A method and system for classifying retired batteries based on a dynamic knowledge graph

Abstract

The application discloses a kind of based on dynamic knowledge graph's retired battery classification method and system, the method includes: generating the identification code of retired battery, establishes the identity identification code database of the corresponding relationship of identification code and manufacturer, new energy vehicle, operation log and maintenance log;Set charging and discharging strategy, obtain analysis processing sample, extract the node and node relationship of information stored in identity identification code database, construct dynamic knowledge graph;According to frequent subgraph algorithm, obtain the feature tensor of dynamic knowledge graph;Analysis processing sample is regarded as the input layer of long short-term memory network model, obtains the time sequence feature of each retired battery;Time sequence feature and feature tensor are regarded as the input layer of fully connected network layer model, obtain label vector;According to label vector, combined with identity identification code database, the quality classification of retired battery is carried out.The retired battery classification method based on dynamic knowledge graph disclosed in the application can improve the utilization rate of retired battery.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of secondary utilization of retired batteries, in particular to a retired battery classification method and system based on a dynamic knowledge graph. BACKGROUND

[0002] Power storage batteries are important energy storage components of new energy vehicles. They are usually connected in series and parallel by a large number of single cells to form a module, and then connected in series to form a battery pack to provide power for vehicles. After a long time of vehicle use and retirement, the single batteries in the battery pack will have great inconsistency in maximum available capacity, ohmic resistance, polarization resistance, etc. This will directly affect the output performance of the battery pack and bring great inconvenience to the secondary utilization of the retired batteries. Therefore, consistency analysis needs to be performed on the single batteries or modules, and the batteries are classified and selected based on the analysis.

[0003] At present, common retired battery screening methods can be divided into: 1) manually using professional equipment such as a direct current resistance tester, a multimeter, etc. to measure the parameters of the same batch of batteries, and classifying them according to the numerical values. Considering that the state of charge (SOC) of the batteries has a great influence on various battery parameters, this method is only suitable for batteries at the same SOC. 2) using a charge-discharge instrument to fully charge and discharge the batteries, and classifying them according to the capacity of the complete discharge or charge of the batteries. This method takes a long time and consumes a lot of energy. 3) using specific charge-discharge conditions to test the performance of the batteries, simulating and calculating the internal parameters of the batteries according to the test results, and classifying them accordingly. This method is efficient, but requires strong professional knowledge in the fields of electrochemistry and control, and is difficult to implement. Moreover, the above methods only consider the data and parameters of the retired batteries, ignoring the correlation factors of the batteries and new energy vehicles, manufacturers, etc. external factors, affecting the accuracy of the classification of the retired batteries.

[0004] In the prior art, patent application publication No. CN106423919A, entitled "Retired lithium battery sorting method and system", classifies and sorts retired power lithium batteries by specific conditions, comprehensive calculation scores, and grading results, creating conditions for their cascade utilization and reasonable recycling. SUMMARY

[0005] The technical problem to be solved by the present application is to solve the problem that only the data and parameters of the retired batteries are considered when classifying the retired batteries, ignoring the correlation factors of the retired batteries and new energy vehicles, manufacturers, etc. external factors, affecting the accuracy of the classification of the batteries.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] A retired battery classification method based on a dynamic knowledge graph, comprising the following steps:

[0008] Generating an identification code for each retired battery, and establishing an identity identification code database of the identification code and the corresponding relationship between the manufacturer, new energy vehicle, operation log and maintenance log;

[0009] Setting a charge-discharge strategy to perform charge-discharge testing on the retired battery, and obtaining charge-discharge data and external physical parameter data of each retired battery;

[0010] According to the charge-discharge data and external physical parameter data, an analysis processing sample is obtained, and the analysis processing sample of each retired battery is stored in the identity identification code database;

[0011] Extract the nodes and node relationships of the stored information in the identity identification code database to construct a dynamic knowledge graph;

[0012] According to the frequent subgraph algorithm, a feature tensor of the dynamic knowledge graph is obtained;

[0013] The analysis processing sample is used as the input layer of the long short-term memory network model, and the time sequence feature of each retired battery is obtained;

[0014] The time sequence feature and the feature tensor are used as the input layer of the full connection network layer model, and a label vector is obtained;

[0015] According to the label vector combined with the identity identification code database, the quality of the retired battery is classified.

