Method and device for predicting initial capacity of energy storage system, equipment and storage medium

Abstract

The invention provides an energy storage system initial capacity prediction method and device, equipment and a storage medium. The method comprises the following steps: acquiring parameter values of various parameter types of a plurality of battery cells of the energy storage system; the multiple parameter types comprise multiple items in the following types: battery cell capacity, battery cell voltage, battery cell temperature and battery cell resistance; based on the parameter values of the plurality of parameter types, constructing first input feature maps corresponding to the plurality of parameter types; wherein the parameter types of the parameter values in the first input feature map are the same; and inputting the constructed first input feature map into a pre-trained AI model, predicting the initial capacity of the energy storage system, and determining a predicted value of the initial capacity.

Description

Technical Field

[0001] The present application relates to the field of battery technology, and is related to but not limited to a method and apparatus, device, and storage medium for predicting the initial capacity of an energy storage system. Background Art

[0002] An energy storage system integrated with multiple cells refers to a system that combines multiple single cells in series, parallel or mixed connection to form a complete battery module or battery pack. This integration method can improve the voltage, capacity and power of the battery system to meet the needs of different application scenarios. Summary of the invention

[0003] The present application provides a method and apparatus, device, and storage medium for predicting the initial capacity of an energy storage system, wherein: In a first aspect, an embodiment of the present application provides a method for predicting the initial capacity of an energy storage system, the method comprising: obtaining parameter values ​​of multiple parameter types of multiple battery cells of the energy storage system; the multiple parameter types include multiple items of the following types: battery cell capacity, battery cell voltage, battery cell temperature, and battery cell resistance; based on the parameter values ​​of the multiple parameter types, constructing first input feature graphs corresponding to the multiple parameter types; wherein the parameter types of the parameter values ​​in the first input feature graph are the same; inputting the constructed first input feature graph into a pre-trained AI model, predicting the initial capacity of the energy storage system, and determining the predicted value of the initial capacity.

[0004] It can be understood that in an embodiment of the present application, a method is provided that can predict the initial capacity of an energy storage system before multiple battery cells are integrated into an energy storage system, filling the technical gap in predicting the initial capacity of an energy storage system. The method predicts the predicted value of the initial capacity of the energy storage system based on a first input characteristic diagram of multiple parameter values ​​of the same parameter type including multiple battery cells. In this way, there is no need to integrate multiple battery cells into an energy storage system, and the initial capacity of the energy storage system can be obtained by measurement, which is beneficial for the manufacturer of the energy storage system to determine the number of battery cells to be integrated for reference based on a more accurate prediction value, and thus is beneficial to reduce the hardware cost of the energy storage system while avoiding complaints caused by the initial capacity not meeting customer requirements.

[0005] It can be understood that in the embodiment of the present application, the inventor of the present application found that when multiple battery cells are integrated into an energy storage system, the battery cell capacity, battery cell voltage, battery cell temperature or battery cell resistance of the multiple battery cells will affect the initial capacity of the energy storage system due to mutual influence. Therefore, in the embodiment of the present application, the initial capacity of the energy storage system is determined based on the first input characteristic diagram including the above parameter types, which is beneficial to improve the accuracy of the predicted value of the predicted initial capacity by determining the key factors affecting the initial capacity of the energy storage system.

[0006] In some embodiments, the energy storage system includes a first number of battery cells; the first input characteristic diagram includes parameter values ​​of the same parameter type of the first number of battery cells; the parameter values ​​in the first input characteristic diagram are arranged in sequence according to the size of the battery cell capacity of the first number of battery cells.

[0007] It can be understood that in the embodiment of the present application, the reason why the parameter values ​​in the first input characteristic diagram are arranged in sequence according to the size of the cell capacity of the first number of battery cells is that the inventor of the present application has found through experimental verification that arranging the parameter values ​​in the first input characteristic diagram according to the size of the cell capacity of the battery cells can improve the accuracy of the predicted value of the initial capacity of the determined energy storage system, which is beneficial for the manufacturer of the energy storage system to determine the number of battery cells required for the energy storage system to be integrated for reference based on the predicted value with higher accuracy, and further helps to ensure that the initial capacity of the energy storage system based on the integration of multiple battery cells meets customer requirements while reducing the hardware cost of the energy storage system.

[0008] In some embodiments, the constructed first input feature map is input into a pre-trained AI model to predict the initial capacity of the energy storage system, and the predicted value of the initial capacity is determined, including: before the energy storage system is integrated, the constructed first input feature map is input into a pre-trained AI model to predict the initial capacity of the energy storage system, and the predicted value of the initial capacity is determined.

[0009] It can be understood that in the embodiment of the present application, before integrating multiple battery cells into an energy storage system, the initial capacity of the energy storage system can be predicted based on multiple first input characteristic diagrams, thereby determining the predicted value of the initial capacity. In this way, it is not necessary to integrate multiple battery cells into an energy storage system before obtaining the initial capacity of the energy storage system, which is beneficial for the manufacturer of the energy storage system to determine the number of battery cells to be integrated for reference based on the predicted value with higher accuracy, and thus is beneficial to reduce the hardware cost of the energy storage system while avoiding complaints caused by the initial capacity not meeting customer requirements.

[0010] In some embodiments, the method further includes: based on the relationship between the predicted value of the initial capacity and a preset expected value, recommending a target number of battery cells required for the integrated energy storage system.

[0011] It can be understood that in the embodiment of the present application, based on the size relationship between the predicted value of the initial capacity and the preset expected value, the target number of cells required for the integrated energy storage system is recommended. In this way, the target number of cells required for the integrated energy storage system can be automatically or intelligently recommended, thereby improving the efficiency of determining the target number of cells required for the integrated energy storage system and reducing the human cost of determining the target number of cells required for the integrated energy storage system.

[0012] In some embodiments, based on the size relationship between the predicted value of the initial capacity and the preset expected value, a target number of battery cells required for the integrated energy storage system is recommended, including: when the predicted value of the initial capacity meets a first condition, the recommended target number is a first number; wherein the first number is the number of battery cells in the energy storage system; the first condition includes: the predicted value of the initial capacity is greater than the preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is within a first numerical range.

[0013] It can be understood that in the embodiment of the present application, when the predicted value of the initial capacity is greater than the preset expected value and the difference between the predicted value of the initial capacity and the preset expected value is within the first numerical range, the recommended target quantity is the first quantity. That is to say, when the portion of the predicted value of the initial capacity that is higher than the preset expected value is within the acceptable over-limit range, the recommended target quantity is the first quantity, so that the recommended target quantity is more reasonable, then, when the energy storage system is integrated based on a reasonable number of battery cells, the initial capacity of the energy storage system can meet the customer's requirements while reducing the hardware cost of the energy storage system.

[0014] In some embodiments, based on the size relationship between the predicted value of the initial capacity and a preset expected value, a target number of battery cells required for the integrated energy storage system is recommended, including: when the predicted value of the initial capacity does not meet the first condition, determining multiple second input feature graphs; wherein the second input feature graph includes parameter values ​​of a second number of battery cells, and the second number is different from the first number; the parameter types of the parameter values ​​in the second input feature graph are the same; the same element position of different second input feature graphs corresponds to different parameter types of the same battery cell; based on the multiple second input feature graphs and a pre-trained AI model, re-determining the predicted value of the initial capacity until the predicted value of the initial capacity meets the first condition, and the recommended target number is the number of battery cells corresponding to the predicted value that meets the first condition.

[0015] It can be understood that in an embodiment of the present application, when the predicted value of the initial capacity does not meet the first condition, multiple second input characteristic diagrams including parameter values ​​of a second number of battery cells are determined, and the second number is different from the first number. That is, when the predicted value of the initial capacity does not meet the first condition, the number of battery cells to be integrated in the energy storage system is updated, and the predicted value of the initial capacity is re-predicted until the predicted value of the initial capacity meets the first condition, and then the target number is recommended as the number of battery cells corresponding to the predicted value that meets the first condition. In other words, the target number of battery cells required for the integrated energy storage system is recommended only when it is ensured that the portion of the predicted value of the initial capacity that is higher than the preset expected value is within an acceptable over-limit range. In this way, the recommended target number is more reasonable, that is, the initial capacity of the energy storage system integrated based on the target number of battery cells can meet customer requirements while reducing the hardware cost of the energy storage system.

