Networking type battery energy storage system and control method thereof

Abstract

The invention discloses a network construction type battery energy storage system and a control method thereof, and relates to the technical field of battery energy storage. Electrical parameters and operation state parameters are preprocessed, and feature vectors are generated; analyzing offset features based on the feature vectors, and performing feature drift marking; analyzing the spacing characteristics of each battery energy storage device, and performing clustering analysis according to the spacing characteristics; analyzing the equivalent capacity of each cluster in the clustering result; obtaining a historical regulation and control record set, and analyzing the association degree between any two moments; the composite energy fluctuation value at the current moment is analyzed, and a dynamic regulation and control threshold value is analyzed in combination with the correlation degree of the real-time moment and the historical moment; matching degree indexes of all clusters are analyzed in combination with the equivalent capacity and the total regulation and control instruction, an ordered cluster index sequence is generated after sorting, and hierarchical regulation and control instruction analysis is conducted on the ordered cluster index sequence; the allocation proportion is dynamically adjusted, resource allocation is optimized, and the regulation and control risk is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of battery energy storage, and in particular to a network-forming battery energy storage system and a control method thereof. Background Art

[0002] With the increasing penetration rate of new energy and power electronic devices, the power system has a tendency of decreasing inertia and weakening system strength, and the stability problem becomes more and more serious. The functions of the energy storage system in the power grid include: providing active or reactive support for the system, improving the grid connection ability of new energy, participating in peak shaving and frequency modulation, and supplying power for a short time during faults. Adopting network-forming control technology to control the energy storage converter can improve the stability of the new power system.

[0003] In scenarios such as equipment impedance fluctuation and response time change, static cluster division is likely to cause some equipment to be overloaded or idle, and it is impossible to achieve balanced capacity distribution. The correlation between historical regulation experience and real-time working conditions has not been effectively explored. The regulation threshold often depends on manual setting or fixed rules, and it is difficult to cope with the randomness and suddenness of power grid power fluctuations. Existing strategies lack dynamic quantitative analysis of cluster matching degree and priority, which are likely to cause conflicts in regulation instructions or mismatches in resource allocation, exacerbating equipment loss and system oscillation.

[0004] Therefore, the present invention discloses a network-forming battery energy storage system and a control method thereof to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a network-forming battery energy storage system and a control method thereof to solve the problems raised in the prior art.

[0006] To achieve the above purpose, the present invention provides the following technical solution: a network-forming battery energy storage control method, the method includes the following steps: S1: Real-time collect the electrical parameters and operation state parameters of the energy storage device, preprocess the electrical parameters and operation state parameters to generate feature vectors; analyze the offset features based on the feature vectors, and perform feature drift marking; S2: Analyze the spacing features of each battery energy storage device, and perform clustering analysis according to the spacing features; analyze the equivalent capacity of each cluster in the clustering result; S3: Obtain the set of historical regulation records, analyze the degree of association between any two moments; extract the real-time power fluctuation, analyze the composite energy fluctuation value at the current moment, and analyze the dynamic regulation threshold in combination with the degree of association between the real-time moment and the historical moment; S4: Analyze the matching degree indexes of each cluster in combination with the equivalent capacity and the total regulation instruction, generate an ordered cluster index sequence after sorting, and perform hierarchical regulation instruction analysis on the ordered cluster index sequence.

[0007] According to the above solution, in S1, the following contents are included: S101: Obtain the electrical parameters and dynamic operation parameters of the network-forming battery energy storage device. The electrical parameters include equivalent impedance and phase angle; the dynamic operation parameters include power change rate and response time; perform normalization processing on the electrical parameters and dynamic operation parameters respectively to generate the feature vector of the battery energy storage device, and denote the feature vector of the i-th device at time t as: V i (t) = [(logZ i (t)) / Z0, θ i (t) / π, tanh((dP i / dt) / △P max ), 1 / T i ; where, Z i (t) represents the equivalent impedance of the i-th battery energy storage device at time t; Z0 represents the impedance reference value; θ i (t) represents the phase angle of the i-th battery energy storage device at time t; the phase angle represents the phase difference between the voltage and current when the battery energy storage device is connected to the power grid; dP i / dt represents the power change rate of the i-th battery energy storage device at time t; △P max represents the maximum value of the power change rate; T i represents the average response time of the i-th battery energy storage device; splice the feature vectors of all battery energy storage devices to obtain the feature vector matrix at time t; This application linearizes the impedance change, limits the power change rate within a range, compresses the parameter range, suppresses outliers, and improves the model stability.

[0008] S102: Preset the time window length of the system, and analyze the offset feature D(t) between the feature matrix at the end of the current window and the matrix at the end of the previous window based on the time window: ; where, V （i，l） (t) represents the l-th feature of the i-th battery energy storage device at time t;, V （i，l） (t - △t) represents the l-th feature of the i-th battery energy storage device at time t - △t; i ∈ [1, I], I represents the total number of battery energy storage devices; l ∈ [1, L], L represents the total number of features; △t represents the time window length; If the offset feature is greater than the offset feature threshold D th (t), mark the feature drift F drift (t) = 1; otherwise F drift (t) = 0; ; ; Among them, μ D (t) represents the first shift feature threshold at time t, and υ D (t) represents the second shift feature threshold at time t; σ D (t) represents the third shift feature threshold at time t.

