Vanadium redox flow battery energy storage power distribution method, device and system

Abstract

The application relates to a vanadium flow battery energy storage power distribution method, device and system, and relates to the technical field of energy storage energy management. The method comprises the following steps: performing online parameter identification processing on an acquired equivalent loss model to determine battery identification parameters of a plurality of battery groups; determining the health states of the plurality of battery groups according to the battery identification parameters, original battery resistances and resistance values at a time of scrapping; determining the charging and discharging priorities of the plurality of battery groups according to the battery states of charge and the health states; selecting a plurality of target battery groups participating in power distribution from the plurality of battery groups according to total scheduling demand power and the charging and discharging priorities; determining a multi-objective function with the minimum battery working loss, the minimum state of charge uniformity and the minimum total life loss as targets according to the battery identification parameters and the battery states of charge, and constructing a multi-objective optimization model; and solving the multi-objective optimization model based on constraint conditions to obtain target power distribution results of the plurality of target battery groups.

Description

Technical Field

[0001] This application relates to the field of energy storage management technology, and in particular to a method, device and system for power distribution in vanadium redox flow battery energy storage. Background Technology

[0002] Research has revealed that, with the advancement of the "dual carbon" goal, vanadium redox flow battery (VRB) energy storage systems have become a key supporting technology for addressing the volatility and intermittency of renewable energy sources such as wind and solar power. However, power distribution strategies for energy storage batteries often focus on state of charge (SOC) equalization control, failing to adequately consider the capacity degradation issues that occur under frequent charge-discharge conditions. This leads to overcharging and discharging of batteries with better state of health (SOH), resulting in accelerated degradation rates, while batteries with severely degraded performance are underutilized. Ultimately, this significantly shortens the overall service life of the energy storage system, making it difficult to achieve long-term optimization goals. Currently, most methods for battery pack charging and discharging use fixed weights or single index ranking, which cannot dynamically adapt to the synergistic mechanism of SOC and SOH under different charging and discharging scenarios. This results in insufficient rationality and adaptability of power allocation. Furthermore, the goals for power allocation optimization are singular, mostly focusing only on peak-valley arbitrage, power fluctuation smoothing, or single loss control. They fail to simultaneously consider the multi-objective synergistic optimization of minimizing battery operating losses, optimizing SOC balance, and minimizing lifespan degradation, leading to poor system operating economy and low reliability. Summary of the Invention

[0003] This invention provides a method, apparatus, and system for power allocation in vanadium redox flow batteries, aiming to at least solve the problem that battery packs cannot dynamically adapt to different charging and discharging scenarios, resulting in insufficient power allocation adaptability and low operational economy and reliability. The technical solution of this invention is as follows: According to a first aspect of the present invention, a method for power allocation in a vanadium redox flow battery energy storage system is provided, which includes multiple battery packs. The method includes: acquiring an equivalent loss model of the vanadium redox flow battery; and performing online parameter identification processing on the equivalent loss model to determine battery identification parameters for the multiple battery packs. The equivalent loss model characterizes the correlation between battery parameters within the vanadium redox flow battery. Based on the battery identification parameters, the original resistance of the battery, and the resistance value at the time of disposal, the health status of the multiple battery packs is determined. Based on the state of charge and health status of each battery pack, the charging and discharging priority of the multiple battery packs is determined. The priority characterizes the order in which multiple battery packs are called up based on their current states when they operate simultaneously. Based on the total power demand and charging / discharging priorities, multiple target battery packs are selected from among the multiple battery packs to participate in power allocation. A multi-objective function is determined based on battery identification parameters and battery state of charge (SOC), with the objectives of minimizing battery operating losses, minimizing SOC balance, and minimizing total lifetime loss. A multi-objective optimization model is constructed based on the multi-objective function. The multi-objective optimization model is solved based on constraints to obtain the target power allocation results for the multiple target battery packs. The constraints include total power constraints, SOC constraints, and maximum battery output constraints.

[0004] As one implementation method, online parameter identification processing is performed on the equivalent loss model to determine the battery identification parameters of multiple battery packs. This includes: constructing a state-space model based on the battery equivalent loss model to determine the battery output; the state-space model characterizes the energy storage changes of the battery's internal state; the battery output characterizes the correlation between the battery's external terminal voltage and its internal state; discretizing the state-space model and battery output to obtain a discretized state-space model and a discretized battery output relationship; generating a linear relationship between data vectors and parameter vectors based on the discretized state-space model and discretized battery outputs; recursively updating the parameter vectors based on the linear relationship to obtain the target parameter vector; and determining the battery identification parameters based on the target parameter vector.

[0005] In this implementation, the battery is a time-varying system. By identifying dynamically changing parameters online, the model can always remain consistent with the current real state of the battery, thus enabling dynamic, accurate, and real-time capture of changes in the battery's internal loss characteristics.

[0006] One implementation method involves identifying battery parameters, including ohmic internal resistance. Based on these parameters, the original battery resistance, and the resistance value at the point of failure, the health status of multiple battery packs is determined. This includes: determining the maximum usable capacity of each battery pack based on the difference between the resistance value at the point of failure and the ohmic internal resistance; determining the rated battery capacity based on the difference between the resistance value at the point of failure and the original battery resistance; and determining the health status of multiple battery packs based on the percentage of the maximum usable capacity to the rated battery capacity.

[0007] In this embodiment, the characteristic parameters obtained by online identification of ohmic internal resistance are used to determine the SOH value in real time during normal battery operation. This enables rapid, low-cost, and real-time estimation of battery health status, realizing real-time monitoring of the battery degradation process and providing a key decision-making basis for subsequent power allocation.

[0008] As one implementation method, the charging and discharging priority of multiple battery packs is determined based on the state of charge and health of each battery pack. This includes: determining the comprehensive score of each battery pack based on its state of charge and health; arranging the multiple battery packs in descending order based on their comprehensive scores to determine the charging and discharging priority; and assigning the highest charging and discharging priority to the battery pack with the highest comprehensive score.

[0009] In this embodiment, by integrating SOC and SOH to determine charging and discharging priorities, energy management and lifespan management are organically combined. Under the premise of ensuring system safety, energy utilization efficiency is maximized and the overall service life is extended, thus achieving efficient, reliable and refined control of multiple battery packs.

[0010] One implementation method involves determining the comprehensive score of each battery pack based on its state of charge (SOC) and state of health (SQH). This includes: constructing an original evaluation matrix based on the SOC and SQH of each battery pack; normalizing the original evaluation matrix according to the current charging / discharging scenario to obtain a normalized evaluation matrix; the normalized evaluation matrix includes a SOC matrix and a SQH matrix; determining the weight of each battery pack under different indicators based on the normalized evaluation matrix; the indicators include SOC indicators and SQH indicators; determining the first information entropy corresponding to the SOC indicator and the second information entropy corresponding to the SQH indicator based on the weights; determining the SOC weight and SQH weight based on the first and second information entropies; and determining the weighted sum of the SOC matrix and SQH matrix based on the SOC weight and SQH weight to obtain the comprehensive score of each battery pack.

[0011] As one implementation method, based on the total power demand and charging / discharging priority, multiple target battery packs are selected from multiple battery packs to participate in power allocation. This includes: accumulating the available power of multiple battery packs sequentially according to charging / discharging priority until the accumulated value of available power is greater than or equal to the total power demand; and determining the multiple battery packs participating in the accumulation as target battery packs.

[0012] In this implementation, while ensuring real-time scheduling, the system lifespan and energy efficiency are maximized by prioritizing the use of healthy batteries and avoiding redundant scheduling.

[0013] As one implementation method, battery identification parameters also include reactive resistance and parasitic resistance. Based on the battery identification parameters and the battery state of charge (SOC), a multi-objective function is determined, aiming to minimize battery operating losses, SOC balance, and total lifetime loss. This includes: determining battery operating losses based on reactive resistance, ohmic internal resistance, parasitic resistance, and the SOC of each battery pack; determining equipment losses based on equipment operating losses, standby losses, and the operating status of each battery pack; determining operating losses based on battery operating losses, equipment losses, and operating status; determining SOC balance based on the squared difference between the SOC of each battery pack and the mean SOC; determining total lifetime loss based on the sum of lifetime losses of each battery pack; and determining the multi-objective function based on the weighted sum of operating losses, SOC balance, and total lifetime loss.

