An electrochemical energy storage system situational awareness method, apparatus, device, and medium

Abstract

The application discloses a situation awareness method and device of an electrochemical energy storage system, equipment and a medium, relates to the field of power grid management, and comprises the following steps: collecting operation data of a battery in the electrochemical energy storage system in a charging and discharging cycle, combining local laboratory battery aging measured data, and constructing a battery operation database; the battery operation database comprises state evolution data sets; based on the battery operation database, first aging characteristic information of the battery and inconsistency characteristic information of a battery module are extracted to determine a characteristic mapping model from a single body level to a module level; based on a preset hybrid neural network model, second aging characteristic information of the battery module and an equivalent circuit model, a battery pack power state estimation model under multiple constraint conditions is determined; and based on the characteristic mapping model and the battery pack power state estimation model, a situation awareness result is determined. The application can realize accurate state perception of the whole life cycle of the energy storage system in the scene that the energy storage system frequently participates in power grid regulation.

Description

Technical Field

[0001] This invention relates to the field of power grid management, and in particular to a situational awareness method, device, equipment, and medium for an electrochemical energy storage system. Background Technology

[0002] With the increasing application of electrochemical energy storage systems in grid regulation, peak shaving, and black start scenarios, these systems face more complex and demanding dynamic operating conditions. In this context, the batteries in these systems are subjected to high-frequency, high-rate charge-discharge cycles for extended periods, which can easily lead to rapid performance degradation and potential safety hazards, resulting in decreased system efficiency and even safety incidents.

[0003] However, traditional solutions rely heavily on historical data modeling. While they can collect basic battery parameters such as voltage, current, and temperature in real time, their generalization ability is limited and they are unable to adapt to the dynamic stress environment brought about by frequently changing operating conditions (such as grid frequency regulation and high-speed charging and discharging). This results in low accuracy and precision. In particular, under the circumstances of multi-level coupling, enhanced inconsistency, and diversified battery aging paths, these solutions are unable to meet the requirements for accurate monitoring and real-time safety management of the entire life cycle of energy storage systems.

[0004] Therefore, how to achieve accurate state perception of the entire life cycle of energy storage systems in scenarios where energy storage systems frequently participate in grid regulation is an urgent problem to be solved. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a situational awareness method, device, equipment, and medium for electrochemical energy storage systems. This method enables accurate situational awareness of the entire lifecycle of the energy storage system, especially in scenarios where the system frequently participates in grid regulation. It addresses the problem of perception bias and distortion caused by the cross-level transmission characteristics of energy storage system states, improving the accuracy, precision, and reliability of situational awareness, thereby enhancing the safety and intelligent management level of the energy storage system under complex regulation scenarios. The specific solution is as follows:

[0006] In a first aspect, this application provides a situational awareness method for an electrochemical energy storage system, including:

[0007] The battery operation data of the electrochemical energy storage system during the charge and discharge cycle is collected and combined with the local laboratory battery aging test data to construct a battery operation database; the battery operation database includes an initial state identification dataset and a state evolution dataset.

[0008] Feature extraction is performed based on the battery operation database, and a feature mapping model from the single cell level to the module level is determined by using the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module.

[0009] Based on the preset hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information and the equivalent circuit model, a battery pack power state estimation model under multiple constraints is determined.

[0010] Based on the feature mapping model, the battery pack power state estimation model, and the topology information corresponding to the electrochemical energy storage system, a multi-state joint estimation is performed to determine the situational awareness result.

[0011] Optionally, the battery operation data collected in the electrochemical energy storage system during charge-discharge cycles, combined with local laboratory battery aging test data, is used to construct a battery operation database, including:

[0012] Based on sensors, determine the operating data of batteries in an electrochemical energy storage system during charge and discharge cycles;

[0013] Obtain local laboratory battery aging test data corresponding to the aforementioned operational data;

[0014] Based on the laboratory battery aging test data and battery aging data, an evolution analysis of battery performance and battery parameters is conducted to determine the battery operation database; the battery parameters include voltage, current and capacity.

[0015] Optionally, the step of extracting features based on the battery operation database and determining a feature mapping model from the individual cell level to the module level using the corresponding first aging feature information of the battery and the inconsistency feature information of the battery module includes:

[0016] The individual charge-discharge curves of the battery are determined based on the battery operation database;

[0017] Feature extraction is performed based on the single-cell charge-discharge curve to determine the first aging characteristic information corresponding to the battery.

[0018] Feature extraction is performed based on the battery operation database to determine the inconsistency feature information corresponding to the battery module; the battery module includes a plurality of batteries, and the inconsistency feature information includes the range of the voltage curves of each battery in the module and the distance between the voltage curves and the average voltage curves of all batteries in the module;

[0019] Based on the first aging feature information and the inconsistency feature information, a feature mapping model from the individual unit level to the module level is determined.

[0020] Optionally, the step of determining the battery pack power state estimation model under multiple constraints based on a preset hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information, and the equivalent circuit model includes:

[0021] The second aging characteristic information corresponding to the battery module is determined based on the battery operation database.

[0022] Based on the health features in the second aging feature information and the inconsistency feature information, a preset hybrid neural network model is trained to determine the battery data-driven model at the module level when the training is completed; the preset hybrid neural network model is a neural network model determined based on a bidirectional long short-term memory network and an attention mechanism.

[0023] Based on the average model and the difference model corresponding to the battery module, the equivalent circuit model corresponding to the battery module is determined;

[0024] Based on the battery data-driven model and the equivalent circuit model, a battery pack power state estimation model under multiple constraints is determined.

[0025] Optionally, determining the equivalent circuit model corresponding to the battery module based on the average model and the difference model corresponding to the battery module includes:

[0026] Based on the battery parameters corresponding to each battery in the battery module, an average model and a difference model corresponding to the battery module are determined; the average model is used to characterize the average state of the battery module, and the difference model is used to characterize the deviation between the state of each battery in the battery module and the average state;

[0027] Based on the average model, the difference model, and the extended Kalman filter algorithm, the equivalent circuit model corresponding to the battery module is determined.

[0028] Optionally, determining the battery pack power state estimation model under multiple constraints based on the battery data-driven model and the equivalent circuit model includes:

[0029] Obtain preset multiple constraints; the preset multiple constraints include the charging and discharging current constraints, state of charge constraints, and health state constraints of the battery module.

[0030] The battery pack power state estimation model is determined by calculation and deduction based on the preset multiple constraints, the battery data-driven model, and the equivalent circuit model.

