BMS management system and management method of flow battery

Abstract

The invention discloses a BMS management system and management method of a flow battery. A BMS management system of a flow battery comprises an external information collection module, an internal information collection module, a battery information collection module, an information preprocessing module, a historical SOH information collection unit and an SOH prediction device, and a current SOH value of the battery is generated based on a feature set; the SOH prediction device includes: a prediction unit that predicts an SOH value of a generated battery based on an input feature value; and the correction unit is used for generating influence information of each element in the feature set and the SOH value based on the historical SOH value, and sending the influence information to the prediction unit so as to guide the generation of the SOH value of the battery. According to the technical scheme provided by the invention, the convergence rate is higher during training, and the prediction unit can be continuously corrected according to the actual conditions of different flow batteries during use.

Description

Technical Field

[0001] The present application relates to the field of batteries, and in particular to a BMS management system and management method for liquid flow batteries. Background Art

[0002] Liquid flow batteries use two independent large tanks to store positive and negative electrolyte solutions. When the battery starts to work, the circulation pump drives the electrolyte to flow to the battery stack, and redox reactions occur on the electrode surface. Because of the presence of ion exchange membranes, the total amount of ions in the positive and negative electrolytes remains unchanged, so in principle the energy storage capacity of the battery will not change. However, during the use of the battery, it is affected by various factors, and side reactions of varying degrees will occur in the electrolyte, causing changes in the composition of the electrolyte, so the battery will age as the battery cycles.

[0003] In the battery BMS management system, the key part is the prediction of the battery's SOH value, which is actually the prediction of the battery's aging degree. The BMS management system will calculate the current battery's charge storage and release capabilities based on the battery's SOH value to perform charge and discharge management.

[0004] In the BMS management of flow batteries, the prediction of SOH value is very difficult because it is affected by factors such as electrode corrosion, material aging, ion conduction membrane aging, etc. At present, the prediction of SOH value of flow batteries generally adopts the following two methods.

[0005] The first method is to predict the SOH value based on the electrochemical model and the equivalent circuit model. This method analyzes the SOH value based on the internal electrochemical principle and circuit principle of the flow battery. Although it has high accuracy, in practice, a large amount of data needs to be collected, and the model is complex and the calculation time is long.

[0006] The second method is based on a data-driven approach, which collects some state information directly related to the SOH value and analyzes the SOH value of the battery based on the intrinsic relationship between the state information and the SOH value. The data model constructed in this way requires a large number of data samples for training. Because the factors affecting the SOH value are complex, the training time of the data model is long and the model convergence speed is slow.

[0007] In summary, the current BMS management system lacks a method that can efficiently predict the SOH value. Summary of the invention

[0008] The content of this application is used to introduce concepts in a brief form, which will be described in detail in the detailed implementation section below. The content of this application is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

[0009] As a first aspect of the present application, in order to solve the technical problems mentioned in the above background technology section, some embodiments of the present application provide a BMS management system for a flow battery, including:

[0010] The external information collection module is used to collect the ambient temperature, ambient humidity, charge and discharge rate, and vibration data of the battery when it is working, so as to generate external information;

[0011] Internal information collection module, used to obtain battery aging information;

[0012] A battery information collection module is used to collect the coulombic efficiency, voltage efficiency, energy efficiency and battery internal resistance of each battery cycle to generate battery information;

[0013] An information preprocessing module receives external information, aging information and battery information, and preprocesses the external information, aging information and battery information to generate a feature set;

[0014] A historical SOH information collection unit is used to collect the historical SOH value of the battery at each cycle;

[0015] An SOH prediction device generates a current SOH value of the battery based on a feature set;

[0016] The SOH prediction device includes:

[0017] A prediction unit, which predicts and generates a battery SOH value based on the input feature value;

[0018] The correction unit generates the influence information of each element in the feature set and the SOH value based on the historical SOH value, and sends the influence information to the prediction unit to guide the generation of the SOH value of the battery.