[0016] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is less than or equal to a preset expected value; and the second number is greater than the first number.

[0017] It can be understood that in the embodiment of the present application, when the predicted value of the initial capacity does not meet the preset expected value or just meets the preset expected value, the number of cells of the energy storage system to be integrated is increased, and then the predicted value of the initial capacity is re-determined based on the multiple second input feature maps of the increased number of cells and the pre-trained AI model. This is helpful to avoid the problem that the energy storage system integrated with the recommended target number of cells does not meet the customer's requirements, thereby causing customer complaints.

[0018] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is greater than a preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is less than the lower limit of the first numerical range; the second number is greater than the first number.

[0019] It can be understood that in the embodiment of the present application, when the predicted value of the initial capacity is higher than the preset expected value, but the higher part is not too much, the number of cells to be integrated into the energy storage system is increased, and then the predicted value of the initial capacity is re-determined based on the multiple second input feature maps with the increased number of cells and the pre-trained AI model. This is helpful to ensure that the energy storage system integrated with the recommended target number of cells meets the customer's requirements, thereby avoiding customer complaints.

[0020] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is greater than a preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is greater than an upper limit value of the first numerical range; the second number is less than the first number.

[0021] It can be understood that in the embodiment of the present application, when the predicted value of the initial capacity far exceeds the preset expected value, the number of cells to be integrated into the energy storage system is reduced, and then the predicted value of the initial capacity is re-determined based on the multiple second input feature maps with reduced number of cells and the pre-trained AI model. This is helpful to avoid the energy storage system with the recommended target number of cells being far higher than the customer's requirements, thereby reducing the hardware cost of the energy storage system.

[0022] In a second aspect, an embodiment of the present application provides an initial capacity prediction device for an energy storage system, the device comprising: an acquisition module, configured to acquire parameter values ​​of multiple parameter types of multiple battery cells of the energy storage system; the multiple parameter types include multiple items of the following types: battery cell capacity, battery cell voltage, battery cell temperature, and battery cell resistance; a construction module, configured to construct first input feature graphs corresponding to the multiple parameter types based on the parameter values ​​of the multiple parameter types; wherein the parameter types of the parameter values ​​in the first input feature graph are the same; and a prediction module, configured to input the constructed first input feature graph into a pre-trained AI model, predict the initial capacity of the energy storage system, and determine the predicted value of the initial capacity.

[0023] In a third aspect, an embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor implements the method of the first aspect when executing the program.

[0024] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which implements the method of the first aspect when executed by a processor.

[0025] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program or instructions, which implements the method of the first aspect when the computer program or instructions are executed by a processor.

[0026] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application and are used together with the specification to illustrate the technical solution of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0029] Figure 1 A schematic diagram of an implementation flow of a method for predicting initial capacity of an energy storage system provided in an embodiment of the present application; Figure 2 A schematic diagram of multiple first input feature maps provided in an embodiment of the present application; Figure 3 A schematic diagram of a first input feature map provided in an embodiment of the present application; Figure 4 A schematic diagram of an energy storage system provided in an embodiment of the present application; Figure 5 A schematic diagram of a mathematical model of an energy storage system provided in an embodiment of the present application; Figure 6 A schematic diagram of the structure of an AI model provided in an embodiment of the present application; Figure 7 A schematic diagram of a convolution operation provided in an embodiment of the present application; Figure 8 A schematic diagram of a pooling operation provided in an embodiment of the present application; Fig. 9 A schematic diagram of an activation function provided in an embodiment of the present application; Fig.10 A schematic diagram of a back propagation process of a convolutional neural network provided in an embodiment of the present application; Fig.11 A schematic diagram of a process flow for implementing a target number of cells required for a recommended integrated energy storage system provided in an embodiment of the present application; Fig.12 A schematic diagram of the structure of an initial capacity prediction device for an energy storage system provided in an embodiment of the present application; Fig.13 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0031] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0032] In the following description, reference is made to “some embodiments\other embodiments”, which describe a subset of all possible embodiments, but it can be understood that “some embodiments\other embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0033] In the following description, the terms "first\second" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0035] In related technology 1, a training data set is used to train a seasonal autoregressive moving average model, and the model parameters are updated to obtain a target seasonal autoregressive moving average model. The current prediction data set is input into the target seasonal autoregressive moving average model to obtain prediction curves of wind output, solar output and load. The load difference is determined based on the prediction curves of wind output, solar output and load, the target prediction error is obtained, and the energy storage capacity prediction value is determined based on the target prediction error and the load difference.

[0036] In related technology 2, a target voltage range is obtained; historical sample data of multiple battery cells within the target voltage range are obtained, and linear fitting is performed to obtain a fitting formula; a first battery cell operating voltage and a second battery cell operating voltage are selected within the target voltage range, and a difference ΔY between the state of charge (SOC) of the first battery cell and the SOC of the second battery cell is obtained according to the fitting formula; the discharge capacity of the battery cell to be tested under the first battery cell operating voltage and the second battery cell operating voltage is obtained, and the predicted capacity of the battery cell is determined according to the difference ΔC between the first discharge capacity and the second discharge capacity, and ΔY.

[0037] However, in the related technologies 1 and 2, the prediction of the process capacity of the battery cell is mainly aimed at, and the capacity of the battery cell integrated cabinet is lacking. In other words, the related technologies only provide a prediction method for the process capacity of a single battery cell, and do not provide a capacity prediction method for the cabinet; wherein multiple battery cells are integrated in the cabinet.

[0038] Among other related technologies, process capacity prediction technology solutions are mainly concentrated in the following four aspects: (1) Curve fitting prediction: By collecting the capacity values ​​of the battery cell during the initial cycle, the exponential curve is fitted with the least squares method to optimize the correlation coefficient of the curve, thereby obtaining a prediction model; (2) Machine learning model prediction: By collecting historical data of battery cells, taking all relevant parameters of the battery cells as input and capacity value as the target, supervised machine learning is performed to build a regression model. Commonly used models include random forest, extreme gradient boosting (eXtreme Gradient Boosting, XGboost), adaptive boosting (Adaptive Boosting, Adaboost), etc. (3) Neural network model prediction: By collecting historical data of battery cells, taking all relevant parameters of the battery cells and the capacity of the battery cells at the previous moment as input, and taking the battery capacity value at the next moment as the target, the recurrent neural network is used as the basic framework, and the long short-term memory network (LSTM) is used to optimize the parameters of the hidden layer of the neural network, so as to build a model for predicting capacity; (4) Fusion prediction of mechanism model and data-driven: Combine electrochemical theory to select factors that affect the battery cell, collect relevant data and capacity target value data corresponding to the characteristic factors, use machine learning models or neural network models for supervised machine learning, and build a regression prediction model. In addition, the latest physical neural network uses the differential equation model in electrochemistry as the loss function for improvement, further improving the ability of regression prediction.

[0039] The inventors of the present application have analyzed and studied the above four related technologies and found that the above four related technologies have certain problems respectively; among which, the above four related technologies have the following corresponding problems respectively: (1) For curve fitting prediction: This method implies the premise that the prediction model conforms to the exponential function, and the optimization of model parameters based on data is based on this premise. However, the assumption of the exponential function is not reliable, and the prediction accuracy is low. (2) For machine learning model predictions: the key factors derived from the model's special engineering will change due to changes in the data set, and the generalization ability and accuracy of the model are in great conflict; (3) For neural network model prediction: There is great ambiguity in how to select key features for neural network modeling, and the computational complexity of the model is relatively large; (4) For the fusion prediction of mechanism model and data-driven: When the mechanism is not well studied, the prediction effect of the model cannot meet the expectation.

