A method for predicting and regulating state of energy storage system based on multi-source data fusion

Abstract

The application discloses a kind of based on multi-source data fusion's energy storage system state prediction and regulation method, and the application relates to energy storage control technical field, by real-time acquisition power grid dispatching instruction, combine cloud historical operation data with each energy storage node real-time state of charge, construct the state characteristic evaluation system of fusion aging factor, average power and state of charge, realize the accurate quantification of energy storage node health state;Introduce the best bearing power evaluation mechanism based on state characteristics, through the strategy of combination of proportional distribution and dynamic control, establish the iterative optimization model with the goal of bearing time balance, real-time prediction each node's bearing time, through the dynamic calculation of power adjustment amount and the feedback optimization of revised allocation power, ensure that each energy storage node realizes time synchronization convergence in charge and discharge operation, avoid single node early overrun, to achieve balance power grid dispatching demand and energy storage system life.

Description

Technical Field

[0001] This invention belongs to the field of energy storage control technology, specifically, it relates to a method for predicting and regulating the state of energy storage systems based on multi-source data fusion. Background Technology

[0002] Energy storage technology, as a key means to support the large-scale grid connection of renewable energy and improve the stability of grid operation, is ushering in a strategic opportunity period for large-scale development.

[0003] In the current technological context, the scheduling and control methods of energy storage systems often exhibit static, lagging, and uncoordinated characteristics. Traditional control methods rely on fixed power allocation strategies or threshold management based on state of charge, resulting in numerous drawbacks when facing complex grid dispatch instructions. After receiving dispatch instructions, existing technologies ignore the differences in aging, historical operating conditions, and real-time state of charge among energy storage nodes. That is, they allocate power proportionally based on rated power or current power, ignoring the actual carrying capacity of different energy storage nodes. This leads to some severely aged or poorly functioning nodes being allocated excessive power, accelerating their performance degradation, while nodes in good condition are not fully utilized, resulting in a decrease in the overall efficiency of the system. In traditional control processes, once power allocation is complete, each energy storage node operates at a fixed power until a node exits due to reaching its state of charge limit. At this point, emergency adjustments or interruptions are required, which not only affects the efficiency of dispatch command execution but also impacts grid stability. Furthermore, the over-limit times of each energy storage node cannot be synchronized; the node reaching its limit first will terminate operation prematurely, while other nodes still have spare capacity. However, the overall task cannot continue, resulting in a waste of energy storage capacity and a disconnect between power allocation and actual carrying capacity. This not only affects the lifespan of the energy storage system but also reduces its response speed and execution accuracy to grid dispatch commands, making it difficult to meet the core requirements of modern power grids for efficient, intelligent, and long-life operation of energy storage systems.

[0004] To address the aforementioned issues, this invention proposes a method for predicting and controlling the state of energy storage systems based on multi-source data fusion. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for predicting and regulating the state of energy storage systems based on multi-source data fusion, which solves the problems of asynchronous operation and uneven utilization of energy storage nodes in the regulation of energy storage systems in existing technologies.

[0006] The objective of this invention can be achieved through the following technical solutions: A method for state prediction and control of energy storage systems based on multi-source data fusion, the method comprising: Step 1: Determine the dispatching instructions based on the power grid signals, and lock in the operation type and total power demand; The energy storage system control process is initiated to determine the state of charge of each energy storage node in real time. Based on the cloud database, the historical operating data associated with each energy storage node under the corresponding operation type is extracted and analyzed to construct the real-time associated state characteristics of each energy storage node. Step 2: Evaluate the optimal power capacity of each energy storage node based on its state characteristics; Based on the optimal power carrying capacity ratio of each energy storage node, the total demand power is allocated to determine the allocated power of each energy storage node, and the carrying capacity time of each energy storage node when it exceeds the limit is predicted in combination with the state of charge. Step 3: Based on the current state of charge and predicted carrying time of each energy storage node, dynamically adjust the power allocation of each energy storage node until the carrying time of each energy storage node converges to be consistent. When the state of charge of each energy storage node exceeds the limit, terminate the current operation and complete the scheduling instruction.

[0007] As a further aspect of the present invention, the specific method for determining the dispatch instruction based on the power grid signal and locking the operation type and total power demand in step one is as follows: Real-time locking of the grid signals received by the energy storage system; Obtain dispatch instructions from the power grid signals, including the locking operation type and total power demand; The operation types include charging operation and discharging operation, and the total power demand is denoted as P_all.

