention mechanism.

[0168] In the embodiment, the to-be-trained model can be a sequence-to-sequence model with an attention mechanism, which is used for time series prediction. The network structure of the to-be-trained model can include an LSTM encoder and an LSTM decoder, and the LSTM encoder and the LSTM decoder include an attention mechanism.

[0169] When training the to-be-trained model, the LSTM encoder can read the capacity sequence in the capacity information and convert the capacity sequence into an intermediate state vector. Then, the attention mechanism can calculate the attention distribution of the intermediate state vector, and then calculate the weighted average information corresponding to the capacity sequence according to the attention distribution, and aggregate the weighted average information to obtain an input sequence. At this time, the LSTM encoder can focus on part of the input sequence to obtain a hidden state, and the hidden state is superimposed with the hidden state of the LSTM decoder, and the output of the time step can be obtained through the LSTM decoder to start generating the output sequence step by step.

[0170] And the execution subject can set the time step length of the input sequence and the output sequence in advance, and train the to-be-trained model according to the time step length. Compared with single-step forward prediction, the model training method with self-defined time step length can reduce the number of iterations, thereby reducing the cumulative error of iterations and improving the accuracy of model training. Moreover, the introduction of the attention mechanism in the time prediction model can capture important information in the input sequence and increase the training weight, so that the time prediction model has better extraction ability for important information in the input sequence, and the accuracy of the single sequence prediction result is improved.

[0171] Optionally, after each iteration prediction, the output sequence of this time can be integrated into the input data set to construct a new input sequence, thereby updating the training of the to-be-trained model and further improving the prediction reliability.

[0172] Step S213, predicting the remaining useful life of the lithium-ion power battery by the trained sequence prediction model to obtain a prediction result.

[0173] In the embodiment, the execution subject can train the to-be-trained model using the data before the prediction starting point to obtain a trained sequence prediction model. Then, the execution subject can predict the remaining useful life of the lithium-ion power battery by the trained sequence prediction model to obtain a prediction result.

[0174] The prediction of the remaining useful life of the lithium-ion power battery by the trained sequence prediction model to obtain a prediction result can include: obtaining a current capacity sequence of the lithium-ion power battery, inputting the current capacity sequence into the trained sequence prediction model, if the current capacity sequence indicates that the capacity value has dropped to a preset capacity threshold, outputting an RUL prediction result based on the sequence prediction model. Wherein, RUL refers to the SOH (capacity) under the current charge and discharge cycle (at the prediction starting point), and the remaining available charge and discharge cycle number when the decline drops to the set capacity threshold. The RUL prediction result can be obtained when the capacity drops to the threshold. Moreover, if the current capacity sequence indicates that the capacity value has not dropped to the preset capacity threshold, the current capacity sequence can be used as the input sequence for subsequent prediction to update the training sequence prediction model.

[0175] According to the difference between the SOC values corresponding to each adjacent sampling period, a plurality of charging segments can be divided, and according to the charging condition statistical information of each charging segment, a common SOC interval is selected in combination with a preset SOC interval determination condition, a capacity information auxiliary sequence prediction model is calculated based on the common SOC interval, and since the capacity information calculation is more accurate and the sequence prediction method has better prediction effect in long-term prediction, the remaining service life prediction is performed by using the sequence prediction model trained in the application, so that the service life prediction accuracy of the lithium ion power battery can be improved.

[0176] Please refer to Figure 3 , Figure 3 is a block diagram structure schematic diagram of a service life prediction device of a lithium ion power battery according to an exemplary embodiment. The service life prediction device 3 of the lithium ion power battery comprises:

[0177] The data acquisition unit 301 is configured to acquire charging data, wherein the charging data at least includes SOC values corresponding to each sampling period.

[0178] The segment division unit 302 is configured to divide each target charging segment according to the difference between the SOC values corresponding to adjacent sampling periods.

[0179] The statistical unit 303 is configured to determine statistical information corresponding to each target charging segment.

[0180] The interval determination unit 304 is configured to select a common SOC interval for each target charging segment according to the statistical information and a preset SOC interval determination condition.

[0181] The capacity calculation unit 305 is configured to calculate capacity information corresponding to the common SOC interval according to a preset ampere-hour integral calculation formula.