[0016] Advantages: By setting a charge-discharge strategy, only the local charge-discharge curve and the external physical parameter change during the corresponding test are obtained to obtain a feature vector, which replaces the directly measured performance parameters, avoiding full charge-discharge testing and multiple charge-discharge cycle testing, reducing power consumption and testing time. Through the establishment of the dynamic knowledge graph of the retired battery, the key data visualization of the whole life cycle management of the battery is realized. In addition, in the analysis of the charge-discharge data and external physical parameter data of the retired battery, the data of the manufacturer, new energy vehicle, operation log and maintenance log are also considered, which makes the classification of the retired battery more accurate.

[0017] In an embodiment of the present application, the setting of the charge-discharge strategy for the retired battery charge-discharge testing comprises the following steps:

[0018] Constant current charging the retired battery for a certain period of time;

[0019] Rest for a certain period of time;

[0020] The retired battery is discharged at constant current, and the discharge time is the same as the charging time; in the constant current discharge stage, the pulse current of different rates is used for charging and discharging;

[0021] Rest for a certain time;

[0022] The retired battery is charged at constant voltage for a certain time;

[0023] Rest for a certain time;

[0024] The retired battery is discharged at constant voltage, and the discharge time is the same as the charging time;

[0025] The charging and discharging data and external physical parameter data of each retired battery are obtained.

[0026] In an embodiment of the present application, the charging and discharging data and external physical parameter data in a time period with the same time interval and time starting point are selected, a plurality of feature vectors are obtained, the plurality of feature vectors are normalized to obtain analysis processing samples, wherein the plurality of feature vectors include temperature change time series of each charging and discharging stage, voltage change time series of the constant voltage charging stage, and constant current stage incremental capacity.

[0027] In an embodiment of the present application, the temperature change time series of each charging and discharging stage is obtained by the following formula:

[0028]

[0029] In the formula, T represents the temperature change time series of each charging and discharging stage, t represents the sampling time, and when the temperature change time series of each charging and discharging stage is obtained, the value range of t is t=1, 2, ……n1.

[0030] In an embodiment of the present application, the voltage change time series of the constant voltage charging stage is obtained by the following formula:

[0031]

[0032] In the formula, V represents the voltage change time series of the constant voltage charging stage, t represents the sampling time, and when the voltage change time series of the constant voltage charging stage is obtained, the value range of t is t=1, 2, ……n2.

[0033] In an embodiment of the present application, the constant current stage incremental capacity is obtained by the following formula:

[0034]

[0035] In the formula, IC represents the incremental capacity in the constant current stage, Q(t) and V(t) represent the battery capacity and terminal voltage at time t, Q(K) and V(K) represent the discrete form of the battery capacity and the discrete form of the terminal voltage, I represents the current in the constant current stage, N represents the sampling interval, T0 represents the sampling period time, Δt represents the time change, ΔV(K) represents the change of the terminal voltage at time K in the discrete form, and V(K-N) represents the terminal voltage at time K-N T0.

[0036] The analysis processing sample is a set of normalized feature vectors, namely

[0037] S={T', V', IC'};

[0038] In the formula, S represents the analysis processing sample, T', V' and IC' represent the temperature change time sequence of each charge and discharge stage, the voltage change time sequence of the constant voltage charging stage and the incremental capacity in the constant current stage after normalization processing, respectively.