[0014] In this implementation, a multi-objective optimization model is constructed with the goals of minimizing battery operating losses, minimizing state-of-charge balance, and minimizing total lifetime loss. This model takes into account the energy loss during battery power distribution, the lifespan of the battery pack, and the state-of-charge balance of the battery pack from multiple perspectives. As a result, the economy, reliability, and lifespan of the battery system are maximized while ensuring safety.

[0015] As one implementation method, based on constraints, the multi-objective optimization model is solved to obtain the target power allocation results for multiple target battery packs. This includes: constructing a total power constraint based on the sum of the allocated power of each target battery pack and the total power required for scheduling; constructing a state of charge constraint based on the state of charge range of each battery pack; and constructing a maximum output constraint for the battery based on the maximum discharge power and maximum charging power of the allocated power. Based on a dynamic multi-objective evolutionary algorithm using generative adversarial networks, the multi-objective optimization model is iteratively solved to obtain the target power allocation results for multiple target battery packs.

[0016] According to a second aspect of the present invention, a vanadium redox flow battery energy storage power distribution device is provided, the device comprising: The equivalent loss model construction unit is configured to obtain the equivalent loss model of the vanadium redox flow battery, and to perform online parameter identification processing on the equivalent loss model to determine the battery identification parameters of multiple battery packs; the equivalent loss model characterizes the correlation between the battery parameters inside the vanadium redox flow battery.

[0017] The health status determination unit is configured to determine the health status of multiple battery packs based on battery identification parameters, original battery resistance, and resistance value at the time of scrapping.

[0018] The initial power allocation unit is configured to determine the charging and discharging priorities of multiple battery packs based on the battery state of charge and health status of each battery pack; the charging and discharging priority represents the order in which multiple battery packs are called up when they are running simultaneously, based on the current state of each battery pack; and selects multiple target battery packs to participate in power allocation from multiple battery packs based on the total power required for scheduling and the charging and discharging priority.

[0019] The target power allocation unit is configured to determine a multi-objective function based on battery identification parameters and battery state of charge, with the objectives of minimizing battery operating loss, minimizing state of charge balance, and minimizing total lifetime loss; construct a multi-objective optimization model based on the multi-objective function; and solve the multi-objective optimization model based on constraints to obtain the target power allocation results for multiple target battery packs; the constraints include total power constraints, state of charge constraints, and maximum battery output constraints.

[0020] According to a third aspect of the present invention, a vanadium redox flow battery energy storage power distribution system is provided, the system being configured to perform a vanadium redox flow battery energy storage power distribution method as described in the first aspect and any possible implementation thereof.

[0021] According to a fourth aspect of the present invention, a vanadium redox flow battery energy storage power distribution device is provided, the device being configured to perform a vanadium redox flow battery energy storage power distribution method as described in the first aspect and any possible implementation thereof.

[0022] According to a fifth aspect of the present invention, a computer-readable storage medium is provided, on which instructions are stored, such that when the instructions in the computer-readable storage medium are executed by a processor of a vanadium redox flow battery energy storage power distribution device, the vanadium redox flow battery energy storage power distribution device is able to perform a vanadium redox flow battery energy storage power distribution method as described in the first aspect and any possible implementation thereof.

[0023] According to a sixth aspect of the present invention, a computer program product is provided, the computer program product including computer instructions, which, when the computer instructions are executed on a vanadium redox flow battery energy storage power distribution device, cause the vanadium redox flow battery energy storage power distribution device to perform the vanadium redox flow battery energy storage power distribution method described in the first aspect and any possible implementation thereof.

[0024] The technical solutions provided by the embodiments of the present invention bring at least the following beneficial effects: The present invention constructs an equivalent loss model of a vanadium redox flow battery, thereby revealing more intuitively the intrinsic correlation between various basic battery parameters inside the battery. Furthermore, the equivalent loss model of the battery is used for online parameter identification. Since the battery identification parameters inside the battery change with temperature, state of charge, or degree of loss during use, online parameter identification enables real-time capture of battery identification parameters. This provides a real-time and accurate data foundation for subsequent health status assessment and power allocation optimization decisions, allowing power allocation to dynamically adapt to different scenarios. Moreover, the health status of each battery pack is evaluated through battery identification parameters, and the health status of each battery pack is combined with the battery state of charge. The system prioritizes battery packs for initial power allocation, quickly identifying target battery packs for power allocation. Then, a multi-objective optimization model is constructed, aiming to minimize battery operating losses, state-of-charge (POC) balance, and total lifespan loss. This model considers energy loss during power allocation, battery pack lifespan, and POC balance from multiple perspectives, ultimately obtaining the optimal power allocation for each target battery pack. This achieves reasonable power distribution and balanced POC, improving the overall energy conversion efficiency of battery storage allocation, effectively preventing premature failure of some batteries due to long-term overcharging or over-discharging, extending the battery pack replacement cycle, reducing total lifespan costs, ensuring system reliability and safety, and ultimately improving overall operational economy.

[0025] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0026] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0027] Figure 1 This is a schematic diagram of an all-vanadium redox flow battery energy storage power distribution system according to an exemplary embodiment; Figure 2 This is a flowchart illustrating a method for all-vanadium redox flow battery energy storage power distribution according to an exemplary embodiment; Figure 3This is a circuit schematic diagram illustrating an equivalent loss model according to an exemplary embodiment; Figure 4 This is a chart showing a comparison of average random consistency index (RI) values ​​according to an exemplary embodiment; Figure 5 This is a flowchart illustrating a multi-objective optimization evolutionary algorithm based on the NSGA-II framework, according to an exemplary embodiment. Figure 6 This is a flowchart illustrating a GAN-DMOEA algorithm according to an exemplary embodiment; Figure 7 This is a schematic diagram of a vanadium redox flow battery energy storage power distribution device according to an exemplary embodiment; Figure 8 This is a schematic diagram of a vanadium redox flow battery energy storage power distribution device according to an exemplary embodiment. Detailed Implementation

[0028] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0029] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0030] Before providing a detailed description of the vanadium redox flow battery energy storage power allocation method provided in this application embodiment, let's briefly introduce the application scenarios and implementation environment involved in this application embodiment.

[0031] Vanadium redox flow batteries (VRBs), as a large-scale electrochemical energy storage technology, have broad application prospects in renewable energy integration and grid peak shaving and valley filling due to their advantages such as high safety, long cycle life, and decoupling of power and capacity. With the advancement of the "dual carbon" goal, VRB energy storage systems have become a key supporting technology for addressing the volatility and intermittency of renewable energy sources such as wind and solar power. However, VRBs still face challenges in practical applications, including high initial investment costs, complex operation and maintenance, and difficulties in optimizing system efficiency.

[0032] Currently, power allocation strategies for energy storage batteries mostly focus on State of Charge (SOC) equalization control, failing to adequately consider the capacity degradation issues that occur under frequent charge-discharge conditions. This leads to overcharging and discharging of batteries with better State of Health (SOH), resulting in accelerated degradation rates, while batteries with severely degraded performance are underutilized, ultimately significantly shortening the overall service life of the energy storage system and making it difficult to achieve long-term optimization goals. Furthermore, existing research lacks scientific quantitative methods for prioritizing battery charge and discharge operations. Existing solutions often use fixed weights or single-index ranking, failing to dynamically adapt to the synergistic effects of SOC and SOH under different charge-discharge scenarios, resulting in insufficient rationality and adaptability of power allocation. In addition, existing power allocation optimization models have relatively singular objectives, mostly focusing only on peak-valley arbitrage, power fluctuation smoothing, or single loss control, failing to simultaneously consider multi-objective synergistic optimization of minimizing battery operating losses, optimizing SOC equalization, and minimizing lifespan degradation, making it difficult to balance system operating economy and reliability.

