CN117129879B

CN117129879B - Threshold adjustment method and training method of battery state of health prediction model

Info

Publication number: CN117129879B
Application number: CN202311393683.7A
Authority: CN
Inventors: 陈培育; 李斌; 王洋; 郑骁麟; 袁中琛; 崇志强; 李振斌; 于天一; 姚程; 王峥; 李冠争; 李超; 李树鹏
Original assignee: Tianjin University; State Grid Corp of China SGCC; State Grid Tianjin Electric Power Co Ltd; Electric Power Research Institute of State Grid Tianjin Electric Power Co Ltd
Current assignee: Tianjin University; State Grid Corp of China SGCC; State Grid Tianjin Electric Power Co Ltd; Electric Power Research Institute of State Grid Tianjin Electric Power Co Ltd
Priority date: 2023-10-26
Filing date: 2023-10-26
Publication date: 2024-01-26
Anticipated expiration: 2043-10-26
Also published as: CN117129879A

Abstract

The invention provides a threshold adjustment method and a training method of a battery health state prediction model, which can be applied to the technical field of data processing and the technical field of lithium ion batteries. The threshold adjustment method comprises the following steps: extracting a characteristic charge capacity corresponding to a predetermined state of charge parameter from the initial charge capacity information; inputting the characteristic charge capacity into a trained battery health state prediction model, and outputting real-time battery health state prediction information of the lithium ion battery; based on the objective function, obtaining objective ohmic resistance information corresponding to the real-time battery health state prediction information according to the real-time battery health state prediction information; obtaining target short-circuit current information according to the target ohmic resistance information and the open-end voltage information; based on the target short circuit current information, a protection threshold of a direct current side circuit associated with the lithium ion battery is adjusted.

Description

Threshold adjustment method and training method of battery state of health prediction model

Technical Field

The invention relates to the technical field of data processing and the technical field of lithium electronic batteries, in particular to a threshold adjustment method, a training method and device of a battery health state prediction model, electronic equipment and a storage medium.

Background

In the process of converting the global energy structure from relying on traditional fossil energy to pursuing clean and high-efficiency energy, as a core for maintaining system stability, energy conversion and buffering, regulation and efficiency improvement, transmission and scheduling, management and operation, the number of lithium ion energy storage power stations is increasing along with the development of new energy in the future. And because the capacity scale of the lithium ion energy storage power station is large and the structure is complex, the loss caused by the lithium ion energy storage power station is large after the safety accident occurs.

In lithium ion energy storage power stations, lithium ion batteries are the primary devices that cause fire explosions within the station, and are also the core devices of high value in lithium ion energy storage power stations. The state of health of the lithium ion battery can be estimated to determine the condition of the lithium ion battery, so that the safe operation of the lithium ion energy storage power station can be ensured.

In the process of realizing the inventive concept, the inventor finds that the internal resistance of the lithium ion battery is gradually increased in the aging process, so that the inter-cluster short-circuit current and the intra-cluster short-circuit current of the direct current side circuit of the lithium ion energy storage power station become smaller, and therefore, the direct current side fuse is refused to operate.

Disclosure of Invention

In view of the above, the present invention provides a threshold adjustment method and a training method, device, electronic apparatus and storage medium of a battery state of health prediction model.

According to a first aspect of the present invention, there is provided a method for adjusting a threshold value of a dc side circuit of a lithium ion energy storage power station, comprising: responding to the received initial charge capacity information of the lithium ion battery corresponding to a plurality of charge state parameters in the constant current charging process and the starting voltage information of the lithium ion battery in a non-working state, and extracting the characteristic charge capacity corresponding to the preset charge state parameters from the initial charge capacity information; inputting the characteristic charge capacity into a trained battery health state prediction model, and outputting real-time battery health state prediction information of the lithium ion battery; obtaining target ohmic resistance information corresponding to real-time battery health state prediction information according to the real-time battery health state prediction information based on an objective function, wherein the objective function represents the association relationship between the battery health state information and ohmic resistance parameters of an equivalent circuit model, and the equivalent circuit model is used for simulating the working state of the lithium ion battery under any working condition; obtaining target short-circuit current information according to the target ohmic resistance information and the open-end voltage information; and adjusting a protection threshold of a direct current side circuit associated with the lithium ion battery based on the target short circuit current information.

According to the embodiment of the invention, the battery state of health prediction model comprises N prediction modules which are connected in parallel, wherein the output ends of the N prediction modules are connected with the linear regression module, and N is an integer greater than 1; inputting the characteristic charge capacity into a trained battery state of health prediction model, outputting real-time battery state of health prediction information of the lithium ion battery, comprising: the characteristic charging capacity is respectively input into the input ends of N prediction modules, and N battery health state prediction information is output; based on N weights corresponding to the N prediction modules, the linear regression module is utilized to process the N battery health state prediction information, and the real-time battery health state prediction information of the lithium ion battery is obtained.

According to an embodiment of the invention, the N prediction modules comprise at least two of: the prediction module is constructed based on a light gradient lifting algorithm, an extreme gradient lifting algorithm, a random forest algorithm, a support vector regression algorithm and a Gaussian process regression algorithm.

According to an embodiment of the present invention, the method for adjusting the threshold value of the dc side circuit of the lithium ion energy storage power station further includes: acquiring a plurality of electrochemical impedance spectrums and M sample charging capacities of the lithium ion battery corresponding to a plurality of charge state parameters in M times of cyclic charge and discharge processes, wherein M is an integer greater than 1; determining ohmic resistance parameters of a plurality of equivalent circuit models corresponding to the plurality of state-of-charge parameters according to the plurality of electrochemical impedance spectrums; obtaining the health state information of M sample batteries of the lithium ion battery according to the charging capacity of the M samples and the standard capacity of the lithium ion battery; and fitting the M sample battery health state information with ohmic resistance parameters of a plurality of equivalent circuit models to obtain an objective function.

According to an embodiment of the present invention, fitting the M sample battery state of health information with ohmic resistance parameters of a plurality of equivalent circuit models to obtain an objective function includes: constructing a function to be fitted by taking the state of health information of M sample batteries as independent variables and ohmic resistance parameters as dependent variables; inputting the state of health information of the M sample batteries and ohmic resistance parameters of a plurality of equivalent circuit models into a function to be fitted to obtain a target fitting coefficient; and obtaining an objective function according to the objective fitting coefficient and the function to be fitted.

According to a second aspect of the present invention, there is provided a training method of a battery state of health prediction model, comprising: acquiring a sample data set, wherein the sample data set comprises actual battery health state information of a sample lithium ion battery corresponding to sample characteristic charging capacity information; inputting sample characteristic charging capacity information of the sample lithium ion battery into an initial model, and outputting battery health state sample prediction information of the sample lithium ion battery; obtaining a loss value based on the loss function according to the battery health state sample prediction information and the battery health state actual information; based on the loss value, model parameters of the initial model are adjusted, and a trained battery health state prediction model is obtained.

According to the embodiment of the invention, the initial model comprises N prediction modules which are connected in parallel, wherein the output ends of the N prediction modules are connected with the linear regression module, and N is an integer greater than 1; the training method of the battery state of health prediction model further comprises the following steps: respectively carrying out parameter optimization on each initial prediction module to obtain prediction modules; wherein the parameters correspond to algorithms running within each prediction module; the algorithm includes any one of the following: light gradient lifting algorithm, extreme gradient lifting algorithm, random forest algorithm, support vector regression algorithm and gaussian process regression algorithm.

According to an embodiment of the present invention, parameter optimization is performed on each initial prediction module to obtain a prediction module, including: inputting sample characteristic charging capacity information into an initial prediction module to obtain first probability distribution of battery health state, wherein the initial prediction module is constructed based on initial parameters; processing the first probability distribution and initial parameters of the initial prediction module to obtain Gaussian distribution of the first parameters and the battery health state; determining an intermediate parameter from the first parameter and a gaussian distribution of battery state of health; changing initial parameters based on the intermediate parameters to obtain an intermediate prediction module; inputting sample characteristic charging capacity information into an intermediate prediction module to obtain second probability distribution of the battery health state; processing the first probability distribution, the second probability distribution, the initial parameters and the intermediate parameters to obtain Gaussian distribution of the second parameters and the battery health state; returning to execute the operation of determining the intermediate parameter in the case that the number of times of optimization is determined to be smaller than the predetermined threshold value; and under the condition that the optimization times are equal to a preset threshold value, obtaining a prediction module.