[0031] Optionally, the multi-state joint estimation based on the feature mapping model, the battery pack power state estimation model, and the topology information corresponding to the electrochemical energy storage system includes:

[0032] The analysis results are determined by analyzing the series and parallel topologies of the battery modules and batteries within the electrochemical energy storage system using topology information.

[0033] Based on the feature mapping model and the battery pack power state estimation model, a multi-level state awareness model is determined.

[0034] Based on the analysis results and the multi-level state perception model, the electrochemical energy storage system is jointly sensed for multiple state parameters to determine the situational awareness results; the multiple state parameters include state of charge parameters, health state parameters, and power state parameters.

[0035] Secondly, this application provides a situational awareness device for an electrochemical energy storage system, comprising:

[0036] The database construction module is used to collect the operating data of batteries in the electrochemical energy storage system during the charge and discharge cycle, and combine it with local laboratory battery aging test data to construct a battery operation database; the battery operation database includes an initial state identification dataset and a state evolution dataset.

[0037] The mapping model construction module is used to extract features based on the battery operation database and use the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module to determine the feature mapping model from the single cell level to the module level.

[0038] The model determination module is used to determine the battery pack power state estimation model under multiple constraints based on a preset hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information and the equivalent circuit model.

[0039] The situation awareness module is used to perform multi-state joint estimation based on the feature mapping model, the battery pack power state estimation model, and the topology information corresponding to the electrochemical energy storage system, so as to determine the situation awareness result.

[0040] Thirdly, this application provides an electronic device, comprising:

[0041] Memory, used to store computer programs;

[0042] A processor is used to execute the computer program to implement the steps of the aforementioned situational awareness method for electrochemical energy storage systems.

[0043] Fourthly, this application provides a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the steps of the aforementioned situational awareness method for an electrochemical energy storage system.

[0044] As can be seen, in this application, operational data of batteries in the electrochemical energy storage system during charge-discharge cycles are collected, and combined with local laboratory battery aging test data, a battery operation database is constructed. The battery operation database includes an initial state identification dataset and a state evolution dataset. Feature extraction is performed based on the battery operation database, and a feature mapping model from the individual cell level to the module level is determined using the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module. Based on a preset hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information, and an equivalent circuit model, a battery pack power state estimation model under multiple constraints is determined. Based on the feature mapping model, the battery pack power state estimation model, and the topology information corresponding to the electrochemical energy storage system, multi-state joint estimation is performed to determine the situational awareness result. In other words, this application first constructs a battery operation database based on local laboratory battery aging test data and battery operation data during charge-discharge cycles in the electrochemical energy storage system. Then, feature extraction is performed based on this database. Using the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module, a feature mapping model from the individual cell level to the module level is determined. Next, based on a pre-set hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information, and the equivalent circuit model, a battery pack power state estimation model under multiple constraints is determined. Finally, based on the feature mapping model, the battery pack power state estimation model, and the system's topology information, the situational awareness result is determined. This enables accurate state awareness of the entire lifecycle of the energy storage system, especially in scenarios where the energy storage system frequently participates in grid regulation. It solves the problem of perception bias and distortion caused by the cross-level transmission characteristics of the energy storage system's state, improving the accuracy, precision, and reliability of state awareness, thereby enhancing the safety and intelligent management level of the energy storage system in complex regulation scenarios. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0046] Figure 1A flowchart of a situational awareness method for an electrochemical energy storage system provided in this application;

[0047] Figure 2 A schematic diagram of the battery charging and discharging voltage and current variation curves provided in this application;

[0048] Figure 3 A schematic diagram of a hybrid neural network structure is provided in this application;

[0049] Figure 4 This application provides an equivalent circuit topology diagram of a series-parallel battery module.

[0050] Figure 5 A schematic diagram of the SOH estimation curve of a battery pack constructed with a B1 battery provided in this application;

[0051] Figure 6 A schematic diagram showing the comparison between the estimated and actual SOH values ​​of an electrochemical energy storage system provided in this application;

[0052] Figure 7 A schematic diagram of the SOC estimation curve of a B1-type battery pack provided in this application;

[0053] Figure 8 A schematic diagram showing the comparison curve between the estimated and actual SOC values ​​of an electrochemical energy storage system provided in this application;

[0054] Figure 9 A schematic diagram of the charge-discharge peak power curves of a B3 battery under different aging conditions provided in this application;

[0055] Figure 10 A schematic diagram of the situational awareness device for an electrochemical energy storage system provided in this application;

[0056] Figure 11 This application provides a structural diagram of an electronic device. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Traditional solutions often rely on historical data modeling. Although they can collect basic battery parameters such as voltage, current, and temperature in real time, their generalization ability is limited and they are difficult to adapt to the dynamic stress environment brought about by frequently changing operating conditions (such as grid frequency regulation and high-speed charging and discharging). This results in low accuracy and precision. In particular, under the circumstances of multi-level coupling, enhanced inconsistency, and diversified battery aging paths, these solutions are difficult to meet the needs of accurate monitoring and real-time safety management of the entire life cycle of energy storage systems.

[0059] To this end, this application provides a situational awareness scheme for an electrochemical energy storage system, which can achieve accurate state awareness of the entire life cycle of the energy storage system in scenarios where the energy storage system frequently participates in grid regulation. This solves the problem of perception bias and distortion caused by the cross-level transmission characteristics of the energy storage system state, and improves the accuracy, precision and reliability of state awareness.

[0060] See Figure 1 As shown in the figure, an embodiment of the present invention discloses a situational awareness method for an electrochemical energy storage system, comprising:

[0061] Step S11: Collect the battery operation data in the electrochemical energy storage system during the charge and discharge cycle, and combine it with the local laboratory battery aging test data to construct a battery operation database; the battery operation database includes an initial state identification dataset and a state evolution dataset.

[0062] In this embodiment, a battery operation database is first constructed using local laboratory battery aging test data. That is, the battery operation data in the electrochemical energy storage system during the charge and discharge cycle is determined based on sensors; local laboratory battery aging test data corresponding to the operation data is obtained; based on the laboratory battery aging test data and the battery aging data, the evolution analysis of battery performance and battery parameters is performed to determine the battery operation database; the battery parameters include voltage, current and capacity.

[0063] Taking lithium-ion batteries as an example, lithium-ion batteries are a general term for batteries that use various lithium-ion intercalation compounds as positive electrode materials. It can be understood that in this embodiment, operational data of lithium-ion batteries during charge-discharge cycles is collected, along with locally stored battery measurement data accumulated by the local laboratory. These two types of data are then combined to construct a battery operation database, covering key operational information of the battery during constant-current / constant-voltage charging and constant-current discharging processes. For example... Figure 2 The diagram shows the voltage and current variation curves of a lithium-ion battery during charging and discharging. Then, using the constructed database, the evolution of battery parameters such as voltage, current, and capacity during the aging cycle is analyzed. This analysis also examines the performance degradation patterns of lithium-ion batteries during cyclic charging and discharging, the main aging factors, and the causes of inconsistencies among individual cells within the battery pack.