[0033] To address the aforementioned issues, this application proposes a power allocation method for vanadium redox flow battery (VRB) energy storage. The method involves online parameter identification using a VRB equivalent model, estimating the battery's state of health, and prioritizing battery charging and discharging based on key state-of-charge and health indicators. Furthermore, during VRB operation, a multi-objective collaborative optimization model is constructed, comprehensively considering objectives such as operating losses, SOC balancing, and lifespan degradation. Finally, a Dynamic Multi-Objective Evolutionary Algorithm (GAN-DMOEA) based on generative adversarial networks is used to solve the model, obtaining the optimal power allocation for each target battery pack. This achieves reasonable power allocation and balanced state of charge, improving the overall energy conversion efficiency of the battery storage allocation, effectively preventing premature failure of some batteries due to long-term overcharging or over-discharging, extending the battery pack replacement cycle, reducing the total lifespan cost, ensuring the reliability and safety of the system operation, and ultimately improving the overall operational economy.

[0034] Secondly, the implementation architecture involved in this application will be briefly introduced below.

[0035] Figure 1 This is a schematic diagram of a vanadium redox flow battery energy storage power distribution system provided in this application. Figure 1 As shown, the all-vanadium redox flow battery energy storage power distribution system includes a parameter identification module 11, a health status assessment module 12, a charge / discharge priority ranking module 13, a model building module 14, and a solution module 15.

[0036] The parameter identification module 11, health status assessment module 12, charge / discharge priority ranking module 13, model building module 14, and solution module 15 are connected via communication.

[0037] The parameter identification module 11 is configured to acquire the equivalent loss model of the vanadium redox flow battery, and to perform online parameter identification processing on the equivalent loss model to determine the battery identification parameters of multiple battery packs.

[0038] The health status assessment module 12 is configured to determine the health status of multiple battery packs based on battery identification parameters, original battery resistance, and resistance value at the time of scrapping.

[0039] The charge / discharge priority sorting module 13 is configured to determine the charge / discharge priority of multiple battery packs based on the state of charge and health status of each battery pack. Based on the total power demand and charge / discharge priority, it selects multiple target battery packs from the multiple battery packs to participate in power allocation.

[0040] The model building module 14 is configured to determine a multi-objective function based on the battery identification parameters and the battery state of charge, with the objectives of minimizing battery operating loss, minimizing state of charge balance, and minimizing total lifetime loss; and to build a multi-objective optimization model based on the multi-objective function.

[0041] The solver module 15 is configured to solve the multi-objective optimization model based on constraints to obtain the target power allocation results for multiple target battery packs.

[0042] For ease of understanding, the power distribution method for all-vanadium redox flow battery energy storage provided in this application will be described in detail below with reference to the accompanying drawings.

[0043] Figure 2 This is a flowchart illustrating a power distribution method for an all-vanadium redox flow battery energy storage system according to an exemplary embodiment, such as... Figure 2 As shown, the power distribution method for vanadium redox flow battery energy storage includes the following steps.

[0044] S21, obtain the equivalent loss model of the vanadium redox flow battery, and perform online parameter identification processing on the equivalent loss model to determine the battery identification parameters of multiple battery packs.

[0045] The equivalent loss model characterizes the correlation between internal battery parameters of the vanadium redox flow battery.

[0046] Because the internal mechanism of a vanadium redox flow battery (VRB) is complex, involving many physical quantities that are difficult to measure directly, such as pump loss, internal resistance, and open-circuit voltage variation, the equivalent loss model uses mathematical equations to correlate these invisible internal characteristics with measurable external characteristics such as terminal voltage, current, and temperature. The circuit diagram of its equivalent loss model is shown below. Figure 3 Characterization.

[0047] like Figure 3 As shown, This is the voltage across the electrode capacitor; Parasitic resistance; Reaction resistance; The internal resistance is ohmic; The charging and discharging current of the battery; This refers to the pump loss current. This is the open-circuit voltage, which is related to the state of charge (SOC). This refers to the current in the fuel cell stack. The current flowing through the parasitic resistance; Electrode capacitance; This is the terminal voltage across VRB; This represents the current flowing through the reactive resistor.

[0048] Based on the equivalent loss model of a vanadium redox flow battery, a state-space model expressing the internal and external states of the battery pack and the battery output are established. Using this model, online parameter identification processing is performed on the equivalent loss model to determine the battery identification parameters for multiple battery packs.

[0049] In one implementation, online parameter identification of the equivalent loss model is performed, which specifically includes the following three steps.

[0050] First, based on the battery equivalent loss model, a state-space model is constructed to determine the battery output.

[0051] Among them, the state-space model represents the changes in energy storage within the battery's internal state.

[0052] Battery output characterizes the relationship between the external terminal voltage of the battery and the internal state of the battery.

[0053] Based on the above Figure 3 The relationship between the parameters in the equivalent loss model shown is used to construct the VRB state space model, and its state space expression is specifically represented by formula (1).

[0054] (1).

[0055] in, The voltage across the electrode capacitor; Parasitic resistance; Reaction resistance; The internal resistance is ohmic; The charging and discharging current of the battery; This refers to the pump loss current. Electrode capacitance; This refers to the number of batteries connected in series within a single battery cell. It is the gas constant; The standard electrode potential of the battery is SOC; the state of charge (SOC) is the battery's state of charge. This refers to the current in the fuel cell stack. The current flowing through the parasitic resistance; For temperature; It is Faraday's constant; The rated capacity of the battery; This is the terminal voltage across VRB.

[0056] Based on the above state-space expression, its battery output expression is determined as follows: Formula (2) for specific characterization.

[0057] (2).

[0058] in, This is the terminal voltage across VRB; Parasitic resistance; The internal resistance is ohmic; The charging and discharging current of the battery; This refers to the pump loss current. This is the voltage across the electrode capacitor.

[0059] Secondly, the state-space model and battery output are discretized to obtain a discretized state-space model and a discretized battery output. Based on the discretized state-space model and the discretized battery output, a linear relationship between the data vector and the parameter vector is generated.

[0060] Specifically, the Forgetting Factor Recursive Least Square (FFRLS) method is used to identify the parameters of the equivalent loss model online. By discretizing the state-space expression and the battery output, real-time parameter data is accurately obtained, laying the foundation for battery health status assessment and power allocation.

[0061] First, based on the preset sampling period, the backward difference method is used to discretize the above state space expression and battery output expression to obtain the discretized state space model and the discretized battery output. The discretized state space model is specifically represented by formula (3).

[0062] (3).

[0063] in, This represents the voltage coefficient at the electrode capacitor terminals; A is the current coefficient at the previous moment; A is the open-circuit voltage coefficient; T is the sampling period; This represents the voltage across the electrode capacitor at the current moment. This represents the current state of battery charge. This is the open-circuit voltage. Wherein, , A and The specific expressions are represented by the following formulas (4) to (7).

[0064] (4).

[0065] (5).

[0066] (6).

[0067] (7).

[0068] The discretized battery output is specifically characterized by the following formula (8).

[0069] (8).

[0070] in, B is the current coefficient at the current moment; B is the voltage coefficient across the electrode capacitor. This represents the current battery terminal voltage. The specific expressions for B are represented by the following formulas (9) and (10).

[0071] (9).

[0072] (10).

[0073] Substituting the discretized formula (3) into formula (8), we obtain the standard form required by the FFRLS algorithm, as specifically represented by the following formula (11).

[0074] (11).

[0075] in, This represents the voltage coefficient at the electrode capacitor terminals; The current coefficient at the previous moment; The current coefficient at the current moment; is a constant coefficient. The specific expression is represented by the following formula (12).

[0076] (12).

[0077] Then, based on the discretized state-space model and the discretized battery output, a linear relationship between the data vector and the parameter vector is generated.

[0078] The core of the FFRLS algorithm is to transform a nonlinear system into a linear parameter estimation problem. The standard form of its linear relationship is represented by the following formulas (13) to (15).

[0079] (13).

[0080] (14).

[0081] (15).

[0082] in, This is the system output, that is... Battery terminal voltage; It is a data vector; The parameter vector to be identified.

[0083] Third, based on the linear relationship, the parameter vector is recursively updated to obtain the target parameter vector. Based on the target parameter vector, the battery identification parameters are determined. The main process of using the FFRLS algorithm to identify parameters is as follows: four steps.