[0040] The common problems of the above four related technologies are: (1) They are all technologies for predicting the capacity of a single battery cell process, and there is no description of the prediction of multiple battery cells and their mutual combination; (2) Predicting the process capacity of a single battery cell and predicting the initial capacity of a multi-cell combination are completely different problems.

[0041] Therefore, there is a huge bottleneck in the applicability of related technologies. The initial capacity test of the energy storage box (an example of an energy storage system) needs to consider the following issues: (1) Test accuracy: High-precision test equipment and standardized test procedures are required to ensure the accuracy of capacity testing; (2) Dynamic testing: The energy storage box will be affected by various working conditions during actual operation, and it is necessary to simulate the actual usage scenario for dynamic testing; (3) The capacity of the electronic stability system (ESS) cable must be sampled before the product is offline. The test time is long, the sampling frequency is high, and energy is wasted during the test process. (4) Low energy storage capacity will lead to customer complaints, and redundant design will lead to increased hardware costs; (5) Engineering practice has proved that the initial capacity of energy storage is not calculated by simply adding up the capacities of all battery cells. The factors that affect the initial capacity of an energy storage cabinet (an example of an energy storage system) are complex, and the interaction between numerous battery cells makes the prediction of initial capacity very challenging.

[0042] Based on the above analysis, it can be seen that at this stage, it is impossible to accurately calculate the initial capacity of an energy storage system integrated with multiple battery cells based on experience.

[0043] The initial capacity of an electric cabinet or electric box is generally determined by integrating multiple battery cells into the electric cabinet or electric box, and then using a dedicated measuring device to measure the initial capacity of the electric cabinet or electric box to obtain the initial capacity of the electric cabinet or electric box.

[0044] However, this method cannot determine the initial capacity of the cabinet or box before integrating multiple battery cells into the cabinet or box. Therefore, in order to meet the customer's requirements for the initial capacity of the cabinet or box, more battery cells are usually used to integrate the cabinet or box, which leads to an increase in hardware costs. However, if fewer battery cells are used to integrate the cabinet or box in order to reduce hardware costs, and because the size of the cabinet or box is fixed, it is impossible to add more battery cells to the cabinet or box, which will cause customer complaints because the initial capacity of the cabinet or box does not meet the customer's requirements.

[0045] Based on this, an embodiment of the present application provides a method for predicting the initial capacity of an energy storage system. Figure 1 A schematic diagram of an implementation flow of a method for predicting the initial capacity of an energy storage system provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method may include the following steps 101 to 103: Step 101, obtaining parameter values ​​of multiple parameter types of multiple battery cells of the energy storage system; the multiple parameter types include multiple items of the following types: battery cell capacity, battery cell voltage, battery cell temperature, and battery cell resistance; Step 102: constructing first input feature graphs corresponding to the plurality of parameter types based on the parameter values ​​of the plurality of parameter types; wherein the parameter values ​​in the first input feature graph have the same parameter type; Step 103: Input the constructed first input feature map into a pre-trained AI model to predict the initial capacity of the energy storage system and determine the predicted value of the initial capacity.

[0046] It can be understood that in an embodiment of the present application, a method is provided that can predict the initial capacity of an energy storage system before multiple battery cells are integrated into an energy storage system, filling the technical gap in predicting the initial capacity of an energy storage system. The method predicts the predicted value of the initial capacity of the energy storage system based on a first input characteristic diagram of multiple parameter values ​​of the same parameter type including multiple battery cells. In this way, there is no need to integrate multiple battery cells into an energy storage system, and the initial capacity of the energy storage system can be obtained by measurement, which is beneficial for the manufacturer of the energy storage system to determine the number of battery cells to be integrated for reference based on a more accurate prediction value, and thus is beneficial to reduce the hardware cost of the energy storage system while avoiding complaints caused by the initial capacity not meeting customer requirements.

[0047] It can be understood that in the embodiment of the present application, the inventor of the present application found that when multiple battery cells are integrated into an energy storage system, the battery cell capacity, battery cell voltage, battery cell temperature or battery cell resistance of the multiple battery cells will affect the initial capacity of the energy storage system due to mutual influence. Therefore, in the embodiment of the present application, the initial capacity of the energy storage system is determined based on the first input characteristic diagram including the above parameter types, which is beneficial to improve the accuracy of the predicted value of the predicted initial capacity by determining the key factors affecting the initial capacity of the energy storage system.

[0048] The following describes further optional implementations and related terms of each of the above steps.

[0049] In step 101, parameter values ​​of multiple parameter types of multiple battery cells of an energy storage system are obtained; the multiple parameter types include multiple items of the following types: battery cell capacity, battery cell voltage, battery cell temperature, and battery cell resistance.

[0050] It should be understood that in the embodiments of the present application, the energy storage system is not limited, and the energy storage system refers to a device for storing and releasing electrical energy, and the energy storage system is integrated by multiple battery cells. In some embodiments, the energy storage system includes an electrical box, an electrical cabinet, and the like.

[0051] It should be understood that in the embodiments of the present application, the parameter types of the battery cells are not limited; wherein the battery cell parameters are key indicators for measuring the performance and characteristics of the battery cells. In some embodiments, the parameter types also include but are not limited to at least one of the following: battery cell energy, battery cell charge and discharge rate, battery cell cycle life, battery cell self-discharge rate, battery cell operating temperature range, battery cell maximum charge current, battery cell maximum discharge current, battery cell overcharge protection voltage, battery cell over-discharge protection voltage, battery cell thermal runaway temperature, battery cell chemical type, etc.

[0052] In step 102, based on the parameter values ​​of the multiple parameter types, first input feature maps corresponding to the multiple parameter types are constructed; wherein the parameter values ​​in the first input feature map have the same parameter type.

[0053] It should be understood that in the embodiments of the present application, there is no limitation on the input feature map, which is the initial input data of the convolutional neural network, and is usually a multi-dimensional data, such as two-dimensional data or a three-dimensional array or a higher-dimensional tensor.

[0054] In some embodiments, the constructed first input characteristic graph includes one or more of the following first input characteristic graphs: a first input characteristic graph of battery cell capacity; a first input characteristic graph of battery cell voltage; a first input characteristic graph of battery cell temperature; and a first input characteristic graph of battery cell resistance.

[0055] It should be understood that in the embodiments of the present application, the first input characteristic diagram of battery cell capacity can be understood as the first input characteristic diagram including parameter values ​​of a first number of battery cell capacities; the first input characteristic diagram of battery cell voltage can be understood as the first input characteristic diagram including parameter values ​​of a first number of battery cell voltages; the first input characteristic diagram of battery cell temperature can be understood as the first input characteristic diagram including parameter values ​​of a first number of battery cell temperatures; the first input characteristic diagram of battery cell resistance can be understood as the first input characteristic diagram including parameter values ​​of a first number of battery cell resistances.

[0056] It can be understood that in the embodiment of the present application, the first input feature map constructed includes but is not limited to the following two combination schemes: Combination scheme 1: The constructed first input characteristic diagram includes: a first input characteristic diagram of battery cell capacity; a first input characteristic diagram of battery cell voltage; a first input characteristic diagram of battery cell temperature; Combination scheme 2: The constructed first input characteristic diagram includes: a first input characteristic diagram of battery cell capacity; a first input characteristic diagram of battery cell voltage; a first input characteristic diagram of battery cell temperature; and a first input characteristic diagram of battery cell resistance.

[0057] It can be understood that in the embodiment of the present application, the more the number of first input characteristic graphs, the more parameter types involved in predicting the initial capacity of the energy storage system. Then, when predicting the initial capacity of the energy storage system based on the first input characteristic graph, the more accurate the learning of the correspondence between the parameters of multiple battery cells and the initial capacity, the more accurate the predicted value of the initial capacity of the energy storage system is.

[0058] In the embodiment of the present application, there is no limitation on the number of cells of the energy storage system, and the number of cells can be greater than 1. In the embodiment of the present application, there is no limitation on the arrangement of parameter values ​​of the same parameter type of the cells of the energy storage system in the first input characteristic diagram.