[0064] Specifically, regarding the analysis of the evolution of battery parameters such as voltage, current, and capacity during the aging cycle, the charging phase first uses a constant current of 1.5A until the voltage reaches the cutoff value of 4.2V, followed by a constant voltage charging phase until the current drops below 20mA. During the constant current charging process, the battery temperature gradually rises and reaches its peak, while it begins to decrease during the constant voltage phase. In one specific implementation, the battery grades used are set as B1, B2, and B3. Local experimental data shows that the battery capacity generally decreases with the increase of charge-discharge cycles. It is worth noting that during the resting period after each charge-discharge cycle, a capacity regeneration phenomenon may occur inside the battery. That is, due to the dissolution of some of the aggregated reactants that slow down the chemical reaction on the electrode surface, a short-term capacity rebound occurs in the next cycle.

[0065] This section analyzes the performance degradation patterns, main aging factors, and causes of inconsistency among individual cells within a battery pack (i.e., battery module) during the cyclic charging and discharging process of lithium-ion batteries. During charging, lithium ions are extracted from the surface of the positive electrode material, enter the electrolyte, and are embedded in the negative electrode through the separator. During discharging, lithium ions are extracted from the negative electrode and return to the positive electrode. The charging and discharging process of lithium batteries is reversible. After multiple charge-discharge cycles, a series of irreversible physical and chemical changes occur within a single battery cell, including the formation of a solid electrolyte interface film, the loss of lithium ions and active materials, and complex electrochemical reactions triggered by operating modes and external conditions. Specifically, high temperatures, deep charge / discharge cycles, high-rate currents, and external impacts can all accelerate the performance degradation of lithium-ion batteries. Furthermore, in battery packs, individual cells are combined in series and parallel to achieve higher system voltage and current. From battery manufacturing, assembly, and use to performance degradation and eventual end-of-life, consistency issues are a constant concern. Specifically, the inconsistency of individual cells within a battery pack mainly stems from two stages: first, differences in individual cells during production and storage; and second, the exacerbation of these differences in individual cell performance during battery pack assembly. These two factors are coupled and jointly affect the consistency and overall performance of the battery pack.

[0066] After completing the above analysis, the results can be organized to obtain the initial state identification dataset and the state evolution dataset.

[0067] Step S12: Based on the battery operation database, feature extraction is performed, and the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module are used to determine the feature mapping model from the single cell level to the module level.

[0068] In this embodiment, after obtaining the battery operation database and performing feature extraction, the database is used to determine a feature mapping model from the individual cell level to the module level. That is, the individual cell charge-discharge curve of the battery is determined based on the battery operation database; feature extraction is performed based on the individual cell charge-discharge curve to determine the first aging feature information corresponding to the battery; feature extraction is performed based on the battery operation database to determine the inconsistency feature information corresponding to the battery module; the battery module includes a plurality of the batteries, and the inconsistency feature information includes the range of the voltage curves of each battery in the module and the distance between the voltage curve and the average voltage curve of all batteries in the module; based on the first aging feature information and the inconsistency feature information, a feature mapping model from the individual cell level to the module level is determined.

[0069] Specifically, the process of determining the cross-level feature mapping model is as follows:

[0070] Because the battery discharge process exhibits highly dynamic characteristics and fragmented charging and discharging states during operation involving frequent grid regulation, the charging process is relatively more stable. Therefore, this embodiment selects to extract the first aging characteristic information from the charging process. During constant current charging, the battery's charging capacity Q can be calculated using the following formula:

[0071] ;

[0072] In the formula, I represents the magnitude of the charging current, in A (amperes); t represents the time, in s (seconds).

[0073] The incremental capacity (IC) curve of a battery, where IC is represented as:

[0074] ;

[0075] In the formula, The charging capacity is within the corresponding voltage range, in Ah (ampere-hours). This represents the voltage range, measured in volts (V). (Setting) =0.05V, the charging capacity of each interval in each cycle was calculated. As the number of battery cycles increases, the charging capacity of each interval gradually decreases. Therefore, only a few consecutive fixed voltage intervals are selected. As the number of cycles increases, the sum of the charging capacities of these intervals gradually decreases. From a statistical perspective, according to the standard deviation formula, their standard deviation also gradually decreases as the number of cycles increases.

[0076] To estimate the health status of any charging segment, it is necessary to extract the health indicators (HI) of all charging segments to train the model. Taking the i-th cycle as an example, given the initial voltage... and cutoff voltage ,according to The constant voltage charging stage is divided into intervals of 0.05V. According to the following formula, the i-th cycle can be divided into n intervals, resulting in a voltage interval sequence. ;

[0077] ;

[0078] During battery charging, the capacity sequence corresponding to the voltage range sequence. If the charging segment length is set to... indivual Then we can obtain the number of charging segments m in the i-th cycle:

[0079] ;

[0080] Based on the above settings, the charging segment sequence is obtained in the i-th iteration: For charging segments Contains capacity sequence: , Contains capacity sequence: And so on. Contains capacity sequence Ultimately, this covers the entire constant current charging stage, with spacing... The charging segment sequence P is 0.05V.

[0081] With charging segments For example, HI1 can be extracted as the sum of the charging capacity within the segment. The formula is as follows:

[0082] ;

[0083] In the formula, Let q be the q-th capacity segment.

[0084] HI2 is the standard deviation of the charging capacity within a segment. The formula is as follows:

[0085] ;

[0086] In the formula, It is the average of all charging capacities in the corresponding segment, in Ah.

[0087] At the same time, in order to distinguish the voltage segment in which the feature is located, the initial voltage corresponding to the voltage segment is... Size is also considered as one of the features, namely feature HI3.

[0088] Before studying the correlation between fragment length and HIs (Health Status) and SOH (State of Health), it is necessary to determine... and For the battery studied in this embodiment, the voltage range for feature extraction is [3.5V, 4.2V]. Generally, the higher the correlation between the input features and output of a data-driven model, the more accurate the model. The Pearson correlation coefficient... It is widely used to quantify the degree of linear correlation between two variables, and its formula is:

[0089] ;

[0090] In the formula, For characteristic sequences, It is its average value; y is the battery capacity. It is its average value.