[0084] 1. Calculate the gain matrix. Based on the forgetting factor, the covariance matrix of the previous time step, and the data vector of the current time step, determine the current gain matrix, which is represented by the following formula (16).

[0085] (16).

[0086] in, This is the current gain matrix; Let be the covariance matrix of the previous time step; This is the data vector at the current moment; for The transpose of is used to calculate the predicted value; The forgetting factor is mainly used to gradually discount the contribution of historical data through exponential weighting, while increasing the influence of new and recent data in parameter updates. Its value is generally set between 0.955 and 0.99.

[0087] 2. Update the parameter estimation and determine the target parameter vector. Based on the current gain matrix and the parameter vector of the previous time step, the parameter vector is corrected and updated to obtain the parameter vector at the current time step, as represented by the following formula (17).

[0088] (17).

[0089] in, This is the parameter vector at the current time. This is the parameter vector from the previous time step; This is the current gain matrix; This is the system output.

[0090] 3. Update the covariance matrix. The covariance matrix is ​​updated based on the covariance matrix of the previous time step, the data vector of the current time step, the current gain matrix, and the forgetting factor to obtain the current covariance matrix. This prepares for the next iteration. Specifically, it is represented by the following formula (18).

[0091] (18).

[0092] in, This is the current covariance matrix; This is the current gain matrix; Let be the covariance matrix of the previous time step; This is the data vector at the current moment; It is a forgetting factor.

[0093] IV. Determine the battery identification parameters based on the target parameter vector.

[0094] The target parameter vector obtained by the above formula (17) is mapped onto the parameters to be identified, thereby performing reverse calculation to determine the battery identification parameters. The following formula (19) specifically represents this.

[0095] (19).

[0096] in, Parasitic resistance; Reaction resistance; The internal resistance is ohmic; This refers to the electrode capacitance.

[0097] In this implementation, the battery is a time-varying system. A state-space model is constructed to characterize the changes in energy storage within the battery, focusing on the static relationship of the battery terminal voltage and the dynamic loss process within the battery. The FFRLS algorithm is then used to identify dynamically changing parameters online, ensuring that the model always remains consistent with the battery's current true state. This enables the dynamic, accurate, and real-time capture of changes in the battery's internal loss characteristics. This not only improves the accuracy of battery state estimation but also enhances adaptability to different operating conditions and aging levels, thereby ensuring the safety and efficiency of the battery system.

[0098] S22 determines the health status of multiple battery packs based on battery identification parameters, original battery resistance, and resistance value at the time of scrapping.

[0099] The health status of a battery pack can be defined as the percentage of maximum available capacity to rated capacity or expressed by the increase in internal resistance.

[0100] In one implementation, the maximum usable capacity of each battery pack is determined based on the difference between the resistance value at the time of scrapping and the ohmic internal resistance; the rated capacity of the battery is determined based on the difference between the resistance value at the time of scrapping and the original resistance of the battery; and the health status of multiple battery packs is determined based on the percentage of the maximum usable capacity to the rated capacity of the battery.

[0101] Based on the definition of internal resistance, the specific expression is represented by the following formula (20).

[0102] (20).

[0103] in, The resistance at the moment when the battery is scrapped is generally considered to be when the battery's health is only 80% and it can be scrapped. This is the current state resistance; This is the original resistance of the battery.

[0104] The resistance at the moment of battery failure is defined by the following formula (21).

[0105] (twenty one).

[0106] in, The resistance at the moment the battery is worn out. This is the original resistance of the battery.

[0107] In summary, the battery health is assessed based on the parameters of the battery identified online by FFRLS in step S21, as specifically represented by the following formula (22).

[0108] (twenty two).

[0109] in, Estimate the health of each battery pack; The parameters identified correspond to those identified online. Ohmic internal resistance.

[0110] In this embodiment, the characteristic parameters obtained by online identification of ohmic internal resistance are used to determine the SOH value in real time during normal battery operation. This enables rapid, low-cost, and real-time estimation of battery health status, realizing real-time monitoring of the battery degradation process and providing a key decision-making basis for subsequent power allocation.

[0111] S23, determine the charging and discharging priority of multiple battery packs based on the state of charge and health status of each battery pack.

[0112] Charging and discharging priority indicates the order in which multiple battery packs are called when they are running simultaneously, based on the current state of each battery pack.

[0113] Based on two important factors, SOH and SOC, the entropy weight method is used to calculate the scores of different battery packs by considering both factors simultaneously, thereby determining the battery packs that participate in power distribution, so as to select the optimal battery pack and extend the overall life of the system.

[0114] In one implementation, determining the charging and discharging priority of the battery pack includes the following two steps.

[0115] First, the overall score of each battery pack is determined based on the state of charge and health status of each battery pack.

[0116] First, based on the state of charge and health of each battery pack, an original evaluation matrix is ​​constructed, as represented by the following formula (23).

[0117] (twenty three).

[0118] Where Q is the original evaluation matrix; SOC is the state of charge of each battery pack; SOH is the state of health of each battery pack; and N is the number of battery packs.

[0119] Secondly, based on the current charging and discharging scenario, the original evaluation matrix is ​​normalized to obtain a normalized evaluation matrix.

[0120] The positive or negative orientation of the indicators is determined based on the charging and discharging scenarios, and normalization is performed to eliminate the influence of dimensions, resulting in a normalized matrix. The normalized evaluation matrix includes a state-of-charge matrix and a health state matrix.

[0121] When scheduling demand power When the system needs to discharge, the battery pack with high SOC and high SOH is most suitable for discharge. This is specifically represented by the following formulas (24) and (25).

[0122] (twenty four).

[0123] (25).

[0124] in, This is the charge state matrix under discharge conditions; This is the health state matrix under discharge conditions.

[0125] When scheduling demand power When the system needs charging, the battery pack with low SOC and high SOH is most suitable for charging. This is specifically represented by the following formulas (26) and (27).

[0126] (26).

[0127] (27).

[0128] in, This is the charge state matrix under charging conditions; This is the health status matrix during charging.

[0129] After completing the above steps, the proportion of each battery pack under different indicators is determined based on the normalized evaluation matrix. The indicators include the state of charge indicator and the state of health indicator. Specifically, they are represented by the following formula (28).

[0130] (28).

[0131] in, For the first The battery pack in the first The proportion under each indicator; State of charge (SOC) index It is an indicator of SOH health status.

[0132] Then, based on the proportions, the first information entropy corresponding to the state of charge index and the second information entropy corresponding to the health state index are determined. Specifically, they are represented by the following formula (29).

[0133] (29).

[0134] in, Information entropy; For the first The battery pack in the first The proportion of each indicator; N is the number of battery packs.

[0135] Then, based on the first and second information entropies, the weights of the charged state and the health state are determined. Specifically, they are represented by the following formula (30).

[0136] (30).

[0137] in, As weight, This refers to information entropy. The smaller the information entropy, the greater the variability of the indicator, and the greater its weight. This is used to determine the weights for the state of charge and the state of health. and ,and .

[0138] Finally, based on the state of charge weights and health state weights, the weighted sum of the state of charge matrix and health state matrix is ​​determined to obtain the comprehensive score of each battery pack. This is specifically represented by the following formula (31).

[0139] (31).

[0140] in, For the overall score; Weights for the state of charge; Weighted by health status; This is the charged state matrix; This is a health status matrix.

[0141] In this way, by considering both SOC and SOH, two key state indicators, scheduling decisions are made more scientific, ensuring a balance between energy availability and lifespan protection for the battery pack, and improving the energy utilization efficiency of the entire system.

[0142] Secondly, based on the comprehensive score, the multiple battery packs are sorted in descending order to determine the charging and discharging priority.

[0143] Among them, the battery pack with the highest overall score has the highest charging and discharging priority.

[0144] Both State of Charge (SOC) and State of Health (SOH) are time-varying parameters, with their priorities updated in real time according to the battery state, providing dynamic response capabilities and allowing for flexible adaptation to different application scenarios. Thus, if the system urgently needs energy throughput (such as grid frequency regulation), the SOC weight can be appropriately increased to prioritize the use of batteries with sufficient energy. If the system operates for long periods and has high lifespan requirements, the SOH weight can be increased to prioritize the protection of healthy batteries.