[0059] In some embodiments, the energy storage system includes a first number of battery cells; the first input characteristic diagram includes parameter values ​​of the same parameter type of the first number of battery cells; the parameter values ​​in the first input characteristic diagram are arranged in sequence according to the size of the battery cell capacity of the first number of battery cells.

[0060] It can be understood that in the embodiment of the present application, the reason why the parameter values ​​in the first input characteristic diagram are arranged in sequence according to the size of the cell capacity of the first number of battery cells is that the inventor of the present application has found through experimental verification that arranging the parameter values ​​in the first input characteristic diagram according to the size of the cell capacity of the battery cells can improve the accuracy of the predicted value of the initial capacity of the determined energy storage system, which is beneficial for the manufacturer of the energy storage system to determine the number of battery cells required for the energy storage system to be integrated for reference based on the predicted value with higher accuracy, and further helps to ensure that the initial capacity of the energy storage system based on the integration of multiple battery cells meets customer requirements while reducing the hardware cost of the energy storage system.

[0061] It should be understood that in the embodiment of the present application, when the parameter values ​​in the first input characteristic diagram are arranged according to the size of the cell capacity of the first number of battery cells, the order of arrangement is not limited. In some embodiments, according to the size of the cell capacity of the first number of battery cells, the parameter values ​​in the first input characteristic diagram are arranged in order from small to large from left to right and / or from top. In other embodiments, according to the size of the cell capacity of the first number of battery cells, the parameter values ​​in the first input characteristic diagram are arranged in order from large to small from left to right and / or from top.

[0062] In some embodiments, the energy storage system includes a first number of battery cells; the first input characteristic diagram includes parameter values ​​of the same parameter type of the first number of battery cells; the parameter values ​​in the first input characteristic diagram are arranged according to the planned arrangement positions of the first number of battery cells.

[0063] In the embodiment of the present application, the planned arrangement position can be understood as the actual arrangement position of the first number of battery cells when they are integrated into the energy storage system. It should be understood that in the embodiment of the present application, the planned arrangement position of the first number of battery cells is not limited, that is, the actual arrangement position of the first number of battery cells is not limited, and the actual arrangement position of the first number of battery cells can be arranged arbitrarily.

[0064] In the embodiment of the present application, “the arrangement of the parameter values ​​in the first input characteristic diagram” does not affect “the arrangement of the first number of battery cells when the first number of battery cells are integrated into an energy storage system”. That is to say, even if the parameter values ​​in the first input characteristic diagram are arranged in sequence according to the size of the battery cell capacity of the first number of battery cells, when the first number of battery cells are integrated into the energy storage system, the first number of battery cells can be arranged arbitrarily to integrate the energy storage system.

[0065] For example, in one possible implementation, there are multiple types of parameters for a single battery cell, such as battery cell voltage, battery cell temperature, battery cell capacity, battery cell resistance, etc. The energy storage system is composed of hundreds of battery cells, thus forming a high-dimensional array. Each row of the array represents a battery cell with a different battery cell number (unique identification of the battery cell), and each column of the array identifies a dimension of the battery cell parameter, such as battery cell voltage. The mathematical model of the energy storage system can be abstracted as a "cube" (a three-dimensional data matrix), such as Figure 2 As shown, 201 and 202 represent cells with different cell numbers, and each "layer" from the outside to the inside represents a type of cell parameter identification, such as 203 represents the cell voltage, and 204 represents the cell capacity; each "layer" (i.e., an example of a first input feature map) is arranged from left to right and from top to bottom in order in all the cells in the energy storage system. The arrangement of all the cells on the "layer" needs to be arranged from left to right and from top to bottom in order according to the size of the cell capacity. Figure 3 A schematic diagram of a first input feature map provided in an embodiment of the present application, such as Figure 3 As shown, 301 represents the parameter value of a cell in the first input characteristic graph, and Figure 3 The parameter values ​​of each cell shown in are of the same parameter type.

[0066] In step 103, the constructed first input feature map is input into a pre-trained AI model to predict the initial capacity of the energy storage system and determine the predicted value of the initial capacity.

[0067] In the embodiments of the present application, there is no limitation on the pre-trained AI model, and the pre-trained AI model includes a machine learning model and / or a deep learning model. In some embodiments, the pre-trained machine learning model includes a random forest, a support vector machine, a gradient boosting regression, an extreme gradient boosting, an adaptive enhancement, etc. In some embodiments, the pre-trained deep learning model includes but is not limited to at least one of the following: a recurrent convolutional neural network, a generative adversarial convolutional neural network, a long short-term memory network, a feedforward neural network, a multi-layer perceptron, a convolutional neural network including an attention mechanism, a deep belief network, etc. In the embodiments of the present application, there is no limitation on the training method of the pre-trained AI model, and there is no limitation on the platform on which the pre-trained AI model is deployed.

[0068] In some embodiments, the structure of the pre-trained AI model includes a first convolutional layer, a first activation layer, a second convolutional layer, a second activation layer, a first pooling layer, a third convolutional layer, a third activation layer, a fourth convolutional layer, a fourth activation layer, a second pooling layer and a fully connected layer, which are connected in sequence.

[0069] It can be understood that in the embodiments of the present application, the inventors of the present application have experimentally verified that setting the structure of the AI ​​model to a first convolutional layer, a first activation layer, a second convolutional layer, a second activation layer, a first pooling layer, a third convolutional layer, a third activation layer, a fourth convolutional layer, a fourth activation layer, a second pooling layer and a fully connected layer connected in sequence is beneficial to improving the accuracy of the predicted value of the predicted initial capacity.

[0070] It should be understood that in the embodiments of the present application, the size of the convolution kernel, the number of convolution kernels, the step size, the number of padding, the number of channels, and other parameters in the first convolution layer, the second convolution layer, the third convolution layer, and the fourth convolution layer are not limited. In some embodiments, the number of convolution kernels in the first convolution layer, the second convolution layer, and the third convolution layer is different, the number of channels in the first convolution layer and the second convolution layer is the same, and the number of channels in the third convolution layer and the fourth convolution layer is the same.

[0071] In the embodiment of the present application, the activation functions used by the first activation layer, the second activation layer, the third activation layer, and the fourth activation layer are not limited, and the activation functions used by the first activation layer, the second activation layer, the third activation layer, and the fourth activation layer may be the same or different. In some embodiments, the activation functions used by each activation layer include, but are not limited to, at least one of the following: a sigmoid function, a hyperbolic tangent function (Tanh function), a linear rectifier function (ReLU function), a leaky ReLU function, a variant of a ReLU function, etc.

[0072] It should be understood that in the embodiments of the present application, the pooling operations used by the first pooling layer and the second pooling layer are not limited, and the pooling operations used by the first pooling layer and the second pooling layer may be the same or different. In some embodiments, the pooling operations used by each pooling layer include but are not limited to at least one of the following: Max Pooling, Average Pooling, Global Max Pooling, Global Average Pooling, L2 Pooling, Overlapping Pooling, Stochastic Pooling, Mixed Pooling, etc.

[0073] For example, in one possible implementation, Figure 4 A schematic diagram of an energy storage system provided in an embodiment of the present application, such as Figure 4 As shown, 401 represents a single battery cell. Figure 5 A schematic diagram of a mathematical model of an energy storage system provided in an embodiment of the present application, such as Figure 5 As shown, the X-axis and Y-axis represent the cell position, and the Z-axis represents the cell parameter type (i.e., the cell key parameters in the figure). Based on this mathematical model, a three-dimensional matrix of the energy storage system can be obtained, for example: [[[3.287911, 313361.2, 28.897, 1366.3, 420.31], ... [3.288811, 313470.9, 31.283, 1366.3, 420.31]] ... [[3.287911, 313361.2, 28.897, 1366.3, 420.31], ... [3.288811, 313470.9, 31.283, 1366.3, 420.31]]]. The parameters of the battery cells are sorted from left to right and from top to bottom according to the capacity of the battery cells of the single battery cells, and the three-dimensional matrix is ​​reconstructed to obtain a three-dimensional matrix of [20, 20, 5] (i.e., an example of the first input feature map).