[0008] As a further aspect of the present invention, the specific method for determining the state of charge of each energy storage node in real time in step one is as follows: Extract all energy storage nodes in the energy storage system, and denote the total number as j; Randomly arrange j energy storage nodes, denoted as the energy storage node sequence C1, C2, ..., Cj; When a dispatch instruction is received from the power grid signal, the state of charge of each of the j energy storage nodes is extracted synchronously and arranged in the order of C1, C2, ..., Cj, denoted as the state of charge sequence H1, H2, ..., Hj.

[0009] As a further aspect of the present invention, the specific method for constructing the real-time associated state features of each energy storage node in step one is as follows: Extract any energy storage node Ci, where i is the counting index, 1≤i≤j; Extract all historical records from the cloud database where the absolute value of the difference between the state of charge of energy storage node Ci and Hi is ≤ ΔH, to form a reference dataset, where ΔH is a preset threshold for the absolute value of the difference. If the sample size in the reference dataset is less than N_min, records are supplemented based on linear interpolation until N_min is reached, where N_min is the minimum sample size; Calculate the average operating power of energy storage node Ci in the reference dataset, denoted as P_avg_i; Extract the rated operating power EPi, cumulative charge-discharge cycle count Si, and cumulative throughput Qi of the energy storage node Ci; Take the rated total throughput EQi of the energy storage node Ci, and calculate the aging factor Di using Di=Qi / EQi, where Di∈[0,1]; The state characteristics Fi of energy storage node Ci are calculated using Fi=α×(P_avg_i / EPi)+β×(1-Di)+γ×(Hi / H_max), where Fi∈[0,1], H_max is the theoretical maximum state of charge of energy storage node Ci at the current time, and α, β, and γ are preset weight coefficients, α+β+γ=1; Similarly, the real-time associated state characteristics of each energy storage node are determined and denoted as the state characteristic sequence F1, F2, ..., Fj.

[0010] As a further aspect of the present invention, the specific method for evaluating the optimal power capacity of each energy storage node based on the state characteristics of each energy storage node in step two is as follows: Extract the state characteristics Fi and rated operating power EPi of the energy storage node Ci; The optimal power P_opt_i to be carried by energy storage node Ci at the current time is calculated using P_opt_i = EPi × Fi; Similarly, the optimal power to be undertaken by each energy storage node is determined based on the state feature sequence F1, F2, ..., Fj, and arranged into the optimal power undertaking sequence P_opt_1, P_opt_2, ..., P_opt_j.

[0011] As a further aspect of the present invention, in step two, the specific method for allocating the total required power and determining the allocated power for each energy storage node is as follows: Extract the optimal power to be undertaken by each energy storage node, sum them up, and denot the total optimal power to be undertaken as P_opt_all; To obtain the optimal power P_opt_i for energy storage node Ci, the allocated power P_asg_i for energy storage node Ci is calculated using P_asg_i = P_opt_i / P_opt_all × P_all. The direction of P_asg_i is consistent with the operation type: P_asg_i is positive for charging operations and negative for discharging operations.

[0012] As a further aspect of the present invention, the specific method for predicting the duration of overload for each energy storage node in step two, based on the state of charge, is as follows: Extract the rated capacity Xi and the safe operating range of state of charge [H_min, H_max] of any energy storage node Ci at the current time; Calculate the current achievable capacity ΔXi of energy storage node Ci based on the operation type; Charging operation: ΔXi = (H_max - Hi) × Xi; Discharge operation: ΔXi = (Hi - H_min) × Xi; The time Ti corresponding to the energy storage node Ci when it is overloaded is calculated using Ti=ΔXi / |P_asg_i|, where if |P_asg_i|=0, Ti is infinite. Similarly, determine the capacity and duration of each energy storage node, denoted as the capacity sequence ΔX1, ΔX2, ..., ΔXj and the duration sequence T1, T2, ..., Tj.