[0145] In this implementation, by integrating SOC and SOH to determine charging and discharging priorities, energy management and lifespan management are organically combined. Under the premise of ensuring system safety, energy utilization efficiency is maximized and the overall service life is extended, achieving efficient, reliable and refined control of multiple battery packs.

[0146] S24: Based on the total power required for scheduling and the charging / discharging priority, select multiple target battery packs from multiple battery packs to participate in power allocation.

[0147] In a real-time mode, the available power of multiple battery packs is accumulated sequentially according to charging and discharging priorities until the accumulated value of available power is greater than or equal to the total power required for scheduling; and the multiple battery packs participating in the accumulation are identified as the target battery packs.

[0148] Understandably, starting with the highest priority battery pack, its available power is added to the current total available power, and this battery pack is marked as the target battery pack, i.e., included in the set participating in power allocation. This process is repeated, adding battery packs of the next higher priority in turn, until the current total available power is greater than or equal to the total power required for scheduling for the first time, or until all battery packs have been traversed, but the current total available power is still less than the total power required for scheduling.

[0149] Optionally, if the total available power exceeds the demand after adding a battery pack, a refined power allocation is performed on that battery pack and the previously selected target battery packs according to a multi-objective optimization model. Furthermore, unselected battery packs are placed in standby mode and temporarily excluded from this power allocation.

[0150] In this implementation, the charging and discharging priority screening mechanism ensures the optimal priority scheduling state, preferentially selecting high-SOC and healthy battery packs for discharging, or preferentially selecting low-SOC and healthy battery packs for charging, thereby achieving balanced use and lifespan maintenance of battery packs while meeting the total power demand.

[0151] S25. Based on the battery identification parameters and battery state of charge, determine a multi-objective function with the objectives of minimizing battery operating loss, minimizing state of charge balance, and minimizing total lifetime loss. Based on this multi-objective function, construct a multi-objective optimization model.

[0152] Battery identification parameters also include reaction resistance and parasitic resistance.

[0153] The multi-objective optimization model is specifically a multi-objective optimization model that takes into account the decay characteristics.

[0154] In one implementation, after determining the battery pack's charging and discharging priorities, the objective function for power allocation is designed. The optimization objectives include minimizing battery operating losses, achieving SOC balance, and minimizing battery lifespan degradation. Establishing the multi-objective function involves the following four steps.

[0155] First, we need to define the first objective function: battery operating loss. Battery operating loss is divided into battery operating loss and equipment loss.

[0156] First, the battery operating losses are determined based on the reaction resistance, ohmic internal resistance, parasitic resistance, and the state of charge of each battery pack. Battery operating losses include internal resistance losses, parasitic losses, and pump-up losses. Battery operating losses are specifically characterized by formulas (32) and (33).

[0157] (32).

[0158] in, This is due to the internal resistance loss of the battery; For parasitic loss; This is for pump lift losses; Reaction resistance; The internal resistance is ohmic; This is the terminal voltage across VRB; Parasitic resistance; This is the pump rise loss constant; I is the stack current; I is the reaction resistance. The current; This represents the state of charge of the i-th battery pack.

[0159] (33).

[0160] in, Battery operating losses; This is due to the internal resistance loss of the battery; For parasitic loss; This is for pump lift losses.

[0161] Secondly, the equipment loss is determined based on the equipment's operating loss, standby loss, and the operating status of each battery pack. Equipment loss is divided into equipment operating loss and standby loss. The specific equipment loss is represented by formula (34).

[0162] (34).

[0163] in, To allocate power; For equipment wear and tear; This is standby power consumption; For the first The status of each battery pack This indicates that the device is powered on. This indicates that it is in standby mode; Rated power of the DC / DC converter; For optimal conversion efficiency, the conversion efficiency of a DC / DC converter can be set at 95%.

[0164] Finally, the operating loss is determined based on the battery operating loss, equipment loss, and operating status. The operating loss is specifically characterized by formula (35).

[0165] (35).

[0166] in, For work-related losses; For the first The status of each battery pack; Battery operating losses; Equipment wear and tear.

[0167] Secondly, determine the second objective function: the state of charge balance.

[0168] The state-of-charge balance of a battery pack is usually assessed by calculating the dispersion of the SOC of each cell within the pack relative to its mean.

[0169] Specifically, the state of charge balance is determined based on the difference between the state of charge of each battery pack and the mean state of charge, and is characterized by the following formula (36).

[0170] (36).

[0171] Where F2 is the state of charge balance; This represents the state of charge of the i-th battery pack. This is the average value of the state of charge.

[0172] Third, determine the third objective function: total lifetime loss.

[0173] The total lifespan loss is determined by the sum of the lifespan losses of each battery pack. Specifically, it is represented by the following formula (37).

[0174] (37).

[0175] in, Total lifespan loss; For the first The lifespan of each battery pack is depleted.

[0176] The process for determining the life loss of a battery pack is specifically represented by the following formulas (38) to (40).

[0177] (38).

[0178] in, For the first The lifespan loss of each battery pack; 0.5 indicates "half a cycle", which is equivalent to half of a full cycle, that is, half of the charging and discharging; This indicates the decay rate per cycle.

[0179] (39).

[0180] in, This represents the percentage of capacity decay after one complete cycle (one charge and one discharge) at 100% DOD, and needs to be obtained by fitting cycle life test data. Indicates the equivalent DOD; This represents the power constant, which is obtained by fitting experimental data.

[0181] (40).

[0182] in, Indicates the first Rated capacity of each battery pack; To distribute power; Δt represents the time of one cycle.

[0183] Fourth, establish the final multi-objective function.

[0184] First, extreme value normalization is used to linearly map the objective function values ​​to the 0-1 interval, thus determining the normalized values ​​of each objective function. This is specifically represented by formula (41).

[0185] (41).

[0186] in, Let 1, 2, 3 represent the k-th objective function mentioned above; Characterizes the normalized value of the k-th objective function; Let k be the objective function; This represents the minimum value of the k-th objective function; This represents the maximum value of the k-th objective function.

[0187] The multi-objective function is determined by the weighted sum of working loss, state-of-charge balance, and total lifetime loss. It is specifically represented by the following formula (42).

[0188] (42).

[0189] in, The minimum weighting coefficient for work loss; The minimum weighting coefficient for SOC balance; The coefficient with the lowest weight for health degradation; , , These are the normalized values ​​for work loss, SOC balance, and health decay, respectively.

[0190] In this way, by using a weighted summation method, the optimal balance is found among the three dimensions of energy efficiency, battery state of charge consistency, and overall battery pack lifespan, thereby maximizing the economy, reliability, and lifespan of the battery system while ensuring safety.

[0191] Furthermore, after constructing the objective function, it is necessary to determine the weight coefficients of the three objectives. The Analytic Hierarchy Process (AHP) is used to determine the weight coefficients, and the specific steps are as follows.

[0192] First, construct a judgment matrix to determine the importance of each objective function. The judgment matrix is ​​obtained using the 1-9 scaling method of AHP. Specifically, it is represented by formula (43).

[0193] (43).

[0194] Secondly, weight calculation is performed by judging the matrix. The largest eigenvalue and its corresponding eigenvector are calculated. Then, the eigenvector is normalized to obtain weights of 0.4286, 0.1429, and 0.4286, and the largest eigenvalue is obtained. .

[0195] Finally, a consistency check is performed, where the check formula is represented by formula (44).

[0196] (44).

[0197] in, To determine the random consistency ratio of the matrix; This serves as a general consistency index for the discrimination matrix; This is used as an average consistency index for judging the matrix.

[0198] The general consistency index of the discriminant matrix is ​​specifically represented by formula (45).

[0199] (45).

[0200] in, This refers to the number of indicators, i.e., the number of weights used for verification. It is the largest eigenvalue.

[0201] Then, through Figure 4 The chart showing the average random consistency index (RI) values ​​is used to determine the RI values ​​of judgment matrices of orders 1-9.

