Battery SOC prediction method and device, monitoring system and storage medium

Abstract

The invention discloses a battery SOC prediction method and device, a monitoring system and a storage medium. The method comprises the following steps: determining the charging and discharging stage of a battery according to the current accumulated charging and discharging duration of the battery; selecting a corresponding pre-trained SOC prediction model according to the charging and discharging stage of the battery; wherein each charging and discharging stage corresponds to one SOC prediction model; and inputting the current operation parameters of the battery into the corresponding SOC prediction model to obtain the SOC prediction value of the battery. According to the scheme, the influence of the accumulated charging and discharging duration on the SOC prediction of the battery is fully considered, and different SOC prediction models are trained in advance for different charging and discharging stages of the battery, so that the accuracy of SOC value prediction can be greatly improved.

Description

Technical Field

[0001] This application relates to, but is not limited to, the field of state of charge assessment of battery energy storage systems, and particularly relates to a method and device for predicting the state of charge (SOC) of a battery, a monitoring system, and a storage medium. Background Art

[0002] With the rapid development of new clean energy in recent years, energy storage technology has developed rapidly in the industry. The all-vanadium redox flow battery is widely used in the energy storage industry and has become the mainstream technical solution in the energy storage field due to its good stability and safety. The redox flow battery is based on the redox reaction generated by the flow of the electrolyte, enabling the vanadium ions to continuously change their valence states, thereby achieving the purpose of charging and discharging. During the use of the all-vanadium redox flow battery, it is necessary to monitor the state of charge (SOC) of the battery in real time to avoid battery damage caused by overcharging or over-discharging. Therefore, real-time and accurate prediction of the SOC of the all-vanadium redox flow battery is crucial for the operation of the all-vanadium redox flow battery system.

[0003] Currently, the mainstream ampere-hour integration method and the reference battery method have certain limitations in predicting SOC: The ampere-hour integration method calculates the remaining power by integrating the current, which only considers the influence of the current as a single factor on SOC, resulting in a large prediction deviation; The reference battery method measures the SOC of the actual stack by designing a branch pipeline in the main pipeline and connecting a proportionally reduced stack. However, due to the small internal resistance of the battery, it is difficult to quickly and accurately measure the SOC of the battery during the dynamic operation of the battery. At the same time, affected by the precipitation of electrolyte crystallization substances, blockage is likely to occur in the liquid path of the reference battery, causing the reference battery to fail. With the continuous development of neural network technology, there are also many new applications in predicting the SOC value of the vanadium redox flow battery system. However, the existing methods only consider factors such as voltage, current, and temperature, and do not consider the influence of charge and discharge time on the SOC prediction of the all-vanadium redox flow battery. Summary of the Invention

[0004] The following is an overview of the subject matter described in detail in this article. This overview is not intended to limit the scope of protection of the claims.

[0005] This application provides a method and device for predicting the SOC of a battery, a monitoring system, and a storage medium, which can more accurately predict the SOC of the battery.

[0006] An embodiment of the present application provides a method for predicting the state of charge (SOC) of a battery, including: determining the charge and discharge stage of the battery according to the current cumulative charge and discharge duration of the battery; selecting a corresponding pre-trained SOC prediction model according to the charge and discharge stage of the battery; wherein, each charge and discharge stage corresponds to an SOC prediction model respectively; inputting the current operating parameters of the battery into the corresponding SOC prediction model to obtain the SOC prediction value of the battery.

[0007] An embodiment of the present application further provides a device for predicting the SOC of a battery, including: a memory and a processor; the memory is used to store a program for predicting the SOC of the battery; the processor is used to read the program for predicting the SOC of the battery and execute the method for predicting the SOC of the battery as described in any embodiment of the present application.

[0008] An embodiment of the present application further provides a battery monitoring system, including: a memory and a processor; the memory is used to store a program for predicting the SOC of the battery; the processor is used to read the program for predicting the SOC of the battery and execute the method for predicting the SOC of the battery as described in any embodiment of the present application.

[0009] An embodiment of the present application further provides a non-transitory computer-readable storage medium, which stores a computer program, wherein the computer program can implement the method for predicting the SOC of the battery as described in any embodiment of the present application when executed by a processor.