[0009] The calculation of the offset feature based on the time window can capture the dynamic changes of the system state in real time. The triggering mechanism provides a reliable basis for subsequent clustering updates, ensuring the system's rapid response to mutations. According to the above scheme, in S2, the following contents are included: S201: When it is monitored that the feature drift F drift (t) = 1, trigger the clustering update; otherwise, use the clustering result of the previous moment; analyze the first spacing feature of each battery energy storage device, and denote the first spacing feature of the i-th battery energy storage device as ρ i : ; Among them, exp() represents the exponential function with the natural number as the base; ||V i -V j ||2 represents the spacing between the feature vector V i and the feature vector V j , and ||·||2 represents the L2 norm calculation function; A represents the spacing feature coefficient; the spacing feature coefficient is equal to the product of the standard deviation of the spacing between the feature vector V i and the feature vector V j and a preset constant; Analyze the second spacing feature of each battery energy storage device, and denote the second spacing feature of the i-th battery energy storage device as δ i ; If the first spacing feature of the i-th battery energy storage device is not the maximum value, extract all the battery energy storage devices whose first spacing feature is greater than that of the i-th battery energy storage device to form the spacing analysis set of the i-th battery energy storage device; the second spacing feature of the i-th battery energy storage device is equal to the maximum value of the spacing between the feature vector of the i-th battery energy storage device and the feature vectors of the battery energy storage devices in the corresponding spacing analysis set; if the first spacing feature of the i-th battery energy storage device is the maximum value, the second spacing feature of the i-th battery energy storage device is equal to the maximum value of the spacing between the feature vector of the i-th battery energy storage device and the feature vector of any battery energy storage device; Denote the battery energy storage devices that simultaneously satisfy that the first spacing feature and the second spacing feature are greater than the corresponding thresholds as the central battery energy storage devices; divide the non-central battery energy storage devices into the corresponding central battery energy storage devices according to the minimum value of the feature vector spacing; traverse all the non-central battery energy storage devices to generate the final clustering result; This application uses an exponential function weighted distance to amplify the local density difference, highlight the high-density areas, and enhance the accuracy of clustering center recognition; combined with the maximum spacing analysis, it effectively distinguishes edge devices from isolated points and avoids noise interference.

[0010] S202: Analyze the centroid vectors of each cluster, and denote the centroid vector of the m-th cluster as V centroid m ; ; where, C m represents the m-th cluster; |C m | represents the number of battery energy storage devices in the m-th cluster, V （m，n） represents the feature vector of the n-th battery energy storage device in the m-th cluster; Adopt weight decay accumulation to analyze the equivalent capacity of each cluster based on the centroid vector, ; where, C m ceff represents the equivalent capacity of the m-th cluster; C （m，n） represents the rated capacity of the n-th battery energy storage device in the m-th cluster, B represents the decay coefficient, and the decay coefficient is equal to the product of the spacing feature coefficient and a preset constant; normalize the equivalent capacities of all clusters.

[0011] Dynamically adjust the capacity weight according to the distance between the device and the centroid. The farther the device is, the smaller its contribution, reflecting the principle of "proximity and balance"; normalizing the equivalent capacity simplifies subsequent regulation and distribution to ensure a reasonable capacity ratio for each cluster.

[0012] According to the above scheme, in S3, it includes the following content: S301: Obtain the set of historical regulation records from the database, denoted as {H r = (t r , △P r , △Q r , C m (t r )) | r ∈ [1, R]}, where R represents the total number of historical regulation records; where t r represents the r-th historical regulation time, △P r represents the r-th active power regulation amount, △Q r represents the r-th reactive power regulation amount, C m (t r ) represents the set of all clustering groups at the corresponding time; obtain the set of clustering groups at the current time, denoted as {C m (t) | m ∈ [1, M]}, where M represents the total number of clusters at the current time; analyze the correlation value between any two times t and t'; ; Wherein, λ represents the correlation coefficient, and the correlation coefficient is a system - preset constant; K() represents the index function; if C m (t) ∩ C m (t') has at least one same battery energy storage device, then K(C m (t) ∩ C m (t')) = 1; if C m (t) ∩ C m (t') is an empty set, then K(C m (t) ∩ C m (t')) = 0; Denote the correlation matrix corresponding to the historical moment sequence {t r |r ∈ [1, R]} as K(t) = {K(t, t r )|r ∈ [1, R]}; S302: Extract the real - time power fluctuation, analyze the composite energy fluctuation value at the current moment, and the composite energy fluctuation value is equal to the product of the arithmetic square root of the sum of the squares of the real - time active power and the real - time reactive power and the maximum correlation value; Analyze the dynamic regulation threshold based on the composite energy fluctuation at the current moment, and the dynamic regulation threshold is equal to the product of the first smoothing coefficient and the composite energy fluctuation value plus the product of the second smoothing coefficient and the exponential moving average of the composite energy fluctuation value; The first smoothing coefficient and the second smoothing coefficient are system - preset constants.

[0013] Mine the regulation experience under similar working conditions through the correlation matrix of the historical regulation records and the current clustering, and provide a reference for the dynamic threshold; Adjust the influence weight of the historical data on the current decision - making to balance real - time performance and experience; Combine the real - time power fluctuation and the maximum correlation value to highlight the influence of key historical events; Through the smoothing coefficient, fuse the current fluctuation and the long - term trend, so that the threshold can not only respond quickly to mutations but also avoid short - term noise interference.