[0039] In an embodiment of the present application, the dynamic knowledge graph comprises nodes, node relationships and time sequence attributes, wherein the nodes are the retired battery, the manufacturer, the new energy vehicle, the operation log, the maintenance log and the analysis processing sample, the node relationships comprise the relationship between the retired battery and the manufacturer, the relationship between the retired battery and the new energy vehicle, the relationship between the retired battery and the operation log, the relationship between the retired battery and the maintenance log and the relationship between the retired battery and the analysis processing sample, and the time sequence attributes are the attributes of the nodes and the node relationships based on time.

[0040] The present application also provides a retired battery classification system based on a dynamic knowledge graph, comprising:

[0041] A code scanner is configured to generate an identification code for each retired battery.

[0042] A charge and discharge instrument is configured to perform charge and discharge tests on the retired batteries according to a set charge and discharge strategy.

[0043] A data acquisition instrument is configured to acquire charge and discharge data and external physical parameter data of each retired battery.

[0044] The storage and operation device is used for establishing an identity recognition code database for establishing a correspondence relationship between the recognition code and a manufacturer, a new energy vehicle, an operation log and a maintenance log; according to the charge-discharge data and external physical parameter data, an analysis processing sample is obtained, and the analysis processing sample of each retired battery is stored in the identity recognition code database; a node and a node relationship of the stored information in the identity recognition code database are extracted to construct a dynamic knowledge graph; according to a frequent subgraph algorithm, a feature tensor of the dynamic knowledge graph is obtained; the analysis processing sample is taken as an input layer of a long short-term memory network model to obtain a time sequence feature of each retired battery; the time sequence feature and the feature tensor are taken as an input layer of a full connection network layer model to obtain a label vector; and according to the label vector combined with the identity recognition code database, the retired batteries are classified in quality.

[0045] In an embodiment of the present application, in the charge-discharge instrument, the setting of the charge-discharge strategy for the retired battery to perform the charge-discharge test includes the following steps:

[0046] The retired battery is charged at a constant current for a certain time;

[0047] The retired battery is rested for a certain time;

[0048] The retired battery is discharged at a constant current, and the discharge time is the same as the charge time, and in the constant current discharge stage, the charge-discharge is performed through different pulse currents;

[0049] The retired battery is rested for a certain time;

[0050] The retired battery is charged at a constant voltage for a certain time;

[0051] The retired battery is rested for a certain time;

[0052] The retired battery is discharged at a constant voltage, and the discharge time is the same as the charge time;

[0053] The charge-discharge data and external physical parameter data of each retired battery are obtained.

[0054] In an embodiment of the present application, the dynamic knowledge graph includes nodes, node relationships and time sequence attributes, wherein the nodes are the retired batteries, the manufacturers, the new energy vehicles, the operation logs, the maintenance logs and the analysis processing samples, the node relationships include relationships between the retired batteries and the manufacturers, relationships between the retired batteries and the new energy vehicles, relationships between the retired batteries and the operation logs, relationships between the retired batteries and the maintenance logs and relationships between the retired batteries and the analysis processing samples, and the time sequence attributes are attributes of the nodes and the node relationships based on time.

[0055] Compared with the prior art, the beneficial effects of the present application are: by constructing a dynamic knowledge graph, the visualization of key data of the whole life cycle of the battery is realized. At the same time, by using the long short-term memory network model and the full connection network layer model to classify the retired batteries, the state of the retired batteries can be quickly and non-destructively evaluated with high precision, thereby improving the evaluation level of the usability of the retired batteries. The user can decide the next use of the retired battery according to the label vector, thereby maximizing the economic benefit, such as using the retired battery in good condition for electric bicycles, using the retired battery in medium condition for energy storage, and recycling and destroying the retired battery in poor condition, thereby minimizing the safety risk. BRIEF DESCRIPTION OF DRAWINGS

[0056] Figure 1 A flow chart of a retired battery classification method based on a dynamic knowledge graph for an embodiment of the present application.

[0057] Figure 2 A flow chart of setting a charging and discharging strategy for an embodiment of the present application.