[0010] Compared with the related art, a method, a device, a monitoring system and a storage medium for predicting the SOC of a battery provided by the embodiments of the present application determine the charge and discharge stage of the battery according to the current cumulative charge and discharge duration of the battery, then select a corresponding pre-trained SOC prediction model according to the charge and discharge stage of the battery, and obtain the SOC prediction value of the battery based on the corresponding SOC prediction model. This solution fully considers the influence of the cumulative charge and discharge duration on the prediction of the battery SOC, and pre-trains different SOC prediction models for different charge and discharge stages of the battery, which can greatly improve the accuracy of the SOC value prediction.

[0011] Other features and advantages of the present application will be described in the subsequent description, and some of them will become obvious from the description, or be understood by implementing the present application. Other advantages of the present application can be realized and obtained through the solutions described in the description and the drawings. Description of the Drawings

[0012] The drawings are used to provide an understanding of the technical solutions of the present application, and constitute a part of the description. They are used together with the embodiments of the present application to explain the technical solutions of the present application, and do not constitute a limitation to the technical solutions of the present application.

[0013] Figure 1 This is a brief flowchart of the battery SOC prediction method according to the embodiments of the present application; Figure 2 This is a schematic diagram of the BP neural network structure according to the embodiments of the present application; Figure 3 This is a flowchart of the battery SOC prediction method according to the embodiments of the present application; Figure 4 This is a schematic diagram of the battery SOC prediction model construction process according to the embodiments of the present application; Figure 5 This is a schematic diagram of the battery SOC prediction device according to the embodiments of the present application. Detailed implementation manners

[0014] The present application describes multiple embodiments, but the description is exemplary rather than restrictive, and it is obvious to those of ordinary skill in the art that there can be more embodiments and implementation solutions within the scope covered by the embodiments described in the present application. Although many possible feature combinations are shown in the drawings and discussed in the detailed implementation manners, many other combination ways of the disclosed features are also possible. Unless specifically restricted, any feature or element of any embodiment can be combined with any other feature or element in any other embodiment, or can replace any other feature or element in any other embodiment.

[0015] The present application includes and contemplates combinations with features and elements known to those of ordinary skill in the art. The embodiments, features, and elements disclosed in the present application can also be combined with any conventional features or elements to form a unique invention solution. Any feature or element of any embodiment can also be combined with features or elements from other invention solutions to form another unique invention solution. Therefore, it should be understood that any feature shown and / or discussed in the present application can be implemented alone or in any suitable combination. Therefore, except for the limitations according to the appended claims and their equivalent replacements, the embodiments are not subject to other limitations. In addition, various modifications and changes can be made within the protection scope of the appended claims.

[0016] In addition, when describing representative embodiments, the specification may have presented the method and / or process as a specific sequence of steps. However, to the extent that the method or process does not depend on the specific order of the steps described herein, the method or process should not be limited to the specific order of steps described. As those of ordinary skill in the art will understand, other step orders are possible. Therefore, the specific order of steps set forth in the specification should not be construed as a limitation on the claims. In addition, the claims directed to the method and / or process should not be limited to performing their steps in the order written, and those skilled in the art can readily understand that these orders can vary and still remain within the spirit and scope of the embodiments of the present application.

[0017] With the continuous development of neural network technology, there are also many new applications in the prediction of the SOC value of vanadium redox flow battery systems. However, existing methods only consider factors such as voltage, current, and temperature, and do not comprehensively consider the influence of electrolyte flow rate and charge-discharge time on the SOC prediction of all-vanadium redox flow batteries.

[0018] For this reason, an embodiment of the present application provides a method for predicting the SOC of a battery, as Figure 1 shown, which may include the following steps: Step S110: Determine the charge-discharge stage of the battery according to the current cumulative charge-discharge duration of the battery; Step S120: Select a corresponding pre-trained SOC prediction model according to the charge-discharge stage of the battery; wherein, each charge-discharge stage corresponds to an SOC prediction model; Step S130: Input the current operating parameters of the battery into the corresponding SOC prediction model to obtain the SOC prediction value of the battery.

[0019] Exemplarily, the battery may be an all-vanadium redox flow battery.

[0020] In the battery SOC prediction method of this embodiment, by determining the charge-discharge stage of the battery according to the current cumulative charge-discharge duration of the battery, and then selecting a corresponding pre-trained SOC prediction model according to the charge-discharge stage of the battery, and obtaining the SOC prediction value of the battery based on the corresponding SOC prediction model. This shows that the solution of this embodiment fully considers the influence of the cumulative charge-discharge duration on the SOC prediction of the battery, and pre-trains different SOC prediction models for different charge-discharge stages of the battery, which can greatly improve the accuracy of the SOC value prediction.