[0014] According to the above - mentioned scheme, in S4, it includes the following content: S401: Obtain the total regulation instruction at time t, and the total regulation instruction includes the total active and reactive regulation amounts; Analyze the matching degree index for each cluster, and denote the matching degree index of the m - th cluster as M m ; ; Wherein, △P req represents the total active regulation amount, △Q req represents the total reactive regulation amount, P total represents the total rated active power capacity, and the total rated active power capacity is equal to the sum of the rated capacities of all battery energy storage devices, Q total represents the total rated reactive power capacity, and the total rated reactive power capacity is equal to the total rated active power capacity; Cm * (t) represents the normalized value of the equivalent capacity of the m-th cluster; all matching degree indicators are sorted in descending order to obtain an ordered cluster index sequence; S402: When the arithmetic square root of the sum of the squares of the total active and reactive power regulation amounts is greater than the corresponding dynamic regulation threshold, hierarchical regulation is triggered. Based on the ordered cluster index sequence, hierarchical regulation instruction analysis is performed, and the hierarchical regulation instruction for cluster m k is denoted as ; where k represents the serial number of the cluster in the ordered cluster index sequence, represents the active power regulation amount of cluster m k , represents the reactive power regulation amount of cluster m k ; ; ; where, R P represents the remaining active power regulation demand, R Q represents the remaining reactive power regulation demand, represents the maximum available active capacity of cluster m k , represents the maximum available reactive capacity of cluster m k ; represents the variance of the capacity distribution of cluster m k ; sign() represents the direction function. If the content of the direction function is greater than 0, the output is +1. If the content of the direction function is less than 0, the output is -1. If the content of the direction function is equal to 0, the output is 0; the maximum available active capacity is equal to the product of the normalized value of the corresponding equivalent capacity and the total active rated capacity; the maximum available reactive capacity is equal to the product of the normalized value of the corresponding equivalent capacity and the total reactive rated capacity.

[0015] Match the regulation demand with the cluster capacity to ensure that large-capacity clusters give priority to undertaking regulation tasks and optimize resource allocation; combine the remaining demand, available capacity and variance to dynamically adjust the allocation ratio, and reduce the allocation of clusters with large variance to reduce the regulation risk.

[0016] In another aspect of the present application, a network-forming battery energy storage system is provided. The system is implemented by applying the above-mentioned network-forming battery energy storage control method. The system includes an energy storage device offset analysis module, a device clustering capacity analysis module, a correlation threshold analysis module and a regulation analysis module; The energy storage device offset analysis module is used to collect the electrical parameters and operating state parameters of the energy storage device in real time, preprocess the electrical parameters and operating state parameters to generate feature vectors; analyze the offset features based on the feature vectors and perform feature drift marking; The device clustering capacity analysis module is used to analyze the spacing characteristics of each battery energy storage device, and perform clustering analysis according to the spacing characteristics; analyze the equivalent capacity of each cluster in the clustering result; The correlation threshold analysis module is used to obtain the historical regulation record set, analyze the correlation degree between any two moments; extract the real-time power fluctuation, analyze the composite energy fluctuation value at the current moment, and analyze the dynamic regulation threshold in combination with the correlation degree between the real-time moment and the historical moment; The regulation analysis module is used to analyze the matching degree index of each cluster by combining the equivalent capacity and the total regulation instruction, generate an ordered cluster index sequence after sorting, and perform hierarchical regulation instruction analysis on the ordered cluster index sequence.

[0017] According to the above solution, the energy storage device offset analysis module includes a feature vector analysis unit and an offset analysis unit; The feature vector analysis unit is used to obtain the electrical parameters and dynamic operation parameters of the network-forming battery energy storage device, perform normalization processing on the electrical parameters and dynamic operation parameters respectively, generate the feature vector of the battery energy storage device, and splice the feature vectors of all battery energy storage devices to obtain a feature vector matrix; The offset analysis unit is used to analyze the offset characteristics between the feature matrix at the end of the current window and the matrix at the end of the previous window based on the time window. If the offset characteristic is greater than the offset characteristic threshold, the feature drift is marked.

[0018] According to the above solution, the device clustering capacity analysis module includes a device clustering unit and an equivalent capacity analysis unit; The device clustering unit is used to analyze the first spacing characteristic and the second spacing characteristic of the battery energy storage device; record the battery energy storage device that simultaneously satisfies that the first spacing characteristic and the second spacing characteristic are greater than the corresponding thresholds as the central battery energy storage device; divide the non-central battery energy storage devices into the corresponding central battery energy storage devices according to the minimum value of the feature vector spacing; traverse all non-central battery energy storage devices to generate the final clustering result; The equivalent capacity analysis unit is used to analyze the centroid vector of each cluster, and use weighted decay accumulation to analyze the equivalent capacity of each cluster based on the centroid vector, and perform normalization processing on the equivalent capacities of all clusters.

[0019] According to the above solution, the correlation threshold analysis module includes a correlation analysis unit and a dynamic regulation threshold analysis unit; The correlation analysis unit is used to obtain the historical regulation record set from the database and analyze the correlation value based on the clustering grouping sets at two moments; The dynamic regulation threshold analysis unit is used to extract the real-time power fluctuation, analyze the composite energy fluctuation value at the current moment, and analyze the dynamic regulation threshold based on the composite energy fluctuation at the current moment. The dynamic regulation threshold is equal to the product of the first smoothing coefficient and the composite energy fluctuation value plus the product of the second smoothing coefficient and the exponential moving average of the composite energy fluctuation value; the first smoothing coefficient and the second smoothing coefficient are preset constants of the system.