[0091] Furthermore, to achieve a deep integration of the data-driven model and the equivalent circuit model, the ohmic resistance identified in the equivalent circuit model of a single battery cell is further selected. As one of the health characteristics, it is defined as the health indicator HI4, which is used to reflect the evolution trend of battery internal resistance as it ages.

[0092] Based on the above formula, the correlation between HIs and SOH extracted for different segment lengths is calculated, thereby selecting the appropriate segment length and the optimal voltage sampling interval.

[0093] Based on the extraction of the first aging characteristic information of individual battery cells, the inconsistencies between modules are considered to achieve cross-level migration modeling of features from individual cells to modules. Since inconsistencies within the battery pack not only reduce the overall performance of the battery pack but also affect the accuracy of state perception, this embodiment needs to analyze the differences between individual cells within the battery module and extract the inconsistencies within the group to improve the state perception capability at the module level.

[0094] The inconsistency features extracted in this embodiment include the range of the individual cell voltage curves within the battery pack. Euclidean distance between the voltage curve of a single cell within the group and the average voltage curve of all cells Assume the voltage curves of the battery cells within the battery pack are determined by vector... This is represented by [the vector]. Furthermore, the average voltage curve of all battery cells is represented by [the vector]. The inconsistency characteristic is represented by the following formula.

[0095] ;

[0096] In the formula, N is the length of the battery voltage sequence. This represents the voltage value of the i-th battery at time j, in volts (V). This represents the average voltage of all batteries at time point j, in volts (V).

[0097] Step S13: Based on the preset hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information and the equivalent circuit model, determine the battery pack power state estimation model under multiple constraints.

[0098] In this embodiment, a battery pack power state estimation model under multiple constraints is determined using a preset hybrid neural network model, the second aging feature information corresponding to the battery module, and the inconsistency feature information. Specifically, the second aging feature information corresponding to the battery module is determined based on the battery operation database. The preset hybrid neural network model is trained based on the health features in the second aging feature information and the inconsistency feature information to determine the battery data-driven model at the module level upon completion of training. The preset hybrid neural network model is a neural network model determined based on a bidirectional long short-term memory network and an attention mechanism. The equivalent circuit model corresponding to the battery module is determined based on the average model and the difference model corresponding to the battery module. Finally, the battery pack power state estimation model under multiple constraints is determined based on the battery data-driven model and the equivalent circuit model.

[0099] Furthermore, regarding the determination of the equivalent circuit model, in this embodiment, firstly, based on the battery parameters corresponding to each battery in the battery module, an average model and a difference model corresponding to the battery module are determined; the average model is used to characterize the average state of the battery module, and the difference model is used to characterize the deviation between the state of each battery in the battery module and the average state; then, based on the average model, the difference model, and the extended Kalman filter algorithm, the equivalent circuit model corresponding to the battery module is determined. Regarding the determination of the battery pack power state estimation model, firstly, preset multiple constraints are obtained; the preset multiple constraints include the charging and discharging current constraints, state of charge constraints, and health state constraints of the battery module; then, based on the preset multiple constraints, the battery data-driven model, and the equivalent circuit model, calculations and deductions are performed to determine the battery pack power state estimation model.

[0100] It is important to understand the specific process for determining the battery pack power state estimation model under multiple constraints, as shown below:

[0101] Combination Figure 3As shown, this embodiment employs a hybrid neural network model combining a Bidirectional Long Short-Term Memory (BiLSTM) network and an attention mechanism. This model leverages the advantages of BiLSTM in handling long dependencies in time-series data while incorporating the attention mechanism to enhance the role of important time steps. This addresses the potential issue of neglecting important information at certain time points when the model has many input items and a large dataset, thereby improving the network's accuracy and computational efficiency. The model uses aging and inconsistency features extracted from voltage and current curves as input and battery capacity as output. The training and test sets are divided in a 7:3 ratio to achieve high-precision estimation of the battery module's State of Health (SOH). In this embodiment, health factors describing battery degradation trends from the battery charging voltage curve, charging current curve, and charging temperature change curve are selected as BiLSTM inputs to construct the time-series matrix. The expanded form is shown in the following equation:

[0102] ;

[0103] In the above formula, Let be the set of parameters for all battery health characteristics at time t. Let T be the historical time-series data set of feature m, and M be the number of parameters in the set. The battery health feature at time t... As input, the hidden states are defined as follows: and The final hidden state at time t is output by the BiLSTM layer. The calculation is as follows:

[0104] ;

[0105] in, Represents the LSTM network function; The hidden state is represented by t-1; U is the weight matrix of the hidden state; W is the weight matrix of the input; b represents the bias term; the arrow represents the direction of LSTM transition.

[0106] Attention mechanisms assign different weights to different input features to enhance important features and prevent irrelevant information from affecting the final result. Specifically, a scoring function is first used... Its expression is:

[0107] ;

[0108] In the formula, For the bias term of the hidden layer; This is the weight matrix from hidden layer to hidden layer; Weighted information for the sequence; The hidden state after bidirectional LSTM fusion; For sequence weights; For activation functions; The weights for Attention; The bias of the attention; v is the attention value; This represents the output of the Attention layer at time t. These are the weighting coefficients.

[0109] Subsequently, an average model based on the average voltage and current of the module was constructed, and a difference model was introduced to correct the inconsistency of individual units, compensate for the loss of model accuracy, and further establish an equivalent circuit model at the module level.

[0110] (1) Construction of the average battery pack model: The average model represents the average state of the battery pack. Its parameters are the average values ​​of the parameters of each individual cell, which are unique. Therefore, its computational cost is only equivalent to that of the individual cell model. In addition, since it is the benchmark for calculating the differences between individual cells, the accuracy of the model is particularly important. First, the Thevenin equivalent circuit model of the power battery is constructed using one ohmic internal resistance and one RC circuit (Resistor-Capacitor Network, a circuit composed of resistors and capacitors). Its mathematical expression is as follows:

[0111] ;

[0112] in, This indicates the polarization resistance of a lithium-ion battery, measured in Ω (ohms). Polarization capacitance is expressed in pF (picofarad); the RC network characterizes the polarization phenomenon of a lithium-ion battery. This indicates the load current, in amperes (A). This indicates the terminal voltage when the battery is externally connected, in volts (V). Let be the state of charge at time t; This represents the maximum usable capacity of the battery pack, in Ah. Indicates the start time of the current integral, in seconds; Indicates coulomb efficiency; for The state of charge at any given moment; Polarization voltage; This refers to the current flowing through the polarization capacitor; The current flowing through the polarization resistor is denoted as OCV(); OCV() is the open-circuit voltage-state-of-charge function. Let be the ohmic resistance. By applying the Laplace transform and discretization to the equations emphasized earlier, we can obtain the following expression:

[0113] ;

[0114] In the formula, The sampling period is expressed in seconds (s). This is the cell polarization time constant; This represents the state of charge at the k-th sampling time. The state of charge at the (k-1)th sampling time; k represents the sampling time; Let be the polarization voltage at the k-th sampling time; The polarization voltage at the (k-1)th sampling time; Let be the load current at the k-th sampling time. Let be the load current at the (k-1)th sampling time. This represents the open-circuit voltage at the k-th sampling time.