[0182] The model training unit 306 is configured to train a to-be-trained model by taking the capacity information as training data, to obtain a trained sequence prediction model.

[0183] The prediction unit 307 is configured to predict the remaining service life of the lithium ion power battery by using the trained sequence prediction model, to obtain a prediction result.

[0184] Further, the segment division unit 302 is specifically configured to:

[0185] For each sampling period, calculate the difference between the SOC value of the sampling period and the SOC value of the previous sampling period.

[0186] If the difference between the SOC value of the sampling period and the SOC value of the previous sampling period is equal to 0 or 1, it is determined that the sampling period and the previous sampling period belong to the same charging segment;

[0187] If the difference between the SOC value of the sampling period and the SOC value of the previous sampling period is not equal to 0 and not equal to 1, it is determined that the sampling period and the previous sampling period do not belong to the same charging segment.

[0188] According to the sampling periods belonging to the same charging segment, each initial charging segment is generated.

[0189] The initial charging segments are selected to obtain the target charging segments.

[0190] Further, the segment division unit 302 is specifically configured to:

[0191] Obtain the SOC value distribution data corresponding to each initial charging segment;

[0192] According to the SOC value distribution data, a target SOC threshold is determined; wherein the target SOC threshold at least includes one of the following: a maximum initial SOC value corresponding to a preset percentile, a minimum cut-off SOC value, and a minimum charging SOC change value.

[0193] According to the target SOC threshold, the initial charging segments are selected to obtain the target charging segments.

[0194] Further, the statistical information includes statistical information of original data features in the target charging segments, and the original data features include at least one of the following: time, temperature, current, voltage, SOC value, and charging frequency; and

[0195] The preset SOC interval determination condition includes:

[0196] The interval range of the SOC interval is greater than a preset range threshold; and

[0197] The SOC interval corresponds to the same charging current; and

[0198] The number of target charging segments containing the SOC interval is greater than a preset number threshold.

[0199] Further, the capacity calculation unit 305 is specifically configured to:

[0200] Determine the charging initial state battery charge state amount, the charging final state battery charge state amount, the charging current, and the charging time corresponding to the common SOC interval;

[0201] The initial capacity sequence is subjected to capacity value correction processing and outlier elimination processing to obtain a target capacity sequence.

[0202] The initial capacity sequence is subjected to capacity value correction processing and outlier elimination processing to obtain a target capacity sequence.

[0203] The target capacity sequence is subjected to normalization processing to obtain a normalized capacity sequence, and the normalized capacity sequence is determined as the capacity information.

[0204] Further, the trained sequence prediction model comprises an LSTM encoder, an LSTM decoder, and an attention mechanism.

[0205] Further, the model training unit 306 is specifically configured to:

[0206] select a first capacity sequence of a first time step length and a second capacity sequence of a second time step length from the capacity information;

[0207] use the first capacity sequence and a charging frequency sequence of the first time step length as input data of the to-be-trained model, and use the second capacity sequence as output data of the to-be-trained model;

[0208] input the input data into the to-be-trained model to obtain prediction data of the to-be-trained model;

[0209] based on the prediction data and the output data, iteratively train the to-be-trained model to obtain the trained sequence prediction model.

[0210] According to the difference between the SOC values corresponding to each adjacent sampling period, a plurality of charging segments can be divided, and a common SOC interval is selected according to the charging condition statistical information of each charging segment and in combination with a preset SOC interval determination condition. The capacity information auxiliary sequence prediction model is trained based on the common SOC interval. Since the capacity information calculation is more accurate and the sequence prediction method has better prediction effect in long-term prediction, the use life prediction precision of the lithium-ion power battery can be improved by using the trained sequence prediction model in the present application for remaining use life prediction.

[0211] As to the use life prediction device 3 of the lithium-ion power battery in the above-mentioned embodiments, the specific manner in which each module performs operation has been described in detail in the embodiments of the above-mentioned related method, and will not be described in detail here.

[0212] Please refer to Figure 4 ,Figure 4 is a block diagram structure schematic diagram of an electronic device according to an exemplary embodiment, the electronic device 4 comprises:

[0213] at least one processor 41; and

[0214] a memory 42 connected in communication with the at least one processor 41; wherein,

[0215] The memory 42 stores instructions executable by the at least one processor 41, and the instructions are executed by the at least one processor 41 to enable the at least one processor 41 to perform the above-mentioned related lithium ion power battery life prediction method.