[0020] According to the above solution, the regulation analysis module includes a regulation sequence analysis unit and a hierarchical regulation instruction analysis unit; The regulation sequence analysis unit is used to obtain the total regulation instruction and analyze the matching degree index for each cluster in combination with the equivalent capacity normalization value; The hierarchical regulation instruction analysis unit is used to trigger hierarchical regulation when the arithmetic square root of the sum of the squares of the total active and reactive regulation amounts is greater than the corresponding dynamic regulation threshold, and perform hierarchical regulation instruction analysis based on the ordered cluster index sequence.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows: This application linearizes the impedance change, limits the power change rate within a range, compresses the parameter range, suppresses outliers, and improves the model stability; The calculation based on the offset feature of the time window can capture the dynamic changes of the system state in real time, and the triggering mechanism provides a reliable basis for subsequent clustering updates, ensuring the system's rapid response to mutations; This application amplifies the local density difference through the exponential function weighted distance, highlights the high-density area, and enhances the accuracy of clustering center recognition; Combining the maximum spacing analysis effectively distinguishes edge devices from isolated points and avoids noise interference; Dynamically adjust the capacity weight according to the distance between the device and the centroid, and the farther the device is, the smaller its contribution, reflecting the principle of "proximity and balance"; The equivalent capacity normalization simplifies the subsequent regulation allocation and ensures a reasonable capacity ratio for each cluster; Through the correlation matrix between the historical regulation record and the current clustering, excavate the regulation experience under similar working conditions and provide a reference for the dynamic threshold; Adjust the influence weight of historical data on the current decision-making, balance real-time performance and experience; Combine real-time power fluctuation and the maximum correlation value to highlight the impact of key historical events; Through the smoothing coefficient, fuse the current fluctuation and the long-term trend, so that the threshold can quickly respond to mutations and avoid short-term noise interference; Match the regulation demand with the cluster capacity to ensure that large-capacity clusters give priority to undertaking regulation tasks and optimize resource allocation; Combine the remaining demand, available capacity and variance, dynamically adjust the allocation ratio, and reduce the allocation of clusters with large variance to reduce the regulation risk. Brief Description of the Drawings

[0022] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the drawings: Figure 1Schematic flow chart of a network - forming battery energy storage control method of the present invention; Figure 2 Schematic structural diagram of a network - forming battery energy storage system of the present invention. Detailed implementation manners

[0023] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0024] Please refer to Figure 1 , the present invention provides a technical solution: a network - forming battery energy storage control method, which includes the following steps: S1: Real - time collect the electrical parameters and operating state parameters of the energy storage device, pre - process the electrical parameters and operating state parameters to generate feature vectors; analyze the offset features based on the feature vectors and perform feature drift marking; S2: Analyze the spacing features of each battery energy storage device, perform clustering analysis according to the spacing features; analyze the equivalent capacity of each cluster in the clustering result; S3: Obtain the set of historical regulation records, analyze the degree of association between any two moments; extract the real - time power fluctuation, analyze the composite energy fluctuation value at the current moment, and analyze the dynamic regulation threshold in combination with the degree of association between the real - time moment and the historical moment; S4: Analyze the matching degree index of each cluster in combination with the equivalent capacity and the total regulation instruction, generate an ordered cluster index sequence after sorting, and perform hierarchical regulation instruction analysis on the ordered cluster index sequence.

[0025] In S1, it includes the following content: S101: Obtain the electrical parameters and dynamic operating parameters of the network - forming battery energy storage device. The electrical parameters include equivalent impedance and phase angle; the dynamic operating parameters include power change rate and response time; perform normalization processing on the electrical parameters and dynamic operating parameters respectively to generate the feature vector of the battery energy storage device. Denote the feature vector of the i - th device at time t as: V i (t)=[ (logZ i (t)) / Z0, θ i (t) / π, tanh((dP i / dt) / △P max ), 1 / T i ; where Z i (t) represents the equivalent impedance of the i - th battery energy storage device at time t; Z0 represents the impedance reference value; θ iThe phase angle of the \(i\)-th battery energy storage device at time \(t\) is denoted as \(\theta(t)\); the phase angle represents the phase difference between the voltage and current when the battery energy storage device is connected to the grid; \(dP\) i / dt represents the power change rate of the \(i\)-th battery energy storage device at time \(t\); \(\Delta P\) max represents the maximum value of the power change rate; \(T\) i represents the average response time of the \(i\)-th battery energy storage device; the eigenvectors of all battery energy storage devices are concatenated to obtain the eigenvector matrix at time \(t\); S102: The system presets the time window length, and based on the time window, analyzes the offset feature \(D(t)\) between the feature matrix at the end of the current window and the matrix at the end of the previous window: ; where, \(V\) （i，l） (t) represents the \(l\)-th feature of the \(i\)-th battery energy storage device at time \(t\); \(V\) （i，l） (t - \(\Delta t\)) represents the \(l\)-th feature of the \(i\)-th battery energy storage device at time \(t - \Delta t\); \(i\in[1, I]\), where \(I\) represents the total number of battery energy storage devices; \(l\in[1, L]\), where \(L\) represents the total number of features; \(\Delta t\) represents the time window length; If the offset feature is greater than the offset feature threshold \(D\) th (t), mark the feature drift \(F\) drift (t) = 1; otherwise \(F\) drift (t) = 0; ; ; where, \(\mu\) D (t) represents the first offset feature threshold at time \(t\), \(\upsilon\) D (t) represents the second offset feature threshold at time \(t\); \(\sigma\) D (t) represents the third offset feature threshold at time \(t\).