[0019] In the technical solution provided by the present application, when predicting the SOH value of the liquid flow battery, it is based on various types of information related to SOH, from which the intrinsic connection between each characteristic information and the SOH value is found to accurately predict the SOH value. In order to solve the problem of low convergence rate of the prediction unit, a correction unit is also introduced. The correction unit will calculate the influence information between each element and the SOH value, and use the influence information to guide the changes of the liquid flow battery. Therefore, in practice, the liquid flow battery will continuously correct the prediction unit during the cycle process, so that the prediction unit will converge faster during training, and when in use, it can also continuously correct the prediction unit according to the actual situation of different liquid flow batteries.

[0020] Compared with lithium batteries, flow batteries have a large storage energy per cell. During the battery aging process, the characteristic data, such as battery internal resistance, current, voltage and other information, change in a small range. It is difficult to capture the battery aging information using this information. To address this problem, this application provides the following technical solutions:

[0021] Furthermore, the coulombic efficiency is the ratio of the actual discharge amount of the battery to the actual charge amount of the battery during one charge and discharge cycle;

[0022] Voltage efficiency is the ratio of the average discharge voltage to the average charge voltage in one charge and discharge cycle;

[0023] Energy efficiency is the ratio of the energy output during the battery discharge process to the energy consumed during the battery charging process during a charge and discharge cycle.

[0024] In the technical solution provided in this application, coulombic efficiency, voltage efficiency and energy efficiency are used as important bases for predicting the SOH value of liquid flow batteries. The reason is that compared with lithium batteries, the chemical reactions occurring in liquid flow batteries will be different during the charging process, discharging process and under different temperatures. Therefore, the ratio of voltage to current of the liquid flow battery during charging and discharging will show corresponding characteristics due to battery aging. Capturing these characteristics can find the changing pattern of the battery SOH value.

[0025] The collected external information, aging information, and battery information are relatively inconsistent in their collection methods, and different information needs to consider different factors when used to identify battery aging, so the collected original information is directly used as the SOH value.

[0026] Furthermore, the information preprocessing module includes:

[0027] Information collection unit: collects external information, aging information, and battery information of the battery;

[0028] Information extraction unit: processes external information, aging information, and battery information into a sequence related to a complete charge and discharge cycle i ; a i represents the sequence at the i-th charge and discharge cycle;

[0029] a i = {b 1i 、b 2i ,…b 9i}, b 1i 、b 2i ,…b 9i Respectively represent the characteristic values ​​of each element in the external information, aging information, and battery information at the i-th charge and discharge cycle;

[0030] Collection units, collect all a i To obtain feature set A;

[0031] A={a1, a2, ...a i …a n}, n represents the number of sequences.

[0032] In the technical solution provided in the present application, external information, aging information and battery information are collected and organized according to the charge and discharge cycle, so in the final feature set A, the sequence sorting method can better reflect the battery aging process.

[0033] The aging process of flow batteries is affected by internal electrochemical reactions, which are difficult to predict. They are affected by the concentration of ions in the electrolyte, trace elements precipitated from the electrodes, and the reaction temperature. In practice, the battery works at two different temperatures, and the substances precipitated from the electrolyte are basically different. Therefore, the aging of each battery is affected by different factors. If the same method is used to predict the aging of the battery, the model accuracy will be affected. To this end, this application provides the following technical solutions:

[0034] The correction unit calculates the influence information U in the following manner;

[0035] S1: Obtain feature set A and the historical SOH value Y corresponding to feature set A;

[0036] Y={y1, y2, ...y i ,…y n},y i Represents the SOH value of the battery at the i-th charge and discharge cycle;

[0037] S2: Convert the feature set A into the feature matrix AP and convert the historical SOH value Y into the numerical matrix YP;

[0038]

[0039]

[0040] S3: Convert the feature matrix AP into the augmented matrix ZA, where the feature matrix AP has a size of n*9 and the augmented matrix ZA has a size of n*2 9 ;

[0041] In the augmented matrix AP;