[0074] Figure 6 A schematic diagram of the structure of an AI model provided in an embodiment of the present application is shown in FIG. Figure 6As shown in the figure, the constructed three-dimensional matrix [20, 20, 5] is input into the convolution layer with a convolution kernel size of 3×3, a number of convolution kernels of 5, a step size of 1, a padding number of 1, a number of channels of 64, and an activation layer with an activation function of ReLU, and the feature matrix is ​​[20, 20, 64]; the feature matrix [20, 20, 64] is input into the convolution layer with a convolution kernel size of 3×3, a number of convolution kernels of 64, a step size of 1, a padding number of 1, a number of channels of 64, and an activation layer with an activation function of ReLU, and the feature matrix is ​​[20, 20, 64]; the feature matrix [20, 20, 64] is input into the pooling layer with a pooling window of 3×3, and a maximum pooling operation is performed, and the output feature matrix is ​​[18, 18, 64]; the feature matrix [18, 18, 64] is input into The feature matrix [18, 18, 128] is input into a convolution layer with a convolution kernel size of 3×3, a number of convolution kernels of 5, a stride of 1, a padding number of 1, a number of channels of 128, and an activation layer with an activation function of ReLU, and the feature matrix [18, 18, 128] is obtained; the feature matrix [18, 18, 128] is input into a convolution layer with a convolution kernel size of 3×3, a number of convolution kernels of 64, a stride of 1, a padding number of 1, a number of channels of 128, and an activation layer with an activation function of ReLU, and the feature matrix [18, 18, 128] is obtained; the feature matrix [18, 18, 128] is input into a pooling layer with a pooling window of 3×3, and a maximum pooling operation is performed to obtain a feature matrix [16, 16, 128]; the feature matrix [16, 16, 128] is input into the fully connected layer to obtain the predicted value.

[0075] Figure 7 A schematic diagram of a convolution operation provided in an embodiment of the present application is shown in FIG. Figure 7 As shown, 701 represents input, 702 represents convolution kernel, and 703 represents output result. Figure 8 A schematic diagram of a pooling operation provided in an embodiment of the present application is shown in FIG. Figure 8 As shown, 801 represents input, 802 represents a maximum pooling operation (Max Pooling) with a window size of 2×2, and 803 represents the output of the maximum pooling operation. Fig. 9 A schematic diagram of an activation function provided in an embodiment of the present application, such as Fig. 9 As shown, when the input z is less than or equal to 0, the output is 0; when the input z is greater than 0, the output is z; wherein the horizontal axis represents the input value and the vertical axis represents the output value.

[0076] Fig.10 A schematic diagram of the back propagation process of a convolutional neural network provided in an embodiment of the present application, such as Fig.10As shown in the figure, the convolution kernel matrix is ​​a matrix used to extract features in a convolutional neural network. The convolution kernel slides over the input data and generates a feature map through a convolution operation. The true value is the target output, and the predicted value is the output generated by the model based on the input data. The difference refers to the difference between the true value and the predicted value, which is used to calculate the loss function and then perform backpropagation. The learning rate coefficient is a hyperparameter in the optimization algorithm that controls the step size of the weight update. The gradient is the partial derivative of the loss function with respect to the model parameters. The calculation formula for the gradient is:

[0077] in, It is l +1 level The convolution kernels are located at The weight of is the loss function E relative to the l +1 activation value In position ( ) and The gradient of the convolution kernel, is the activation function in The derivative at represents the size of the convolution kernel, represents the step length (i.e. stride), Indicates l The number of convolution kernels of the layer.

[0078] The update formula of the convolution kernel is:

[0079] in, It is l The weight of the layer, It is l +1 layer weight, represents the learning rate, Represents the loss function E relative to the k Convolution kernel weights The gradient of Indicates l +1 layer activation value, Represents the loss function E relative to the l +1 activation value gradient.

[0080] In some embodiments, the learning rate factor is set to 0.1.

[0081] In some embodiments, the constructed first input feature map is input into a pre-trained AI model to predict the initial capacity of the energy storage system, and the predicted value of the initial capacity is determined, including: before the energy storage system is integrated, the constructed first input feature map is input into a pre-trained AI model to predict the initial capacity of the energy storage system, and the predicted value of the initial capacity is determined.

[0082] It can be understood that in the embodiment of the present application, before integrating multiple battery cells into an energy storage system, the initial capacity of the energy storage system can be predicted based on multiple first input characteristic diagrams, thereby determining the predicted value of the initial capacity. In this way, it is not necessary to integrate multiple battery cells into an energy storage system before obtaining the initial capacity of the energy storage system, which is beneficial for the manufacturer of the energy storage system to determine the number of battery cells to be integrated for reference based on the predicted value with higher accuracy, and thus is beneficial to reduce the hardware cost of the energy storage system while avoiding complaints caused by the initial capacity not meeting customer requirements.

[0083] In some embodiments, the energy storage system initial capacity prediction method further includes: based on the size relationship between the predicted value of the initial capacity and a preset expected value, recommending a target number of battery cells required for the integrated energy storage system.

[0084] It can be understood that in the embodiment of the present application, based on the size relationship between the predicted value of the initial capacity and the preset expected value, the target number of cells required for the integrated energy storage system is recommended. In this way, the target number of cells required for the integrated energy storage system can be automatically or intelligently recommended, thereby improving the efficiency of determining the target number of cells required for the integrated energy storage system and reducing the human cost of determining the target number of cells required for the integrated energy storage system.

[0085] In some embodiments, based on the size relationship between the predicted value of the initial capacity and the preset expected value, a target number of battery cells required for the integrated energy storage system is recommended, including: when the predicted value of the initial capacity meets a first condition, the recommended target number is a first number; wherein the first number is the number of battery cells in the energy storage system; the first condition includes: the predicted value of the initial capacity is greater than the preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is within a first numerical range.

[0086] It can be understood that in the embodiment of the present application, when the predicted value of the initial capacity is greater than the preset expected value and the difference between the predicted value of the initial capacity and the preset expected value is within the first numerical range, the recommended target quantity is the first quantity. That is to say, when the portion of the predicted value of the initial capacity that is higher than the preset expected value is within the acceptable over-limit range, the recommended target quantity is the first quantity, so that the recommended target quantity is more reasonable, then, when the energy storage system is integrated based on a reasonable number of battery cells, the initial capacity of the energy storage system can meet the customer's requirements while reducing the hardware cost of the energy storage system.

[0087] It should be understood that in the embodiments of the present application, the first numerical range is not limited. In some embodiments, the first numerical range is preset. In other embodiments, the first numerical range can be determined based on a variety of factors such as customer preferences, application scenarios of the energy storage system, the price of the energy storage system, the prediction accuracy of a pre-trained AI model, or the prediction error of a pre-trained AI model. In one possible implementation, the first numerical range is [5, 10].

[0088] In some embodiments, Fig.11 A schematic diagram of a process flow for implementing a target number of cells required for a recommended integrated energy storage system according to an embodiment of the present application is shown in FIG. Fig.11 As shown, the target number of cells required for the integrated energy storage system can be recommended through the following steps 1101 and 1102: Step 1101, when the predicted value of the initial capacity does not satisfy the first condition, determine a plurality of second input characteristic graphs; wherein the second input characteristic graphs include parameter values ​​of a second number of battery cells, the second number being different from the first number; the parameter types of the parameter values ​​in the second input characteristic graphs are the same; and the same element position in different second input characteristic graphs corresponds to different parameter types of the same battery cell; Step 1102, based on multiple second input feature maps and a pre-trained AI model, re-determine the predicted value of the initial capacity until the predicted value of the initial capacity meets the first condition, and the recommended target number is the number of battery cells corresponding to the predicted value that meets the first condition.