[0115] (2) Construction of battery pack difference model: The difference model is used to characterize the deviation between the state of each cell in the battery pack and the average state. Therefore, the number of difference models is the same as the number of cells in series. This deviation is mainly due to the difference in ohmic internal resistance caused by the deterioration of battery pack consistency. In order to balance the computational complexity, the Rint model is selected.

[0116] ;

[0117] ;

[0118] in, This is the reference battery's port voltage, in volts (V). This represents the port voltage of the non-reference cell i, i.e., the differential cell cell i, in V; The current SOC state of the reference battery; This represents the current SOC state of the non-reference battery i. Indicates the polarization voltage of the reference cell; Represents the polarization voltage of non-reference cell i, in V; This is the ohmic internal resistance of the reference cell, in Ω; Represents the ohmic internal resistance of non-reference cell i, in Ω; It is the SOC difference between differential cell i and the reference cell; This indicates that when the SOC difference between the two batteries is Voltage difference at time, in V; This represents the load current, in amperes (A). Combining the above equations, the solution is:

[0119] ;

[0120] In the formula, It is the voltage difference between the differential cell i and the reference cell, in volts (V). This represents the internal resistance difference between the reference cell and the differential cell i, expressed in V. The Rint model, in addition to containing internal random variables... A varying open-circuit voltage source. Taking the battery discharge current as the positive direction, solve the above problem for... for:

[0121] ;

[0122] In the formula, This represents the current remaining available capacity of differential battery cell i, in Ah. Indicates the current remaining available capacity of the reference battery, in Ah; This indicates the maximum available capacity of differential cell i in its current state of equilibrium (SOH); This indicates the maximum usable capacity of the reference battery in its current state of equilibrium (SOH). This represents the Coulomb efficiency. Therefore, the discrete-time approximate recursion can be obtained as follows:

[0123] ;

[0124] In the formula, k represents the sampling time; This represents the difference between the SOC of individual cell i at time k and the average SOC; This represents the difference between the open-circuit voltage of cell i at time k and the average open-circuit voltage. This represents the difference between the open-circuit voltage and the average ohmic resistance of cell i at time k.

[0125] (3) Based on the equivalent circuit model, the extended Kalman filter algorithm is used to obtain the state equation of the battery pack average model based on the above content. and output observation equation As shown below:

[0126] ;

[0127] In the formula, an extended Kalman filter is used. To estimate the transient quantity x; the system input and output vectors are represented by u and y, respectively; Indicates the use of estimating model parameters The estimation error covariance matrix of the extended Kalman filter; Let represent the estimation error covariance matrix of the extended Kalman filter used to estimate the state vector x; Indicates the sampling time; k represents the sampling point; L represents the time scale for parameter identification, in seconds; For the k-th sampling point, the +1 battery state vectors at different time scales; For the k-th sampling point, the The system input vector at each time scale; This is the model parameter vector at the kth sampling point; For the k-th sampling point, the +1 timescale of state of charge; For the k-th sampling point, the +1 timescale polarization voltage; For the k-th sampling point, the Load current at a time scale; For the k-th sampling point, the Load voltage at a given time scale; For k sampling points, the first Polarization voltage at a time scale; For the k-th sampling point, the Open-circuit voltage at a time scale; The ohmic resistance of the battery; For the k-th sampling point, the The state of charge at several time scales.

[0128] ;

[0129] In the formula, For the k-th sampling point, the A battery state vector at each time scale. and Differentiation into:

[0130] ;

[0131] ;

[0132] in, , as well as Represents the system matrix. For the k-1th sampling point, the State transition matrix at -1 time scales; For the k-1th sampling point, the Observation matrix at multiple time scales; For the k-1th sampling point, the The estimated battery state vector at each time scale; For the k-1th sampling point, the The estimated battery state vector at a time scale of -1; This represents the estimated model parameter vector at the k-th sampling point; For the k-1th sampling point, the State of charge at various time scales; The open-circuit voltage is used. In the differential model, current is used as the input parameter, and the voltage difference between the reference battery and the differential battery cell is used as the output. Discrete equations for the voltage difference and the SOC (State of Charge) difference are established, yielding the state equation and observation equation of the differential model:

[0133] ;

[0134] ;

[0135] In the formula, and All use micro-timescale Kalman filtering; For the i-th battery at the k-th sampling point, the... The difference in state of charge at each time scale represents the difference in state of charge with respect to the reference cell. For the i-th battery at the k-th sampling point, the... The resistance difference over a time scale represents the difference in resistance compared to the reference cell; For the k-th sampling point, the +1 battery state vectors at different time scales; For the k-th sampling point, the Battery state vectors at various time scales; For the i-th battery at the k-th sampling point, the... +1 time scale difference in state of charge; For the i-th battery at the k-th sampling point, the... The voltage difference over a time scale represents the voltage difference from the reference cell; Let be the model parameter vector at k sampling points of the i-th battery. By linearizing the state equation and observation equation, the following derivatives can be obtained:

[0136] ;

[0137] ;

[0138] In the formula, For the difference model at the k-1th sampling point of the i-th battery, the... State transition matrix at -1 time scales; This is the prior estimate of the i-th battery parameter at the k-th sampling point of the difference model; This represents the state vector of the i-th battery in the difference model. This is the estimated value of the i-th battery parameter at the k-th sampling point of the difference model; For the k-1th sampling point of the difference model, the... The estimated value of the state of charge difference of the i-th battery at each time scale.

[0139] Figure 4 This section presents the equivalent circuit topology of a series-parallel battery pack, showcasing the topology of the equivalent circuit model under the series-parallel structure. The average model represents the average state of the battery pack, while the differential state equivalent circuit model reflects the differences between individual cells within the battery pack. Furthermore... Figure 4 In For the cell in the Nth column and Mth row; For the Nth column of parallel battery modules; The internal resistance of the battery is ohmic. This is the battery open-circuit voltage; This refers to the battery polarization voltage. Battery polarization resistor; For battery polarization capacitors; This is the battery load current; Let be the resistance difference of the i-th battery; This represents the state-of-charge difference of the i-th battery. This represents the difference in battery load voltage.