A third aspect of the present invention provides a threshold adjustment device for a dc side circuit of a lithium ion energy storage power station, comprising: the extraction module is used for responding to the received initial charge capacity information of the lithium ion battery corresponding to the plurality of charge state parameters in the constant current charging process and the initial voltage information of the lithium ion battery in the non-working state, and extracting the characteristic charge capacity corresponding to the preset charge state parameters from the initial charge capacity information; the first input module is used for inputting the characteristic charge capacity into the trained battery health state prediction model and outputting real-time battery health state prediction information of the lithium ion battery; the first obtaining module is used for obtaining target ohmic resistance information corresponding to the real-time battery health state prediction information based on an objective function according to the real-time battery health state prediction information, wherein the objective function represents the association relationship between the battery health state information and ohmic resistance parameters of an equivalent circuit model, and the equivalent circuit model is used for simulating the working state of the lithium ion battery under any working condition; the second obtaining module is used for obtaining target short-circuit current information according to the target ohmic resistance information and the open-end voltage information; and a first adjustment module for adjusting a protection threshold of a direct current side circuit associated with the lithium ion battery based on the target short circuit current information.

A fourth aspect of the present invention provides a training apparatus of a battery state of health prediction model, comprising: the third obtaining module is used for obtaining a sample data set, wherein the sample data set comprises actual battery health state information of a sample lithium ion battery corresponding to sample characteristic charging capacity information; the second input module is used for inputting sample characteristic charging capacity information of the sample lithium ion battery into the initial model and outputting battery health state sample prediction information of the sample lithium ion battery; the fourth obtaining module is used for obtaining a loss value based on the loss function according to the battery health state sample prediction information and the battery health state actual information; and the second adjusting module is used for adjusting the model parameters of the initial model based on the loss value to obtain a trained battery health state prediction model.

A fifth aspect of the present invention provides an electronic device comprising: one or more processors; and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method for adjusting the threshold of the direct current side circuit of the lithium ion energy storage power station or the method for training the battery state of health prediction model.

The sixth aspect of the present invention also provides a computer readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the above-described method for adjusting the threshold value of the dc side circuit of a lithium ion energy storage power station or the above-described method for training a battery state of health prediction model.

According to the threshold adjustment method, the training method of the battery state of health prediction model, the training device of the battery state of health prediction model, the electronic equipment and the storage medium, the characteristic charging capacity is extracted from the charging capacity information obtained in the lithium ion charging process. Based on the method, only the characteristic charge capacity is input into the trained battery health state prediction model, and the use of the full charge capacity information to obtain the real-time battery health state prediction information of the lithium ion battery is avoided. Therefore, the influence of inaccurate charge capacity information on the accuracy of the real-time battery health state prediction information is avoided, the information quantity processed by the battery health state prediction model is reduced, the efficiency of determining the real-time battery health state prediction information is further improved, and the accuracy of the real-time battery health state prediction information is improved.

And the target ohmic resistance information is obtained according to the real-time battery health state prediction information through the association relation between the battery health state information and the ohmic resistance parameters of the equivalent circuit model, so that the efficiency of obtaining the target ohmic resistance information is improved.

Based on this, by using the target ohmic resistance information and the actually measured open-end voltage, target short-circuit current information corresponding to the real-time lithium ion health state is obtained. The efficiency of determining the real-time battery health state prediction information is improved, and the accuracy of the real-time battery health state prediction information is improved. Therefore, the efficiency of determining the target short-circuit current information is improved, and the accuracy of the target short-circuit current information is improved.

Therefore, the protection threshold value of the direct current side circuit related to the lithium ion battery can be adjusted based on accurate target short circuit current information, the accuracy of the adjusted protection threshold value can be improved, and the protection refusal condition caused by the reduction of the short circuit current after the short circuit fault of the direct current side circuit due to the reduction of the health degree of the lithium ion battery can be avoided.

Drawings

The foregoing and other objects, features and advantages of the invention will be apparent from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

fig. 1 shows an application scenario diagram of a method for adjusting a threshold value of a direct current side circuit of a lithium ion energy storage power station or a training method of a battery health state prediction model according to an embodiment of the invention;

FIG. 2 illustrates a flow chart of a method of threshold adjustment of a DC side circuit of a lithium ion energy storage power station in accordance with an embodiment of the invention;

fig. 3 shows a block diagram of an equivalent circuit model of a lithium ion battery according to an embodiment of the present invention;

FIG. 4 illustrates a schematic diagram of output implementation battery state of health prediction information, according to an embodiment of the present invention;

FIG. 5 illustrates a flowchart of a method of training a battery state of health prediction model, according to an embodiment of the present invention;

FIG. 6 shows a schematic diagram of a prediction module optimization method according to an embodiment of the invention;

FIG. 7 shows a block diagram of a threshold adjustment device for a DC side circuit of a lithium ion energy storage power station in accordance with an embodiment of the invention;

FIG. 8 shows a block diagram of a training device of a battery state of health prediction model according to an embodiment of the present invention; and

fig. 9 shows a block diagram of an electronic device suitable for implementing a method for threshold adjustment of a direct current side circuit of a lithium ion energy storage power station or a training method of a battery state of health prediction model according to an embodiment of the invention.

Detailed Description

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. It should be understood that the description is only illustrative and is not intended to limit the scope of the invention. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the invention. It may be evident, however, that one or more embodiments may be practiced without these specific details. In addition, in the following description, descriptions of well-known structures and techniques are omitted so as not to unnecessarily obscure the present invention.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The terms "comprises," "comprising," and/or the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.

All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It should be noted that the terms used herein should be construed to have meanings consistent with the context of the present specification and should not be construed in an idealized or overly formal manner.

Where expressions like at least one of "A, B and C, etc. are used, the expressions should generally be interpreted in accordance with the meaning as commonly understood by those skilled in the art (e.g.," a system having at least one of A, B and C "shall include, but not be limited to, a system having a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.).

In addition, the inventor finds that the measured values of voltage and the like are inaccurate due to the measurement error and the interference of environmental noise when the lithium ion battery is actually used, so that the accuracy of the predicted battery health state information is low.

In view of this, an embodiment of the present invention provides a method for adjusting a threshold value of a dc side circuit of a lithium ion energy storage power station, including: and responding to the received initial charge capacity information of the lithium ion battery corresponding to the plurality of charge state parameters in the constant current charging process and the starting voltage information of the lithium ion battery in the non-working state, and extracting the characteristic charge capacity corresponding to the preset charge state parameters from the initial charge capacity information. And inputting the characteristic charge capacity into a trained battery health state prediction model, and outputting real-time battery health state prediction information of the lithium ion battery. And obtaining target ohmic resistance information corresponding to the real-time battery health state prediction information based on the objective function according to the real-time battery health state prediction information, wherein the objective function represents the association relationship between the battery health state information and ohmic resistance parameters of an equivalent circuit model, and the equivalent circuit model is used for simulating the working state of the lithium ion battery under any working condition. And obtaining target short-circuit current information according to the target ohmic resistance information and the open-end voltage information. Based on the target short circuit current information, a protection threshold of a direct current side circuit associated with the lithium ion battery is adjusted.

Fig. 1 shows an application scenario diagram of a method for adjusting a threshold value of a direct current side circuit of a lithium ion energy storage power station or a training method of a battery health state prediction model according to an embodiment of the invention.

As shown in fig. 1, an application scenario 100 according to this embodiment may include a terminal device 101, a server 102, and a lithium ion energy storage power station 103. The medium of the communication link may be provided between the terminal device 101, the server 102 and the lithium ion energy storage power station 103 via a network. The network may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

A user may interact with at least one of the server 102 and the lithium ion energy storage power station 103 over a network using the terminal device 101 to receive or send messages or the like. Various communication client applications, such as shopping class applications, web browser applications, search class applications, instant messaging tools, mailbox clients, social platform software, etc., may be installed on the terminal device 101 (by way of example only).

The terminal device 101 may be a variety of electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablets, laptop and desktop computers, and the like.

The server 105 may be a server providing various services, such as a background management server (for example only) providing support for websites browsed by the user using the terminal device 101. The background management server may analyze and process the received data such as the user request, and feed back the processing result (e.g., the web page, information, or data obtained or generated according to the user request) to the terminal device.

For example, the server 105 may be utilized to collect information such as voltage and current on the dc side of the lithium ion energy storage power station.

It should be noted that, the method for adjusting the threshold value of the direct current side circuit of the lithium ion energy storage power station or the training method of the battery health state prediction model provided in the embodiment of the invention may be generally executed by the server 102. Accordingly, the threshold adjustment device of the direct current side circuit of the lithium ion energy storage power station or the training device of the battery health state prediction model provided in the embodiment of the present invention may be generally disposed in the server 102. The method for adjusting the threshold value of the direct current side circuit of the lithium ion energy storage power station or the method for training the battery health state prediction model provided by the embodiment of the invention may also be executed by a server or a server cluster which is different from the server 102 and can communicate with the terminal device 101, the lithium ion energy storage power station 103 and/or the server 102. Accordingly, the threshold adjustment device of the direct current side circuit of the lithium ion energy storage power station or the training device of the battery health state prediction model provided in the embodiment of the present invention may also be provided in a server or a server cluster different from the server 102 and capable of communicating with the terminal device 101, the lithium ion energy storage power station 103 and/or the server 102.