[0216] In practical applications, the electronic device 4 can be a server, and it should be noted that the electronic device 4 is not limited to be embodied in the form of a server. As for the electronic device 4 in the above-mentioned embodiments, the specific way in which the processor 41 executes the program in the memory 42 has been described in detail in the embodiments related to the method, and will not be described in detail here.

[0217] In addition, the present application also provides a non-transitory computer readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to perform the above-mentioned related lithium ion power battery life prediction method.

[0218] The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD), etc. The storage medium can also include a combination of the above-mentioned types of memories.

[0219] In addition, the present application also provides a computer program product comprising a computer program, which, when executed by a processor, implements the above-mentioned related lithium ion power battery life prediction method.

[0220] It can be understood that the same or similar parts in the above-mentioned embodiments can be mutually referred to, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0221] It should be noted that in the description of the present application, the terms "first", "second" and the like are only for the purpose of description, and cannot be understood as indicating or implying relative importance. In addition, in the description of the present application, unless otherwise specified, the meaning of "a plurality of" or "many" is at least two.​

[0222] It will be understood that when an element is referred to as being "on" or "connected to" another element, it can be directly on or connected to the other element or intervening elements can be present; and that when an element is referred to as being "connected to" another element, it can be directly connected to the other element or intervening elements can be present. Also, the use of "a" or "an" to describe elements can include the plural, and vice versa.

[0223] Any process or method described in flow chart form or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing specific logic functions (or steps) of the process, and alternate implementations are possible. The various embodiments of the application can be implemented in software, hardware, firmware, or a combination thereof, and various preferred embodiments of the application can be realized in a variety of ways.

[0224] It should be understood that aspects of the application can be implemented in hardware, software, firmware, or combinations thereof, and various preferred embodiments of the application can be realized in a variety of ways. In the foregoing embodiments, various steps or methods can be implemented in software, firmware, or hardware, or a combination thereof. For example, if implemented in software, the steps or methods can be stored in or implemented as one or more instructions or code on a computer-readable medium, which can be executed by a processing unit. Computer-readable media can include computer- readable storage media, which refers to media code that carries computer program code for use by the processing unit. Computer-readable storage media can be any available media that can be accessed by a general purpose or special purpose computing system. By way of example, and not limitation, such computer-readable storage media can comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired computer program code in a non-transitory computer- readable medium, which can be accessed by a general purpose or special purpose computing system, which can include the processing unit. When implemented in software, the steps or methods can be implemented as one or more instructions or code on a computer-readable medium, which can be executed by the processing unit. A computer-readable medium can include a non-transitory computer-readable medium that includes a computer-readable storage medium to

[0225] It should be understood that any number of hardware or software components can be used to implement the preferred embodiments of the application. For example, software components can be implemented as computer-executable instructions on a computer-readable storage medium, which can be executed by a processing unit. In this regard, the terms "module" or "functionality" as used herein includes but is not limited to, hardware, firmware, software, combinations thereof, or electronic representations of all or a portion of the foregoing.

[0226] Further, functional units in the various embodiments of the present application can be integrated in one processing module, or each of the functional units can be physically present alone, or two or more of the functional units can be integrated in one module. The above-mentioned integrated module can be realized in the form of hardware or software function module. When the integrated module is realized in the form of software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0227] The above-mentioned storage medium can be a read-only memory, a magnetic disk or an optical disk, etc.

[0228] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are contained in at least one embodiment or example of the present application. In the specification, the illustrative description of the above terms does not necessarily mean the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0229] Although the embodiments of the present application have been shown and described above, it is understood that the above-described embodiments are exemplary, and cannot be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above-described embodiments within the scope of the present application.

Claims

1. A method for predicting the lifespan of a lithium-ion power battery, characterized in that, The method includes: Acquire charging data; wherein the charging data includes at least the SOC value corresponding to each sampling period; Based on the difference between the SOC values ​​corresponding to adjacent sampling periods, each target charging segment is obtained; Determine the statistical information corresponding to each target charging segment; Based on the statistical information and the preset SOC interval determination conditions, a common SOC interval is selected for each target charging segment; the preset SOC interval determination conditions include: the range of the SOC interval is greater than a preset range threshold. Calculate the capacity information corresponding to the common SOC interval according to the preset ampere-hour integral calculation formula; The capacity information is used as training data to train the model to be trained, and a trained sequence prediction model is obtained. The remaining lifespan of lithium-ion power batteries is predicted using the sequence prediction model that has been trained, and the prediction results are obtained.