[0042]

[0043] Among them, j represents the index of the column in the feature matrix AP, k represents the index of the column in the augmented matrix ZA,

[0044] M[j]=min(X j ), X j ={x 1j 、x 2j 、x 2j ,…x 9j}, M[j] is X j The minimum value among them, k∈(1,2 9 -1), l∈(1,2 9 -1), j∈(1,9), b j represents the binary value of the bth combination; k j is the result of normalizing attribute j;

[0045] S4: Use the augmented matrix and the actual SOH value to build a linear regression model to solve the interaction measure set U;

[0046] Y=ZT*U+δ 2 ;

[0047] Among them, Y is the historical SOH value, U is the interaction measure set, δ 2 is a random noise term;

[0048] S5: Denormalize the obtained interaction measure to the original scale to obtain the interaction measure w of the relevant elements on the original scale;

[0049]

[0050] t l is the lth intermediate variable, u l is the lth interaction measure in the interaction measure set U;

[0051]

[0052] The interaction measures reflect the interaction between different health characteristic indices on the SOH estimation results. At the original scale, these interaction measures are easier to understand and interpret because they are directly related to the actual battery health characteristic values, so they can reflect the complex interactions in the battery aging process, thereby increasing the effect on SOH estimation. And denormalizing the interaction measures back to the original scale can ensure that the same scale of data is used in model training and actual application, thereby avoiding errors and confusion caused by inconsistent data scales.

[0053] Furthermore, median alignment normalization is performed on each element in the feature matrix AP.

[0054] Through median alignment normalization, all health characteristic indices are converted to the same scale level, which makes the comparison and combination of different characteristics more fair and accurate when building estimation models. And because the interaction measures are denormalized back to the original scale in this scheme, some characteristics of the original data can be retained, thereby improving the accuracy and applicability of the model.

[0055] Furthermore, U = (ZAPZA) -1 Z T Y, T represents the transpose operation of the matrix.

[0056] When building a prediction model for SOH value based on big data, an overly complex model structure can take into account more types of information, but it will lead to long calculation time and high training difficulty of the model. To this end, this application provides the following technical solutions:

[0057] The prediction unit includes:

[0058] Generator, with a built-in multi-directional convolutional network, the feature values ​​are input into the generator to generate the SOH value;

[0059] The corrector,receives the SOH value generated by the generator, and judges the accuracy of the SOH value based on the interaction measure on the original scale, and sends the accuracy to the generator.

[0060] In the technical solution provided in this application, the accuracy of the SOH value is judged based on the interaction measurement on the original scale. Therefore, in practice, the generator can be trained and improved during the use of the liquid battery, so that the learning direction of the generator gradually approaches the aging condition of the liquid battery.

[0061] Furthermore, the aging information is related to the electrode material and the operating time of the battery.

[0062] As a second aspect of the present application, the present application provides a BMS management method for a liquid flow battery, wherein the BMS management method uses a BMS management system of the liquid flow battery to predict the SOH value of the battery.

[0063] In summary: This technical solution analyzes various types of information related to the SOH (health state) of the flow battery, uses the prediction unit to predict the SOH value, and introduces the correction unit to calculate the impact of each element on the SOH value to guide the model correction and improve the convergence rate. In practice, the model can be continuously corrected according to the actual situation of the flow battery to ensure the accuracy of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The drawings constituting a part of this application are used to provide a further understanding of this application, so that other features, purposes and advantages of this application become more obvious. The illustrative embodiment drawings and their descriptions of this application are used to explain this application and do not constitute an improper limitation on this application.

[0065] In addition, throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that the components and elements are not necessarily drawn to scale.

[0066] In the attached picture:

[0067] Figure 1 This is a schematic diagram of the structure of the BMS management system of the flow battery.

[0068] Figure 2 Schematic diagram of the structure of the SOH prediction device. DETAILED DESCRIPTION

[0069] Embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.