[0058] Figure 3 A dynamic knowledge graph block diagram for an embodiment of the present application.

[0059] Figure 4 A system block diagram of a retired battery classification system based on a dynamic knowledge graph for an embodiment of the present application. DETAILED DESCRIPTION

[0060] In order to facilitate those skilled in the art to understand the technical solutions of the present application, the technical solutions of the present application will be further described in conjunction with the drawings of the present application.

[0061] The terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0062] Please refer to Figure 1 As shown in the drawings, the present application provides a retired battery classification method based on a dynamic knowledge graph, which comprises the following steps:

[0063] S100, generate an identification code for each retired battery, and establish an identity identification code database of the identification code and the corresponding relationship of the production manufacturer, new energy vehicle, operation log and maintenance log.

[0064] In step S100, an identification code is generated for each retired single battery by a code scanner, such as a two-dimensional code, a bar code, or an electronic tag, etc. At the same time, an identity recognition code database is established, which stores the correspondence between the identification code and the manufacturer, new energy vehicle, operation log, and maintenance log.

[0065] In step S200, the retired batteries are tested by setting a charging and discharging strategy, and the charging and discharging data and external physical parameter data of each retired battery are obtained.

[0066] In step S200, the retired batteries are tested by setting a charging and discharging strategy, and the charging and discharging data and external physical parameter data of each retired battery are obtained.

[0067] Please refer to Figure 1 and Figure 2 In an embodiment of the present application, in step S200, the testing of the retired batteries by setting a charging and discharging strategy includes the following steps:

[0068] In step S210, the retired batteries are charged at a constant current for a certain time.

[0069] In step S220, the retired batteries are rested for a certain time.

[0070] In step S230, the retired batteries are discharged at a constant current, and the discharge time is the same as the charging time. In the constant current discharging stage, different pulse currents are used for charging and discharging.

[0071] In step S240, the retired batteries are rested for a certain time.

[0072] In step S230, the retired batteries are discharged at a constant current, and the discharge time is the same as the charging time. In the constant current discharging stage, different pulse currents are used for charging and discharging.

[0073] In step S250, the retired batteries are charged at a constant voltage for a certain time.

[0074] In step S260, the retired batteries are rested for a certain time.

[0075] In step S270, the retired batteries are discharged at a constant voltage, and the discharge time is the same as the charging time.

[0076] In step S230, the retired batteries are discharged at a constant current, and the discharge time is the same as the charging time. In the constant current discharging stage, different pulse currents are used for charging and discharging.

[0077] In step S280, the charging and discharging data and external physical parameter data of each retired battery are obtained.

[0078] Through the pulse current of different discharge rates or the simplified incomplete charge and discharge experiment conditions, the test time required is short, the electric energy consumed is less, and the electric power cost is low, since the complete charge and discharge is not required, and only the charging is performed in fragments.

[0079] In the process of charging and discharging the retired single battery, the charging and discharging data and external physical parameter data of the retired single battery are collected by the data acquisition instrument, wherein the charging and discharging data are, for example, current, voltage, time, etc., and the external physical parameter data are, for example, temperature change and size change, etc.

[0080] The data collected by the data acquisition instrument are stored in the database, which corresponds to the identification code of the retired battery.

[0081] S300, according to the charging and discharging data and external physical parameter data, an analysis processing sample is obtained, and the analysis processing sample of each retired battery is stored in the identity recognition code database.

[0082] The charging and discharging data and external physical parameter data in the time period of the same time interval and time starting point are selected, a plurality of feature vectors are obtained, the plurality of feature vectors are normalized to obtain an analysis processing sample, wherein the plurality of feature vectors include temperature change time series of each charging and discharging stage, voltage change time series of the constant voltage charging stage, and incremental capacity of the constant current stage.