[0026] In S2, it includes the following content: S201: When it is monitored that the feature drift \(F\) drift (t) = 1, trigger the clustering update; otherwise, use the clustering result of the previous moment; analyze the first spacing feature of each battery energy storage device, and denote the first spacing feature of the \(i\)-th battery energy storage device as \(\rho\) i : where, \(\exp()\) represents the exponential function with the natural number as the base; \(\left\lVert V\right\rVert^2\) i - V j \(\left\lVert\right\rVert^2\) represents the eigenvector \(V\) i and the eigenvector \(V\) jThe spacing, ||·||2 represents the L2 norm calculation function; A represents the spacing feature coefficient; the spacing feature coefficient is equal to the eigenvector V i and the eigenvector V j The product of the standard deviation of the spacing of and the preset constant; Analyze the second spacing feature of each battery energy storage device, and denote the second spacing feature of the i-th battery energy storage device as δ i ; If the first spacing feature of the i-th battery energy storage device is not the maximum value, extract all battery energy storage devices whose first spacing feature is greater than that of the i-th battery energy storage device to form the spacing analysis set of the i-th battery energy storage device; the second spacing feature of the i-th battery energy storage device is equal to the maximum value of the spacing between the eigenvector of the i-th battery energy storage device and the eigenvectors of the battery energy storage devices in the corresponding spacing analysis set; if the first spacing feature of the i-th battery energy storage device is the maximum value, the second spacing feature of the i-th battery energy storage device is equal to the maximum value of the spacing between the eigenvector of the i-th battery energy storage device and the eigenvector of any battery energy storage device; Denote the battery energy storage devices that simultaneously satisfy the first spacing feature and the second spacing feature being greater than the corresponding threshold as the central battery energy storage devices; divide the non-central battery energy storage devices into the corresponding central battery energy storage devices according to the minimum value of the eigenvector spacing; traverse all non-central battery energy storage devices to generate the final clustering result; S202: Analyze the centroid vector of each cluster, and denote the centroid vector of the m-th cluster as V centroid m ; ; Among them, C m represents the m-th cluster; |C m | represents the number of battery energy storage devices in the m-th cluster, V （m，n） represents the eigenvector of the n-th battery energy storage device in the m-th cluster; Based on the centroid vector, use weighted decay accumulation to analyze the equivalent capacity of each cluster, ; Among them, C m ceff represents the equivalent capacity of the m-th cluster; C （m，n） represents the rated capacity of the n-th battery energy storage device in the m-th cluster, B represents the decay coefficient, and the decay coefficient is equal to the product of the spacing feature coefficient and the preset constant; normalize the equivalent capacities of all clusters.

[0027] In S3, it includes the following content: S301: Obtain the historical regulation record set from the database, denoted as {H r = (tr , △P r , △Q r , C m (t r ))| r ∈ [1, R]}, where R represents the total number of historical regulation records; where t r represents the r-th historical regulation moment, △P r represents the r-th active power regulation amount, △Q r represents the r-th reactive power regulation amount, C m (t r ) represents the set of all clustering groups at the corresponding moment; obtain the clustering group set at the current moment, denoted as {C m (t)| m ∈ [1, M]}, where M represents the total number of clusters at the current moment; analyze the correlation value between any two moments t and t’; ; Among them, λ represents the correlation coefficient, and the correlation coefficient is a system preset constant; K() represents the index function; if C m (t)∩C m (t’) has at least one same battery energy storage device, then K(C m (t)∩C m (t’)) = 1; if C m (t)∩C m (t’) is an empty set, then K(C m (t)∩C m (t’)) = 0; denote the correlation matrix corresponding to the historical moment sequence {t r | r ∈ [1, R]} as K(t) = {K(t, t r )| r ∈ [1, R]}; S302: Extract the real-time power fluctuation, analyze the composite energy fluctuation value at the current moment, and the composite energy fluctuation value is equal to the product of the arithmetic square root of the sum of the squares of the real-time active power and the real-time reactive power and the maximum correlation value; analyze the dynamic regulation threshold based on the composite energy fluctuation at the current moment, and the dynamic regulation threshold is equal to the product of the first smoothing coefficient and the composite energy fluctuation value plus the product of the second smoothing coefficient and the exponential moving average of the composite energy fluctuation value; the first smoothing coefficient and the second smoothing coefficient are system preset constants.