It should be understood that the number of terminal devices, servers and lithium ion energy storage power stations in fig. 1 is merely illustrative. Any number of terminal devices, servers, and lithium ion energy storage power stations may be provided, as desired.

The method for adjusting the threshold value of the direct current side circuit of the lithium ion energy storage power station or the method for training the battery health state prediction model according to the embodiment of the invention will be described in detail below based on the scenario described in fig. 1 through fig. 2 to 6.

Fig. 2 shows a flow chart of a method for threshold adjustment of a dc side circuit of a lithium ion energy storage power station according to an embodiment of the invention.

As shown in fig. 2, the method for adjusting the threshold of the dc side circuit of the lithium ion energy storage power station in this embodiment includes operations S210 to S250.

In operation S210, characteristic charge capacities corresponding to predetermined state of charge parameters are extracted from initial charge capacity information in response to received initial charge capacity information of the lithium ion battery corresponding to a plurality of state of charge parameters in a constant current charge process and start voltage information of the lithium ion battery in a non-operating state.

In operation S220, the characteristic charge capacity is input into the trained battery state of health prediction model, and real-time battery state of health prediction information of the lithium ion battery is output.

In operation S230, based on the objective function, according to the real-time battery health status prediction information, the objective ohmic resistance information corresponding to the real-time battery health status prediction information is obtained, wherein the objective function characterizes the association relationship between the battery health status information and the ohmic resistance parameter of the equivalent circuit model, and the equivalent circuit model is used for simulating the working status of the lithium ion battery under any working condition.

In operation S240, target short-circuit current information is obtained from the target ohmic resistance information and the open-end voltage information.

In operation S250, a protection threshold of a direct current side circuit associated with the lithium ion battery is adjusted based on the target short-circuit current information.

According to an embodiment of the invention, the lithium-ion battery may be a rechargeable battery deployed in a lithium-ion energy storage power station, operating by lithium ions moving between a positive electrode and a negative electrode.

The State of Charge (SOC) may represent the ratio between the remaining Charge capacity of the lithium ion battery after aging and the Charge capacity of the lithium ion battery in a fully charged State. Wherein the remaining charge capacity may be related to the health of the lithium ion battery. For example, the higher the health of a lithium ion battery, the correspondingly higher the remaining charge capacity of the lithium ion battery; the lower the health of the lithium ion battery, the lower the remaining charge capacity of the lithium ion battery may correspond.

The initial charge capacity corresponding to the plurality of state-of-charge parameters may be a charge capacity corresponding to each of the plurality of state-of-charge parameters obtained when the lithium ion battery is charged. For example, the state of charge parameter may include 1% -100%. For example, in the case where the state of charge parameter is 90%, a charge capacity corresponding to 90% may be obtained from the charge capacity of the lithium ion battery, and the charge capacity may be determined as the above-described initial charge capacity.

The predetermined state of charge parameter may represent a predetermined value of the state of charge parameter. For example, the predetermined state of charge parameter may include 30% -70%. The characteristic charge capacity corresponding to the predetermined state of charge parameter may be a charge capacity corresponding to the predetermined state of charge parameter in the initial charge capacity information. For example, the charging capacity may be a charging capacity corresponding to a state of charge parameter of 30% -70%.

According to the embodiment of the invention, the battery state of health prediction model can be constructed according to the association relation between the charging capacity and the battery state of health information. The trained battery state of health prediction model may be a battery state of health prediction model that is optimized a number of times equal to a predetermined threshold. The trained battery state of health prediction model may be used to determine real-time battery state of health prediction information corresponding to the charge capacity.

The real-time battery state of health prediction information may be real-time battery state of health information of the lithium ion battery predicted by the battery state of health prediction model. The battery state of health information may be used to characterize the health of the lithium ion battery.

According to an embodiment of the present invention, the target ohmic resistance information may be ohmic resistance information of an equivalent circuit model of the lithium ion battery corresponding to the real-time battery state of health prediction information. The ohmic resistance information may include ohmic resistance parameters in the equivalent circuit model.

The ohmic resistance parameter may include a resistance value of an ohmic resistance, etc. The working state of the lithium ion battery under any working condition can comprise a battery discharging state and the like.

According to an embodiment of the present invention, the non-operating state may include an undischarged state of the lithium ion battery. The open-end voltage information may be a measured potential difference between the positive and negative electrodes of the lithium ion battery, i.e., an open-end voltage, in a non-operating state. For example, the open-end voltage information may be measured after the lithium ion battery is left to stand for a first predetermined period of time, whereby the target short-circuit current information of the aged lithium ion battery may be determined from the open-end voltage information; alternatively, the open-ended voltage information may be measured after the lithium-ion battery is used for a second predetermined period of time, whereby the target short-circuit current information of the aged lithium-ion battery may be determined from the open-ended voltage information. The first predetermined time length and the second predetermined time length can be set according to requirements.

According to an embodiment of the present invention, the target short-circuit current information may be short-circuit current information of the lithium ion battery corresponding to the real-time battery state of health prediction information. The protection threshold of the dc side circuit may be a current threshold. The protection threshold of the direct-current side circuit may be adjusted according to the current value of the target short-circuit current. For example, when the current value of the target short-circuit current information decreases, the protection threshold of the dc-side circuit may be correspondingly decreased. Therefore, the protection refusal condition caused by the reduction of the short-circuit current after the short-circuit fault of the direct-current side circuit due to the reduction of the health degree of the lithium ion battery can be avoided.

According to the embodiment of the present invention, the characteristic charge capacity can be extracted from the initial charge capacity information corresponding to the plurality of state of charge parameters in accordance with the predetermined state of charge parameters. The characteristic charge capacity may include an initial charge capacity corresponding to 30% -70% of the state of charge parameter. The characteristic charge capacity is extracted from the initial charge capacity information obtained in the constant current charge process of the lithium ion battery, so that the characteristic charge capacity can be obtained.

According to the embodiment of the invention, the real-time battery health state prediction information can be processed according to the objective function to obtain the objective ohmic resistance parameter. The target ohmic resistance information of the equivalent circuit model of the lithium ion battery is determined by using the real-time battery state-of-health prediction information output by the battery state-of-health prediction model. Furthermore, calculation of target short-circuit current information and fault diagnosis of a direct-current side circuit of the lithium ion energy storage power station can be performed, and therefore basis can be provided for equipment type selection of a direct-current side fuse and the like of the lithium ion energy storage power station.

According to the embodiment of the invention, the target short-circuit current information can be calculated according to ohm law and the target ohm resistance information and the open-end voltage information.

Fig. 3 shows a block diagram of an equivalent circuit model of a lithium ion battery according to an embodiment of the present invention.

As shown in fig. 3, the equivalent circuit model may include an open-end voltage U _oc Target ohmic resistance R of lithium ion battery ₀ Polarization resistor R in lithium ion battery polarization process ₁ Polarization capacitor C in lithium ion battery polarization process ₁ Parallel terminal voltage U of loop ₁ And terminal voltage U of lithium ion battery _t . In fig. 3, i may represent a current in the equivalent circuit model, and an arrow below i may be used to represent a direction of the current.

The current of the lithium ion battery increases sharply after the occurrence of a short-circuit fault. Based on this, the short-circuit current equivalent can be regarded as a pulse function, which in turn can be regarded as consisting of innumerable high-frequency signals. Further, R is ₁ And C ₁ The amplitude response value of the formed parallel loop is large under the low frequency condition, and the amplitude response value is small under the high frequency condition. Thus, R is ₁ And C ₁ The parallel loop is negligible after short circuit. Based on this, the target short-circuit current information can be calculated by the following formula (1).

（1）

Wherein I is _sc Target short circuit current information may be represented. U (U) _oc Can represent open end voltage information, R ₀ Target ohmic resistance information may be represented.

According to the embodiment of the invention, according to the target short-circuit current information, the setting of the protection threshold value of the direct current side can be adjusted through the energy management system of the lithium ion energy storage power station, so that the situation that the protection refuses due to the reduction of the short-circuit current after the short-circuit fault of the direct current side after the aging of the lithium ion battery is prevented.

According to the embodiment of the invention, the protection threshold value of the direct-current side circuit associated with the lithium ion battery can be reduced when the current value of the target short-circuit current is reduced.

According to the embodiment of the invention, the characteristic charging capacity is extracted from the charging capacity information obtained in the lithium ion charging process. Based on the method, only the characteristic charge capacity is input into the trained battery health state prediction model, and the use of the full charge capacity information to obtain the real-time battery health state prediction information of the lithium ion battery is avoided. Therefore, the influence of inaccurate charge capacity information on the accuracy of the real-time battery health state prediction information is avoided, the information quantity processed by the battery health state prediction model is reduced, the efficiency of determining the real-time battery health state prediction information is further improved, and the accuracy of the real-time battery health state prediction information is improved.