[0070] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0071] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0072] Reference Figure 1, the BMS management system of the flow battery of the present application includes: an external information collection module, an internal information collection module, a battery information collection module, an information preprocessing module, a historical SOH information collection unit and an SOH prediction device. Among them, the external information collection module is used to collect the ambient temperature, ambient humidity, charge and discharge rate, and vibration data of the battery when it is working to generate external information. The internal information collection module is used to obtain the aging information of the battery. The battery information collection module is used to collect the coulomb efficiency, voltage efficiency, energy efficiency and battery internal resistance of the battery during each cycle to generate battery information. The information preprocessing module is used to receive external information, aging information and battery information, and preprocess the external information, aging information and battery information to generate a feature set. The historical SOH information collection unit is used to collect the historical SOH value of the battery at each cycle.

[0073] In addition, the BMS management system of the flow battery also includes sensors that collect external information, aging information, battery information, and historical SOH values. The specific types of sensors can be selected by those skilled in the art according to actual needs, and will not be discussed here. For ease of understanding, the characteristics of all information are discussed, and the selection of sensors can be selected according to needs.

[0074] External information mainly includes ambient temperature, ambient humidity, charge and discharge rate, and vibration data. Temperature and humidity refer to the external temperature of the battery and have nothing to do with the temperature inside the battery. The charge and discharge rate here is based on the rated power as a ratio. For example, if the rated power is 100P and the charging power is 200P, the charging power is 2. Vibration data mainly refers to the vibration of the battery. Because vibration mainly affects the chemical reaction inside the electrolyte, vibration data only needs to collect vibration energy.

[0075] Aging information is related to the battery's electrode material and working time. Aging information only needs to record the battery's working time and the corresponding electrode material. In practice, aging information cannot be collected directly and can only be corrected based on the experimental data provided by the manufacturer. For example, the data provided by an electrode manufacturer shows that the degree of corrosion of the electrode in the electrolyte for 1 year, 2 years, and 3 years is 1 / 2 / 3 respectively. Aging information is an empirical value and needs to be set in advance according to time.

[0076] Battery information is related to each cycle of battery charging. Coulomb efficiency is the ratio of the actual discharge amount of the battery to the actual charge amount of the battery in one charge and discharge cycle. Therefore, to measure the coulomb efficiency, you need to set up a current sensor to calculate the current of the battery during each discharge and each charge. Voltage efficiency is the ratio of the average discharge voltage to the average charge voltage in one charge and discharge cycle. Therefore, to measure voltage efficiency, you need to set up a voltage sensor to measure the voltage of the battery during each charge and discharge. Energy efficiency is the ratio of the energy output during the battery discharge process to the energy consumed during the battery charging process in one charge and discharge cycle. Energy efficiency is related to voltage and current. When calculating, the time integral of the current is multiplied by the time integral of the voltage, which is the corresponding charging energy. For example, to calculate the energy during charging, you only need to calculate that the integral of the current over time is 10 and the integral of the voltage over time is 10 during the charging time, so the charging energy is 100.

[0077] The historical SOH value is calculated by combining the capacity obtained after each discharge of the battery with the rated capacity. For example, after the battery completes a charge and discharge cycle, the battery capacity is calculated to be 99 based on the current and voltage of the battery during discharge, but the rated capacity of the battery is 100, so the SOH value corresponding to this charge and discharge cycle is 99. When the battery is managed by BMS, the current energy storage capacity of the battery is predicted, and after the battery is discharged, the SOH value of the battery can be obtained based on the discharge amount of the battery. In this solution, the corresponding SOH value is calculated after each discharge of the battery, which is used to predict the battery SOH value during the next charge and discharge cycle, so as to facilitate battery charging management or discharge management.

[0078] The collection and collection methods of all the above information, the information collected under the above scheme lacks regularity and lacks relevance to each other. Therefore, this scheme adopts the following scheme:

[0079] The information preprocessing module includes: an information collection unit, an information extraction unit, and a collection unit. The information collection unit collects external information, aging information, and battery information of the battery. The information extraction unit processes the external information, aging information, and battery information into a sequence related to a complete charge and discharge cycle. i .a i Represents the sequence at the i-th charge and discharge cycle.