[0013] As a further aspect of the present invention, the specific method for dynamically adjusting the power allocation of each energy storage node in step three until the carrying time of each energy storage node converges to a consistent level is as follows: S81, take the capacity sequence ΔX1,ΔX2,...,ΔXj and the capacity time sequence T1,T2,...,Tj; Extract the capacity ΔXi and the duration Ti that any energy storage node Ci can handle; S82, using T_avg=(1 / j)×∑ j i=1 Ti calculates the average operating time T_avg for all energy storage nodes; S83, use ei=Ti-T_avg to calculate the time deviation ei of energy storage node Ci, and use ΔPi=K×ei to calculate the power adjustment amount ΔPi of energy storage node Ci. If ΔPi=0, it means no adjustment is needed; if ΔPi>0, the current power of energy storage node Ci is increased; if ΔPi<0, the current power of energy storage node Ci is decreased. S84, calculate the control and distribution power P_tum_i using P_tum_i=P_asg_i+ΔPi, where 0≤P_tum_i≤EPi; S85, using ΔP_sum=∑ j i=1 Calculate the total power adjustment deviation ΔP_sum using (P_tum_i-P_asg_i), and perform a correction based on P_new_i=P_tum_i-ΔP_sum / j to obtain the corrected power allocation P_new_i. Where 0≤P_new_i≤EPi, if P_new_i>EPi, then the energy storage node Ci is considered to have exceeded its limit, and the portion exceeding EPi will be extracted and evenly distributed to the energy storage nodes that have not exceeded their limits, until the corrected power allocation of all energy storage nodes does not exceed their rated operating power. S86, update the undertaking time series based on the corrected allocation power of all energy storage nodes, iterate steps S81 to S85 until the absolute value of the deviation between the undertaking time of all energy storage nodes and the average undertaking time is less than the preset convergence threshold, and output the corrected allocation power of all energy storage nodes.

[0014] As a further aspect of the present invention, in step three, after determining the corrected allocation power of all energy storage nodes, all energy storage nodes are instructed to execute the corresponding operation type based on their respective corrected allocation power, and when the charge state of each energy storage node simultaneously exceeds the limit, the current operation type is terminated and the scheduling instruction is completed.

[0015] The beneficial effects of this invention are: (1) This invention achieves the synergy of the energy storage system in response to grid dispatch by deeply integrating multi-source data and dynamic control mechanism; it improves the rationality and prediction accuracy of the power allocation undertaken by each energy storage node by using historical operating data and real-time state of charge to construct feature analysis, and avoids the premature termination of operation of a single node due to overload; secondly, by dynamically adjusting the allocated power in real time, it drives the state of charge undertaking time of all nodes to converge and be consistent, ensuring that all energy storage nodes of the entire energy storage system reach the state of charge over-limit point synchronously; it maximizes the utilization of the overall energy storage capacity of the system, avoids resource idleness or local overconsumption, and enhances the stability and continuous power supply capability of the energy storage system in response to dispatch instructions. (2) This invention generates a state feature that integrates multi-dimensional information for each energy storage node; captures grid dispatch instructions and the current state of charge of the node in real time; creatively introduces the average operating power based on historical similar operating conditions; deeply analyzes the future output capacity of the node; at the same time, it accurately quantifies the wear and tear on the equipment life by the number of cycles and throughput through the aging factor; combined with the theoretical maximum state of charge, it enables the dispatch decision to automatically avoid overuse of aging or poorly performing nodes; guides the system to allocate charging and discharging tasks in the optimal way; while responding to grid demand efficiently, it balances the operating pressure of each node to the maximum extent. (3) The present invention dynamically calculates the optimal power to be undertaken based on the state characteristics and rated power of each node, ensuring that the power allocation matches the actual capacity of the node; by allocating the total demand power according to the optimal power to be undertaken, the load is balanced; by combining the state of charge and the safe operating range, the capacity and duration that each node can undertake when it exceeds the limit are accurately predicted, thereby providing early warning of potential risks, preventing overcharging or over-discharging, extending the life of energy storage equipment, and optimizing power flow management. (4) This invention achieves refined control of power distribution in energy storage systems by introducing time deviation feedback and closed-loop iterative correction mechanism; real-time deviation signal is generated by using average load time, preliminary adjustment is calculated by proportional adjustment, and then the total power deviation is corrected and the over-limit power is redistributed to ensure accurate tracking of the total power command of the system and strict constraint of the physical limits of each node; the problem of overload or underload of some nodes under traditional fixed proportional distribution is avoided. Secondly, the rate of change of the state of charge of each energy storage unit is forced to be synchronized by the time convergence mechanism, eliminating the barrel effect caused by the difference in initial capacity or different aging degree, delaying the time for a single node to reach the upper and lower limits of the state of charge, maximizing the utilization of the overall throughput capacity, thereby improving the cycle life and energy efficiency of the energy storage system. Attached Figure Description