[0021] In an exemplary embodiment, the SOC prediction model can be trained through the following process: For multiple batteries, obtain multiple first data sets respectively according to the following steps: The battery data of the entire life cycle of the battery is collected at a preset collection time interval, and the battery data is preprocessed to obtain the charge and discharge data corresponding to different cumulative charge and discharge times; wherein the entire life cycle refers to the entire period from first use to scrapping, and the charge and discharge data at least includes the cumulative charge and discharge time; illustratively, scrapping means that the battery can no longer continue to charge and discharge; the cumulative charge and discharge time may be recorded according to the collection rule when collecting data. For example, since the battery data of the entire life cycle of the battery is collected at a preset collection time interval, this means that the cumulative charge and discharge time corresponding to the data collected for the first time is 1*the preset collection time interval, the cumulative charge and discharge time corresponding to the data collected for the second time is 2*the preset collection time interval, and so on; it may also be that when the battery management system itself has the function of recording the cumulative charge and discharge time of the battery, the cumulative charge and discharge time recorded by the battery management system is directly collected each time data is acquired.

[0022] The charge and discharge data of the battery are divided into a plurality of first data sets according to the accumulated charge and discharge time range corresponding to each charge and discharge stage; wherein the first data sets correspond to the charge and discharge stages one by one; the charge and discharge stages are obtained by dividing the entire life cycle of the battery according to preset accumulated time intervals, and each charge and discharge stage corresponds to a accumulated charge and discharge time range; After the first data sets of multiple batteries are collected, for each charge and discharge stage, the SOC prediction model corresponding to the charge and discharge stage is trained according to the first data sets of multiple batteries corresponding to the charge and discharge stage.

[0023] Exemplarily, the preset accumulation time interval and the preset collection time interval can be set according to experience or actual needs.

[0024] For example, corresponding sensors may be arranged at the inlet and outlet positions of the battery stack to collect data such as voltage, current, inlet and outlet stack temperatures, and electrolyte flow rate.

[0025] It can be seen from this embodiment that each battery has multiple first data sets corresponding to different charge and discharge stages; after the first data sets of multiple batteries are concentrated, for each charge and discharge stage, there are multiple first data sets of batteries corresponding to the charge and discharge stage.

[0026] In an example of the present embodiment, before dividing the charge and discharge data of the battery into multiple first data sets according to the cumulative charge and discharge time range corresponding to each charge and discharge stage, it also includes: dividing the entire life cycle of the battery into different charge and discharge stages according to preset cumulative time intervals, and determining the cumulative charge and discharge time range corresponding to each charge and discharge stage.

[0027] The battery SOC prediction method of this embodiment is as follows. First, the entire life cycle of the battery is divided into different charge and discharge stages according to a preset cumulative duration interval, and the cumulative charge and discharge duration range corresponding to each charge and discharge stage is determined. Then, according to the cumulative charge and discharge duration range corresponding to each charge and discharge stage, the battery data of the entire life cycle of multiple batteries collected is divided into multiple first data sets (that is, the battery data stored in each first data set is the battery data of the charge and discharge stage corresponding to the first data set for the multiple batteries collected). Then, an SOC prediction model is trained for each first data set to obtain an SOC prediction model corresponding to each charge and discharge stage. In this way, when predicting the SOC value, the battery in different charge and discharge stages can be selected to use the corresponding SOC prediction model to predict the SOC value, so as to obtain a more accurate SOC prediction value.

[0028] In an example of this embodiment, the preprocessing of the battery data to obtain charge and discharge data corresponding to different times includes: The battery data is divided into multiple second data sets according to a preset average duration interval; wherein, the preset average duration interval is greater than the preset acquisition time interval and less than the preset cumulative duration interval; For each second data set, calculate the average value of all data in the second data set from any dimension, and use the average value as the new data corresponding to the second data set; Combine all the new data corresponding to the second data sets in order as the charge and discharge data.

[0029] Exemplarily, the preset average duration interval can be set according to experience or actual requirements.