[0080] For example, external information and aging information are both time-related data. You only need to use the node at the end of each charge and discharge cycle of the battery as the basis for collection, and then use the average value within the charge and discharge cycle as the corresponding element value. For example, the first charge and discharge cycle is from January 1 to January 2, and you only need to use the average value of all external information and aging information from January 1 to January 2 as the basis for the corresponding element, such as the average ambient temperature and average ambient humidity.

[0081] a i = {b 1i 、b 2i ,…b 9i}, b 1i 、b 2i ,…b 9i They represent the characteristic values ​​of each element in the external information, aging information, and battery information at the i-th charge and discharge cycle. In this scheme, the external information, aging information, and battery information include a total of 9 elements listed. In practice, they can be appropriately increased or decreased. Regardless of the increase or decrease, the battery information is the most critical information and is indispensable.

[0082] The collection unit is used to collect all a i To obtain the feature set A. A={a1, a2, ...a i …a n}, n represents the number of sequences.

[0083] In a feature set A, there are n sequences, each of which includes 9 elements. Each sequence is the value of the relevant information collected by the battery during a charge and discharge cycle.

[0084] Corresponding to the historical SOH values, this solution also collects data based on the nodes of the battery's charge and discharge cycle, so the number of historical SOH values ​​corresponds to the number of sequences.

[0085] refer to Figure 2 , The SOH prediction device includes: a prediction unit and a correction unit.

[0086] A prediction unit, which predicts and generates a battery SOH value based on the input feature value;

[0087] The correction unit generates the influence information of each element in the feature set and the SOH value based on the historical SOH value, and sends the influence information to the prediction unit to guide the generation of the SOH value of the battery.

[0088] S1: Obtain feature set A and the historical SOH value Y corresponding to feature set A.

[0089] Y={y1, y2, ...y i ,…y n},y iRepresents the SOH value of the battery at the i-th charge and discharge cycle;

[0090] S2: Convert the feature set A into the feature matrix AP and convert the historical SOH value Y into the numerical matrix YP;

[0091]

[0092] Median alignment normalization is performed for each element in the feature matrix AP. Median alignment normalization is normalization based on the median. For example, the element of ambient temperature has three values ​​in three cycles, 31, 32, and 33. 32 is used as the median, and then these three values ​​are normalized. And so on, all elements are normalized.

[0093] S3: Convert the feature matrix AP into the augmented matrix ZA, where the feature matrix AP has a size of n*9 and the augmented matrix ZA has a size of n*2 9 ;

[0094] In the augmented matrix AP;

[0095]

[0096] Among them, j represents the index of the column in the feature matrix AP, k represents the index of the column in the augmented matrix ZA,

[0097] M[j]=min(X j ), X j ={x 1j 、x 2j 、x 2j ,…x 9j}, M[j] is X j The minimum value among them, k∈(1,2 9 -1), l∈(1,2 9 -1), j∈(1,9), b j represents the binary value of the bth combination; k j It is the result of normalizing attribute j.

[0098] Constructing the augmented matrix ZA is actually to convert the 9 columns in AP into a matrix, because each column represents an element, such as ambient temperature, ambient humidity, etc. The combination of these elements may be 2 9 . So the number of columns of the augmented matrix ZA is 2 9 , and the number of rows is the same as the feature matrix AP.

[0099] S4: Use the augmented matrix and the actual SOH value to build a linear regression model to solve the interaction measure set U;

[0100] Y=ZT*U+δ 2 ;

[0101] Among them, Y is the historical SOH value, U is the interaction measure set, δ 2 is a random noise term.

[0102] Introducing Y and ZT, the actual linear regression model is:

[0103]

[0104] By using the least squares method, a set of interaction measures U can be found so that the random noise term between the predicted SOH value and the actual SOH value is minimized.