[0016] The invention will now be further described with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart illustrating the method described in this invention; Figure 2 This is a flowchart illustrating the method described in Embodiment 2 of the present invention; Figure 3 This is a flowchart illustrating the method described in Embodiment 3 of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, this application provides a method for state prediction and control of energy storage systems based on multi-source data fusion; As an embodiment 1 of this application, it specifically includes: Step 1: Determine the dispatching instructions based on the power grid signals, and lock in the operation type and total power demand; The energy storage system control process is initiated to determine the state of charge of each energy storage node in real time. Based on the cloud database, the historical operating data associated with each energy storage node under the corresponding operation type is extracted and analyzed to construct the real-time associated state characteristics of each energy storage node. Step 2: Evaluate the optimal power capacity of each energy storage node based on its state characteristics; Based on the optimal power carrying capacity ratio of each energy storage node, the total demand power is allocated to determine the allocated power of each energy storage node, and the carrying capacity time of each energy storage node when it exceeds the limit is predicted in combination with the state of charge. Step 3: Based on the current state of charge and predicted carrying time of each energy storage node, dynamically adjust the power allocation of each energy storage node until the carrying time of each energy storage node converges to be consistent. When the state of charge of each energy storage node exceeds the limit, terminate the current operation and complete the scheduling instruction.

[0020] Example 2 This embodiment, based on embodiment 1, further discloses a method for constructing real-time associated state characteristics of each energy storage node, such as... Figure 2 As shown, it specifically includes the following: This embodiment, as a key preliminary step in the intelligent scheduling method of energy storage system, mainly includes data acquisition, status perception and feature extraction, with the aim of laying a data foundation for subsequent operations; The energy storage system described in this solution includes several energy storage nodes. Each energy storage node can be understood as a battery cluster plus a battery cluster monitoring and control device. All energy storage nodes in the energy storage system are not limited in specifications and types, as long as they can achieve network interconnection. First, the energy storage system monitors and receives grid signals transmitted from the grid in real time. These grid signals include specific dispatch instructions, operation types, and total power demand. The operation types include charging and discharging operations. The total power demand represents the total power that the grid requires the energy storage system to generate. It should be noted that the total power demand described in this scheme is determined considering the amount that the energy storage system itself can absorb. In other words, the total power demand sent from the grid to the current energy storage system will not exceed the amount that the energy storage system itself can absorb. Here, the total power demand is denoted as P_all for ease of subsequent calculation.

[0021] Next, all energy storage nodes in the energy storage system are counted, and the total number is recorded as j. Then, the counted j energy storage nodes are randomly arranged to obtain a sequence, denoted as the energy storage node sequence C1, C2, ..., Cj, to avoid priority deviations caused by fixed sorting of some nodes. Next, the current percentage of electricity, i.e. the state of charge, of each node is synchronously obtained (this operation is performed when the dispatch command in the power grid signal is obtained). The state of charge of each of the j energy storage nodes is then sorted according to the energy storage node sequence C1, C2, ..., Cj. The resulting sequence is denoted as the state of charge sequence H1, H2, ..., Hj.

[0022] Next, an arbitrary energy storage node Ci is extracted from the determined energy storage node sequence C1, C2, ..., Cj for example processing. The remaining energy storage nodes are processed in the same and synchronous manner as energy storage node Ci. i is the counting index, 1≤i≤j.

[0023] First, extract all historical records from the cloud database where the absolute value of the difference between the state of charge (SBC) of energy storage node Ci and its current SBC Hi is ≤ ΔH. These historical records include all operational data of energy storage node Ci during charging or discharging operations in the past period. These data are then grouped and used as samples to form a reference dataset. (The number of operational data in the reference dataset can also be set by the operator, specifying a corresponding number of operational data, such as taking only one week's worth of operational data or one day's worth of operational data to form a sample.) The samples include SBC, operating power, cumulative charge / discharge cycles, and cumulative throughput. Here, ΔH is a threshold value of the absolute difference preset by the operator based on the actual situation.

[0024] It should be noted that if the sample size in the reference dataset is less than the minimum sample size N_min preset by the operator, the sample size in the reference dataset is supplemented by linear interpolation, that is, the operating power of the corresponding energy storage node under different states of charge is supplemented. However, the cumulative number of charge-discharge cycles and the cumulative throughput do not need to be supplemented by interpolation.