[0089] It can be understood that in an embodiment of the present application, when the predicted value of the initial capacity does not meet the first condition, multiple second input characteristic diagrams including parameter values ​​of a second number of battery cells are determined, and the second number is different from the first number. That is, when the predicted value of the initial capacity does not meet the first condition, the number of battery cells to be integrated in the energy storage system is updated, and the predicted value of the initial capacity is re-predicted until the predicted value of the initial capacity meets the first condition, and then the target number is recommended as the number of battery cells corresponding to the predicted value that meets the first condition. In other words, the target number of battery cells required for the integrated energy storage system is recommended only when it is ensured that the portion of the predicted value of the initial capacity that is higher than the preset expected value is within an acceptable over-limit range. In this way, the recommended target number is more reasonable, that is, the initial capacity of the energy storage system integrated based on the target number of battery cells can meet customer requirements while reducing the hardware cost of the energy storage system.

[0090] It should be understood that in the embodiments of the present application, when the predicted value of the initial capacity does not meet the first condition, the method for determining the number of battery cells of the energy storage system to be integrated is not limited. In some embodiments, when the predicted value of the initial capacity does not meet the first condition, the change in the number of battery cells of the energy storage system to be integrated is determined; based on the change, the number of battery cells of the energy storage system to be integrated is determined.

[0091] In some embodiments, when the predicted value of the initial capacity does not satisfy the first condition, determining the change in the number of battery cells of the energy storage system to be integrated includes: when the predicted value of the initial capacity does not satisfy the first condition, determining the change in the number of battery cells of the energy storage system to be integrated to be a preset change.

[0092] It should be understood that in the embodiments of the present application, the preset amount of change is not limited. In some embodiments, the preset amount of change can be determined based on an empirical value. In other embodiments, the preset amount of change can be determined based on the difference between the predicted value of the initial capacity and the preset expected value, the relationship between the difference and the first numerical range, and the cell capacity of the single cell. In one possible implementation, the preset amount of change is 1.

[0093] In other embodiments, when the predicted value of the initial capacity does not meet the first condition, the change in the number of battery cells to be integrated in the energy storage system is determined based on the difference between the predicted value of the initial capacity and a preset expected value, the relationship between the difference and the first numerical range, and the battery cell capacity of a single battery cell.

[0094] Exemplarily, in one possible implementation, the predicted value of the initial capacity is 98, the preset expected value is 100, the capacity value of a single cell is 1, the first numerical range is [5, 10], and the change in the number of cells of the energy storage system to be integrated is determined to be any integer between 7 and 12.

[0095] It should be understood that in the embodiment of the present application, "until the predicted value of the initial capacity meets the first condition, the recommended target number is the number of battery cells corresponding to the predicted value that meets the first condition" can be understood as: when the re-determined predicted value of the initial capacity meets the first condition, the recommended target number is the second number; when the re-determined initial capacity does not meet the first condition, the second input feature map is used as the first input feature map, and when the predicted value of the initial capacity does not meet the first condition, multiple second input feature maps are determined, and the steps of re-determining the predicted value of the initial capacity based on the multiple second input feature maps and the pre-trained AI model are executed.

[0096] Exemplarily, in one possible implementation, the first input feature graph includes parameter values ​​of 100 battery cells, the predicted value of the initial capacity is 98, the preset expected value is 100, the capacity value of a single battery cell is 1, the first numerical range is [5, 10], and the determined second input feature graph includes parameter values ​​of 107 (i.e., an example of the second number) battery cells. Based on multiple second input feature graphs and a pre-trained AI model, the re-determined predicted value of the initial capacity is 104. 104 does not meet the first condition, and the second input feature graph is used as the first input feature graph, and the second input feature graph is re-determined, and the re-determined second input feature graph includes parameter values ​​of 110 battery cells. Based on multiple second input feature graphs and a pre-trained AI model, the re-determined predicted value of the initial capacity is 107, which meets the first condition, and the recommended target number is 107.

[0097] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is less than or equal to a preset expected value; and the second number is greater than the first number.

[0098] It can be understood that in the embodiment of the present application, when the predicted value of the initial capacity does not meet the preset expected value or just meets the preset expected value, the number of cells of the energy storage system to be integrated is increased, and then the predicted value of the initial capacity is re-determined based on the multiple second input feature maps of the increased number of cells and the pre-trained AI model. This is helpful to avoid the problem that the energy storage system integrated with the recommended target number of cells does not meet the customer's requirements, thereby causing customer complaints.

[0099] For example, in a possible implementation, when the predicted value of the initial capacity is 98 and the preset expected value is 100, the number of cells in the energy storage system to be integrated is increased.

[0100] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is greater than a preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is less than the lower limit of the first numerical range; the second number is greater than the first number.

[0101] It can be understood that in the embodiment of the present application, when the predicted value of the initial capacity is higher than the preset expected value, but the higher part is not too much, the number of cells to be integrated into the energy storage system is increased, and then the predicted value of the initial capacity is re-determined based on the multiple second input feature maps with the increased number of cells and the pre-trained AI model. This is helpful to ensure that the energy storage system integrated with the recommended target number of cells meets the customer's requirements, thereby avoiding customer complaints.

[0102] For example, in a possible implementation, when the predicted value of the initial capacity is 102 and the preset expected value is 100, the number of cells in the energy storage system to be integrated is increased.

[0103] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is greater than a preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is greater than an upper limit value of the first numerical range; the second number is less than the first number.

[0104] It can be understood that in the embodiment of the present application, when the predicted value of the initial capacity far exceeds the preset expected value, the number of cells to be integrated into the energy storage system is reduced, and then the predicted value of the initial capacity is re-determined based on the multiple second input feature maps with reduced number of cells and the pre-trained AI model. This is helpful to avoid the energy storage system with the recommended target number of cells being far higher than the customer's requirements, thereby reducing the hardware cost of the energy storage system.

[0105] For example, in a possible implementation, when the predicted value of the initial capacity is 118 and the preset expected value is 100, the number of cells to be integrated into the energy storage system is reduced.

[0106] Based on this, an exemplary application of an embodiment of the present application in a practical application scenario will be described below.

[0107] An embodiment of the present application provides a method for predicting the initial capacity of an energy storage system. After a plurality of battery cells are integrated into a battery cabinet or a battery box, it is difficult to determine the initial capacity due to the mutual influence between the plurality of battery cells. The method can screen the key factors that affect the initial capacity of the battery cabinet or the battery box.

[0108] A convolutional neural network framework is used to extract features of the parameters involved in each battery cell in a group of battery cells, that is, the relevant parameters affecting the initial capacity, so as to solve the problem of selecting the key factors affecting the initial capacity of the energy storage system after the multi-battery cell integrated energy storage system.

[0109] There are two key steps in the calculation of the convolution layer, one is local association and the other is window sliding. Local association is to regard each neuron as a filter, and window sliding is the filter calculating the local data. The implementation of convolution involves three parameters: stride (an example of step length), filter size and padding. The calculation process of convolution is that the stride controls the filter to perform convolution operations. For example, the size of the input image is 7x7, the filter size is 3x3, and the stride is 1. The change of stride causes the change of receptive field. In biology, the receptive field usually refers to the area where a certain perceptual neuron in the visual system (such as retinal neurons or visual neurons in the cerebral cortex) responds to light stimulation in a certain part of the visual field. In the field of computer science, the receptive field is used to describe which area of ​​the input image a neuron in a certain layer of the network is sensitive to. The size of the receptive field determines how much contextual information the neuron can capture. The larger the convolution kernel (for example: 3×3), the larger the receptive field and the more contextual information captured. Convolution calculations will reduce the spatial dimension of the data. In the early layers of the network, if you want to retain more information about the original input content and facilitate the extraction of underlying features, you can use padding. For example, for a 32×32×3 input image, a 5×5×5 filter is used for convolution, and the output is 28×28×3. In order to make the output image dimension 32×32×3, you can use 0 to pad the outside of the input image. After applying zero padding of size 2, the input image is updated to 36×36×3, so that a 5×5×3 filter can still get an output of 32×32×3.

[0110] The function of the activation layer is to perform nonlinear mapping on the convolution result; the pooling layer is sandwiched between consecutive convolution layers to compress the amount of data and parameters and reduce overfitting. In short, if the input is an image, the main function of the pooling layer is to compress the image.