[0083] The temperature change time series of each charging and discharging stage is obtained by the following formula:

[0084]

[0085] In the formula, T represents the temperature change time series of each charging and discharging stage, t represents the sampling time, and when the temperature change time series of each charging and discharging stage is obtained, the value range of t is t=1, 2, ……n1.

[0086] The voltage change time series of the constant voltage charging stage is obtained by the following formula:

[0087]

[0088] In the formula, V represents the voltage change time series of the constant voltage charging stage, t represents the sampling time, and when the voltage change time series of the constant voltage charging stage is obtained, the value range of t is t=1, 2, ……n2.

[0089] The incremental capacity of the battery is defined as the ratio of the capacity change to the terminal voltage change, i.e. dQ / dV, and the IC curve is the incremental capacity curve, and the change curve is drawn with the dQ / dV value as the ordinate and the terminal voltage V value corresponding to the abscissa, and the value of dQ / dV can be obtained as follows:

[0090]

[0091] Further, the incremental capacity of the constant current stage is obtained by the following formula:

[0092]

[0093] In the formula, IC represents the incremental capacity of the constant current stage, Q(t) and V(t) represent the battery capacity and terminal voltage at time t, Q(K) and V(K) represent the discrete form of the battery capacity and the discrete form of the terminal voltage, I represents the current of the constant current stage, N represents the sampling interval, T0 represents the sampling period time, Δt represents the time change, ΔV(K) represents the change of the terminal voltage at time K in the discrete form, and V(K-N) represents the terminal voltage at time K-N T0.

[0094] The analysis processing sample is a set of normalized characteristic vectors, i.e.

[0095] S={T',V',IC'};

[0096] In the formula, S represents the analysis processing sample, T', V' and IC' represent the temperature change time sequence of each charge and discharge stage, the voltage change time sequence of the constant voltage charging stage and the incremental capacity of the constant current stage after normalization processing, respectively.

[0097] Because different characteristic vectors have different scales due to different dimensions and units, the clustering weight is distorted. The clustering weight is distorted because the clustering algorithm based on distance as a measure is biased towards the index with larger value. For example, the capacity and internal resistance of the battery are used as clustering indexes, the capacity is in mAh, and the internal resistance is in Ω. The aging degree of the batteries with 2000 mAh and 1980 mAh is not much different, while the batteries with internal resistance of 0.1 Ω and 0.3 Ω may be the difference between new batteries and scrap batteries. However, the value of 2000 is 20 larger than that of 1980, and the value of 0.1 is 0.2 smaller than that of 0.3. The former value is much larger than the latter value, and when the Euclidean distance is calculated, the distance value is mainly contributed by the former, so that the clustering result is biased towards the distribution of the index with larger value, and the distribution of the index with less importance is almost ignored. This is the clustering weight distortion, so the normalization processing is needed for the characteristic vectors.

[0098] S400, extract the nodes and node relationships of the stored information in the identity recognition code database, and construct a dynamic knowledge graph.

[0099] Please see Figure 1 , Figure 3 As shown in Table 1, in one embodiment of the present invention, the dynamic knowledge graph includes nodes, node relationships, and temporal attributes. The nodes are: retired battery 300, manufacturer 200, new energy vehicle 100, operation log 400, maintenance log 600, and analysis and processing sample 500. The node relationships include the relationship between retired battery 300 and manufacturer 200, the relationship between retired battery 300 and new energy vehicle 100, the relationship between retired battery 300 and operation log 400, the relationship between retired battery 300 and maintenance log 600, and the relationship between retired battery 300 and analysis and processing sample 500. The temporal attributes are time-based attributes of the nodes and node relationships. By establishing the dynamic knowledge graph, key data visualization for the full lifecycle management of retired battery 300 is achieved.

[0100] Table 1 shows the nodes, relationships, and temporal attributes of the dynamic knowledge graph.

[0101]

[0102] S500, according to the frequent subgraph algorithm, obtain the feature tensor of the dynamic knowledge graph.