[0028] In S4, it includes the following content: S401: Obtain the total regulation command at moment t, and the total regulation command includes the total active and reactive regulation amounts; analyze the matching degree index for each cluster, and denote the matching degree index of the m-th cluster as M m ; ; Among them, △P reqRepresents the total active power regulation amount, △Q req Represents the total reactive power regulation amount, P total Represents the total rated active power capacity. The total rated active power capacity is equal to the sum of the rated capacities of all battery energy storage devices, Q total Represents the total rated reactive power capacity. The total rated reactive power capacity is equal to the total rated active power capacity; C m * (t) represents the normalized value of the equivalent capacity of the m-th cluster; all matching degree indicators are sorted in descending order to obtain an ordered cluster index sequence; Example 1: S402: In this example, the actual value P actual = 950kW, Q actual = 300kVar, reference value: P ref = 1000kW, Q ref = 250kVar; Therefore, △P = 950 - 1000 = -50kW, ΔQ = 300 - 250 = +50kVar; [(-50) 2 +(+50) 2 1 / 2 ≈ 70.7kVA; the dynamic regulation threshold in this example is 60kVA; then hierarchical regulation is triggered; Based on the ordered cluster index sequence, hierarchical regulation instruction analysis is performed, and the hierarchical regulation instruction of cluster m k is denoted as ; where k represents the serial number of the cluster in the ordered cluster index sequence, represents the active power regulation amount of cluster m k , represents the reactive power regulation amount of cluster m k ; ; ; Among them, R P represents the remaining active power regulation demand, R Q represents the remaining reactive power regulation demand, represents the maximum available active power capacity of cluster m k , represents the maximum available reactive power capacity of cluster m k , represents cluster m k ​The variance of the capacity distribution; sign() represents the direction function. If the content of the direction function is greater than 0, the output is +1. If the content of the direction function is less than 0, the output is -1. If the content of the direction function is equal to 0, the output is 0; The available active maximum capacity is equal to the product of the normalized value of the corresponding equivalent capacity and the total active rated capacity; The available reactive maximum capacity is equal to the product of the normalized value of the corresponding equivalent capacity and the total reactive rated capacity.

[0029] Please refer to Figure 2 , The present invention provides a technical solution: a network-forming battery energy storage system, which includes an energy storage device offset analysis module, a device clustering capacity analysis module, a correlation threshold analysis module, and a regulation analysis module; The energy storage device offset analysis module is used to collect electrical parameters and operating state parameters of the energy storage device in real time, preprocess the electrical parameters and operating state parameters to generate feature vectors; analyze offset features based on the feature vectors and perform feature drift marking; The device clustering capacity analysis module is used to analyze the spacing characteristics of each battery energy storage device, perform clustering analysis according to the spacing characteristics; analyze the equivalent capacity of each cluster in the clustering result; The correlation threshold analysis module is used to obtain a set of historical regulation records, analyze the correlation degree between any two moments; extract real-time power fluctuations, analyze the composite energy fluctuation value at the current moment, and analyze the dynamic regulation threshold in combination with the correlation degree between the real-time moment and the historical moment; The regulation analysis module is used to analyze the matching degree indexes of each cluster in combination with the equivalent capacity and the total regulation instruction, generate an ordered cluster index sequence after sorting, and perform hierarchical regulation instruction analysis on the ordered cluster index sequence.

[0030] The energy storage device offset analysis module includes a feature vector analysis unit and an offset analysis unit; The feature vector analysis unit is used to obtain the electrical parameters and dynamic operating parameters of the network-forming battery energy storage device, perform normalization processing on the electrical parameters and dynamic operating parameters respectively to generate feature vectors of the battery energy storage device, splice the feature vectors of all battery energy storage devices to obtain a feature vector matrix; The offset analysis unit is used to analyze the offset feature between the feature matrix at the end of the current window and the matrix at the end of the previous window based on a time window. If the offset feature is greater than the offset feature threshold, mark the feature drift.

[0031] The device clustering capacity analysis module includes a device clustering unit and an equivalent capacity analysis unit; The device clustering unit is used to analyze the first spacing feature and the second spacing feature of the battery energy storage devices; mark the battery energy storage devices that simultaneously satisfy the conditions that the first spacing feature and the second spacing feature are greater than the corresponding thresholds as the central battery energy storage devices; divide the non - central battery energy storage devices into the corresponding central battery energy storage devices according to the minimum value of the feature vector spacing; traverse all non - central battery energy storage devices to generate the final clustering result; The equivalent capacity analysis unit is used to analyze the centroid vector of each cluster, adopt weighted decay accumulation to analyze the equivalent capacity of each cluster based on the centroid vector, and normalize the equivalent capacities of all clusters.

[0032] The association threshold analysis module includes an association analysis unit and a dynamic regulation threshold analysis unit; The association analysis unit is used to obtain the historical regulation record set from the database and analyze the association value based on the clustering grouping sets at two moments; The dynamic regulation threshold analysis unit is used to extract the real - time power fluctuation, analyze the composite energy fluctuation value at the current moment, analyze the dynamic regulation threshold based on the composite energy fluctuation at the current moment, and the dynamic regulation threshold is equal to the product of the first smoothing coefficient and the composite energy fluctuation value plus the product of the second smoothing coefficient and the exponential moving average of the composite energy fluctuation value; the first smoothing coefficient and the second smoothing coefficient are system - preset constants.

[0033] The regulation analysis module includes a regulation sequence analysis unit and a hierarchical regulation instruction analysis unit; The regulation sequence analysis unit is used to obtain the total regulation instruction and analyze the matching degree index for each cluster in combination with the equivalent capacity normalization value; The hierarchical regulation instruction analysis unit is used to trigger hierarchical regulation when the arithmetic square root of the sum of the squares of the total active and reactive regulation amounts is greater than the corresponding dynamic regulation threshold, and perform hierarchical regulation instruction analysis based on the ordered cluster index sequence.