According to the embodiment of the invention, the battery state of health prediction model comprises N prediction modules which are connected in parallel, wherein the output ends of the N prediction modules are connected with the linear regression module, and N is an integer greater than 1. Inputting the characteristic charge capacity into a trained battery state of health prediction model, outputting real-time battery state of health prediction information of the lithium ion battery, comprising: and respectively inputting the characteristic charge capacity into the input ends of the N prediction modules, and outputting N battery health state prediction information. Based on N weights corresponding to the N prediction modules, the linear regression module is utilized to process the N battery health state prediction information, and the real-time battery health state prediction information of the lithium ion battery is obtained.

According to an embodiment of the present invention, the N prediction modules may include a prediction module constructed based on a light gradient lifting algorithm, a prediction module constructed based on an extreme gradient lifting algorithm, and the like.

According to the embodiment of the invention, the characteristic charge capacity can be input into the input end of the prediction module constructed based on the light gradient lifting algorithm, and the battery health state prediction information corresponding to the prediction module constructed based on the light gradient lifting algorithm is output. And, the characteristic charge capacity may be input to an input end of a prediction module constructed based on the extreme gradient lifting algorithm, and the battery state of health prediction information corresponding to the prediction module constructed based on the extreme gradient lifting algorithm may be output.

According to an embodiment of the present invention, it is possible to process battery state of health prediction information corresponding to a prediction module constructed based on a light gradient lifting algorithm using a linear regression module according to a weight corresponding to a prediction module constructed based on a light gradient lifting algorithm, and to process battery state of health prediction information corresponding to a prediction module constructed based on an extreme gradient lifting algorithm according to a weight corresponding to a prediction module constructed based on an extreme gradient lifting algorithm. Thus, real-time battery state of health prediction information can be obtained.

According to the embodiment of the invention, the N weights corresponding to the N prediction modules are used, and the battery health state prediction information output by each of the N prediction modules is combined, so that the real-time battery health state prediction information of the lithium ion battery is obtained by combining the characteristics of each of the N prediction modules, the situation that only a single type of prediction module is used to obtain the real-time battery health state prediction information is avoided, and the accuracy of the real-time battery health state prediction information is improved.

According to the embodiment of the invention, the N different prediction modules are combined to output the real-time battery health state prediction information, so that the prediction of the battery health state of the lithium ion battery fused by the multi-data driving model is realized, and the accuracy of the real-time battery health state prediction information and the robustness of the output battery health state prediction information can be improved.

Fig. 4 shows a schematic diagram of output implementation battery state of health prediction information according to an embodiment of the present invention.

As shown in fig. 4, the feature charge capacity 410 may be input to the inputs of a prediction module 421 constructed based on a light gradient lifting algorithm, a prediction module 422 constructed based on an extreme gradient lifting algorithm, a prediction module 423 constructed based on a random forest algorithm, a prediction module 424 constructed based on a support vector regression algorithm, and a prediction module 425 constructed based on a gaussian process regression algorithm, respectively.

Thus, the battery state of health prediction information respectively output by the prediction module 421 constructed based on the light gradient lifting algorithm, the prediction module 422 constructed based on the extreme gradient lifting algorithm, the prediction module 423 constructed based on the random forest algorithm, the prediction module 424 constructed based on the support vector regression algorithm, and the prediction module 425 constructed based on the gaussian process regression algorithm can be obtained.

The linear regression module 430 may be utilized to process the above-mentioned battery state of health prediction information respectively output by the prediction module 421 constructed based on the light gradient lifting algorithm, the prediction module 422 constructed based on the extreme gradient lifting algorithm, the prediction module 423 constructed based on the random forest algorithm, the prediction module 424 constructed based on the support vector regression algorithm, and the prediction module 425 constructed based on the gaussian process regression algorithm, according to weights respectively corresponding to the battery state of health prediction information respectively output by the prediction module 421 constructed based on the light gradient lifting algorithm, the prediction module 422 constructed based on the extreme gradient lifting algorithm, the prediction module 423 constructed based on the random forest algorithm, the prediction module 424 constructed based on the support vector regression algorithm, and the prediction module 425 constructed based on the gaussian process regression algorithm. Thus, real-time battery state of health prediction information 440 may be obtained.

According to an embodiment of the present invention, the method for adjusting the threshold value of the dc side circuit of the lithium ion energy storage power station further includes: and acquiring a plurality of electrochemical impedance spectrums and M sample charging capacities of the lithium ion battery corresponding to the plurality of charge state parameters in the M-cycle charge and discharge processes, wherein M is an integer greater than 1. Ohmic resistance parameters of a plurality of equivalent circuit models corresponding to the plurality of state of charge parameters are determined according to the plurality of electrochemical impedance spectra. And obtaining the health state information of the M sample batteries of the lithium ion battery according to the charging capacity of the M samples and the standard capacity of the lithium ion battery. And fitting the M sample battery health state information with ohmic resistance parameters of a plurality of equivalent circuit models to obtain an objective function.

According to an embodiment of the present invention, the sample charge capacity may be a charge capacity of a lithium-ion battery actually measured during lithium ion charging in M cycles of charge and discharge. The sample battery State Of Health (SOH) information may be calculated by the following formula (2).

（2）

Wherein C is _n The sample charge capacity may be represented. C (C) ₀ Can represent the standard capacity of a lithium ion battery.

According to embodiments of the invention, electrochemical impedance spectroscopy may be used to determine ohmic resistance parameters of an equivalent circuit model corresponding to state of charge.

In the process of realizing the inventive concept, the inventor finds that accurate target short-circuit current information can be calculated only by ohmic resistance parameters of an equivalent circuit model of the lithium ion battery. The accurate ohmic resistance parameter can be obtained through electrochemical impedance spectroscopy, but the electrochemical impedance spectroscopy is difficult to obtain in the normal charge and discharge process of the lithium ion battery.

Based on this, the inventors found that since the electrochemical impedance spectrum is closely related to the battery state of health information.

Thus, in some embodiments, the state of charge parameters of the lithium ion battery during the cyclic charge and discharge may be measured by the electrochemical workstation to be 0%, 10%, 20%,..100% electrochemical impedance spectrum during the M cyclic charge and discharge.

According to the embodiment of the invention, the charging process of the lithium ion battery is relatively fixed, the charging process is mostly constant-current charging-constant-voltage charging process, and the discharging process is relatively complex. Based on the above, the working condition multiplying power of the lithium ion energy storage power station is 0.5C, so that the battery can be charged in a constant current charging-constant voltage charging mode of 0.5C. Wherein, the C value can represent the ratio of the discharge current of the lithium ion battery to the capacity of the lithium ion battery, and the common unit is a multiple.

According to an embodiment of the invention, the battery is discharged by adopting any discharging working condition of 0.5C, and the battery is cycled until the capacity of the lithium ion battery reaches below 70% of the nominal capacity in the charging process. Thus, the above battery state of health information can be measured.

According to the embodiment of the invention, in the cyclic charge and discharge process, the voltage and current of the lithium ion battery in the cyclic charge and discharge process can be collected through the high-precision voltage information collecting device and the current information collecting device at the sampling frequency of at least 10 Hz.

According to the embodiment of the invention, the sample charging capacity of the lithium ion battery in the charging process can be obtained by an ampere-hour integration method, and the ampere-hour integration method has the following expression:

（3）

wherein C is _n The sample charge capacity may be represented. i may represent the current of the lithium ion battery during charging. T_charge is the time of the charging process. τ may represent a time constant.

According to the embodiment of the invention, the ohmic resistance parameters of an accurate equivalent circuit model can be determined according to the electrochemical impedance spectrum obtained in the M-cycle charge and discharge processes of the lithium ion battery. And according to the sample charging capacity obtained in the M-cycle charging and discharging processes of the lithium ion battery, accurate M sample battery health state information can be determined by combining the standard capacity of the lithium ion battery. Therefore, the method can fit the state of health information and ohmic resistance parameters of the M sample batteries to obtain an accurate objective function.

According to an embodiment of the present invention, fitting the M sample battery state of health information with ohmic resistance parameters of a plurality of equivalent circuit models to obtain an objective function includes: and constructing a function to be fitted by taking the state of health information of the M sample batteries as independent variables and ohmic resistance parameters as dependent variables. And inputting the information of the health states of the M sample batteries and ohmic resistance parameters of a plurality of equivalent circuit models into a function to be fitted to obtain a target fitting coefficient. And obtaining an objective function according to the objective fitting coefficient and the function to be fitted.

According to the embodiment of the invention, the sample battery health state information may be battery health state information of a lithium-ion battery obtained in a lithium ion charging process in an M-cycle charging and discharging process.

