Cluster energy storage cooperative control cloud service platform

Abstract

The invention, which relates to the technical field of energy storage cooperative control, discloses a cluster energy storage cooperative control cloud service platform comprising a data acquisition and edge uploading module, a data preprocessing module, a sensitivity vector construction module, a power flow risk prediction module and an energy storage weight regulation and control module. And the data acquisition and edge uploading module acquires reverse power flow data of each power transformation node in real time through a high-precision current sensor, a voltage sensor and an active power sensor. According to the method, the reverse power flow data of the power transformation nodes are acquired in real time, and the characteristic engineering and the support vector machine model are combined, so that the platform can intelligently identify the power flow susceptibility degree of each node, and the discharge proportion of the energy storage unit is dynamically adjusted according to the power flow susceptibility degree; the problems of abnormal voltage rise and protection misoperation caused by concentrated reverse injection of multiple energy storage stations into the same node are avoided from the source, and the safety and stability of power grid operation and the economical efficiency of energy storage resource scheduling are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage collaborative control, and in particular to a cluster energy storage collaborative control cloud service platform. Background Art

[0002] The cluster energy storage collaborative control cloud service platform is an intelligent management and control platform for multi-station and multi-type energy storage systems. Relying on cloud computing, big data and edge intelligence technologies, it realizes unified access, real-time monitoring, dynamic power dispatch and collaborative optimization control of distributed energy storage equipment. The platform builds a digital twin model of the energy storage cluster, integrates multi-source information such as grid load forecasting, electricity price fluctuations, battery health status, and uses AI algorithms to intelligently formulate charging and discharging strategies, and coordinates various energy storage units on demand to participate in various business scenarios such as grid peak shaving, demand response, and frequency regulation assistance. It not only improves the operating efficiency and economic benefits of the energy storage system, but also enhances the flexibility and stability of grid operation. It is suitable for a high proportion of new energy access environments in new power systems.

[0003] The dynamic power dispatch of the cluster energy storage collaborative control cloud service platform refers to the process in which the platform makes real-time decisions and adjustments on the timing, coordination, and optimality of the charging and discharging behaviors of multiple energy storage units based on a variety of real-time data such as the external power grid operation status, load forecasts, electricity price changes, new energy output fluctuations, and energy storage system status. Its core purpose is to achieve efficient utilization of energy storage resources and maximize economic benefits while ensuring the stability of the power grid.

[0004] The prior art has the following deficiencies: In the existing dynamic dispatching system of energy storage clusters, the dispatching platform usually does not finely control the reverse power flow direction of the energy storage power station injected into the power grid when multiple energy storage units respond to discharge instructions at the same time. Especially in scenarios with complex power grid topology or densely distributed energy storage access nodes, it is easy for multiple energy storage units to reversely send power to the same substation node at the same time. This type of uncoordinated centralized injection behavior may cause the voltage of local nodes to rise abnormally and rapidly, and then cause faults such as false operation of relay protection devices and forced disconnection of energy storage stations. In severe cases, it may even induce safety hazards such as regional voltage collapse, overload or burning of power equipment, affecting the safety and stability of the power grid system.

[0005] Reverse power flow refers to the direction of electric energy flow from the user side or distributed power sources (such as energy storage systems, photovoltaic power generation, etc.) to the main line or substation of the power grid, which is opposite to the forward power flow from the power generation side to the load side in the traditional power system. In the energy storage system, when multiple energy storage units are in the discharge state, the released electric energy is no longer used only for local consumption, but flows back to the substation or public grid node at a certain power level.