[0034] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "including", "comprising" or any other variant thereof is intended to cover non - exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0035] It is apparent to those skilled in the art that the present invention is not limited to the details of the above-described exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, in all respects, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Accordingly, all changes that fall within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. A grid-forming battery energy storage control method, characterized in that, The method includes the following steps: S1: Collect the electrical parameters and operating state parameters of the energy storage device in real time, preprocess the electrical parameters and operating state parameters, and generate feature vectors; Analyze the offset features based on the feature vectors and perform feature drift marking; S2: Analyze the spacing features of each battery energy storage device, and perform clustering analysis according to the spacing features; analyze the equivalent capacity of each cluster in the clustering result; S3: Obtain the historical regulation record set and analyze the degree of association between any two moments; Extract the real-time power fluctuation, analyze the composite energy fluctuation value at the current moment, and analyze the dynamic regulation threshold in combination with the degree of association between the real-time moment and the historical moment; S4: Analyze the matching degree index of each cluster by combining the equivalent capacity and the total regulation instruction, generate an ordered cluster index sequence after sorting, and perform hierarchical regulation instruction analysis on the ordered cluster index sequence.

2. The network-forming type battery energy storage control method according to claim 1, characterized in that: In S1, the following content is included: S101: Obtain the electrical parameters and dynamic operation parameters of the network-forming battery energy storage device. The electrical parameters include equivalent impedance and phase angle; the dynamic operation parameters include power change rate and response time. Normalize the electrical parameters and dynamic operation parameters respectively to generate the feature vector of the battery energy storage device. Denote the feature vector of the i-th device at time t as: V i (t) = [(logZ i (t)) / Z0, θ i (t) / π, tanh((dP i / dt) / △P max ),1 / T i ; where Z i (t) represents the equivalent impedance of the i-th battery energy storage device at time t; Z0 represents the impedance reference value; θ i (t) represents the phase angle of the i-th battery energy storage device at time t; the phase angle represents the phase difference between the voltage and current when the battery energy storage device is connected to the power grid; dP i / dt represents the power change rate of the i-th battery energy storage device at time t; △P max represents the maximum value of the power change rate; T i represents the average response time of the i-th battery energy storage device. Concatenate the feature vectors of all battery energy storage devices to obtain the feature vector matrix at time t; S102: The system preset time window length, analyze the offset feature D(t) between the feature matrix at the end of the current window and the matrix at the end of the previous window based on the time window; if the offset feature is greater than the offset feature threshold D th (t), mark the feature drift F drift (t) = 1; otherwise F drift (t) = 0.

3. A grid-forming battery energy storage control method according to claim 2, characterized in that: In S2, the following content is included: S201: When the feature drift F drift (t) = 1 is detected, trigger cluster update; Otherwise, use the clustering result of the previous moment; analyze the first spacing feature of each battery energy storage device, and denote the first spacing feature of the i-th battery energy storage device as ρ i ; Analyze the second spacing feature of each battery energy storage device, and denote the second spacing feature of the i-th battery energy storage device as δ i ; If the first spacing feature of the i-th battery energy storage device is not the maximum value, extract all battery energy storage devices with the first spacing feature greater than that of the i-th battery energy storage device to form the spacing analysis set of the i-th battery energy storage device; the second spacing feature of the i-th battery energy storage device is equal to the maximum value of the distance between the feature vector of the i-th battery energy storage device and the feature vectors of the battery energy storage devices in the corresponding spacing analysis set; if the first spacing feature of the i-th battery energy storage device is the maximum value, the second spacing feature of the i-th battery energy storage device is equal to the maximum value of the distance between the feature vector of the i-th battery energy storage device and the feature vector of any battery energy storage device; The battery energy storage devices that simultaneously satisfy the conditions that the first spacing feature and the second spacing feature are greater than the corresponding thresholds are recorded as central battery energy storage devices; the non-central battery energy storage devices are divided into the corresponding central battery energy storage devices according to the minimum value of the feature vector distance; traverse all non-central battery energy storage devices to generate the final clustering result; S202: Analyze the centroid vectors of each cluster, and denote the centroid vector of the m-th cluster among them as V centroid m ; Adopt weight decay accumulation based on the centroid vector to analyze the equivalent capacity of each cluster; perform normalization processing on the equivalent capacities of all clusters.

4. A grid-forming battery energy storage control method according to claim 3, characterized in that: In S3, the following content is included: S301: Obtain the set of historical regulation records from the database, denoted as {H r = (t r , △P r , △Q r , C m (t r )) | r ∈ [1, R]}, where R represents the total number of historical regulation records; among them, t r represents the r-th historical regulation moment, △P r represents the r-th active power regulation amount, △Q r represents the r-th reactive power regulation amount, C m (t r ) represents the set of all clustering groups at the corresponding moment; obtain the clustering group set at the current moment, denoted as {C m (t) | m ∈ [1, M]}, where M represents the total number of clusters at the current moment; analyze the correlation value between any two moments t and t'; denote the correlation matrix corresponding to the historical moment sequence {t r | r ∈ [1, R]} as K(t) = {K(t, t r ) | r ∈ [1, R]}; S302: Extract the real-time power fluctuation, analyze the composite energy fluctuation value at the current moment, and the composite energy fluctuation value is equal to the product of the arithmetic square root of the sum of the squares of the real-time active power and the real-time reactive power and the maximum correlation value; Analyze the dynamic regulation threshold based on the composite energy fluctuation at the current moment, and the dynamic regulation threshold is equal to the product of the first smoothing coefficient and the composite energy fluctuation value plus the product of the second smoothing coefficient and the exponential moving average of the composite energy fluctuation value; the first smoothing coefficient and the second smoothing coefficient are system preset constants.