[0202] If the judgment matrix of Then it is believed It has satisfactory consistency; if it does not, the matrix needs to be changed. The elements satisfy the consistency check.

[0203] Based on the above verification process, the number of indicators was determined to be 3, with a corresponding RI value of 0.58. ,get =0, therefore we know Therefore, the weight allocation is valid.

[0204] S26. Based on the constraints, the multi-objective optimization model is solved to obtain the target power allocation results for multiple target battery packs.

[0205] The constraints include total power constraints, state of charge constraints, and maximum battery output constraints.

[0206] First, determine the constraints, considering the relationship between dispatch power and allocated power, the SOC constraint of the battery pack, and the maximum charging and discharging power of the battery.

[0207] First, a total power constraint is constructed based on the sum of the allocated power of each target battery pack and the total power required for scheduling. The total power required for scheduling is equal to the sum of the allocated power of each battery pack, as specifically represented by the following formula (46).

[0208] (46).

[0209] in, For power scheduling; To allocate power.

[0210] Secondly, based on the state of charge range of each battery pack, state of charge constraints are constructed. The upper and lower limits of the SOC of the battery pack are represented by the following formula (47).

[0211] (47).

[0212] in, This is the lower limit of SOC; This represents the upper limit of SOC.

[0213] Third, based on the maximum discharge power and maximum charging power of the allocated power, the maximum output power constraint of the battery is constructed. The maximum output power constraint of the battery is specifically characterized by the following formula (48).

[0214] (48).

[0215] in, This represents the maximum discharge power. This is the maximum charging power.

[0216] Then, based on the dynamic multi-objective evolutionary algorithm of generative adversarial networks, the multi-objective optimization model is iteratively solved to obtain the target power allocation results of multiple target battery packs.

[0217] After completing the parameter identification of the VRB equivalent model, SOH estimation, and the establishment of a multi-objective optimization model taking into account attenuation characteristics based on the above steps, an efficient solution algorithm is performed based on the dynamic multi-objective evolutionary algorithm of generative adversarial networks.

[0218] In one implementation, a Dynamic Multi-Objective Evolutionary Algorithm (GAN-DMOEA) based on generative adversarial networks is used. The core idea of ​​this algorithm is to train a GAN model using historical Pareto solution sets, enabling its generator to learn the potential distribution of high-quality solutions. When a change in the system's environment is detected, such as a change in scheduling power requirements, the algorithm quickly generates an initial population adapted to the new environment and guides the search towards the optimal solution region. Its main steps consist of the following three steps.

[0219] Firstly, historical knowledge extraction and GAN model training.

[0220] When environmental changes are detected, high-quality, representative knowledge is extracted from historical data through cluster analysis and quality screening, and a GAN model is trained based on the extracted historical knowledge. This historical knowledge includes both high-quality and low-quality solution sets.

[0221] Cluster analysis uses historical Pareto solutions Cluster them to divide them Each of the following clusters. This refers to historical power allocation schemes.

[0222] Quality screening involves calculating the centroid of each cluster, performing a non-dominated sort on all centroids, and then selecting the solutions represented by all centroids belonging to the first non-dominated layer to form a high-quality solution set. The remaining solutions are classified as low-quality solution sets. .

[0223] Among them, high-quality solution set Used to train generative adversarial networks.

[0224] The architecture of a GAN model consists of a generator G and a discriminator D. The generator G aims to learn the latent distribution of high-quality solution sets; the discriminator D aims to distinguish between true solutions and generated solutions. By training the generator G and discriminator D alternately, it is ensured that the generator G will eventually produce high-quality new solutions.

[0225] Secondly, hybrid population generation and multi-objective evolutionary optimization.

[0226] Following the steps outlined above, after extracting historical knowledge and training the GAN, a new round of optimization is performed.

[0227] 1. Generate a mixed population by generating a solution based on the CAN model. Specifically, it is represented by the following formula (49).

[0228] (49).

[0229] in, It is a mixed population; To generate solutions for GANs, the distribution characteristics of historical best solutions are inherited to accelerate convergence; The solution is generated through differential evolution operations, and local mining of superior solutions is performed to improve the solution quality. The solution is a randomly sampled solution, randomly generated within the feasible region, used to maintain population diversity and avoid premature convergence.

[0230] Second, the mixed population is used as the initial population and input into the NSGA-II framework for iterative evolution until the termination condition is met, and the optimal solution set of power allocation multi-objective optimization under the current dynamic environment is output.

[0231] First, calculate each individual The three objective function values ​​(corresponding to a power allocation scheme) are determined while ensuring that the solution satisfies the constraints.

[0232] Then, through non-dominated sorting, a series of non-dominated layers are obtained. , ,...in This is the optimal frontier. Simultaneously, to maintain the distribution diversity of solutions on the Pareto front, the calculation is performed for each individual within the same non-dominated layer. Crowding It is represented by the following formula (50).

[0233] (50).

[0234] in, For each individual The degree of congestion; The value of the k-th objective function is represented.

[0235] Furthermore, offspring populations are generated through selection, crossover, and mutation.

[0236] The binary tournament selection method is adopted, which prioritizes the individual with the higher non-dominant ranking when comparing two individuals. Individuals with smaller indices are selected, and if their ranks are the same, individuals with higher crowding are chosen. Crossover and mutation are performed using simulated binary crossover (SBX) and polynomial mutation (PM) to generate the offspring population from the selected parent individuals. .

[0237] Then, an elite retention operation is performed to generate the next generation of parent population.

[0238] parental population With offspring population merged into Then through the Perform non-dominated sorting and crowding calculation, and prioritize... Take all individuals, then select This process continues until N individuals are selected to form the next generation of the parent population. .

[0239] The multi-objective optimization evolutionary algorithm based on the NSGA-II framework is as follows: Figure 5 As shown.

[0240] S301, generates a mixed population.

[0241] S302, perform non-dominated sorting of each individual in the mixed population and calculate its crowding degree.

[0242] S303 uses selection, crossover, and mutation to generate offspring populations based on non-dominated ordering and crowding.

[0243] S304 uses the offspring population to generate the next generation of parent population through elite preservation.

[0244] S305, determine whether the number of iterations meets the termination condition.

[0245] S306 outputs the optimal solution set for the power allocation multi-objective optimization problem in the final population.

[0246] Third, the optimal solution output and knowledge base update.

[0247] Repeat the above evolutionary steps until the preset maximum number of iterations is reached, then terminate and output the first non-dominated layer of the final population. As an approximate solution to the multi-objective optimization problem of power allocation under the current dynamic environment (time t+1), the Pareto optimal solution set is used. With Pareto Frontier At the same time, the optimal solution set obtained Representative solutions are added to the historical Pareto solution set to update the knowledge base, thereby enhancing the algorithm's ability to cope with future environmental changes.

[0248] In summary, the flowchart of the GAN-DMOEA algorithm is as follows: Figure 6 As shown.

[0249] S401 sets algorithm parameters such as population size and maximum number of iterations.

[0250] S402, Initialize the Pareto solution set.

[0251] S403, detect whether the environment has changed.

[0252] S404 extracts high-quality solution sets through cluster analysis and quality screening.

[0253] S405, train the CAN model based on a high-quality solution set.

[0254] S406, generate a solution based on the CAN model to generate a mixed population.

[0255] S407, based on the NSGA-II framework for multi-objective evolutionary optimization, uses a mixed population as the initial population for iterative evolution.

[0256] S408 checks whether the number of iterations has reached the termination condition.

[0257] S409 outputs the approximate Pareto solution set and frontier at time t+1.

[0258] S410: Update the knowledge base based on representative solutions from the optimal solution set.

[0259] To achieve the above functions, the vanadium redox flow battery energy storage power distribution device includes corresponding hardware structures and / or software modules for performing each function. Those skilled in the art will readily recognize that, based on the algorithmic steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is implemented in hardware or by computer software driving hardware 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.

[0260] This disclosure also provides an embodiment such as Figure 7 The vanadium redox flow battery energy storage power distribution device shown includes: an equivalent loss model construction unit 501, a health status determination unit 502, a preliminary power distribution unit 503, and a target power distribution unit 504.