[0140] Furthermore, the updated state of charge (SOC) and polarization voltage of the battery pack are substituted into the numerical analysis model of the battery pack. Combined with multiple constraints such as battery pack voltage, current, and SOP (State of Power), the peak power current of the battery pack is calculated. Based on this, and combined with the peak voltage, a numerical model for estimating the battery pack's state of power under multiple constraints is established.

[0141] (1) Maximum charge and discharge current constraint: The maximum discharge current specified by the manufacturer in the design of the power battery. Minimum charging current This serves as a constraint to prevent the power battery from operating beyond the capacity of the energy storage unit.

[0142] ;

[0143] In the formula, This is the peak discharge current of the battery pack calculated at time k; The peak charging current of the battery pack at time k; This represents the internal resistance of the battery pack during the discharge process. This refers to the internal resistance of the battery pack during the charging process. This is the charging cutoff voltage; This is the discharge cutoff voltage; Open-circuit voltage, in volts (V). This represents the maximum available capacity of the battery pack, expressed in Ah.

[0144] (2) State of Charge (SOC) Constraints: Overcharging and over-discharging of batteries pose significant safety hazards. Therefore, when the SOC of the battery pack approaches the preset limit, the battery current value needs to be limited to prevent damage to the battery due to excessive current. The interval between L consecutive sampling times is specified. The formula for calculating the maximum battery capacity is:

[0145] ;

[0146] In the formula, Let SOC be the value at time k. Improve battery charging and discharging efficiency; This is the maximum limit of battery SOC. As the minimum limit of battery SOC, the calculation based on the actual limit peak current of battery SOC within the time interval ∆t can ensure the safety performance of the battery during charging and discharging.

[0147] (3) Health status constraints: As aging progresses, the maximum usable capacity of the power battery gradually decreases, and the charging and discharging capabilities weaken. It is necessary to update the maximum usable capacity of the battery pack based on the current health status. .

[0148] Based on the above constraints, the multi-constraint peak dynamic current and the peak power of the power battery are calculated, and the peak power of the battery pack at time k over the duration interval L is obtained accordingly:

[0149] ;

[0150] ;

[0151] in, Let be the peak discharge current of the battery pack at time k. The peak charging current of the battery pack at time k, in amperes (A). The minimum charging power set during battery pack design and application. The maximum discharge power limit set for battery pack design and application, in watts (W). The peak discharge power of the battery pack estimated at time k is obtained by the multi-constraint dynamic continuous peak power estimation method. The peak charging power of the battery pack estimated at time k by the multi-constraint dynamic continuous peak power estimation method is expressed in W. This is the maximum discharge current; Minimum charging current; Let be the cutoff voltage at time k+L.

[0152] Step S14: Perform multi-state joint estimation based on the feature mapping model, the battery pack power state estimation model, and the topology information corresponding to the electrochemical energy storage system to determine the situational awareness result.

[0153] In this embodiment, a feature mapping model and a battery pack power state estimation model are used to perform multi-state joint estimation of the batteries in the system. That is, the series and parallel topology of the battery modules and batteries in the electrochemical energy storage system is analyzed through topology information to determine the analysis results; a multi-level state perception model is determined based on the feature mapping model and the battery pack power state estimation model; and multiple state parameters of the electrochemical energy storage system are jointly perceived based on the analysis results and the multi-level state perception model to determine the situational awareness results. The multiple state parameters include state of charge parameters, health parameters, and power state parameters.

[0154] It is important to understand that in the process of multi-state joint estimation, this embodiment will consider the topological information of the energy storage system and first analyze the series and parallel topology within the electrochemical energy storage system; then, it will establish a mapping relationship between topological parameters and the system state-aware model. For example, a classic battery pack topology of 3 parallel and 6 series can be selected. Based on the physical connection relationship of the battery cells in the energy storage system, the characteristics of voltage accumulation and capacity consistency under series structure and capacity accumulation and voltage consistency under parallel structure are clarified, and a module-level electrical characteristic mapping model is constructed. By analyzing the connection method and number of batteries in the module, the theoretical values ​​of its overall voltage, current, and capacity are determined, and the calculation logic of key states such as SOC, SOP, and SOH at the module level is defined accordingly. Considering that the voltage of each battery cell is consistent when connected in parallel, resulting in high overall consistency; however, under series connection conditions, due to the difference in voltage between individual cells, it is difficult to achieve uniform full charge and discharge, which can easily lead to an increase in the performance difference between cells and a significant decrease in consistency. Therefore, this embodiment sets the consistency of parallel cells to be the same, but the consistency of series cells is not.

[0155] Subsequently, at the modeling level, a multi-state joint estimation model is constructed from the individual unit level to the module level, combining equivalent circuit models and data-driven methods. At the individual unit level, actual collected voltage, current, and capacity are used as inputs, and key state indicators are output. At the module level, dynamic perception of the overall operating state is achieved through battery topology mapping and feature fusion, as well as consideration of inconsistencies within the battery pack. This model can realize online prediction and updating of parameters such as state of charge, state of health, and state of power at the module level. Specifically, in one implementation, the state perception results can be as follows: Figures 5 to 9 As shown, Figure 5 The SOH estimation curve for the battery pack constructed for battery B1 is shown, with the first 70% representing the training set and the last 30% representing the validation set. Figure 6 The curves show the comparison between the estimated and actual SOH values ​​of the electrochemical energy storage system. The SOH estimation error for each battery pack is less than 2%. Figure 7 The SOC estimation curve for the battery pack constructed for battery B1. Figure 8 The curves show the comparison between the estimated and actual SOC values ​​of the electrochemical energy storage system. The SOC estimation error for each battery pack is less than 1.5%. Figure 9 The peak charge and discharge power curves of B3 batteries at different aging stages are shown below. Figure 9 As can be seen, the more severe the aging, the lower the peak charging and discharging power of the battery.