5. A network-forming battery energy storage control method according to claim 4, characterized in that: In S4, the following content is included: S401: Obtain the total regulation command at time t, where the total regulation command includes the total active and reactive regulation amounts; analyze the matching degree index for each cluster, and denote the matching degree index of the m-th cluster as M m ; ； Among them, △P req represents the total active power regulation amount, △Q req represents the total reactive power regulation amount, P total represents the total rated active power capacity, and the total rated active power capacity is equal to the sum of the rated capacities of all battery energy storage devices, Q total represents the total rated reactive power capacity, and the total rated reactive power capacity is equal to the total rated active power capacity; C m * C(t) represents the normalized value of the equivalent capacity of the m-th cluster; arrange all matching degree indexes in descending order to obtain an ordered cluster index sequence; S402: When the arithmetic square root of the sum of the squares of the total active and reactive power regulation amounts is greater than the corresponding dynamic regulation threshold, hierarchical regulation is triggered. Based on the ordered cluster index sequence, hierarchical regulation instructions are analyzed, and the hierarchical regulation instruction for cluster m k is denoted as ; where k represents the serial number of the cluster in the ordered cluster index sequence, represents the active power regulation amount of cluster m k , and k represents the reactive power regulation amount of cluster m 6. A network-forming battery energy storage system, which is implemented by applying the network-forming battery energy storage control method according to any one of claims 1-5, and is characterized in that, The system includes an energy storage device offset analysis module, a device clustering capacity analysis module, an association threshold analysis module, and a regulation analysis module; The energy storage device offset analysis module is used to collect the electrical parameters and operating state parameters of the energy storage device in real time, preprocess the electrical parameters and operating state parameters, and generate feature vectors; Based on Analyze the offset features of the feature vectors and perform feature drift marking; The device clustering capacity analysis module is used to analyze the spacing characteristics of each battery energy storage device and perform clustering analysis according to the spacing characteristics; analyze the equivalent capacity of each cluster in the clustering result; The correlation threshold analysis module is used to obtain the historical regulation record set and analyze the correlation degree between any two moments; Extract the real-time power fluctuation, analyze the composite energy fluctuation value at the current moment, and analyze the dynamic regulation threshold in combination with the correlation degree between the real-time moment and the historical moment; The regulation analysis module is used to analyze the matching degree index of each cluster by combining the equivalent capacity and the total regulation instruction, generate an ordered cluster index sequence after sorting, and perform hierarchical regulation instruction analysis on the ordered cluster index sequence.

7. The network-forming battery energy storage system according to claim 6, characterized in that: The energy storage device offset analysis module includes a feature vector analysis unit and an offset analysis unit; The feature vector analysis unit is used to obtain the electrical parameters and dynamic operation parameters of the network-forming battery energy storage device, perform normalization processing on the electrical parameters and dynamic operation parameters respectively, generate the feature vector of the battery energy storage device, and splice the feature vectors of all battery energy storage devices to obtain a feature vector matrix; The offset analysis unit is used to analyze the offset characteristics between the feature matrix at the end of the current window and the matrix at the end of the previous window based on the time window. If the offset characteristic is greater than the offset characteristic threshold, mark the feature drift.

8. The network-forming type battery energy storage system according to claim 6, wherein: The device clustering capacity analysis module includes a device clustering unit and an equivalent capacity analysis unit; The device clustering unit is used to analyze the first spacing characteristic and the second spacing characteristic of the battery energy storage device; The battery energy storage devices that simultaneously satisfy that the first spacing characteristic and the second spacing characteristic are greater than the corresponding thresholds are recorded as central battery energy storage devices; the non-central battery energy storage devices are divided into the corresponding central battery energy storage devices according to the minimum value of the feature vector spacing; traverse all non-central battery energy storage devices to generate the final clustering result; The equivalent capacity analysis unit is used to analyze the centroid vector of each cluster, and use weighted decay accumulation to analyze the equivalent capacity of each cluster based on the centroid vector, and perform normalization processing on the equivalent capacities of all clusters.

9. A network-forming battery energy storage system according to claim 6, characterized in that: The correlation threshold analysis module includes a correlation analysis unit and a dynamic regulation threshold analysis unit; The correlation analysis unit is used to obtain the historical regulation record set from the database and analyze the correlation value based on the clustering grouping sets of two moments; The dynamic regulation threshold analysis unit is used to extract the real-time power fluctuation, analyze the composite energy fluctuation value at the current moment, and analyze the dynamic regulation threshold based on the composite energy fluctuation at the current moment. The dynamic regulation threshold is equal to the product of the first smoothing coefficient and the composite energy fluctuation value plus the product of the second smoothing coefficient and the exponential moving average of the composite energy fluctuation value; the first smoothing coefficient and the second smoothing coefficient are system preset constants.

10. A network-forming battery energy storage system according to claim 6, characterized in that: The regulation analysis module includes a regulation sequence analysis unit and a hierarchical regulation instruction analysis unit; The regulation sequence analysis unit is used to obtain the total regulation instruction and analyze the matching degree index of each cluster in combination with the equivalent capacity normalization value; The hierarchical control instruction analysis unit is used to trigger hierarchical control when the arithmetic square root of the sum of the squares of the total active and reactive power regulation amounts is greater than the corresponding dynamic regulation threshold, and perform hierarchical control instruction analysis based on the ordered cluster index sequence.

Images