[0025] Next, the reference dataset of energy storage node Ci is extracted, all the operating power of energy storage node Ci is obtained, and an averaging operation is performed to obtain the average operating power of energy storage node Ci, denoted as P_avg_i. The average operating power P_avg_i represents the average carrying capacity exhibited by energy storage node Ci over the past time.

[0026] Next, the rated operating power EPi, cumulative charge-discharge cycle count Si, and cumulative throughput Qi of the energy storage node Ci are extracted from the reference dataset. Next, obtain the rated total throughput EQi of the energy storage node Ci. The rated total throughput EQi is the standard parameter of the energy storage node Ci when it leaves the factory and is regarded as a known value. The aging factor Di of the energy storage node Ci is calculated by using Di=Qi / EQi, where Di∈[0,1]. The aging factor Di reflects the degree of lifespan consumption of the energy storage node Ci. The larger Di is, the more the energy storage node Ci is theoretically consumed and the more severe the aging is. Then, the state characteristics Fi of the energy storage node Ci are calculated by using Fi=α×(P_avg_i / EPi)+β×(1-Di)+γ×(Hi / H_max), and the state characteristics Fi∈[0,1]. Where (P_avg_i / EPi) represents the ratio of the average power that the energy storage node Ci can actually achieve in history to the theoretical maximum power. The higher the ratio, the better the energy storage node Ci has performed in history and the closer it is to its rated capacity. (1-Di) represents the remaining health of the energy storage node Ci. Since Di represents the aging degree that has been depleted, (1-Di) is its remaining health. When it is fully healthy, it is considered to be 1. (Hi / H_max) represents the percentage of the current state of charge of energy storage node Ci, which is used to reflect the current energy level of energy storage node C; The state feature Fi comprehensively reflects the power carrying capacity and health status of the energy storage node Ci under the current state of charge. H_max is the theoretical maximum state of charge of the energy storage node Ci at the current time, which is regarded as a known value. α, β, and γ are preset weighting coefficients, and α+β+γ=1. By repeating the above steps, the real-time associated state characteristics of all energy storage nodes are determined and sorted according to the order of the energy storage node sequence, denoted as the state characteristic sequence F1, F2, ..., Fj.

[0027] Example 3 This embodiment, based on embodiment 2, further discloses a method for determining the allocated power and service time of each energy storage node, such as... Figure 3 As shown, it specifically includes the following: First, extract the state characteristics Fi and rated operating power EPi of the energy storage node Ci determined in Example 2; Next, the optimal power P_opt_i associated with the energy storage node Ci at the current time is calculated by using P_opt_i=EPi×Fi. The optimal power P_opt_i represents the most suitable power value associated with the energy storage node Ci in the current state. Repeat the above steps to process all energy storage nodes in the same and synchronous manner. Based on the state feature sequence F1, F2, ..., Fj, determine the optimal power to be undertaken by each energy storage node and arrange them into the optimal power undertaking sequence P_opt_1, P_opt_2, ..., P_opt_j. This allows each energy storage node to have the optimal power capacity based on its own historical behavior and current state, avoiding a one-size-fits-all allocation method.

[0028] Next, a summation operation is performed based on the optimal power undertaking sequence P_opt_1, P_opt_2, ..., P_opt_j, and the result of the summation operation is denoted as the total optimal power undertaking P_opt_all; Based on the proportion of the optimal power undertaken by any energy storage node Ci to the total optimal power undertaken, the actual allocated power P_asg_i of energy storage node Ci under the given total demand power P_all issued by the grid is locked. Specifically: P_asg_i = P_opt_i / P_opt_all × P_all. In this way, the power of each energy storage node is allocated proportionally, which can naturally allow energy storage nodes in good condition to output more power and energy storage nodes in poor condition to output less power, so that the workload of each energy storage node matches its bearing capacity. The direction of P_asg_i is consistent with the operation type; P_asg_i is positive during charging operations and negative during discharging operations.