[0261] The equivalent loss model construction unit 501 is configured to construct the equivalent loss model of the all-vanadium redox flow battery; online parameter identification processing is performed on the equivalent loss model to determine the battery identification parameters of multiple battery packs; the equivalent loss model characterizes the correlation between the battery parameters inside the all-vanadium redox flow battery. The health status determination unit 502 is configured to determine the health status of multiple battery packs based on battery identification parameters, original battery resistance, and resistance value at the time of scrapping. The preliminary power allocation unit 503 is configured to determine the charging and discharging priorities of multiple battery packs based on the battery state of charge and health status of each battery pack; the charging and discharging priority represents the order of calling multiple battery packs when they are running simultaneously, determined by the current state of each battery pack; and selects multiple target battery packs to participate in power allocation from multiple battery packs based on the total power required for scheduling and the charging and discharging priority. The target power allocation unit 504 is configured to determine a multi-objective function based on battery identification parameters and battery state of charge, with the objectives of minimizing battery operating loss, minimizing state of charge balance, and minimizing total lifetime loss; construct a multi-objective optimization model based on the multi-objective function; and solve the multi-objective optimization model based on constraints to obtain the target power allocation results for multiple target battery packs.

[0262] As one implementation method, the equivalent loss model construction unit 501 is specifically configured to perform online parameter identification processing on the equivalent loss model to determine the battery identification parameters of multiple battery packs, including: constructing a state-space model based on the battery equivalent loss model to determine the battery output; the state-space model characterizes the energy storage changes of the battery's internal state; the battery output characterizes the correlation between the battery's external terminal voltage and the battery's internal state; discretizing the state-space model and the battery output to obtain a discretized state-space model and a discretized battery output relationship; generating a linear relationship between the data vector and the parameter vector based on the discretized state-space model and the discretized battery output; recursively updating the parameter vector based on the linear relationship to obtain the target parameter vector; and determining the battery identification parameters based on the target parameter vector.

[0263] As one implementation, the health status determination unit 502 is specifically configured such that battery identification parameters include ohmic internal resistance. Based on the battery identification parameters, the original battery resistance, and the resistance value at the time of disposal, the health status of multiple battery packs is determined, including: determining the maximum usable capacity of each battery pack based on the difference between the resistance value at the time of disposal and the ohmic internal resistance; determining the rated battery capacity based on the difference between the resistance value at the time of disposal and the original battery resistance; and determining the health status of multiple battery packs based on the percentage of the maximum usable battery capacity to the rated battery capacity.

[0264] As one implementation method, the preliminary power allocation unit 503 is specifically configured to determine the charging and discharging priority of multiple battery packs based on the battery state of charge and health status of each battery pack, including: determining the comprehensive score of each battery pack based on the battery state of charge and health status of each battery pack; arranging the multiple battery packs in descending order based on the comprehensive score to determine the charging and discharging priority; wherein the battery pack with the highest comprehensive score has the highest charging and discharging priority.

[0265] As one implementation method, the preliminary power allocation unit 503 is specifically configured to determine the comprehensive score of each battery pack based on the battery state of charge (SOC) and health status of each battery pack. This includes: constructing an original evaluation matrix based on the SOC and health status of each battery pack; normalizing the original evaluation matrix according to the current charging and discharging scenario to obtain a normalized evaluation matrix; the normalized evaluation matrix includes a SOC matrix and a health status matrix; determining the weight of each battery pack under different indicators based on the normalized evaluation matrix; the indicators include SOC indicators and health status indicators; determining the first information entropy corresponding to the SOC indicator and the second information entropy corresponding to the health status indicator based on the weights; determining the SOC weight and health status weight based on the first and second information entropies; and determining the weighted sum of the SOC matrix and the health status matrix based on the SOC weight and health status weight to obtain the comprehensive score of each battery pack.

[0266] As one implementation method, the preliminary power allocation unit 503 is specifically configured to select multiple target battery packs to participate in power allocation from multiple battery packs according to the total power required for scheduling and the charging and discharging priority. This includes: accumulating the available power of multiple battery packs in sequence according to the charging and discharging priority until the accumulated value of the available power is greater than or equal to the total power required for scheduling; and determining the multiple battery packs participating in the accumulation as target battery packs.

[0267] As one implementation, the target power distribution unit 504 is specifically configured such that the battery identification parameters also include reactive resistance and parasitic resistance. Based on the battery identification parameters and the battery state of charge (SOC), a multi-objective function is determined, aiming to minimize battery operating losses, SOC balance, and total lifetime loss. This includes: determining battery operating losses based on reactive resistance, ohmic internal resistance, parasitic resistance, and the SOC of each battery pack; determining equipment losses based on equipment operating losses, standby losses, and the operating status of each battery pack; determining operating losses based on battery operating losses, equipment losses, and operating status; determining SOC balance based on the squared difference between the SOC of each battery pack and the mean SOC; determining total lifetime loss based on the sum of lifetime losses of each battery pack; and determining the multi-objective function based on the weighted sum of operating losses, SOC balance, and total lifetime loss.

[0268] As one implementation, the target power allocation unit 504 is specifically configured to solve a multi-objective optimization model based on constraints to obtain target power allocation results for multiple target battery packs, including: constructing a total power constraint based on the sum of the allocated power of each target battery pack and the total power required for scheduling; constructing a state of charge constraint based on the state of charge range of each battery pack; and constructing a maximum output constraint for the battery based on the maximum discharge power and maximum charging power of the allocated power; and iteratively solving the multi-objective optimization model based on a dynamic multi-objective evolutionary algorithm using a generative adversarial network to obtain target power allocation results for multiple target battery packs.

[0269] Regarding the apparatus in the above embodiments, the specific manner in which each unit module performs its operations has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0270] Figure 8 This is a schematic diagram of a vanadium redox flow battery energy storage power distribution device provided in this application. Figure 8 The vanadium redox flow battery energy storage power distribution device 60 includes: a first processor 601, a communication bus 602, a memory 603, a communication interface 604, an output device 605, an input device 606, and a second processor 607.

[0271] The vanadium redox flow battery energy storage power distribution device 60 may include at least one first processor 601 and a memory 603 for storing processor-executable instructions. The first processor 601 is configured to execute the instructions in the memory 603 to implement the vanadium redox flow battery energy storage power distribution method in the following embodiments.

[0272] In addition, the vanadium redox flow battery energy storage power distribution device 60 may also include a communication bus 602, at least one communication interface 604, an input device 606, and an output device 605.

[0273] The first processor 601 may be a processor (central processing unit, CPU), a microprocessor unit, an ASIC, or one or more integrated circuits for controlling the execution of programs according to the present application.

[0274] The communication bus 602 may include a path for transmitting information between the aforementioned components.

[0275] Communication interface 604 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0276] Input device 606 is used to receive input signals and output device 605 is used to output signals.

[0277] The memory 603 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed discs, laser discs, optical discs, digital universal discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory may exist independently and be connected to the processing unit via a bus. The memory may also be integrated with the processing unit.

[0278] The memory 603 stores instructions for executing the scheme of this application, and its execution is controlled by the first processor 601. The first processor 601 executes the instructions stored in the memory 603 to realize the functions of the method of this application.

[0279] In a specific implementation, as one example, the first processor 601 may include one or more CPUs, for example... Figure 8 CPU0 and CPU1 in the CPU.

[0280] In a specific implementation, as one example, the all-vanadium redox flow battery energy storage power distribution device 60 may include multiple processors, such as... Figure 8 The first processor 601 and the second processor 607 are described. Each of these processors may be a single-core processor or a multi-core processor. A processor here may refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions).

[0281] The vanadium redox flow battery energy storage power distribution device, such as Figure 8 The diagram shows a first processor 601 and a memory 603 for storing executable instructions of the first processor 601. The first processor 601 is configured to execute the executable instructions to implement the all-vanadium redox flow battery energy storage power distribution method as described in any of the possible embodiments above. Since the same technical effects can be achieved, further details are omitted here to avoid repetition.