[0103] The frequent subgraph algorithm is used to obtain the m×k dimension feature tensor F of the dynamic knowledge graph. G Where, the feature tensor F of the i-th row Gi , which corresponds to the feature tensor of the dynamic knowledge graph of the i-th retired battery. Specifically, the frequent subgraph algorithm is existing technology and will not be described here.

[0104] 600, the analyzed and processed samples are used as the input layer of the Long Short-Term Memory Network model to obtain the temporal characteristics of each retired battery.

[0105] S700, the temporal features and the feature tensor are used as the input layer of the fully connected network layer model to obtain the label vector.

[0106] In steps S600 and S700, retired batteries based on a dynamic real knowledge graph are classified using a classification neural network model. This classification neural network model comprises a two-layer structure: a first layer is a long short-term memory network model, and a second layer is a fully connected network model. In this embodiment, the classification neural network model is existing technology and will not be described further.

[0107] The analysis and processing samples of each decommissioned battery are input into the Long Short-Term Memory network model to obtain the time-series features F of the i-th decommissioned battery. Ti Connectivity feature tensor F Gi and temporal features FTi , to obtain the comprehensive feature F of the i-th retired battery i , to obtain the comprehensive feature F of the i-th retired battery i The full connection network layer model is input, and the final output corresponds to the classification result of the labeled retired battery, that is, a label vector with three dimensions is output. In this embodiment, the feature tensor F Gi For example, (1, 2, 3), the time sequence feature F Ti For example, (7, 8, 0), the connection feature tensor F Gi and the time sequence feature F Ti , to obtain the comprehensive feature F of the i-th retired battery i is (1, 2, 3, 7, 8, 0).

[0108] S800, according to the label vector and the identity recognition code database, classifying the quality of the retired battery.

[0109] According to the classification result, the corresponding identification code is found by combining the data stored in the identity recognition code database, and then each retired battery is classified, and the same retired batteries are grouped to achieve good usability. In this embodiment, for example, the final goal is to divide the quality of the retired battery into three, then according to the remaining capacity of the retired battery, the retired battery can be divided into good, medium and poor according to the remaining capacity. Specifically, the output label vector [1, 0, 0] corresponds to the retired battery state of good, which can be directly recycled and reused. The output label vector [0, 1, 0] corresponds to the retired battery state of medium, which can be further screened or disassembled, and then recycled and reused. The output label vector [0, 0, 1] corresponds to the retired battery state of poor, which cannot be further recycled and reused.

[0110] Please refer to Figure 4As shown, the application also provides a retired battery classification system based on a dynamic knowledge graph, which comprises a code scanner 710, a charge-discharge instrument 720, a data acquisition instrument 730, a storage and operation device 740 and a retired battery 300. The code scanner 710 is used to generate an identification code of each retired battery 300, and the charge-discharge instrument 720 is used to perform charge-discharge test on the retired battery 300 according to a set charge-discharge strategy. The data acquisition instrument 730 is used to acquire charge-discharge data and external physical parameter data of each retired battery 300, and the storage and operation device 740 is used to establish an identification code database of the identification code and the corresponding relationship between the production manufacturer, the new energy vehicle, the operation log and the maintenance log, acquire analysis processing samples according to the charge-discharge data and the external physical parameter data, and store the analysis processing samples of each retired battery 300 in the identification code database correspondingly, extract nodes and node relationships of the stored information in the identification code database, construct a dynamic knowledge graph, acquire a feature tensor of the dynamic knowledge graph according to a frequent subgraph algorithm, take the analysis processing samples as an input layer of a long short-term memory network model, acquire time sequence features of each retired battery 300, take the time sequence features and the feature tensor as an input layer of a full connection network layer model, acquire a label vector, and classify the retired battery 300 according to the label vector and the identification code database.

[0111] Referring to Figure 4 In an embodiment of the application, the charge-discharge instrument 720 performs charge-discharge test on the retired battery according to the set charge-discharge strategy, which comprises the following steps:

[0112] S210, the retired battery is charged at a constant current for a certain time.