[0029] Through the above operations, the sum of the allocated power of all energy storage nodes is equal to the total demand power P_all, which ensures the system's accurate response to grid dispatch, allows high-capacity energy storage nodes to take on more power and low-capacity energy storage nodes to take on less power, avoids some energy storage nodes from aging faster due to overuse, and helps to extend the life of the entire system. Then, the rated capacity Xi and the safe operating range of the state of charge [H_min, H_max] of the energy storage node Ci at the current time are extracted. The rated capacity Xi and the safe operating range of the state of charge [H_min, H_max] are both known values ​​and are bound to the energy storage node Ci itself. Next, the current capacity ΔXi of the energy storage node Ci is calculated based on the different operation types. When the operation type is charging, ΔXi = (H_max - Hi) × Xi, which represents how much electricity can still be charged. When the operation type is discharging, ΔXi = (Hi - H_min) × Xi, which represents how much electricity can still be discharged. Then, the bearing time Ti corresponding to the energy storage node Ci when it is overloaded is calculated using Ti=ΔXi / |P_asg_i|. The calculation of the bearing time Ti is essentially a time prediction. Based on the distance between the current state of charge of the energy storage node and the safety boundary, combined with the power to be applied, the remaining safe operating time is calculated. This is used to quantitatively assess the safety margin during the execution of scheduling instructions. The bearing time Ti can predict which energy storage nodes are about to approach the safety boundary, thus providing a basis for subsequent dynamic adjustments or alarms. It should be noted that if |P_asg_i|=0, Ti is infinite. Repeat the above steps to determine the available capacity and available time for all energy storage nodes, and perform serialization processing based on the sorting order of the energy storage node sequence, denoted as the available capacity sequence ΔX1, ΔX2, ..., ΔXj and the available time sequence T1, T2, ..., Tj.

[0030] Example 4 This embodiment, based on embodiment 3, further discloses a method for determining the corrected power allocation of each energy storage node to execute grid dispatch commands, specifically including the following: This embodiment uses closed-loop feedback control to dynamically adjust the output power of each energy storage node. The ultimate goal is to make the operating time of all energy storage nodes tend to be consistent, while ensuring that the power does not exceed the limit. First, obtain the available capacity sequence ΔX1, ΔX2, ..., ΔXj and the available time sequence T1, T2, ..., Tj as described in Example 3, and extract the available capacity ΔXi and available time Ti of any energy storage node Ci from them; Next, by adopting: T_avg=(1 / j)×∑ j i=1 Ti calculates the average time T_avg for all energy storage nodes. For example, if there are 3 energy storage nodes, and it is estimated that they will take 10 minutes, 20 minutes and 30 minutes to complete the charging operation respectively, then the average time is 20 minutes. Next, the time deviation ei between the energy storage node Ci and the average operating time T_avg is calculated using ei=Ti-T_avg. If the time deviation ei is equal to 0, it means that there is no time deviation between the energy storage node Ci and the average operating time T_avg. If the time deviation ei is greater than 0, it means that the operating time of the energy storage node Ci is longer than the average operating time T_avg, indicating that the corresponding energy storage node is operating slowly, and the power of the energy storage node Ci needs to be increased. Conversely, if the time deviation ei is less than 0, it means that the operating time of the energy storage node Ci is shorter than the average operating time T_avg, indicating that the corresponding energy storage node is operating fast, and the power of the energy storage node Ci needs to be reduced. Then, the power adjustment amount ΔPi of the energy storage node Ci is calculated by ΔPi=K×ei. If ΔPi=0, it means no adjustment is needed. If ΔPi>0 (i.e., time deviation ei is greater than 0), it means the current power of the energy storage node Ci is increased, and the increased power value is the power adjustment amount ΔPi. If ΔPi<0 (i.e., time deviation ei is less than 0), it means the current power of the energy storage node Ci needs to be reduced, and the reduced power value is the power adjustment amount ΔPi. The control allocation power P_tum_i associated with the energy storage node Ci is calculated based on P_tum_i=P_asg_i+ΔPi, where 0≤P_tum_i≤EPi. The purpose of this step is to add the amount to be adjusted, ΔPi, to the original allocation power P_asg_i of the energy storage node Ci. During this process, the power distribution P_tum_i may exceed the limit, so the constraint condition is given: 0≤P_tum_i≤EPi; Then, by using ΔP_sum=∑j i=1 (P_tum_i-P_asg_i) calculates the total power adjustment deviation ΔP_sum after the power allocation calculation for all energy storage nodes as described above, and performs a correction operation on the power allocation of energy storage node Ci by P_new_i=P_tum_i-ΔP_sum / j, finally obtaining the corrected power allocation P_new_i of energy storage node Ci. If 0 ≤ P_new_i ≤ EPi, then in subsequent operations, energy storage node Ci will perform the corresponding operation with the corrected allocated power P_new_i; if P_new_i > EPi, then energy storage node Ci is considered to have exceeded the limit. To prevent energy storage nodes from failing due to exceeding their limits, the portion of the corrected allocation power P_new_i of energy storage node Ci that exceeds EPi is extracted (in actual operation, all energy storage nodes Ci that exceed their limits are operated simultaneously) and evenly distributed to energy storage nodes that do not exceed their limits. This process is repeated until the corrected allocation power of all energy storage nodes does not exceed their rated operating power. Once the corrected allocation power of all energy storage nodes is determined, the undertaking time series of all energy storage nodes is updated based on the corrected allocation power, and the above steps are repeated until the absolute value of the deviation between the undertaking time of all energy storage nodes and the average undertaking time is less than the convergence threshold preset by the operator. Then, the corrected allocation power of all energy storage nodes is output, and all energy storage nodes are instructed to execute the corresponding operation type based on their respective corrected allocation power. When the charge state of each energy storage node is synchronously overloaded, the current operation type is terminated and the scheduling instruction is completed. This avoids premature depletion or full charging of certain energy storage nodes, which could lead to a decrease in the stability of the energy storage system and thus maximize the utilization of the entire energy storage system's regulation capabilities.