According to an inventive embodiment, the function to be fitted may be a polynomial. The target fitting coefficients may include ohmic resistance parameters obtained by fitting. And according to the polynomial fitting, a fitting relation between the sample battery health state information of the lithium ion battery and the ohmic resistance parameter can be established. The polynomial fit may be as shown in equation (4):

（4）

wherein a is ₀ ，a ₁ ，a ₂ ，a ₃ ，a ₄ … is the ohmic resistance parameter of the polynomial fit. SOH may represent battery state of health information. R is R ₀ May be the target ohmic resistance information. Wherein the degree of polynomial fit may be at least 5.

Thus, the objective function can be obtained based on the above formula (4).

According to the embodiment of the invention, the accurate target fitting coefficient and further the accurate target function can be obtained by inputting the M sample battery health state information and the ohmic resistance parameters of the equivalent circuit models into the function to be fitted.

Fig. 5 shows a flowchart of a method of training a battery state of health prediction model according to an embodiment of the present invention.

As shown in fig. 5, the training method of the battery state of health prediction model of this embodiment includes operations S510 to S540.

In operation S510, a sample data set is acquired, wherein the sample data set includes battery state of health actual information of a sample lithium ion battery corresponding to sample characteristic charge capacity information.

In operation S520, sample characteristic charge capacity information of the sample lithium ion battery is input into the initial model, and battery state of health sample prediction information of the sample lithium ion battery is output.

In operation S530, a loss value is obtained from the battery state of health sample prediction information and the battery state of health actual information based on the loss function.

In operation S540, model parameters of the initial model are adjusted based on the loss value, resulting in a trained battery state of health prediction model.

According to an embodiment of the invention, the sample dataset may be used to train an initial model. The sample data set may include actual battery state of health information obtained by performing experimental measurements on a sample lithium ion battery. The battery state of health actual information may be actual measured battery state of health information.

According to an embodiment of the invention, the sample lithium ion battery may be a lithium ion battery for acquiring a sample dataset. The initial model may be constructed based on a light gradient lifting algorithm, but is not limited thereto, and may be constructed based on other algorithms. The trained battery state of health prediction model can be used for outputting battery state of health sample prediction information according to sample characteristic charge capacity information.

According to an embodiment of the present invention, the sample characteristic charge capacity information may be characteristic charge capacity information for training the initial model. For example, the charge capacity of the sample lithium ion battery may be corresponding to a state of charge parameter of 30% -70%.

The battery state of health sample prediction information may be battery state of health information output by the battery state of health prediction model. The accuracy of the battery state of health sample prediction information may correspond to the model effect of the initial model.

According to an embodiment of the invention, a loss function may be used to determine a loss value between actual battery state of health information and battery state of health sample prediction information. The loss functions may include root mean square error (Root Mean Square Error, RMSE) loss functions, mean absolute error (Mean Absolute Error, meana) loss functions, maximum absolute error (Maximum Absolute Error, maxAE) loss functions, and the like. The loss value may be used to determine actual battery state of health information and battery state of health sample prediction information to evaluate the training effect of the initial model.

According to the embodiment of the invention, the actual measured battery health state actual information of the lithium ion battery corresponding to the sample characteristic charging capacity information can be obtained by carrying out experiments on the sample lithium ion battery.

According to the embodiment of the invention, sample characteristic charging capacity information used for representing the battery health state of the lithium ion battery can be extracted from actually measured sample characteristic charging capacity information, and the sample characteristic charging capacity information can be charging capacity information of a constant current charging process with the charge state parameter of the sample lithium ion battery being 30% -70%. The sample characteristic charge capacity information can be calculated by the following formula (5):

（5）

Where F may represent sample characteristic charge capacity information. T_soc1 may represent a time when SOC is 30% during constant current charging. T_soc2 may represent the time at which SOC is 70% during constant current charging. After the battery ages, the charge capacity value mainly in the section of the SOC interval is greatly reduced. Thus, the sample characteristic charge capacity information and the battery state of health actual information of the sample lithium ion battery corresponding to the sample characteristic charge capacity information can be combined together to form a sample data set.

According to the embodiment of the invention, the sample characteristic charge capacity information of the sample lithium ion battery can be input into the initial model, and the battery health state sample prediction information corresponding to the sample characteristic charge capacity information can be output.

According to the embodiment of the invention, the loss value can be determined according to the actual information of the battery health state and the predicted information of the battery health state sample by using the root mean square error loss function, the average absolute error loss function, the maximum absolute error loss function and the like, so that the training effect of the battery health state prediction model can be evaluated according to the loss value.

Based on the model parameters of the initial model can be adjusted according to the loss value under the condition that the model effect of the initial model is determined to not meet the requirement, until the model effect of the initial model meets the requirement, and the trained battery health state prediction model is obtained. Wherein, the model effect of the initial model can be determined to meet the requirement under the condition that the optimization times of the initial model are equal to a preset threshold value.

According to an embodiment of the invention, the prediction accuracy of a multi-data driven model is assessed by analyzing the error cases of the health status estimation result and the true value, and the prediction effect of the model is assessed using root mean square error (Root Mean Square Error, RMSE), mean absolute error (Mean Absolute Error, meana) and maximum absolute error (Maximum Absolute Error, maxAE), the expressions of these three errors being:

（6）

（7）

（8）

where RMSE may represent root mean square error. Meaae may represent the mean absolute error. MaxAE can represent the maximum absolute error. M may represent the number of samples in the sample dataset. i may represent the order of the samples in the sample data set. y is _i Battery state of health sample prediction information may be represented.May represent actual battery state of health information.

According to the embodiment of the present invention, in the case where the loss value is equal to or greater than the predetermined loss threshold value, the value of the predetermined threshold value may be increased, thereby increasing the number of optimizations.

According to the embodiment of the invention, the model effect of the battery health state prediction model after the model parameters are adjusted can be verified until the trained battery health state prediction model is obtained.

According to the embodiment of the invention, by using the battery state of health sample prediction information corresponding to the sample characteristic charge capacity information, the use of the full amount of sample characteristic charge capacity information to obtain the battery state of health actual information and the battery state of health sample prediction information is avoided. Therefore, the influence of inaccurate sample characteristic charging capacity information on the accuracy of battery health state sample prediction information is avoided, the information quantity processed by an initial model is reduced, and the efficiency of obtaining a trained battery health state prediction model through training can be improved.

According to the embodiment of the invention, the initial model comprises N prediction modules which are connected in parallel, wherein the output ends of the N prediction modules are connected with the linear regression module, and N is an integer greater than 1. The training method of the battery state of health prediction model further comprises the following steps: and respectively carrying out parameter optimization on each initial prediction module to obtain the prediction module. Wherein the parameters correspond to algorithms running within each prediction module. The algorithm includes any one of the following: light gradient lifting algorithm, extreme gradient lifting algorithm, random forest algorithm, support vector regression algorithm and gaussian process regression algorithm.

According to the embodiment of the invention, the sample characteristic charge capacity information can be input into a trained prediction module constructed based on a light gradient lifting algorithm, a trained prediction module constructed based on an extreme gradient lifting algorithm, a trained prediction module constructed based on a random forest algorithm, a trained prediction module constructed based on a support vector regression algorithm and a trained prediction module constructed based on a Gaussian process regression algorithm, so that the battery health state sample prediction information corresponding to each prediction module is obtained. A new data set may be constructed from the battery state of health sample prediction information and the battery state of health actual information. The data set can be used for training a linear regression module, adaptively adjusting weights corresponding to battery health state prediction information respectively output by a prediction module constructed based on a light gradient lifting algorithm, a prediction module constructed based on an extreme gradient lifting algorithm, a prediction module constructed based on a random forest algorithm, a prediction module constructed based on a support vector regression algorithm and a prediction module constructed based on a Gaussian process regression algorithm, and fixing the weights. Based on this, a trained linear regression module can be obtained. The expression of the linear regression module may be as shown in equation (9):

（9）

Where X may represent battery state of health sample prediction information. Y may represent real-time battery state of health prediction information.Can represent random errors, is set to 0 as the mean and 0 as the variance + ->The gaussian distribution may refer to a probability distribution about the state of health of the battery in the battery state of health prediction information output by the prediction module. />The solution of (2) may be obtained using a least squares method, and may be as shown in equation (10):

（10）

where X may represent battery state of health sample prediction information. Y may represent real-time battery state of health prediction information.

According to the embodiment of the invention, the prediction modules are obtained by optimizing the N different initial prediction modules, so that the prediction of the battery health state of the lithium ion battery fused by the multi-data driving model can be realized, and the accuracy of the real-time battery health state prediction information and the robustness of the output battery health state prediction information can be further improved.

FIG. 6 shows a schematic diagram of a prediction module optimization method according to an embodiment of the invention.

As shown in FIG. 6, the prediction module optimizing method of the embodiment includes operations S610-S680.