[0282] This application also provides a computer-readable storage medium. When the instructions in the computer-readable storage medium are executed by the processor of the vanadium redox flow battery energy storage power distribution device, the vanadium redox flow battery energy storage power distribution device is able to perform the vanadium redox flow battery energy storage power distribution method as described in any of the above possible embodiments. And it can achieve the same technical effect; to avoid repetition, it will not be described again here.

[0283] This application also provides a computer program product, including a computer program or instructions, which are executed by a processor as described in any of the possible embodiments above for a vanadium redox flow battery energy storage power distribution method. This achieves the same technical effects, and to avoid repetition, will not be repeated here.

[0284] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0285] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for distributing energy storage power in a vanadium redox flow battery, characterized in that, The method is applied to a vanadium redox flow battery energy storage system, the system comprising multiple battery packs; the method includes: An equivalent loss model of a vanadium redox flow battery is obtained, and online parameter identification processing is performed on the equivalent loss model to determine the battery identification parameters of the multiple battery packs; the equivalent loss model characterizes the correlation between the battery parameters inside the vanadium redox flow battery. The health status of the multiple battery packs is determined based on the battery identification parameters, the original battery resistance, and the resistance value at the time of scrapping. The charging and discharging priorities of the multiple battery packs are determined based on the state of charge and health status of each battery pack; the charging and discharging priorities represent the order in which multiple battery packs are called up when they are running simultaneously, based on the current state of each battery pack. Based on the total power required for scheduling and the charging / discharging priority, multiple target battery packs are selected from the multiple battery packs to participate in power allocation; Based on the battery identification parameters and the battery state of charge, a multi-objective function is determined with the objectives of minimizing battery operating loss, minimizing state of charge balance, and minimizing total lifetime loss; a multi-objective optimization model is constructed based on the multi-objective function. Based on the constraints, the multi-objective optimization model is solved to obtain the target power allocation results of the multiple target battery packs; the constraints include total power constraints, state of charge constraints, and maximum output constraints of the batteries.

2. The vanadium redox flow battery energy storage power distribution method according to claim 1, characterized in that, The online parameter identification process for the equivalent loss model, to determine the battery identification parameters of the multiple battery packs, includes: Based on the battery equivalent loss model, a state-space model is constructed to determine the battery output; the state-space model represents the energy storage changes of the battery's internal state; the battery output represents the correlation between the battery's external terminal voltage and the battery's internal state. Discretize the state space model and the battery output to obtain the discretized state space model and the discretized battery output relationship; Based on the discretized state-space model and the discretized battery output, a linear relationship between the data vector and the parameter vector is generated; Based on the linear relationship, the parameter vector is updated recursively to obtain the target parameter vector; The battery identification parameters are determined based on the target parameter vector.

3. The vanadium redox flow battery energy storage power distribution method according to claim 2, characterized in that, The battery identification parameters include ohmic internal resistance; The process of determining the health status of the multiple battery packs based on the battery identification parameters, the original battery resistance, and the resistance value at the time of disposal includes: The maximum usable capacity of each battery pack is determined based on the difference between the resistance value at the scrapping time and the ohmic internal resistance. The rated capacity of the battery is determined based on the difference between the resistance value at the time of scrapping and the original resistance of the battery. The health status of the plurality of battery packs is determined based on the percentage of the maximum usable capacity of the battery to the rated capacity of the battery.

4. The method for all-vanadium redox flow battery energy storage power distribution according to claim 3, characterized in that, The step of determining the charging and discharging priority of the plurality of battery packs based on the state of charge and health status of each battery pack includes: The overall score of each battery pack is determined based on the battery state of charge and the health status of each battery pack. Based on the overall score, the multiple battery packs are sorted in descending order to determine the charging and discharging priority; wherein the battery pack with the highest overall score has the highest charging and discharging priority.

5. The vanadium redox flow battery energy storage power distribution method according to claim 4, characterized in that, The determination of a comprehensive score for each battery pack based on its state of charge and state of health includes: Based on the state of charge and health status of each battery pack, an original evaluation matrix is ​​constructed; Based on the current charging and discharging scenario, the original evaluation matrix is ​​normalized to obtain a normalized evaluation matrix; the normalized evaluation matrix includes a state of charge matrix and a health state matrix. Based on the normalized evaluation matrix, the proportion of each battery pack under different indicators is determined; the indicators include state of charge indicators and state of health indicators. Based on the weight, determine the first information entropy corresponding to the state of charge index and the second information entropy corresponding to the health status index; Based on the first information entropy and the second information entropy, determine the charge state weight and health state weight; Based on the state of charge weights and the health state weights, a weighted sum of the state of charge matrix and the health state matrix is ​​determined to obtain the comprehensive score of each battery pack.

6. The vanadium redox flow battery energy storage power distribution method according to claim 4, characterized in that, The step of selecting multiple target battery packs from the multiple battery packs to participate in power allocation based on the total power demand and the charging / discharging priority includes: According to the charging and discharging priority, the available power of the multiple battery packs is accumulated sequentially until the accumulated value of the available power is greater than or equal to the total power required for scheduling; and the multiple battery packs participating in the accumulation are determined as the target battery packs.

7. The vanadium redox flow battery energy storage power distribution method according to claim 3, characterized in that, The battery identification parameters also include reaction resistance and parasitic resistance; The step of determining a multi-objective function based on the battery identification parameters and the battery state of charge, with the objectives of minimizing battery operating loss, minimizing state of charge balance, and minimizing total lifetime loss, includes: Based on the reactive resistance, the ohmic internal resistance, the parasitic resistance, and the state of charge of each battery pack, the battery operating loss is determined; based on the equipment operating loss, standby loss, and the operating state of each battery pack, the equipment loss is determined; based on the battery operating loss, the equipment loss, and the operating state, the operating loss is determined. Furthermore, the state of charge balance is determined based on the squared difference between the state of charge and the mean state of charge of each of the battery packs. And, based on the sum of the life loss of each of the battery packs, the total life loss is determined; The multi-objective function is determined based on the weighted sum of the operating loss, the state-of-charge balance, and the total lifetime loss.

8. The method for all-vanadium redox flow battery energy storage power distribution according to claim 6, characterized in that, The solution of the multi-objective optimization model based on constraints, to obtain the target power allocation results for the multiple target battery packs, includes: The total power constraint is constructed based on the sum of the allocated power of each of the target battery packs and the total power required for scheduling; the state of charge constraint is constructed based on the state of charge range of each of the battery packs; and the maximum output constraint of the battery is constructed based on the maximum discharge power and the maximum charging power of the allocated power. A dynamic multi-objective evolutionary algorithm based on generative adversarial networks is used to iteratively solve the multi-objective optimization model to obtain the target power allocation results for the multiple target battery packs.

9. A power distribution device for an all-vanadium redox flow battery energy storage system, characterized in that, The device includes: The equivalent loss model construction unit is configured to obtain the equivalent loss model of the vanadium redox flow battery, and to perform online parameter identification processing on the equivalent loss model to determine the battery identification parameters of the multiple battery packs; the equivalent loss model characterizes the correlation relationship of the battery parameters inside the vanadium redox flow battery. The health status determination unit is configured to determine the health status of the plurality of battery packs based on the battery identification parameters, the original resistance of the battery, and the resistance value at the time of scrapping. The preliminary power allocation unit is configured to determine the charging and discharging priorities of the plurality of battery packs based on the battery state of charge and the health state of each battery pack; the charging and discharging priority represents the order of calling determined according to the current state of each battery pack when the plurality of battery packs are running simultaneously; and selects a plurality of target battery packs to participate in power allocation from the plurality of battery packs based on the total power required for scheduling and the charging and discharging priority. The target power allocation unit is configured to determine a multi-objective function based on the battery identification parameters and the battery state of charge, with the objectives of minimizing battery operating loss, minimizing state of charge balance, and minimizing total lifetime loss; construct a multi-objective optimization model based on the multi-objective function; and solve the multi-objective optimization model based on constraints to obtain the target power allocation results for the multiple target battery packs; the constraints include total power constraints, state of charge constraints, and maximum battery output constraints.

10. A power distribution system for an all-vanadium redox flow battery energy storage system, characterized in that, It is configured to perform the all-vanadium redox flow battery energy storage power distribution method as described in any one of claims 1-8.

Images