[0030] It should be noted that each data object (i.e., the data collected each time) can include multiple dimensions (such as voltage, current, inlet and outlet stack temperature, electrolyte flow rate, and each attribute counts as one dimension). Calculating the average value of all data in the second data set from any dimension means that for each second data set, the average value is taken for each dimension of the multiple data objects included in the second data set, and finally a data object formed by the average values of all dimensions is obtained, that is, the data object corresponding to the second data set (for the sake of distinction, here "a data object formed by the average values of all dimensions" is called the second data object, and "the multiple data objects originally included in the second data set" are called the first data objects; the second data object and the first data object have the same dimensions, and the value of each dimension data included in the second data object here is equal to the average value of the corresponding dimension values of the multiple first data objects originally included in the second data set). The following is a specific example for illustration. For example, a certain original second data set includes 3 data objects, which are (voltage 2, current 20, temperature 21, internal resistance 20, electrolyte flow rate 300), (voltage 3, current 30, temperature 22, internal resistance 30, electrolyte flow rate 300), and (voltage 4, current 40, temperature 26, internal resistance 40, electrolyte flow rate 400). Then, after performing the operation of "calculating the average value of all data in the second data set from any dimension and using the average value as the new data corresponding to the second data set", the new data corresponding to the second data set is (voltage 3, current 30, temperature 23, internal resistance 30, electrolyte flow rate 300).

[0031] Exemplarily, preprocessing the battery data may further include: performing operations such as screening and cleaning on the battery data to remove outliers.

[0032] In the battery SOC prediction method of this embodiment, after collecting the data collected at a preset collection time interval, all the collected data will also be divided into multiple parts at a preset average duration interval, and each part includes (preset average duration interval / preset collection time interval) data; then the average value is taken for each part of the data, and all the average values are combined to form charge and discharge data; finally, the SOC prediction model is trained based on the charge and discharge data; in this way, the error in the data collection process can be effectively reduced, the effect of model training can be improved, and further the accuracy of SOC value prediction can be improved.

[0033] The process of obtaining the first data set will be specifically described below with a specific example. For example, the preset cumulative duration interval can be 5 days, the preset acquisition time interval can be set to 1 second, and the preset average duration interval can be set to 1 minute. Assuming that the full life cycle of the battery is 100 days, a total of 100 * 24 * 60 * 60 data are collected from the first use of the battery to its scrapping. The process of processing these data to obtain the first data set can be as follows: First step, divide these data into 100 * 24 * 60 second data sets with 1 minute (i.e., the preset average duration interval) as the unit (that is, in the order of data acquisition, every 60 data form a second data set), and then take the average value of the 60 data in each second data set to obtain the average values corresponding to the 100 * 24 * 60 second data sets, and use the average values corresponding to the 100 * 24 * 60 second data sets as charge and discharge data (equivalent to obtaining 100 * 24 * 60 charge and discharge data, and the cumulative charge and discharge duration interval of each charge and discharge data is 1 minute); Second step, divide the 100-day full life cycle of the battery into 20 charge and discharge stages with 5 days (i.e., the preset cumulative duration interval) as the unit. The cumulative charge and discharge duration range corresponding to the first charge stage is from 1 minute to 5 * 24 * 60 minutes, the cumulative charge and discharge duration range corresponding to the second charge stage is from (5 * 24 * 60 + 1) minutes to 10 * 24 * 60 minutes, ……, the cumulative charge and discharge duration range corresponding to the 19th charge stage is from (90 * 24 * 60 + 1) minutes to 95 * 24 * 60 minutes, and the cumulative charge and discharge duration range corresponding to the 20th charge stage is from (95 * 24 * 60 + 1) minutes to 100 * 24 * 60 minutes; then, divide the 100 * 24 * 60 charge and discharge data obtained in the first step into 20 first data sets according to the range corresponding to each charge and discharge stage.

[0034] In an exemplary embodiment, the charge and discharge data further includes: voltage, current, temperature, internal resistance, electrolyte flow rate; training the SOC prediction model corresponding to each charge and discharge stage according to the first data set corresponding to each charge and discharge stage includes: Using the voltage, current, temperature, internal resistance, and electrolyte flow rate as inputs and the actual SOC value of the battery as a label, training the SOC prediction model until the SOC prediction value output by the SOC prediction model meets the preset conditions.

[0035] Exemplarily, the SOC prediction value output by the SOC prediction model satisfying the preset condition may refer to meeting the target accuracy. For example, when the target accuracy is 1%, when the difference between the SOC prediction value output by the model and the actual SOC value is controlled within 1%, it is considered that the model training is completed. It should be noted that the specific value of the target accuracy in this application is not limited and can be set according to the actual situation.