[0111] The specific functions of the pooling layer include feature invariance and feature dimensionality reduction. Feature invariance can be understood as removing only some irrelevant information when compressing an image, while retaining important features. For example, after a picture of a puppy is reduced in size, we can still recognize that it is a dog. Feature dimensionality reduction is to remove redundant features and extract important features. This can prevent overfitting to a certain extent and facilitate optimization.

[0112] A fully connected layer means that all neurons between two layers are connected by weights. Usually, the fully connected layer is at the end of the convolutional neural network. This is the same as the way neurons are connected in traditional neural networks.

[0113] It can be understood that in the embodiment of the present application, the key factors affecting the initial capacity of the electric cabinet or box are extracted, and the key parameters of the single battery cell and the information of the coupling of multiple batteries are integrated to perform sufficient feature extraction without the guidance of the mechanism. The parameters of the neural network are continuously optimized through the reverse calculation of gradient descent, thereby realizing supervised learning and realizing the prediction of the initial capacity of the electric cabinet or box.

[0114] In one possible implementation, there are multiple types of parameters for a single cell, such as cell voltage, cell temperature, cell capacity, cell resistance, etc. The energy storage system is composed of hundreds of cells, thus forming a high-dimensional array. Each row of the array represents a cell with a different cell number (cell unique identifier), and each column of the array identifies a dimension (i.e., a certain type) of the cell parameter, such as cell voltage. The mathematical model of the energy storage system can be abstracted as a "cube" (a three-dimensional data matrix), such as Figure 2 As shown, 201 and 202 represent cells with different cell numbers, and each "layer" from the outside to the inside represents a type of cell parameter identification, such as 203 represents the cell voltage, and 204 represents the cell capacity; all cells in the energy storage system are arranged in order from left to right and from top to bottom on each "layer". The arrangement of all cells on the "layer" needs to be arranged from left to right and from top to bottom according to the size of the cell capacity.

[0115] It can be understood that in the embodiment of the present application, supervised machine learning feature extraction is implemented in an environment where complex systems (i.e., an example of an energy storage system) are mutually coupled, which solves the bottleneck that related machine learning models and deep learning models cannot find effective feature factors. This method has a more effective prediction effect in a complex interactive environment.

[0116] In one possible implementation, Figure 4 A schematic diagram of an energy storage system provided in an embodiment of the present application, such as Figure 4 As shown, 401 represents a single battery cell. Figure 5 A schematic diagram of a mathematical model of an energy storage system provided in an embodiment of the present application, such as Figure 5As shown, the X-axis and Y-axis represent the cell position, and the Z-axis represents the cell parameter type (i.e., the cell key parameters in the figure). Based on this mathematical model, a three-dimensional matrix of the energy storage system can be obtained, for example: [[[3.287911, 313361.2, 28.897, 1366.3, 420.31], ... [3.288811, 313470.9, 31.283, 1366.3, 420.31]] ... [[3.287911, 313361.2, 28.897, 1366.3, 420.31], ... [3.288811, 313470.9, 31.283, 1366.3, 420.31]]]. The parameters of the battery cells are sorted from left to right and from top to bottom according to the capacity of the battery cells of the single battery cells, and the three-dimensional matrix is ​​reconstructed to obtain a three-dimensional matrix of [20, 20, 5] (i.e., an example of the first input feature map).

[0117] Figure 6 A schematic diagram of the structure of an AI model provided in an embodiment of the present application is shown in FIG. Figure 6 As shown, the constructed three-dimensional matrix [20, 20, 5] is input into a convolution layer with a convolution kernel size of 3×3, a number of convolution kernels of 5, a step size of 1, a padding number of 1, a number of channels of 64, and an activation layer with an activation function of ReLU, and the feature matrix is ​​[20, 20, 64]; the feature matrix [20, 20, 64] is input into a convolution layer with a convolution kernel size of 3×3, a number of convolution kernels of 64, a step size of 1, a padding number of 1, a number of channels of 64, and an activation layer with an activation function of ReLU, and the feature matrix is ​​[20, 20, 64]; the feature matrix [20, 20, 64] is input into a pooling layer with a pooling window of 3×3, and a maximum pooling operation is performed, and the output feature matrix is ​​[18, 18, 64]; the feature matrix [18, 18, 64] is input The convolution kernel size is 3×3, the number of convolution kernels is 128, the stride is 1, the padding number is 1, the number of channels is 128, and the activation function is ReLU. The feature matrix [18, 18, 128] is input into the convolution kernel size is 3×3, the number of convolution kernels is 128, the stride is 1, the padding number is 1, the number of channels is 128, and the activation function is ReLU. The feature matrix [18, 18, 128] is input into the convolution kernel size is 3×3, the number of convolution kernels is 128, the stride is 1, the padding number is 1, the number of channels is 128, and the activation function is ReLU. The feature matrix [18, 18, 128] is input into the pooling layer with a pooling window of 3×3, and the maximum pooling operation is performed to obtain the feature matrix [16, 16, 128]; the feature matrix [16, 16, 128] is input into the fully connected layer to obtain the predicted value.

[0118] Fig.10A schematic diagram of the back propagation process of a convolutional neural network provided in an embodiment of the present application, such as Fig.10 As shown in the figure, the convolution kernel matrix is ​​a matrix used to extract features in a convolutional neural network. The convolution kernel slides over the input data and generates a feature map through a convolution operation. The true value is the target output, and the predicted value is the output generated by the model based on the input data. The difference refers to the difference between the true value and the predicted value, which is used to calculate the loss function and then perform backpropagation. The learning rate coefficient is a hyperparameter in the optimization algorithm that controls the step size of the weight update. The gradient is the partial derivative of the loss function with respect to the model parameters.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical solution of the present application can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present application, which should be included in the scope of protection of the present application.

[0120] It should be noted that although the steps of the method in the present application are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step for execution, and / or one step may be decomposed into multiple steps for execution, etc.; or, the steps in different embodiments may be combined into a new technical solution. Based on the foregoing embodiments, an embodiment of the present application provides a device, which includes the modules included and the units included in the modules, which can be implemented by a processor; of course, it can also be implemented by a specific logic circuit; in the implementation process, the processor may be an AI acceleration engine (such as NPU, etc.), a graphics processor (GPU), a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP) or a field programmable gate array (FPGA), etc.

[0121] Fig.12 A schematic diagram of the structure of an energy storage system initial capacity prediction device provided in an embodiment of the present application is shown in FIG. Fig.12 As shown, the energy storage system initial capacity prediction device 120 includes an acquisition module 1201, a construction module 1202 and a prediction module 1203, wherein: The acquisition module 1201 is configured to acquire parameter values ​​of multiple parameter types of multiple battery cells of the energy storage system; the multiple parameter types include multiple items of the following types: battery cell capacity, battery cell voltage, battery cell temperature, and battery cell resistance; A construction module 1202 is configured to construct first input feature maps corresponding to the plurality of parameter types based on the parameter values ​​of the plurality of parameter types; wherein the parameter values ​​in the first input feature map have the same parameter type; The prediction module 1203 is configured to input the constructed first input feature map into a pre-trained AI model, predict the initial capacity of the energy storage system, and determine the predicted value of the initial capacity.

[0122] In some embodiments, the energy storage system includes a first number of battery cells; the first input characteristic diagram includes parameter values ​​of the same parameter type of the first number of battery cells; the parameter values ​​in the first input characteristic diagram are arranged in sequence according to the size of the battery cell capacity of the first number of battery cells.

[0123] In some embodiments, the prediction module 1203 is configured to input the constructed first input feature map into a pre-trained AI model before integrating the energy storage system, predict the initial capacity of the energy storage system, and determine the predicted value of the initial capacity.

[0124] In some embodiments, the energy storage system initial capacity prediction device 120 further includes a recommendation module; wherein the recommendation module is configured to recommend a target number of battery cells required for the integrated energy storage system based on the size relationship between the predicted value of the initial capacity and a preset expected value.

[0125] In some embodiments, the recommendation module is configured to recommend a first target quantity as a first quantity when the predicted value of the initial capacity satisfies a first condition; wherein the first quantity is the number of battery cells in the energy storage system; and the first condition includes: the predicted value of the initial capacity is greater than a preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is within a first numerical range.