2. The method according to claim 1, characterized in that, The step of dividing the target charging segment based on the difference between the SOC values ​​corresponding to adjacent sampling periods includes: For each sampling period, calculate the difference between the SOC value of that sampling period and the SOC value of the previous sampling period; If the difference between the SOC value of the current sampling period and the SOC value of the previous sampling period is equal to 0 or 1, then the current sampling period and the previous sampling period are determined to belong to the same charging segment. If the difference between the SOC value of the current sampling period and the SOC value of the previous sampling period is not equal to 0 and not equal to 1, then it is determined that the current sampling period and the previous sampling period do not belong to the same charging segment. Each initial charging segment is generated based on the sampling period belonging to the same charging segment; Each initial charging segment is filtered to obtain each target charging segment.

3. The method according to claim 2, characterized in that, The initial charging segments are filtered to obtain the target charging segments, including: Obtain the SOC value distribution data corresponding to each initial charging segment; Based on the SOC value distribution data, a target SOC threshold is determined; wherein, the target SOC threshold includes at least one of the following: the maximum initial SOC value corresponding to the preset percentile, the minimum cutoff SOC value, and the minimum charging SOC change value; Based on the target SOC threshold, each initial charging segment is filtered out to obtain each target charging segment.

4. The method according to claim 1, characterized in that, The statistical information includes statistical information on the original data features of each target charging segment, and the original data features include at least one of the following: time, temperature, current, voltage, SOC value, and number of charging cycles; as well as The SOC range corresponds to the same charging current; and The number of target charging segments containing the SOC range is greater than a preset threshold.

5. The method according to claim 1, characterized in that, According to the preset ampere-hour integral calculation formula, the capacity information corresponding to the common SOC interval is calculated, including: Determine the battery state of charge at the beginning of charging, the battery state of charge at the end of charging, the charging current, and the charging time corresponding to the common SOC interval. Substituting the battery state of charge at the beginning of charging, the battery state of charge at the end of charging, the charging current, and the charging time into the preset ampere-hour integral calculation formula, the initial capacity sequence corresponding to the common SOC interval is obtained. The initial capacity sequence is subjected to capacity value correction and outlier elimination processing to obtain the target capacity sequence; The target capacity sequence is normalized to obtain a normalized capacity sequence, and the normalized capacity sequence is determined as the capacity information.

6. The method according to claim 1, characterized in that, The trained sequence prediction model includes an LSTM encoder, an LSTM decoder, and an attention mechanism.

7. The method according to claim 1, characterized in that, Using the capacity information as training data, the model to be trained is trained to obtain a trained sequence prediction model, including: Select a first capacity sequence with a first time step length and a second capacity sequence with a second time step length from the capacity information; The first capacity sequence and the charging count sequence of the first time step length are used as input data for the model to be trained, and the second capacity sequence is used as output data for the model to be trained. The input data is input into the model to be trained to obtain the prediction data of the model to be trained; Based on the predicted data and the output data, the model to be trained is iteratively trained to obtain the trained sequence prediction model.

8. A device for predicting the lifespan of a lithium-ion power battery, characterized in that, The device includes: A data acquisition unit is used to acquire charging data; wherein the charging data includes at least the SOC value corresponding to each sampling period; The segmentation unit is used to divide the target charging segment according to the difference between the SOC values ​​corresponding to adjacent sampling periods; The statistics unit is used to determine the statistical information corresponding to each target charging segment; An interval determination unit is used to select a common SOC interval for each target charging segment based on the statistical information and preset SOC interval determination conditions; the preset SOC interval determination conditions include: the range of the SOC interval is greater than a preset range threshold. The capacity calculation unit is used to calculate the capacity information corresponding to the common SOC interval according to the preset ampere-hour integral calculation formula. The model training unit is used to train the model to be trained by using the capacity information as training data, so as to obtain the trained sequence prediction model. The prediction unit is used to predict the remaining service life of the lithium-ion power battery through the sequence prediction model that has been trained, and to obtain the prediction result.

9. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-7.

Images