[0105] The interaction measure set U found in this way is the required interaction measure.

[0106] Thus, U = (ZAPZA) -1 Z T Y, T represents the transpose operation of the matrix.

[0107] The following are the elements in the interaction measure set U;

[0108] e is the initial value, which is 1 in this application. is the interaction measure of the corresponding set.

[0109] S5: Denormalize the obtained interaction measure to the original scale to obtain the interaction measure w of the relevant elements on the original scale;

[0110]

[0111] t l is the lth intermediate variable, u l is the lth interaction measure in the interaction measure set U;

[0112]

[0113] k∈(1,2 9 -1), l∈(1,2 9 -1), j∈(1,9), b j Represents the binary value of the b-th combination, b∈(1,2 9 -1), b j It indicates whether the bth combination contains element j. For example, if the bth combination contains element j, then b j It is 1, otherwise it is 0.

[0114] The prediction unit includes: a generator and a corrector.

[0115] The generator has a built-in multi-directional convolutional network, and the feature values ​​are input into the generator to generate the SOH value; the corrector receives the SOH value generated by the generator, and judges the accuracy of the SOH value based on the interaction measure on the original scale, and sends the accuracy to the generator.

[0116] The generator is a multi-directional convolutional network that mainly predicts the trend of the SOH value based on the feature set and then calculates the SOH value. The trimmer calculates the interaction measure on the original scale of the SOH value generated by the generator, and then compares it with the previous interaction measure on the original scale to generate the accuracy of the current SOH value and guide the correction of the generator's SOH value.

[0117] When training the generator and the discriminator, both need to be trained. For the training of the generator, it is only necessary to train according to the convolutional network. For the discriminator, which is also a neural network model, when training the neural network model, the SOH value and the corresponding feature set are used as input data, and the interaction measure of the SOH value and the corresponding feature set on the original scale is used as output data. The input data and output data are used as a training sample. Then multiple training samples are input into the discriminator, so that the discriminator can automatically learn the calculation experience of the interaction measure on the original scale, and then quickly obtain an output data after obtaining the input data.

[0118] Thus, when the generator is trained, the following steps are included:

[0119] Step 1: The feature set is input to the generator, which generates a predicted SOH value;

[0120] Step 2: Input the predicted SOH value and feature set into the discriminator, which will obtain the interaction measure Q1 on the original scale of the predicted SOH value and feature set;

[0121] Step 3: The discriminator compares Q1 with the true SOH value and the interaction measure Q2 on the original scale calculated from the feature set, and then sends the comparison result to the generator.

[0122] In this way, when the generator is corrected, in addition to comparing the predicted SOH value with the actual SOH value, it can also compare with Q1 and Q2 to further correct the generator.

[0123] The multi-directional convolutional network is a common neural network model. The specific construction method is not described here. However, during training, the loss function of the generator is L;

[0124] L = λ1L1 + λ2L2;

[0125] L1 is the original probability loss function, L2 is the element size loss function, λ1 and λ2 are the first and second parameters respectively;

[0126]

[0127] Where n is the number of charge cycles in the feature set, y i is the true SOH value of the i-th sample, p i is the predicted probability of the i-th sample;

[0128]

[0129] Among them, W i The interaction measure on the original scale calculated for the ith sample, W u is the interaction measure on the true original scale.

[0130] The above description is only some preferred embodiments of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present application is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with (but not limited to) the technical features with similar functions disclosed in the embodiments of the present application to form a technical solution.