[0156] In summary, this embodiment provides a safety situation awareness scheme for electrochemical energy storage systems that frequently participate in grid regulation. Addressing issues such as accelerated aging, increased inconsistencies, and inaccurate state estimation in energy storage systems frequently involved in grid regulation, a state awareness framework integrating multi-source data-driven approaches and mechanism modeling is constructed. First, voltage and current sensing technologies are used to collect various types of operational data from lithium-ion batteries under different operating conditions. Combined with publicly available battery aging experimental data and laboratory measured data, an operational database is constructed, covering initial identification data and dynamic state evolution data. Through systematic analysis of multi-source data, key health characteristics during the cycling process are extracted, clarifying the aging evolution patterns and dominant factors. Second, based on the charge-discharge characteristics of individual battery cells, a cell-level aging characteristic model is constructed and migrated to the module level. Simultaneously considering inconsistencies in capacity, internal resistance, etc., between cells, a multi-level battery pack characteristic model is formed. At the module level, a data-driven model integrating a BiLSTM-Attention hybrid neural network is proposed to capture temporal health characteristics. Combined with the battery pack average model and the individual cell difference model, an equivalent circuit model containing descriptions of aging and inconsistencies is constructed. Based on this, a power state estimation model is constructed to estimate the dynamic peak power capability (SOP) of the battery pack based on constraints such as SOC, voltage, current, and polarization characteristics. Finally, focusing on the actual system structure, the series and parallel topology relationships of the battery modules are analyzed. A hybrid modeling system integrating equivalent circuits and data-driven approaches is constructed at the battery pack level to achieve joint estimation and real-time updating of multiple state parameters (SOC, SOH, SOP), providing accurate perception of the system's operational status. This invention realizes a closed-loop state modeling system from individual cells to modules and then to the system, possessing good engineering practicality and scalability, and can effectively support the safety management and performance assurance of energy storage systems in complex grid control scenarios.

[0157] Therefore, this application first constructs a battery operation database based on local laboratory battery aging test data and battery operation data during charge-discharge cycles in the electrochemical energy storage system. Then, feature extraction is performed based on this database. Using the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module, a feature mapping model from the individual cell level to the module level is determined. Next, based on a pre-set hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information, and the equivalent circuit model, a battery pack power state estimation model under multiple constraints is determined. Finally, based on the feature mapping model, the battery pack power state estimation model, and the system's topology information, the situational awareness result is determined. This enables accurate state awareness of the entire lifecycle of the energy storage system, especially in scenarios where the energy storage system frequently participates in grid regulation. It solves the problem of perception bias and distortion caused by the cross-level transmission characteristics of the energy storage system's state, improving the accuracy, precision, and reliability of state awareness, thereby enhancing the safety and intelligent management level of the energy storage system in complex regulation scenarios.

[0158] See Figure 10 As shown in the illustration, this application also discloses a situational awareness device for an electrochemical energy storage system, comprising:

[0159] The database construction module 11 is used to collect the operating data of the battery in the electrochemical energy storage system during the charge and discharge cycle, and combine it with the local laboratory battery aging test data to construct a battery operation database; the battery operation database includes an initial state identification dataset and a state evolution dataset.

[0160] The mapping model construction module 12 is used to extract features based on the battery operation database and use the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module to determine the feature mapping model from the single cell level to the module level.

[0161] The model determination module 13 is used to determine the battery pack power state estimation model under multiple constraints based on the preset hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information and the equivalent circuit model.

[0162] The situation awareness module 14 is used to perform multi-state joint estimation based on the feature mapping model, the battery pack power state estimation model and the topology information corresponding to the electrochemical energy storage system, so as to determine the situation awareness result.

[0163] In some specific embodiments, the database construction module 11 can be used to: determine the operating data of the battery in the electrochemical energy storage system during the charge and discharge cycle based on sensors; acquire local laboratory battery aging test data corresponding to the operating data; and perform evolution analysis of battery performance and battery parameters based on the laboratory battery aging test data and battery aging data to determine the battery operating database; the battery parameters include voltage, current and capacity.

[0164] In some specific embodiments, the mapping model construction module 12 can be used to: determine the individual charge-discharge curves of the battery based on the battery operation database; perform feature extraction based on the individual charge-discharge curves to determine the first aging feature information corresponding to the battery; perform feature extraction based on the battery operation database to determine the inconsistency feature information corresponding to the battery module; the battery module includes a plurality of the batteries, and the inconsistency feature information includes the range of the voltage curves of each battery in the module and the distance between the voltage curves and the average voltage curves of all batteries in the module; and determine a feature mapping model from the individual cell level to the module level based on the first aging feature information and the inconsistency feature information.

[0165] In some specific embodiments, the model determination module 13 can be used to: determine the second aging feature information corresponding to the battery module based on the battery operation database; train a preset hybrid neural network model based on the second aging feature information and the health features in the inconsistency feature information to determine the battery data-driven model at the module level upon completion of training; the preset hybrid neural network model is a neural network model determined based on a bidirectional long short-term memory network and an attention mechanism; determine the equivalent circuit model corresponding to the battery module based on the average model and the difference model corresponding to the battery module; and determine the battery pack power state estimation model under multiple constraints based on the battery data-driven model and the equivalent circuit model.

[0166] In some specific embodiments, the model determination module 13 can be used to: determine an average model and a difference model corresponding to the battery module based on the battery parameters corresponding to each battery in the battery module; the average model is used to characterize the average state of the battery module, and the difference model is used to characterize the deviation between the state of each battery in the battery module and the average state; and determine the equivalent circuit model corresponding to the battery module based on the average model, the difference model, and the extended Kalman filter algorithm.

[0167] In some specific embodiments, the model determination module 13 can be used to: obtain preset multiple constraints; the preset multiple constraints include the charging and discharging current constraints, state of charge constraints, and health state constraints of the battery module; and perform calculations and deductions based on the preset multiple constraints, the battery data-driven model, and the equivalent circuit model to determine the battery pack power state estimation model.

[0168] In some specific embodiments, the situation awareness module 14 can be used to: analyze the series and parallel topology of the battery modules and batteries in the electrochemical energy storage system through topology information to determine the analysis results; determine a multi-level state awareness model based on the feature mapping model and the battery pack power state estimation model; and perform joint sensing of multiple state parameters of the electrochemical energy storage system based on the analysis results and the multi-level state awareness model to determine the situation awareness results; the multiple state parameters include state of charge parameters, health state parameters, and power state parameters.