[0036] Of course, this application does not limit that the "preset condition" in "until the SOC prediction value output by the SOC prediction model satisfies the preset condition" is the "target accuracy", and it can also be other conditions.

[0037] The battery SOC prediction method of this embodiment not only considers the influence of voltage, current, temperature, and internal resistance on SOC prediction, but also considers the influence of electrolyte flow rate on SOC prediction, further improving the accuracy of SOC prediction.

[0038] In an exemplary embodiment, training the SOC prediction model corresponding to each charge-discharge stage according to the first data set corresponding to each charge-discharge stage includes: Step S210: Train the SOC prediction model corresponding to the first charge-discharge stage according to the first data set corresponding to the first charge-discharge stage of multiple batteries, and after the training is completed, proceed to step S220; Step S220: Continue to train the SOC prediction model corresponding to the N-1th charge-discharge stage that has been trained according to the first data set corresponding to the Nth charge-discharge stage of multiple batteries. After the training is completed, determine whether there is still a charge-discharge stage for which the corresponding SOC model has not been trained. If so, N = N + 1 and return to step S220; if not, end the training; where N is a positive integer and the initial value of N is 2.

[0039] The battery SOC prediction method of this embodiment, in the process of constructing the SOC prediction model, trains the current SOC prediction model based on the SOC prediction model of the previous charge-discharge stage, and finally continuously iterates to obtain the SOC prediction model for the entire life cycle. In this way, on the one hand, the training time of the model can be saved; on the other hand, multiple SOC prediction models for different charge-discharge stages can be obtained, and the corresponding SOC prediction models are used for prediction for different charge-discharge stages to improve the accuracy of SOC prediction.

[0040] In an exemplary embodiment, the SOC prediction model is a BP neural network model. As Figure 2As shown in the figure, the BP neural network model described in this embodiment can be a 5-N-1 structure BP neural network model. Its main structure is the input layer - hidden layer - output layer. The input data starts from the input layer, passes through the linear combination of weights and biases, and then is processed by the activation function and transmitted to the hidden layer. The output of each hidden layer is used as the input of the next layer, and so on until the output layer generates the final result. According to the result of error calculation, the error signal is propagated backward from the output layer to the input layer through the network, and the weights and biases are continuously corrected to achieve the error target. In this model, the dimension of the input layer is 5, which are the stack internal resistance, voltage, current, temperature, and electrolyte flow rate respectively; the output layer is the SOC value, and the hidden layer nodes are found to be the optimal hidden layer nodes through continuous model testing.

[0041] The following is a complete example of the battery SOC prediction method of this application. As shown in Figure 3 the figure (the schematic diagram can refer to Figure 4 ), it may include the following steps: Step S310: For multiple batteries, collect battery data throughout the battery life cycle at a preset acquisition time interval. This step can be specifically implemented through the following process: Arrange corresponding sensors at the inlet and outlet positions of the stack in the all-vanadium redox flow battery system to collect data such as voltage, current, inlet and outlet stack temperature, electrolyte flow rate, and cumulative charge and discharge duration. Set to collect data every 1 second (i.e., the preset acquisition time interval). At the same time, measure the internal resistance of each stack before leaving the factory, and extract the SOC value of each stack during operation to establish an original database.

[0042] Step S320: Preprocess the battery data to obtain charge and discharge data corresponding to different cumulative charge and discharge durations. This step can specifically include the following steps S321 - S323: Step S321: Screen and clean the acquired data, and clean the outliers in each group of data, such as the data with a value of 0 in one dimension; Step S322: In order to further eliminate the errors in the numerical values of each dimension of the data, take the average value of the valid data obtained in 1 minute (i.e., the preset average duration interval) in each dimension to obtain a new data. Perform the above operations on all the data, and finally obtain the training data and test data required by the model.

[0043] Step S323: Divide all the training data and test data according to the cumulative charge and discharge duration. The data generated when the cumulative charge and discharge time reaches 5 days (i.e., the preset cumulative duration interval) is a database until the stack can no longer be charged and discharged. In this way, N databases arranged in the order of charge and discharge time are obtained.

[0044] Step S330: For each charge-discharge stage, train the SOC prediction model corresponding to this charge-discharge stage respectively according to the charge-discharge data corresponding to this charge-discharge stage.