[0113] S220, the retired battery is rested for a certain time.

[0114] S230, the retired battery is discharged at a constant current, and the discharge time is the same as the charge time. In the constant current discharge stage, the retired battery is charged and discharged by different pulse currents.

[0115] S240, the retired battery is rested for a certain time.

[0116] S250, the retired battery is charged at a constant voltage for a certain time.

[0117] S260, the retired battery is rested for a certain time.

[0118] S270, the retired battery is discharged at a constant voltage, and the discharge time is the same as the charge time.

[0119] S280, the charge-discharge data and the external physical parameter data of each retired battery are acquired.

[0120] Referring to Figure 4As shown, in an embodiment of the present application, the dynamic knowledge graph comprises nodes, node relationships and time sequence attributes, wherein the nodes are the retired battery, the manufacturer, the new energy vehicle, the operation log, the maintenance log and the analysis processing sample, the node relationships comprise the relationship between the retired battery and the manufacturer, the relationship between the retired battery and the new energy vehicle, the relationship between the retired battery and the operation log, the relationship between the retired battery and the maintenance log and the relationship between the retired battery and the analysis processing sample, and the time sequence attributes are the attributes of the nodes and the node relationships based on time.

[0121] It is apparent for a person skilled in the art that the present application is not limited to the details of the above-described exemplary embodiments, but can be implemented in other concrete forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered in all aspects as illustrative and not restrictive, the scope of the present application being defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and range of equivalency of the claims are embraced therein, and no claim element should be considered to be limited by the reference signs in the claims.

[0122] The above-described embodiments only represent the implementation of the present application, and the protection scope of the present application is not limited to the above-described embodiments. For those skilled in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. A method for classifying retired batteries based on a dynamic knowledge graph, characterized in that, The method comprises the following steps: generating an identification code of each retired battery, and establishing an identification code database corresponding the identification code to a manufacturer, a new energy vehicle, an operation log and a maintenance log; setting a charging and discharging strategy to perform charging and discharging test on the retired battery, and obtaining charging and discharging data and external physical parameter data of each retired battery; obtaining analysis processing samples according to the charging and discharging data and the external physical parameter data, and storing the analysis processing samples of each retired battery in the identification code database; extracting nodes and node relationships of the stored information in the identification code database, and constructing a dynamic knowledge graph; the dynamic knowledge graph comprises nodes, node relationships and time sequence attributes, wherein the nodes are the retired battery, the manufacturer, the new energy vehicle, the operation log, the maintenance log and the analysis processing samples, the node relationships comprise relationships of the retired battery to the manufacturer, the new energy vehicle, the operation log, the maintenance log and the analysis processing samples, and the time sequence attributes are attributes of the nodes and the node relationships based on time; obtaining a feature tensor of the dynamic knowledge graph according to a frequent subgraph algorithm; taking the analysis processing samples as an input layer of a long short-term memory network model to obtain time sequence features of each retired battery; taking the time sequence features and the feature tensor as an input layer of a fully connected network layer model to obtain a label vector; classifying the retired battery according to the label vector and the identification code database. 2.The method of claim 1, wherein, The method of setting a charging and discharging strategy to perform charging and discharging test on the retired battery comprises the following steps: performing constant current charging on the retired battery for a certain time; resting for a certain time; performing constant current discharging on the retired battery, and the discharging time is the same as the charging time; in the constant current discharging stage, different pulse currents are used for charging and discharging; resting for a certain time; performing constant voltage charging on the retired battery for a certain time; resting for a certain time; performing constant voltage discharging on the retired battery, and the discharging time is the same as the charging time; obtaining the charging and discharging data and the external physical parameter data of each retired battery. 3.The method of claim 1, wherein, selecting charging and discharging data and external physical parameter data in a time period with the same time interval and time starting point, obtaining a plurality of feature vectors, performing normalization processing on the plurality of feature vectors to obtain analysis processing samples, wherein the plurality of feature vectors comprise temperature change time sequences of each charging and discharging stage, voltage change time sequences of the constant voltage charging stage and incremental capacity of the constant current stage. 4.The method of claim 3, wherein, The temperature change time sequences of each charging and discharging stage are obtained by the following formula: ； In the formula, temperature change time series of each charge and discharge stage, temperature change time series of each charge and discharge stage, temperature change time series of each charge and discharge stage, 5.The method of claim 4, wherein, The voltage change time sequences of the constant voltage charging stage are obtained by the following formula: ； In the formula, a voltage change time series of the constant-voltage charging phase, is represented as a sampling time, and when the voltage change time series of the constant-voltage charging phase is acquired, its value range is . 6.The method of claim 5, wherein, The incremental capacity of the constant current stage is obtained by the following formula: ； wherein, is expressed as incremental capacity at constant current phase, and is expressed as battery capacity and terminal voltage at time, is expressed as discrete form of terminal voltage, is expressed as current at constant current phase, is expressed as sampling interval, is expressed as sampling period time, is expressed as time change, is expressed as terminal voltage change at time in discrete form, is expressed as is expressed as terminal voltage at time, The analysis processing samples are a set of the plurality of feature vectors after normalization processing, that is, ； In the formula, is expressed as an analysis process sample, , and respectively represent the temperature change time series of each charge and discharge stage, the voltage change time series of the constant voltage charge stage, and the incremental capacity of the constant current stage after normalization processing.