In operation S610, the sample characteristic charge capacity information is input to an initial prediction module, and a first probability distribution of the battery state of health is obtained, wherein the initial prediction module is constructed based on initial parameters.

In operation S620, the first probability distribution and the initial parameters of the initial prediction module are processed to obtain a gaussian distribution of the first parameters and the battery state of health.

In operation S630, an intermediate parameter is determined from the first parameter and the gaussian distribution of the battery state of health.

In operation S640, the initial parameters are changed based on the intermediate parameters, resulting in an intermediate prediction module.

In operation S650, the sample characteristic charge capacity information is input to the intermediate prediction module to obtain a second probability distribution of the battery state of health.

In operation S660, the first probability distribution, the second probability distribution, the initial parameters, and the intermediate parameters are processed to obtain gaussian distributions of the second parameters and the battery state of health.

In operation S670, it is determined whether the number of optimizations is equal to a predetermined threshold? In the case where it is determined that the number of optimizations is less than the predetermined threshold, the operation of determining the intermediate parameter is performed back.

In operation S680, in the case where it is determined that the number of optimizations is equal to the predetermined threshold, a prediction module is obtained.

According to an embodiment of the invention, the initial prediction module may be a prediction module to be trained. The initial parameter may be a super parameter of the initial prediction module. The first probability distribution may be a prediction result output by the initial prediction module for characterizing a battery state of health of the lithium ion battery. The gaussian distribution of the first parameter and the battery state of health may be used to characterize the association between the initial parameter and the battery state of health.

According to an embodiment of the invention, the intermediate parameter may be an adjusted initial parameter. The intermediate prediction module may be an initial prediction module that has been adjusted for model parameters.

According to an embodiment of the present invention, the second probability distribution may be a prediction result output by the intermediate prediction module for characterizing a battery state of health of the lithium ion battery. The gaussian distribution of the second parameter and the battery state of health may be used to characterize the association between the intermediate parameter and the battery state of health.

According to an embodiment of the present invention, the number of optimizations may be the number of optimizations performed on the initial model.

According to the embodiment of the invention, the first probability distribution used for representing the battery health state and the initial parameters of the initial prediction module can be correlated to obtain the Gaussian distribution of the first parameters and the battery health state.

According to the embodiment of the invention, the average probability distribution of the first probability distribution and the second probability distribution can be calculated according to the first probability distribution and the second probability distribution. And calculating average parameters of the initial parameters and the intermediate parameters according to the initial model parameters and the intermediate model parameters. Thus, the average probability distribution and the average parameter used for representing the state of health of the battery can be correlated, and the Gaussian distribution of the second parameter and the state of health of the battery can be obtained.

According to the embodiment of the present invention, in the case where it is determined that the number of optimizations is smaller than the predetermined threshold value, the operation of determining the intermediate parameter is performed back.

According to an embodiment of the invention, the prediction module is obtained in case it is determined that the number of optimizations is equal to a predetermined threshold.

According to the embodiment of the invention, in the training process of the initial module, the model parameters of the initial module have a great influence on the model training effect, so that the model parameters of the initial module can be optimized. Wherein the model parameters may be super parameters.

For example, the superparameter of lightGBM (light Gradient Boosting Machine, light gradient lifting algorithm) has the maximum number of leaf nodes of the tree, the number of iterations, the minimum number of data in the leaf nodes, etc.; super parameters of XGBoost (eXtreme Gradient Boosting, extreme gradient lifting algorithm) are learning rate, maximum depth of tree, minimum loss required for further partitioning of leaf nodes of tree, etc.; the super parameters of RF (Random Forest algorithm) include the number of trees, the maximum depth of the trees, the minimum number of samples required for splitting the internal nodes, etc.; the super parameters of SVR (Support Vector Regression, support vector regression algorithm) have penalty coefficient, insensitivity random time coefficient and width coefficient; the hyper-parameters of GPR (Gaussian Process Regression, gaussian process regression algorithm) are kernel function diagonal values, etc.

According to an embodiment of the invention, a gaussian process may be used to fit the relationship between the hyper-parameters and the probability distribution output by the initial module, i.e. the mean function m (x) and the covariance function k (x, x) ^T ) The relationship is determined.

The intermediate parameter may be determined from the gaussian distribution of the first parameter and the battery state of health by the following formula (11)：

（11）

Wherein,can represent intermediate parameters, θ is a superparameter, < ->To output a value for the model corresponding to the value of theta,jfor the current optimization times, θ _j And theta ⁺ The value of the super-parameter at the j-th time and the value near the neighborhood,qd is a set of hyper-parameters and model output values for the set total optimization times. />Representing a left continuous step function. Can be at the current theta ⁺ Is found in the neighborhood of f (θ ⁺ ) Large candidate points, taking these θ _1:q The point of greatest probability, i.e. the intermediate parameter θ _i+1 . The obtained intermediate parameter theta _i+1 Substituting the initial parameters into the initial model to update the initial parameters of the initial model to obtain an intermediate model. And outputting a second probability distribution corresponding to the sample characteristic charging capacity information by the intermediate model. Thus, the first probability distribution, the second probability distribution, the initial parameter θ and the intermediate parameter θ can be calculated _i+1 Processing to obtain more accurate Gaussian distribution D of the second parameter and the battery health state _1:i+1 =D _1:i ∪(θ _i+1 ,c _i+1 ). Based on this, the model parameters are optimized by repeating until the number of optimizations is equal to a predetermined threshold.

According to the embodiment of the invention, the intermediate parameter is determined from the first parameter and the Gaussian distribution of the battery health state, the intermediate prediction module is obtained by changing the initial parameter according to the intermediate parameter, and the second probability distribution of the battery health state is obtained by the intermediate prediction module according to the sample characteristic charge capacity information.

Based on this, a gaussian distribution of the second parameter and the battery state of health is obtained by applying the first probability distribution, the second probability distribution, the initial parameter, and the intermediate parameter. Thus, the determination of intermediate parameters from the gaussian distribution may be repeated to optimize the intermediate module. The efficiency of determining the intermediate parameters is improved, the model effect can be improved, and the accuracy of the battery health state sample prediction information can be improved.

Based on the threshold adjustment method of the direct current side circuit of the lithium ion energy storage power station, the invention further provides a threshold adjustment device of the direct current side circuit of the lithium ion energy storage power station. The device will be described in detail below in connection with fig. 7.

Fig. 7 shows a block diagram of a threshold adjustment device for a dc side circuit of a lithium ion energy storage power station according to an embodiment of the invention.

As shown in fig. 7, the threshold adjustment device 700 of the dc side circuit of the lithium ion energy storage power station of this embodiment includes an extraction module 710, a first input module 720, a first obtaining module 730, a second obtaining module 740, and a first adjustment module 750.

The extracting module 710 is configured to extract, from the initial charge capacity information, a characteristic charge capacity corresponding to a predetermined state of charge parameter in response to the received initial charge capacity information corresponding to a plurality of state of charge parameters of the lithium ion battery during constant current charging and start voltage information of the lithium ion battery in a non-operating state. In an embodiment, the extracting module 810 may be configured to perform the operation S210 described above, which is not described herein.

The first input module 720 is configured to input the characteristic charge capacity into the trained battery state of health prediction model, and output real-time battery state of health prediction information of the lithium ion battery. In an embodiment, the prediction module 720 may be configured to perform the operation S220 described above, which is not described herein.

The first obtaining module 730 is configured to obtain target ohmic resistance information corresponding to real-time battery health status prediction information according to real-time battery health status prediction information based on an objective function, where the objective function characterizes a correlation between the battery health status information and ohmic resistance parameters of an equivalent circuit model, and the equivalent circuit model is used for simulating a working state of the lithium ion battery under any working condition. In an embodiment, the first obtaining module 730 may be configured to perform the operation S230 described above, which is not described herein.

The second obtaining module 740 is configured to obtain the target short-circuit current information according to the target ohmic resistance information and the open-end voltage information. In an embodiment, the second obtaining module 740 may be configured to perform the operation S240 described above, which is not described herein.

The first adjustment module 750 is configured to adjust a protection threshold of a dc side circuit associated with the lithium ion battery based on the target short circuit current information. In an embodiment, the first adjustment module 750 may be used to perform the operation S250 described above, which is not described herein.

According to an embodiment of the invention, the first input module 720 comprises a first input sub-module and a first processing sub-module. The first input sub-module is used for inputting the characteristic charge capacity into the input ends of the N prediction modules respectively and outputting N battery health state prediction information; the first processing sub-module is used for processing the N battery health state prediction information by utilizing the linear regression module based on N weights corresponding to the N prediction modules to obtain the real-time battery health state prediction information of the lithium ion battery.