[0045] Exemplarily, when the preset cumulative duration interval is 5 days, this step may include the following steps S331 - S333: Step S331: Input the data in the first database (i.e., the database corresponding to the cumulative charge-discharge duration of 1 - 5 days) into the model for training. When the target accuracy is reached, the training ends, and the real-time SOC prediction model of the all-vanadium redox flow battery at the earliest stage of charge-discharge is obtained.

[0046] Step S332: Then input the data in the second database (i.e., the database corresponding to the cumulative charge-discharge duration of 6 - 10 days) into the first model for training again. Since the mapping relationship in the first model has been established, after training the model with new data, the internal mapping relationship will be adjusted, so as to obtain the SOC prediction model for the stage of cumulative charge-discharge time of 6 - 10 days.

[0047] Step S333: Input the data in the third database (i.e., the database corresponding to the cumulative charge-discharge duration of 11 - 15 days) into the second model to obtain the SOC prediction model for the stage of cumulative charge-discharge time of 11 - 15 days. And so on, finally obtaining the SOC prediction model corresponding to each charge-discharge stage of the all-vanadium redox flow battery.

[0048] Step S340: Integrate the trained multiple SOC prediction models into the battery monitoring system. When performing SOC prediction, determine the charge-discharge stage where the battery is located according to the current cumulative charge-discharge duration of the battery, then select the corresponding pre-trained SOC prediction model according to the charge-discharge stage where the battery is located, and then input the current operating parameters of the battery into the corresponding SOC prediction model to obtain the SOC prediction value of this battery.

[0049] Through step S340, integrate the SOC prediction models of different charge-discharge stages into the all-vanadium redox flow system, and then input the voltage, stack, temperature, internal resistance, and electrolyte flow rate of the battery into the corresponding SOC prediction models in this system, and the prediction value of the SOC of this battery can be obtained in real time.

[0050] In summary, based on the powerful non-linear processing ability of the BP neural network, this embodiment accurately and quickly predicts the SOC of the all-vanadium redox flow battery in real time. At the same time, considering the influence of charge-discharge time on SOC, when constructing the model, the SOC prediction model is constructed by segmenting according to charge-discharge time, and finally a set of SOC prediction models for the entire life cycle of the all-vanadium redox flow battery is obtained, which can accurately predict the SOC values of batteries in different charge-discharge stages.

[0051] An embodiment of the present application provides a battery SOC prediction device, as Figure 5 shown, including: a memory and a processor; The memory is used to store a program for battery SOC prediction; The processor is used to read the program for battery SOC prediction and execute the battery SOC prediction method as described in any embodiment of the present application.

[0052] An embodiment of the present application further provides a battery monitoring system, including: a memory and a processor; the memory is used to store a program for battery SOC prediction; the processor is used to read the program for battery SOC prediction and execute the battery SOC prediction method as described in any embodiment of the present application.

[0053] An embodiment of the present application further provides a non-transitory computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program can implement the battery SOC prediction method as described in any embodiment of the present application when executed by a processor.

[0054] In summary, a battery SOC prediction method, device, monitoring system, and storage medium provided by the present application have the following beneficial effects: (1) Compared with the existing SOC prediction methods, it comprehensively considers influencing factors such as voltage, current, inlet and outlet stack temperature, charge and discharge time, electrolyte flow rate, etc., and improves the accuracy of SOC prediction. (2) In the process of constructing the SOC prediction model, the charge and discharge time is divided at fixed time intervals, and its relevant data is collected. Based on the previous SOC prediction model, combined with the data of this time period, the SOC prediction model is further trained to obtain a new SOC prediction model, and finally the SOC prediction model for the entire life cycle is continuously iteratively obtained. In this way, on the one hand, the training time of the model can be saved; on the other hand, multiple SOC prediction models for different charge and discharge stages can be obtained, and the corresponding SOC prediction models are used for prediction in different charge and discharge stages to improve the accuracy of SOC prediction. (3) In terms of data processing, taking the average value of the data obtained at the preset average duration interval can effectively reduce the error in the data acquisition process, and then improve the training effect of the SOC prediction model, thereby improving the accuracy of SOC prediction.

[0055] Those of ordinary skill in the art will understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, and appropriate combinations thereof. In the hardware implementation, the division of the functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, one physical component can have multiple functions, or one function or step can be executed by several physical components in cooperation. Some or all of the components can be implemented as software executed by a processor, such as a digital signal processor or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include a computer storage medium (or non-transitory medium) and a communication medium (or transitory medium). As is well known to those of ordinary skill in the art, the term "computer storage medium" includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cartridges, tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those of ordinary skill in the art that a communication medium typically contains computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and can include any information delivery medium.