7. A retired battery classification system based on a dynamic knowledge graph, characterized in that, The method comprises the following steps: a code scanner is used to generate an identification code of each retired battery; The charge-discharge instrument is used for charging and discharging test of the retired battery according to a set charging and discharging strategy. The data acquisition instrument is used for acquiring the charging and discharging data and external physical parameter data of each retired battery. The storage and operation device is used for establishing an identity recognition code database of corresponding relationship of the identity recognition code, the manufacturer, the new energy vehicle, the operation log and the maintenance log; acquiring analysis processing samples according to the charging and discharging data and the external physical parameter data, and storing the analysis processing samples of each retired battery in the identity recognition code database; extracting nodes and node relationships of stored information in the identity recognition code database, and constructing a dynamic knowledge graph; and acquiring a feature tensor of the dynamic knowledge graph according to a frequent subgraph algorithm. The analysis processing samples are used as an input layer of a long short-term memory network model to acquire time sequence features of each retired battery; the time sequence features and the feature tensor are used as an input layer of a full connection network layer model to acquire a label vector; and the retired batteries are classified according to quality based on the label vector and the identity recognition code database. The dynamic knowledge graph includes nodes, node relationships and time sequence attributes, wherein the nodes are the retired batteries, the manufacturers, the new energy vehicles, the operation logs, the maintenance logs and the analysis processing samples, the node relationships include relationships of the retired batteries and the manufacturers, the retired batteries and the new energy vehicles, the retired batteries and the operation logs, the retired batteries and the maintenance logs and the retired batteries and the analysis processing samples, and the time sequence attributes are attributes of the nodes and the node relationships based on time.

8. The dynamic knowledge graph-based retired battery classification system of claim 7, wherein, In the charge-discharge instrument, the set charging and discharging strategy for the retired battery charging and discharging test includes the following steps: The retired battery is charged with constant current for a certain time; The retired battery is rested for a certain time; The retired battery is discharged with constant current, and the discharge time is the same as the charging time. In the constant current discharge stage, the charging and discharging are performed through different pulse currents; The retired battery is rested for a certain time; The retired battery is charged with constant voltage for a certain time; The retired battery is rested for a certain time; The retired battery is discharged with constant voltage, and the discharge time is the same as the charging time; The charging and discharging data and the external physical parameter data of each retired battery are acquired.

Images