[0031] All data in the formulas described above have been calculated with dimensions removed. Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0032] The above description is merely an example and illustration of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0033] It should be stated that all user data collected in this application was collected with the user's consent and authorization. Furthermore, the uses of user data are legal and compliant, and the use and processing of user data comply with the relevant laws, regulations, and standards of the relevant regions.

Claims

1. A method for state prediction and control of energy storage systems based on multi-source data fusion, characterized in that, The method includes: Step 1: Determine the dispatching instructions based on the power grid signals, and lock in the operation type and total power demand; The energy storage system control process is initiated to determine the state of charge of each energy storage node in real time. Based on the cloud database, the historical operating data associated with each energy storage node under the corresponding operation type is extracted and analyzed to construct the real-time associated state characteristics of each energy storage node. Step 2: Evaluate the optimal power capacity of each energy storage node based on its state characteristics; Based on the optimal power carrying capacity ratio of each energy storage node, the total demand power is allocated to determine the allocated power of each energy storage node, and the carrying capacity time of each energy storage node when it exceeds the limit is predicted in combination with the state of charge. Step 3: Based on the current state of charge and predicted carrying time of each energy storage node, dynamically adjust the power allocation of each energy storage node until the carrying time of each energy storage node converges to be consistent. When the state of charge of each energy storage node exceeds the limit, terminate the current operation and complete the scheduling instruction.

2. The method according to claim 1, characterized in that, In step one, the specific method for determining the dispatch instruction based on the power grid signal and locking the operation type and total power demand is as follows: Real-time locking of the grid signals received by the energy storage system; Obtain dispatch instructions from the power grid signals, including the locking operation type and total power demand; The operation types include charging operation and discharging operation, and the total power demand is denoted as P_all.

3. The method according to claim 2, characterized in that, In step one, the specific method for determining the state of charge of each energy storage node in real time is as follows: Extract all energy storage nodes in the energy storage system, and denote the total number as j; Randomly arrange j energy storage nodes, denoted as the energy storage node sequence C1, C2, ..., Cj; When a dispatch instruction is received from the power grid signal, the state of charge of each of the j energy storage nodes is extracted synchronously and arranged in the order of C1, C2, ..., Cj, denoted as the state of charge sequence H1, H2, ..., Hj.

4. The method according to claim 3, characterized in that, In step one, the specific method for constructing the real-time associated state characteristics of each energy storage node is as follows: Extract any energy storage node Ci, where i is the counting index, 1≤i≤j; Extract all historical records from the cloud database where the absolute value of the difference between the state of charge of energy storage node Ci and Hi is ≤ ΔH, to form a reference dataset, where ΔH is a preset threshold for the absolute value of the difference. If the sample size in the reference dataset is less than N_min, records are supplemented based on linear interpolation until N_min is reached, where N_min is the minimum sample size; Calculate the average operating power of energy storage node Ci in the reference dataset, denoted as P_avg_i; Extract the rated operating power EPi, cumulative charge-discharge cycle count Si, and cumulative throughput Qi of the energy storage node Ci; Take the rated total throughput EQi of the energy storage node Ci, and calculate the aging factor Di using Di=Qi / EQi, where Di∈[0,1]; The state characteristics Fi of energy storage node Ci are calculated using Fi=α×(P_avg_i / EPi)+β×(1-Di)+γ×(Hi / H_max), where Fi∈[0,1], H_max is the theoretical maximum state of charge of energy storage node Ci at the current time, and α, β, and γ are preset weight coefficients, α+β+γ=1; Similarly, the real-time associated state characteristics of each energy storage node are determined and denoted as the state characteristic sequence F1, F2, ..., Fj.