[0169] Furthermore, embodiments of this application also disclose an electronic device, Figure 11 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0170] Figure 11 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the situational awareness method of the electrochemical energy storage system disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0171] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0172] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0173] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the situational awareness method of the electrochemical energy storage system executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0174] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned situational awareness method for the electrochemical energy storage system. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0175] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0176] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0177] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0178] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0179] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A situational awareness method for an electrochemical energy storage system, characterized in that, include: Collect battery operation data during charge and discharge cycles in the electrochemical energy storage system, and combine it with local laboratory battery aging test data to construct a battery operation database; The battery operation database includes an initial state identification dataset and a state evolution dataset; Feature extraction is performed based on the battery operation database, and a feature mapping model from the single cell level to the module level is determined by using the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module. Based on the preset hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information and the equivalent circuit model, a battery pack power state estimation model under multiple constraints is determined. Based on the feature mapping model, the battery pack power state estimation model, and the topology information corresponding to the electrochemical energy storage system, a multi-state joint estimation is performed to determine the situational awareness result. The step of extracting features based on the battery operation database and determining a feature mapping model from the individual cell level to the module level using the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module includes: The individual charge-discharge curves of the battery are determined based on the battery operation database; Feature extraction is performed based on the single-cell charge-discharge curve to determine the first aging characteristic information corresponding to the battery. Feature extraction is performed based on the battery operation database to determine the inconsistency feature information corresponding to the battery module; the battery module includes a plurality of batteries, and the inconsistency feature information includes the range of the voltage curves of each battery in the module and the distance between the voltage curves and the average voltage curves of all batteries in the module; Based on the first aging feature information and the inconsistency feature information, a feature mapping model from the unit level to the module level is determined. The method for determining the battery pack power state estimation model under multiple constraints based on a preset hybrid neural network model, the second aging characteristic information corresponding to the battery module, the inconsistency characteristic information, and the equivalent circuit model includes: The second aging characteristic information corresponding to the battery module is determined based on the battery operation database. Based on the health features in the second aging feature information and the inconsistency feature information, a preset hybrid neural network model is trained to determine the battery data-driven model at the module level when the training is completed; the preset hybrid neural network model is a neural network model determined based on a bidirectional long short-term memory network and an attention mechanism. Based on the average model and the difference model corresponding to the battery module, the equivalent circuit model corresponding to the battery module is determined; Based on the battery data-driven model and the equivalent circuit model, a battery pack power state estimation model under multiple constraints is determined. The multi-state joint estimation based on the feature mapping model, the battery pack power state estimation model, and the topology information corresponding to the electrochemical energy storage system includes: The analysis results are determined by analyzing the series and parallel topologies of the battery modules and batteries within the electrochemical energy storage system using topology information. Based on the feature mapping model and the battery pack power state estimation model, a multi-level state awareness model is determined. Based on the analysis results and the multi-level state perception model, the electrochemical energy storage system is jointly sensed for multiple state parameters to determine the situational awareness results; the multiple state parameters include state of charge parameters, health state parameters, and power state parameters.

2. The situational awareness method for an electrochemical energy storage system according to claim 1, characterized in that, The system collects operational data of batteries during charge-discharge cycles in the electrochemical energy storage system and combines this data with local laboratory battery aging test data to construct a battery operation database, including: Based on sensors, determine the operating data of batteries in an electrochemical energy storage system during charge and discharge cycles; Obtain local laboratory battery aging test data corresponding to the aforementioned operational data; Based on the laboratory battery aging test data and battery aging data, an evolution analysis of battery performance and battery parameters is conducted to determine the battery operation database; the battery parameters include voltage, current and capacity.

3. The situational awareness method for an electrochemical energy storage system according to claim 1, characterized in that, The step of determining the equivalent circuit model corresponding to the battery module based on the average model and the difference model corresponding to the battery module includes: Based on the battery parameters corresponding to each battery in the battery module, an average model and a difference model corresponding to the battery module are determined; the average model is used to characterize the average state of the battery module, and the difference model is used to characterize the deviation between the state of each battery in the battery module and the average state; Based on the average model, the difference model, and the extended Kalman filter algorithm, the equivalent circuit model corresponding to the battery module is determined.

4. The situational awareness method for an electrochemical energy storage system according to claim 1, characterized in that, The determination of the battery pack power state estimation model under multiple constraints based on the battery data-driven model and the equivalent circuit model includes: Obtain preset multiple constraints; the preset multiple constraints include the charging and discharging current constraints, state of charge constraints, and health state constraints of the battery module. The battery pack power state estimation model is determined by calculation and deduction based on the preset multiple constraints, the battery data-driven model, and the equivalent circuit model.

5. A situational awareness device for an electrochemical energy storage system, characterized in that, include: The database construction module is used to collect the operating data of batteries in the electrochemical energy storage system during the charge and discharge cycle, and combine it with local laboratory battery aging test data to construct a battery operation database; the battery operation database includes an initial state identification dataset and a state evolution dataset. The mapping model construction module is used to extract features based on the battery operation database and use the first aging feature information corresponding to the battery and the inconsistency feature information corresponding to the battery module to determine the feature mapping model from the single cell level to the module level. The model determination module is used to determine the battery pack power state estimation model under multiple constraints based on a preset hybrid neural network model, the second aging feature information corresponding to the battery module, the inconsistency feature information and the equivalent circuit model. The situation awareness module is used to perform multi-state joint estimation based on the feature mapping model, the battery pack power state estimation model, and the topology information corresponding to the electrochemical energy storage system, so as to determine the situation awareness result. The mapping model construction module is configured to: determine the individual charge-discharge curves of the battery based on the battery operation database; extract features based on the individual charge-discharge curves to determine the first aging feature information corresponding to the battery; extract features based on the battery operation database to determine the inconsistency feature information corresponding to the battery module; the battery module includes several batteries, and the inconsistency feature information includes the range of the voltage curves of each battery in the module and the distance between the voltage curves and the average voltage curves of all batteries in the module; and determine a feature mapping model from the individual cell level to the module level based on the first aging feature information and the inconsistency feature information. The model determination module is used to: determine the second aging characteristic information corresponding to the battery module based on the battery operation database; Based on the second aging feature information and the health features in the inconsistency feature information, a preset hybrid neural network model is trained to determine the battery data-driven model at the module level upon completion of training. The preset hybrid neural network model is a neural network model determined based on a bidirectional long short-term memory network and an attention mechanism. Based on the average model and the difference model corresponding to the battery module, the equivalent circuit model corresponding to the battery module is determined. Based on the battery data-driven model and the equivalent circuit model, a battery pack power state estimation model under multiple constraints is determined. The situation awareness module is used to: analyze the series and parallel topology of the battery modules and batteries in the electrochemical energy storage system through topology information to determine the analysis results; determine a multi-level state awareness model based on the feature mapping model and the battery pack power state estimation model; and perform joint sensing of multiple state parameters of the electrochemical energy storage system based on the analysis results and the multi-level state awareness model to determine the situation awareness results; the multiple state parameters include state of charge parameters, health state parameters, and power state parameters.

6. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the situational awareness method for the electrochemical energy storage system as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, Used to store a computer program, which, when executed by a processor, implements the situational awareness method for an electrochemical energy storage system as described in any one of claims 1 to 4.

Images