According to an embodiment of the present invention, the threshold adjustment device for a direct current side circuit of a lithium ion energy storage power station further includes a fifth obtaining module, a determining module, a sixth obtaining module and a fitting module. The fifth obtaining module is used for obtaining a plurality of electrochemical impedance spectrums and M sample charging capacities of the lithium ion battery corresponding to a plurality of charge state parameters in the M-cycle charging and discharging processes, wherein M is an integer greater than 1; the determining module is used for determining ohmic resistance parameters of a plurality of equivalent circuit models corresponding to the plurality of charge state parameters according to the plurality of electrochemical impedance spectrums; the sixth obtaining module is used for obtaining the health state information of the M sample batteries of the lithium ion battery according to the charging capacity of the M samples and the standard capacity of the lithium ion battery; the fitting module is used for fitting the state of health information of the M sample batteries and ohmic resistance parameters of the equivalent circuit models to obtain an objective function.

According to an embodiment of the invention, the fitting module comprises a construction sub-module, a second input sub-module and a first obtaining sub-module. The construction submodule is used for constructing a function to be fitted by taking the state of health information of M sample batteries as independent variables and ohmic resistance parameters as dependent variables; the second input submodule is used for inputting the health state information of the M sample batteries and ohmic resistance parameters of the equivalent circuit models into a function to be fitted to obtain a target fitting coefficient; the first obtaining submodule is used for obtaining an objective function according to the objective fitting coefficient and the function to be fitted.

Any of the extraction module 710, the first input module 720, the first obtaining module 730, the second obtaining module 740, and the first adjusting module 750 may be combined in one module to be implemented, or any of the modules may be split into a plurality of modules according to an embodiment of the present invention. Alternatively, at least some of the functionality of one or more of the modules may be combined with at least some of the functionality of other modules and implemented in one module. According to an embodiment of the invention, at least one of the extraction module 710, the first input module 720, the first obtaining module 730, the second obtaining module 740, and the first adapting module 750 may be implemented at least partially as hardware circuitry, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or as hardware or firmware in any other reasonable way of integrating or packaging the circuitry, or as any one of or a suitable combination of three of software, hardware, and firmware. Alternatively, at least one of the extraction module 710, the first input module 720, the first obtaining module 730, the second obtaining module 740, and the first adjusting module 750 may be at least partially implemented as a computer program module, which when executed, may perform the corresponding functions.

Based on the training method of the battery state of health prediction model, the invention also provides a training device of the battery state of health prediction model. The device will be described in detail below in connection with fig. 8.

Fig. 8 shows a block diagram of a training apparatus of a battery state of health prediction model according to an embodiment of the present invention.

As shown in fig. 8, the training apparatus 800 of the battery state of health prediction model of this embodiment includes a third obtaining module 810, a second input module 820, a fourth obtaining module 830, and a second adjusting module 840.

The third obtaining module 810 is configured to obtain a sample data set, where the sample data set includes actual battery health information of a sample lithium ion battery corresponding to sample characteristic charging capacity information. In an embodiment, the third obtaining module 810 may be configured to perform the operation S510 described above, which is not described herein.

The second input module 820 is configured to input sample characteristic charge capacity information of the sample lithium ion battery into the initial model, and output battery health status sample prediction information of the sample lithium ion battery. In an embodiment, the input module 820 may be used to perform the operation S520 described above, which is not described herein.

The fourth obtaining module 830 is configured to obtain a loss value based on the loss function according to the battery state of health sample prediction information and the battery state of health actual information. In an embodiment, the fourth obtaining module 830 may be configured to perform the operation S530 described above, which is not described herein.

The second adjustment module 840 is configured to adjust model parameters of the initial model based on the loss value, to obtain a trained battery state of health prediction model. In an embodiment, the second adjustment module 840 may be used to perform the operation S540 described above, which is not described herein.

According to an embodiment of the present invention, the training device of the battery state of health prediction model further includes an optimization module. The optimization module is used for respectively carrying out parameter optimization on each initial prediction module to obtain prediction modules; wherein the parameters correspond to algorithms running within each prediction module; the algorithm includes any one of the following: light gradient lifting algorithm, extreme gradient lifting algorithm, random forest algorithm, support vector regression algorithm and gaussian process regression algorithm.

According to an embodiment of the invention, the optimization module comprises a second processing sub-module, a second obtaining sub-module, a determining sub-module, a third obtaining sub-module, a third input sub-module, a third processing sub-module, an executing sub-module and a fourth obtaining sub-module. The second processing sub-module is used for inputting sample characteristic charging capacity information into the initial prediction module to obtain first probability distribution of the battery health state, wherein the initial prediction module is constructed based on initial parameters; the second obtaining submodule is used for processing the first probability distribution and the initial parameters of the initial prediction module to obtain Gaussian distribution of the first parameters and the battery health state; the determining submodule is used for determining an intermediate parameter from the first parameter and the Gaussian distribution of the state of health of the battery; the third obtaining submodule is used for changing initial parameters based on intermediate parameters to obtain an intermediate prediction module; the third input sub-module is used for inputting the sample characteristic charging capacity information into the middle prediction module to obtain a second probability distribution of the battery health state; the third processing sub-module is used for processing the first probability distribution, the second probability distribution, the initial parameters and the intermediate parameters to obtain Gaussian distribution of the second parameters and the battery health state; the determining submodule is used for returning to execute the operation of determining the intermediate parameter under the condition that the determined optimization times are smaller than a preset threshold value; the third obtaining sub-module is used for obtaining the prediction module under the condition that the optimization times are equal to a preset threshold value.

Any of the third obtaining module 810, the second input module 820, the fourth obtaining module 830, and the second adjusting module 840 may be combined in one module to be implemented, or any of the modules may be split into a plurality of modules according to an embodiment of the present invention. Alternatively, at least some of the functionality of one or more of the modules may be combined with at least some of the functionality of other modules and implemented in one module. According to an embodiment of the present invention, at least one of the third obtaining module 810, the second input module 820, the fourth obtaining module 830 and the second adjusting module 840 may be implemented at least partially as a hardware circuit, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or as hardware or firmware by any other reasonable way of integrating or packaging the circuits, or as any one of or a suitable combination of any of the three. Alternatively, at least one of the third obtaining module 810, the second input module 820, the fourth obtaining module 830, and the second adjusting module 840 may be at least partially implemented as a computer program module, which when executed, may perform the corresponding functions.

As shown in fig. 9, an electronic device 900 according to an embodiment of the present invention includes a processor 901 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 902 or a program loaded from a storage section 908 into a Random Access Memory (RAM) 903. The processor 901 may include, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or an associated chipset and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), or the like. Processor 901 may also include on-board memory for caching purposes. Processor 901 may include a single processing unit or multiple processing units for performing the different actions of the method flows according to embodiments of the invention.

In the RAM 903, various programs and data necessary for the operation of the electronic device 900 are stored. The processor 901, the ROM 902, and the RAM 903 are connected to each other by a bus 904. The processor 901 performs various operations of the method flow according to an embodiment of the present invention by executing programs in the ROM 902 and/or the RAM 903. Note that the program may be stored in one or more memories other than the ROM 902 and the RAM 903. The processor 901 may also perform various operations of the method flow according to embodiments of the present invention by executing programs stored in the one or more memories.

According to an embodiment of the invention, the electronic device 900 may also include an input/output (I/O) interface 905, the input/output (I/O) interface 905 also being connected to the bus 904. The electronic device 900 may also include one or more of the following components connected to an input/output (I/O) interface 905: an input section 906 including a keyboard, a mouse, and the like; an output portion 907 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and a speaker; a storage portion 908 including a hard disk or the like; and a communication section 909 including a network interface card such as a LAN card, a modem, or the like. The communication section 909 performs communication processing via a network such as the internet. The drive 910 is also connected to an input/output (I/O) interface 905 as needed. A removable medium 911 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is installed as needed on the drive 910 so that a computer program read out therefrom is installed into the storage section 908 as needed.

The present invention also provides a computer-readable storage medium that may be embodied in the apparatus/device/system described in the above embodiments; or may exist alone without being assembled into the apparatus/device/system. The computer-readable storage medium carries one or more programs which, when executed, implement methods in accordance with embodiments of the present invention.

According to embodiments of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but is not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to embodiments of the invention, the computer-readable storage medium may include ROM 902 and/or RAM 903 and/or one or more memories other than ROM 902 and RAM 903 described above.

Embodiments of the present invention also include a computer program product comprising a computer program containing program code for performing the method shown in the flowcharts. When the computer program product runs in a computer system, the program code is used for enabling the computer system to realize the threshold adjustment method of the direct-current side circuit of the lithium ion energy storage power station or the training method of the battery health state prediction model.

The above-described functions defined in the system/apparatus of the embodiment of the present invention are performed when the computer program is executed by the processor 901. The systems, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the invention.