[0056] In the description of this specification, the descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples.

[0057] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A battery SOC prediction method, characterized in that: include: Determine the charging and discharging stage of the battery according to the current accumulated charging and discharging time of the battery; Select a corresponding pre-trained SOC prediction model according to the charging and discharging stage of the battery; wherein each charging and discharging stage corresponds to a SOC prediction model; The current operating parameters of the battery are input into the corresponding SOC prediction model to obtain the SOC prediction value of the battery.

2. The battery SOC prediction method according to claim 1, characterized in that: The SOC prediction model is trained through the following process: For multiple batteries, multiple first data sets are obtained according to the following steps: Collect battery data of the entire life cycle of the battery at a preset collection time interval, and pre-process the battery data to obtain charge and discharge data corresponding to different cumulative charge and discharge durations; wherein the entire life cycle refers to the entire period from first use to scrapping, and the charge and discharge data at least includes the cumulative charge and discharge duration; The charge and discharge data of the battery are divided into a plurality of first data sets according to the accumulated charge and discharge time range corresponding to each charge and discharge stage; wherein the first data sets correspond to the charge and discharge stages one by one; the charge and discharge stages are obtained by dividing the entire life cycle of the battery according to preset accumulated time intervals, and each charge and discharge stage corresponds to a accumulated charge and discharge time range; After the first data sets of multiple batteries are collected, for each charge and discharge stage, the SOC prediction model corresponding to the charge and discharge stage is trained according to the first data sets of multiple batteries corresponding to the charge and discharge stage.

3. The battery SOC prediction method according to claim 2, characterized in that: The preprocessing of the battery data to obtain charge and discharge data corresponding to different accumulated charge and discharge times includes: Dividing the battery data into a plurality of second data sets according to a preset average time interval; wherein the preset average time interval is greater than the preset collection time interval and less than the preset cumulative time interval; For each second data set, calculating the average value of all data in the second data set from any dimension, and using the average value as new data corresponding to the second data set; All new data corresponding to the second data set are combined in sequence as the charge and discharge data.

4. The battery SOC prediction method according to claim 2, characterized in that: The charge and discharge data also include: voltage, current, temperature, internal resistance, and electrolyte flow rate; the training of the SOC prediction model corresponding to each charge and discharge stage according to the first data set corresponding to the charge and discharge stage includes: The voltage, current, temperature, internal resistance, and electrolyte flow rate are used as inputs, and the actual SOC value of the battery is used as a label to train the SOC prediction model until the SOC prediction value output by the SOC prediction model meets the preset conditions.

5. The battery SOC prediction method according to claim 2, characterized in that: The step of training the SOC prediction model corresponding to the charge and discharge stage according to the first data set corresponding to the charge and discharge stage of the plurality of batteries includes:

51. According to the first data set corresponding to the first charge and discharge stage of the plurality of batteries corresponding to the first charge and discharge stage, train the SOC prediction model corresponding to the first charge and discharge stage, and proceed to step 52 after the training is completed; 52. According to the first data set corresponding to the Nth charge and discharge stage of the multiple batteries corresponding to the charge and discharge stage, continue to train the SOC prediction model corresponding to the N-1th charge and discharge stage that has completed training. After the training is completed, determine whether there are any charge and discharge stages whose corresponding SOC models have not been trained. If so, N=N+1 and return to step 52; if not, end the training; wherein the initial value of N is 2.

6. The battery SOC prediction method according to claim 1, characterized in that: The SOC prediction model is a BP neural network model.

7. The battery SOC prediction method according to claim 1, characterized in that: The battery is an all-vanadium liquid flow battery.

8. A battery SOC prediction device, comprising: A memory and a processor, characterized in that: The memory is used to store a program for battery SOC prediction; The processor is used to read the program for battery SOC prediction and execute the battery SOC prediction method as described in any one of claims 1 to 7.

9. A battery monitoring system comprising: A memory and a processor, characterized in that: The memory is used to store a program for battery SOC prediction; The processor is used to read the program for battery SOC prediction and execute the battery SOC prediction method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing a computer program, wherein: When the computer program is executed by a processor, it is capable of implementing the battery SOC prediction method as described in any one of claims 1 to 7.

Images