5. The method according to claim 4, characterized in that, In step two, the specific method for evaluating the optimal power capacity of each energy storage node based on its state characteristics is as follows: Extract the state characteristics Fi and rated operating power EPi of the energy storage node Ci; The optimal power P_opt_i to be carried by energy storage node Ci at the current time is calculated using P_opt_i = EPi × Fi; Similarly, the optimal power to be undertaken by each energy storage node is determined based on the state feature sequence F1, F2, ..., Fj, and arranged into the optimal power undertaking sequence P_opt_1, P_opt_2, ..., P_opt_j.

6. The method according to claim 5, characterized in that, In step two, the total power demand is allocated, and the specific method for determining the allocated power for each energy storage node is as follows: Extract the optimal power to be undertaken by each energy storage node, sum them up, and denot the total optimal power to be undertaken as P_opt_all; To obtain the optimal power P_opt_i for energy storage node Ci, the allocated power P_asg_i for energy storage node Ci is calculated using P_asg_i = P_opt_i / P_opt_all × P_all. The direction of P_asg_i is consistent with the operation type: P_asg_i is positive for charging operations and negative for discharging operations.

7. The method according to claim 6, characterized in that, In step two, the specific method for predicting the time each energy storage node will bear when overloaded, based on its state of charge, is as follows: Extract the rated capacity Xi and the safe operating range of state of charge [H_min, H_max] of any energy storage node Ci at the current time; Calculate the current achievable capacity ΔXi of energy storage node Ci based on the operation type; Charging operation: ΔXi = (H_max - Hi) × Xi; Discharge operation: ΔXi = (Hi - H_min) × Xi; The time Ti corresponding to the energy storage node Ci when it is overloaded is calculated using Ti=ΔXi / |P_asg_i|, where if |P_asg_i|=0, Ti is infinite. Similarly, determine the capacity and duration of each energy storage node, denoted as the capacity sequence ΔX1, ΔX2, ..., ΔXj and the duration sequence T1, T2, ..., Tj. 8.S81, take the capacity sequence ΔX1,ΔX2,...,ΔXj and the capacity time sequence T1,T2,...,Tj; Extract the capacity ΔXi and the duration Ti that any energy storage node Ci can handle; S82, using T_avg=(1 / j)×∑ j i=1 Ti calculates the average operating time T_avg for all energy storage nodes; S83, use ei=Ti-T_avg to calculate the time deviation ei of energy storage node Ci, and use ΔPi=K×ei to calculate the power adjustment amount ΔPi of energy storage node Ci. If ΔPi=0, it means no adjustment is needed; if ΔPi>0, the current power of energy storage node Ci is increased; if ΔPi<0, the current power of energy storage node Ci is decreased. S84, calculate the control and distribution power P_tum_i using P_tum_i=P_asg_i+ΔPi, where 0≤P_tum_i≤EPi; S85, using ΔP_sum=∑ j i=1 Calculate the total power adjustment deviation ΔP_sum using (P_tum_i-P_asg_i), and perform a correction based on P_new_i=P_tum_i-ΔP_sum / j to obtain the corrected power allocation P_new_i. Where 0≤P_new_i≤EPi, if P_new_i>EPi, then the energy storage node Ci is considered to have exceeded its limit, and the portion exceeding EPi will be extracted and evenly distributed to the energy storage nodes that have not exceeded their limits, until the corrected power allocation of all energy storage nodes does not exceed their rated operating power. S86, update the undertaking time series based on the corrected allocation power of all energy storage nodes, iterate steps S81 to S85 until the absolute value of the deviation between the undertaking time of all energy storage nodes and the average undertaking time is less than the preset convergence threshold, and output the corrected allocation power of all energy storage nodes.

9. The method according to claim 8, characterized in that, In step three, after determining the corrected allocation power of all energy storage nodes, all energy storage nodes are instructed to perform the corresponding operation type based on their respective corrected allocation power. When the charge state of each energy storage node is simultaneously overloaded, the current operation type is terminated, and the scheduling instruction is completed.

Images