In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, or the like. In another embodiment, the computer program may also be transmitted, distributed, and downloaded and installed in the form of a signal on a network medium, via communication portion 909, and/or installed from removable medium 911. The computer program may include program code that may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

In such an embodiment, the computer program may be downloaded and installed from the network via the communication portion 909 and/or installed from the removable medium 911. The above-described functions defined in the system of the embodiment of the present invention are performed when the computer program is executed by the processor 901. The systems, devices, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the invention.

According to embodiments of the present invention, program code for carrying out computer programs provided by embodiments of the present invention may be written in any combination of one or more programming languages, and in particular, such computer programs may be implemented in high-level procedural and/or object-oriented programming languages, and/or in assembly/machine languages. Programming languages include, but are not limited to, such as Java, c++, python, "C" or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of remote computing devices, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., connected via the Internet using an Internet service provider).

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

It should be noted that, unless there is an execution sequence between different operations or an execution sequence between different operations in technical implementation, the execution sequence between multiple operations may be different, and multiple operations may also be executed simultaneously.

Those skilled in the art will appreciate that the features recited in the various embodiments of the invention and/or in the claims may be combined in various combinations and/or combinations, even if such combinations or combinations are not explicitly recited in the invention. In particular, the features recited in the various embodiments of the invention and/or in the claims can be combined in various combinations and/or combinations without departing from the spirit and teachings of the invention. All such combinations and/or combinations fall within the scope of the invention.

The embodiments of the present invention are described above. However, these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Although the embodiments are described above separately, this does not mean that the measures in the embodiments cannot be used advantageously in combination. The scope of the invention is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be made by those skilled in the art without departing from the scope of the invention, and such alternatives and modifications are intended to fall within the scope of the invention.

Claims

1. The method for adjusting the threshold value of the direct current side circuit of the lithium ion energy storage power station is characterized by comprising the following steps of:

responding to received initial charge capacity information corresponding to a plurality of charge state parameters of a lithium ion battery in a constant current charging process and starting voltage information of the lithium ion battery in a non-working state, and extracting characteristic charge capacity corresponding to a preset charge state parameter from the initial charge capacity information;

inputting the characteristic charge capacity into a trained battery health state prediction model, and outputting real-time battery health state prediction information of the lithium ion battery;

obtaining target ohmic resistance information corresponding to the real-time battery health state prediction information according to the real-time battery health state prediction information based on an objective function, wherein the objective function represents the association relationship between the battery health state information and ohmic resistance parameters of an equivalent circuit model, and the equivalent circuit model is used for simulating the working state of the lithium ion battery under any working condition;

obtaining target short-circuit current information according to the target ohmic resistance information and the open-end voltage information; and

and adjusting a protection threshold of a direct current side circuit associated with the lithium ion battery based on the target short circuit current information.

2. The method of claim 1, wherein the battery state of health prediction model comprises N prediction modules connected in parallel, the outputs of the N prediction modules are all connected with a linear regression module, and N is an integer greater than 1;

inputting the characteristic charge capacity into a trained battery health state prediction model, and outputting real-time battery health state prediction information of the lithium ion battery, wherein the method comprises the following steps of:

the characteristic charge capacity is respectively input into the input ends of the N prediction modules, and N battery health state prediction information is output;

and processing the N battery health state prediction information by using the linear regression module based on N weights corresponding to the N prediction modules to obtain the real-time battery health state prediction information of the lithium ion battery.

3. The method of claim 2, wherein the N prediction modules comprise at least two of: the prediction module is constructed based on a light gradient lifting algorithm, an extreme gradient lifting algorithm, a random forest algorithm, a support vector regression algorithm and a Gaussian process regression algorithm.

4. The method as recited in claim 1, further comprising:

acquiring a plurality of electrochemical impedance spectrums and M sample charging capacities of the lithium ion battery corresponding to a plurality of charge state parameters in M times of cyclic charge and discharge processes, wherein M is an integer greater than 1;

determining ohmic resistance parameters of a plurality of equivalent circuit models corresponding to the plurality of state-of-charge parameters according to the plurality of electrochemical impedance spectrums;

obtaining M sample battery health state information of the lithium ion battery according to the M sample charging capacities and the standard capacity of the lithium ion battery;

fitting the M sample battery health state information and ohmic resistance parameters of the equivalent circuit models to obtain the objective function.

5. The method of claim 4, wherein the fitting the M sample battery state of health information to ohmic resistance parameters of the plurality of equivalent circuit models results in the objective function, comprising:

constructing a function to be fitted by taking the state of health information of M sample batteries as independent variables and ohmic resistance parameters as dependent variables;

inputting the state of health information of the M sample batteries and ohmic resistance parameters of the equivalent circuit models into the function to be fitted to obtain a target fitting coefficient;

And obtaining the objective function according to the objective fitting coefficient and the function to be fitted.

6. A method for training a battery state of health prediction model, comprising:

acquiring a sample data set, wherein the sample data set comprises actual battery health state information of a sample lithium ion battery corresponding to sample characteristic charging capacity information;

inputting sample characteristic charging capacity information of a sample lithium ion battery into an initial model, and outputting battery health state sample prediction information of the sample lithium ion battery;

obtaining a loss value based on the loss function according to the battery health state sample prediction information and the battery health state actual information;

based on the loss value, adjusting model parameters of the initial model to obtain a trained battery state of health prediction model, wherein the trained battery state of health prediction model is applied to the threshold adjustment method of the direct current side circuit of the lithium ion energy storage power station of any one of claims 1-5.

7. The method of claim 6, wherein the initial model comprises N prediction modules connected in parallel, the outputs of the N prediction modules are all connected with a linear regression module, and N is an integer greater than 1; further comprises:

Respectively carrying out parameter optimization on each initial prediction module to obtain the prediction modules; wherein the parameters correspond to algorithms running within each of the prediction modules; the algorithm includes any one of the following: light gradient lifting algorithm, extreme gradient lifting algorithm, random forest algorithm, support vector regression algorithm and gaussian process regression algorithm.

8. The method of claim 7, wherein the performing parameter optimization on each initial prediction module to obtain the prediction module includes:

inputting the sample characteristic charging capacity information into the initial prediction module to obtain a first probability distribution of the battery health state, wherein the initial prediction module is constructed based on initial parameters;

processing the first probability distribution and the initial parameters of the initial prediction module to obtain Gaussian distribution of the first parameters and the battery health state;

determining an intermediate parameter from the first parameter and a gaussian distribution of battery state of health;

changing the initial parameters based on the intermediate parameters to obtain an intermediate prediction module;

inputting the sample characteristic charging capacity information into the middle prediction module to obtain a second probability distribution of the battery health state;

Processing the first probability distribution, the second probability distribution, the initial parameters and the intermediate parameters to obtain Gaussian distribution of the second parameters and the battery health state;

returning to execute the operation of determining the intermediate parameter in the case that the number of times of optimization is determined to be smaller than the predetermined threshold value;

and under the condition that the optimization times are equal to the preset threshold value, obtaining the prediction module.

9. A threshold adjustment device for a direct current side circuit of a lithium ion energy storage power station, comprising:

the extraction module is used for responding to the received initial charge capacity information of the lithium ion battery corresponding to a plurality of charge state parameters in the constant current charging process and the initial voltage information of the lithium ion battery in a non-working state, and extracting the characteristic charge capacity corresponding to the preset charge state parameters from the initial charge capacity information;

the first input module is used for inputting the characteristic charge capacity into a trained battery health state prediction model and outputting real-time battery health state prediction information of the lithium ion battery;

the first obtaining module is used for obtaining target ohmic resistance information corresponding to the real-time battery health state prediction information based on an objective function according to the real-time battery health state prediction information, wherein the objective function represents the association relationship between the battery health state information and ohmic resistance parameters of an equivalent circuit model, and the equivalent circuit model is used for simulating the working state of the lithium ion battery under any working condition;

The second obtaining module is used for obtaining target short-circuit current information according to the target ohmic resistance information and the open-end voltage information; and

and the first adjusting module is used for adjusting the protection threshold value of the direct current side circuit associated with the lithium ion battery based on the target short-circuit current information.

10. A training device for a battery state of health prediction model, comprising:

the third obtaining module is used for obtaining a sample data set, wherein the sample data set comprises actual battery health state information of a sample lithium ion battery corresponding to sample characteristic charging capacity information;

the second input module is used for inputting sample characteristic charging capacity information of the sample lithium ion battery into the initial model and outputting battery health state sample prediction information of the sample lithium ion battery;

the fourth obtaining module is used for obtaining a loss value based on a loss function according to the battery health state sample prediction information and the battery health state actual information;

and the second adjusting module is used for adjusting the model parameters of the initial model based on the loss value to obtain a trained battery health state prediction model, wherein the trained battery health state prediction model is applied to the threshold value adjusting method of the direct current side circuit of the lithium ion energy storage power station according to any one of claims 1-5.

11. An electronic device, comprising:

one or more processors;

storage means for storing one or more programs,

wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method of any of claims 1-8.

12. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to perform the method according to any of claims 1-8.