[0126] In some embodiments, the recommendation module is configured to determine multiple second input feature maps when the predicted value of the initial capacity does not meet the first condition; wherein the second input feature map includes parameter values ​​of a second number of battery cells, and the second number is different from the first number; the parameter types of the parameter values ​​in the second input feature map are the same; the same element position of different second input feature maps corresponds to different parameter types of the same battery cell; based on the multiple second input feature maps and the pre-trained AI model, the predicted value of the initial capacity is re-determined until the predicted value of the initial capacity meets the first condition, and the recommended target number is the number of battery cells corresponding to the predicted value that meets the first condition.

[0127] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is less than or equal to a preset expected value; and the second number is greater than the first number.

[0128] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is greater than a preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is less than the lower limit of the first numerical range; the second number is greater than the first number.

[0129] In some embodiments, the predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is greater than a preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is greater than an upper limit value of the first numerical range; the second number is less than the first number.

[0130] The description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the device embodiment of the present application, please refer to the description of the method embodiment of the present application for understanding.

[0131] It should be noted that the division of modules in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional unit in each embodiment of the present application may be integrated into a processing unit, or may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit. It may also be implemented in the form of a combination of software and hardware.

[0132] It should be noted that in the embodiments of the present application, if the above method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application can be essentially or partly reflected in the form of a software product, which is stored in a storage medium and includes several instructions for enabling an electronic device to execute all or part of the methods of each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk. In this way, the embodiments of the present application are not limited to any specific combination of hardware and software.

[0133] An embodiment of the present application provides an electronic device, Fig.13 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application is shown in FIG. Fig.13 As shown, the electronic device 130 includes a memory 1301 and a processor 1302. The memory 1301 stores a computer program that can be run on the processor 1302. When the processor 1302 executes the program, the steps in the method provided in the above embodiment are implemented.

[0134] It should be noted that the memory 1301 is configured to store instructions and applications executable by the processor 1302, and can also cache data to be processed or processed by the processor 1302 and various modules in the electronic device 130, which can be implemented through flash memory (FLASH) or random access memory (Random Access Memory, RAM).

[0135] In the embodiments of the present application, there is no limitation on the type of electronic device, and the electronic device may be a smart phone, a laptop computer, a tablet computer, a smart bracelet, an Internet of Things (IoT) device, or a vehicle-mounted device, etc.

[0136] An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps in the method provided in the above embodiment are implemented.

[0137] An embodiment of the present application provides a computer program product including instructions, which, when executed on a computer, enables the computer to execute the steps of the method provided in the above method embodiment.

[0138] It should be noted here that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the description of the method embodiments of this application for understanding.

[0139] It should be understood that "one embodiment" or "an embodiment" or "some embodiments" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in one embodiment" or "in some embodiments" appearing throughout the specification may not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the size of the sequence number of the above-mentioned processes does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application. The above-mentioned sequence numbers of the embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments. The above description of each embodiment tends to emphasize the differences between the various embodiments, and the same or similar aspects can be referenced to each other. For the sake of brevity, this article will not repeat them.

[0140] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there may be three relationships. For example, object A and / or object B can represent three situations: object A exists alone, object A and object B exist at the same time, and object B exists alone.

[0141] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0142] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The embodiments described above are only schematic. For example, the division of modules is only a logical function division. There may be other division methods in actual implementation, such as: multiple modules or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or modules can be electrical, mechanical or other forms.

[0143] The modules described above as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules; they may be located in one place or distributed on multiple network units; some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiment of the present application.

[0144] In addition, all functional modules in the embodiments of the present application may be integrated into one processing unit, or each module may be a separate unit, or two or more modules may be integrated into one unit; the above-mentioned integrated modules may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0145] A person skilled in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, read-only memories (ROM), magnetic disks or optical disks.

[0146] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application can essentially or in other words, the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a display device or a cloud device to execute all or part of the methods of each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0147] The methods disclosed in several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0148] The features disclosed in several product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0149] The features disclosed in several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0150] The above are only implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be covered by the protection scope of the present application.

Claims

1. A method for predicting the initial capacity of an energy storage system, characterized in that: The method comprises: Obtain parameter values ​​of multiple parameter types of multiple battery cells of the energy storage system; the multiple parameter types include multiple items of the following types: battery cell capacity, battery cell voltage, battery cell temperature, and battery cell resistance; Based on the parameter values ​​of the multiple parameter types, construct first input feature maps corresponding to the multiple parameter types respectively; wherein the parameter values ​​in the first input feature maps have the same parameter type; The constructed first input feature map is input into a pre-trained AI model to predict the initial capacity of the energy storage system and determine the predicted value of the initial capacity.

2. The method according to claim 1, characterized in that The energy storage system includes a first number of battery cells; the first input characteristic diagram includes parameter values ​​of the same parameter type of the first number of battery cells; the parameter values ​​in the first input characteristic diagram are arranged in sequence according to the size of the battery cell capacity of the first number of battery cells.

3. The method according to claim 1 or 2, characterized in that: The step of inputting the constructed first input feature graph into a pre-trained AI model, predicting the initial capacity of the energy storage system, and determining the predicted value of the initial capacity includes: Before integrating the energy storage system, the constructed first input feature map is input into a pre-trained AI model to predict the initial capacity of the energy storage system and determine the predicted value of the initial capacity.

4. The method according to claim 3, characterized in that The method further comprises: Based on the relationship between the predicted value of the initial capacity and the preset expected value, a target number of battery cells required for integrating the energy storage system is recommended.

5. The method according to claim 4, characterized in that The recommending a target number of cells required for integrating the energy storage system based on the relationship between the predicted value of the initial capacity and the preset expected value includes: When the predicted value of the initial capacity satisfies a first condition, it is recommended that the target quantity is a first quantity; wherein the first quantity is the number of battery cells in the energy storage system; the first condition includes: the predicted value of the initial capacity is greater than the preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is within a first numerical range.

6. The method according to claim 5, characterized in that The recommending a target number of cells required for integrating the energy storage system based on the relationship between the predicted value of the initial capacity and the preset expected value includes: In the case where the predicted value of the initial capacity does not satisfy the first condition, a plurality of second input characteristic graphs are determined; wherein the second input characteristic graphs include parameter values ​​of a second number of battery cells, the second number being different from the first number; the parameter values ​​in the second input characteristic graphs have the same parameter type; and the same element position in different second input characteristic graphs corresponds to different parameter types of the same battery cell; Based on the multiple second input feature maps and the pre-trained AI model, the predicted value of the initial capacity is re-determined until the predicted value of the initial capacity meets the first condition, and the target number is recommended to be the number of battery cells corresponding to the predicted value that meets the first condition.

7. The method according to claim 6, characterized in that The predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is less than or equal to the preset expected value; and the second quantity is greater than the first quantity.

8. The method according to claim 6, characterized in that The predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is greater than the preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is less than the lower limit of the first numerical range; the second quantity is greater than the first quantity.

9. The method according to claim 6, characterized in that The predicted value of the initial capacity does not satisfy the first condition including: the predicted value of the initial capacity is greater than the preset expected value, and the difference between the predicted value of the initial capacity and the preset expected value is greater than the upper limit of the first numerical range; the second number is less than the first number.

10. An energy storage system initial capacity prediction device, characterized in that: include: An acquisition module is configured to acquire parameter values ​​of multiple parameter types of multiple battery cells of the energy storage system; the multiple parameter types include multiple items of the following types: battery cell capacity, battery cell voltage, battery cell temperature, and battery cell resistance; A construction module configured to construct first input feature maps corresponding to the plurality of parameter types respectively based on the parameter values ​​of the plurality of parameter types; wherein the parameter values ​​in the first input feature maps have the same parameter type; The prediction module is configured to input the constructed first input feature map into a pre-trained AI model, predict the initial capacity of the energy storage system, and determine the predicted value of the initial capacity.

11. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 9 is implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

13. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images