Claims

1. A BMS management system for a liquid flow battery, characterized in that: include: The external information collection module is used to collect the ambient temperature, ambient humidity, charge and discharge rate, and vibration data of the battery when it is working, so as to generate external information; Internal information collection module, used to obtain battery aging information; A battery information collection module is used to collect the coulombic efficiency, voltage efficiency, energy efficiency and battery internal resistance of each battery cycle to generate battery information; An information preprocessing module receives external information, aging information and battery information, and preprocesses the external information, aging information and battery information to generate a feature set; A historical SOH information collection unit is used to collect the historical SOH value of the battery at each cycle; An SOH prediction device generates a current SOH value of the battery based on a feature set; The SOH prediction device includes: A prediction unit, which predicts and generates a battery SOH value based on the input feature value; The correction unit generates the influence information of each element in the feature set and the SOH value based on the historical SOH value, and sends the influence information to the prediction unit to guide the generation of the SOH value of the battery.

2. The BMS management system for a flow battery according to claim 1, characterized in that: Coulomb efficiency is the ratio of the actual discharge capacity of the battery to the actual charge capacity of the battery during one charge and discharge cycle; Voltage efficiency is the ratio of the average discharge voltage to the average charge voltage in one charge and discharge cycle; Energy efficiency is the ratio of the energy output during the battery discharge process to the energy consumed during the battery charging process during one charge and discharge cycle.

3. The BMS management system for a flow battery according to claim 2, characterized in that: The information preprocessing module includes: Information collection unit: collects external information, aging information, and battery information of the battery; Information extraction unit: processes external information, aging information, and battery information into a sequence related to a complete charge and discharge cycle i ; a i represents the sequence at the i-th charge and discharge cycle; a i = {b 1i , b 2i ,…b 9i }, b 1i , b 2i ,…b 9i Respectively represent the characteristic values ​​of each element in the external information, aging information, and battery information at the i-th charge and discharge cycle; Collection units, collect all a i To obtain feature set A; A={a1, a2, ...a i …a n }, n represents the number of sequences.

4. The BMS management system for a flow battery according to claim 3, characterized in that: The correction unit calculates the influence information U in the following manner; S1: Obtain feature set A and the historical SOH value Y corresponding to feature set A; Y={y1, y2, ...y i ,…y n },y i Represents the SOH value of the battery at the i-th charge and discharge cycle; S2: Convert the feature set A into the feature matrix AP and convert the historical SOH value Y into the numerical matrix YP; S3: Convert the feature matrix AP into the augmented matrix ZA, where the feature matrix AP has a size of n*9 and the augmented matrix ZA has a size of n*2 9 ; In the augmented matrix AP; Among them, j represents the index of the column in the feature matrix AP, and k represents the index of the column in the augmented matrix ZA; S4: Use the augmented matrix and the actual SOH value to build a linear regression model to solve the interaction measure; Y=ZT*U+δ 2 ; Among them, Y is the historical SOH value, U is the interaction measure set, δ 2 is a random noise term; S5: Denormalize the obtained interaction measure to the original scale to obtain the interaction measure w of the relevant elements on the original scale; t l is the lth intermediate variable, u l is the lth interaction measure in the interaction measure set U; M[j]=min(X j ), X j ={x 1j 、x 2j 、x 2j ,…x 9j }, M[j] is X j The minimum value in ; k0∈(1,2 9 -1), l∈(1,2 9 -1), j∈(1,9), b j Represents the binary value of the b-th combination.

5. The BMS management system for a liquid flow battery according to claim 4, characterized in that: Median alignment normalization is performed on each element in the feature matrix AP.

6. The BMS management system for a liquid flow battery according to claim 2, characterized in that: U=(ZAPZA) -1 Z T Y, T represents the transpose operation of the matrix.

7. The BMS management system for a flow battery according to claim 1, characterized in that: The prediction unit includes: Generator, with a built-in multi-directional convolutional network, the feature values ​​are input into the generator to generate the SOH value; The corrector,receives the SOH value generated by the generator, and judges the accuracy of the SOH value based on the interaction measure on the original scale, and sends the accuracy to the generator.

8. The BMS management system for a flow battery according to claim 1, characterized in that: Aging information is related to the battery's electrode material and operating time.

9. A BMS management method for a flow battery, characterized in that: The BMS management method uses the BMS management system of the flow battery according to any one of claims 1 to 8 to predict the SOH value of the battery.

Images