[0006] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The object of the present invention is to provide a cluster energy storage collaborative control cloud service platform. By real-time acquiring the reverse power flow data of substation nodes and combining feature engineering and support vector machine models, the platform can intelligently identify the susceptibility degree of the power flow of each node, and dynamically adjust the discharge ratio of energy storage units accordingly, avoiding the problems of abnormal voltage increase and protection misoperation caused by multiple energy storage stations injecting into the same node concentratedly from the source, effectively improving the safety, stability of power grid operation and the economy of energy storage resource scheduling, realizing the coordinated optimization between intelligent scheduling of energy storage systems and safe operation of power grids, so as to solve the problems in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solution: A cluster energy storage collaborative control cloud service platform, including a data acquisition and edge upload module, a data preprocessing module, a sensitivity vector construction module, a power flow risk prediction module, and an energy storage weight regulation module; The data acquisition and edge upload module, in the dynamic scheduling scenario of the energy storage cluster, in order to comprehensively perceive the operating status of each substation node, first deploy high-precision current sensors, voltage sensors, and active power sensors at each substation node, real-time acquire the reverse power flow data of each substation node, and upload the acquired data to the cloud service platform in real-time through the edge gateway; The data preprocessing module preprocesses the reverse power flow data through the cloud service platform to improve the data quality; The sensitivity vector construction module uses feature engineering technology to extract sensitivity features from the preprocessed data that can reflect the reaction intensity of substation nodes to external power injection. After in-depth analysis of the extracted features, the analyzed power flow sensitivity features are constructed into feature vectors to represent the reaction intensity of each substation node when facing external power shocks; The power flow risk prediction module takes the generated sensitivity feature vector as input, sends it into a pre-trained support vector machine model for real-time inference, outputs the power flow susceptibility score of each substation node within the current short-term monitoring window, and makes an intelligent prediction of the power flow risk status of each substation node; The energy storage weight regulation module, according to the power flow susceptibility score output by the support vector machine model, assigns different discharge weights to the energy storage units connected to each substation node, assigns a lower weight to high-susceptibility nodes to limit their discharge ratio; assigns a higher weight to low-susceptibility nodes to enhance their discharge capacity, realizes the intelligent spatial allocation of energy storage resources, reduces the injection intensity in high-risk areas of the system, and avoids voltage shocks and protection misoperations from the source.

[0009] Preferably, the specific steps for deploying high-precision current sensors, voltage sensors, and active power sensors at each substation node to obtain reverse power flow data in real time are as follows: Determine the sensor installation locations according to the primary wiring diagram of the substation and the energy storage access topology; Install current transformers, voltage transformers, and digital power sensors to ensure high sampling accuracy and wide dynamic range; Real-time access the sensor data to the edge gateway device through the communication unit; Preliminarily screen, cache, and timestamp calibrate the collected sensor data in the edge gateway to form a structured power flow data stream.

[0010] Preferably, use feature engineering techniques to extract sensitivity features from the preprocessed data that can reflect the reaction intensity of the substation node to external power injection. Among them, the extracted features include the aggregation degree of the power flow path after reverse injection and the asymmetry degree of three-phase power reverse injection. Within a short-term monitoring window, after in-depth analysis of the extracted features, generate a power flow reverse path compression factor and a node injection imbalance factor respectively, and construct the analyzed power flow reverse path compression factor and node injection imbalance factor into a feature vector to characterize the reaction intensity of each substation node when facing external power impact.

[0011] Preferably, input the sensitivity feature vector composed of the power flow reverse path compression factor and the node injection imbalance factor into a pre-trained support vector machine model for real-time inference, and output the power flow susceptibility score of each substation node within the current short-term monitoring window through the support vector machine model, and conduct intelligent prediction of the power flow risk state of each substation node through the power flow susceptibility score.

[0012] Preferably, according to the power flow susceptibility score output by the support vector machine model, allocate different discharge weights to the energy storage units connected to each substation node. The specific steps are as follows: After obtaining the power flow susceptibility score of each substation node, construct an exponential decay function based on the risk suppression factor, map the power flow sensitivity to the discharge limit factor of the energy storage system, and obtain the dynamic discharge weight. The calculation formula of the dynamic discharge weight is as follows: , where: is the dynamic discharge weight of the energy storage unit connected to the th substation node; is the current available discharge power, representing the maximum power output supported by the energy storage SOC level; is the power flow susceptibility score predicted by the support vector machine model, and the larger the value, the higher the power flow sensitivity; is the risk-sensitive adjustment coefficient, which controls the intensity of the impact of the power flow susceptibility score on the dynamic discharge weight attenuation.

[0013] Preferably, in order to ensure that the system maintains the spatial balance of the substation node distribution while meeting the overall power demand, the generated dynamic discharge weights are normalized to form the final allocation ratio vector. The normalization process uses an adaptive energy constraint model for weighted adjustment. The specific expression is: , where: is the final discharge ratio (normalized weight) of the th energy storage unit; N is the total number of energy storage substation nodes currently participating in the discharge scheduling; the sub-multiplier term is the power flow susceptibility correction factor, which further compresses the discharge ratio of the substation node with higher power flow susceptibility; q is a variable used to index all other substation nodes in the entire energy storage substation node set, used to complete the sum calculation, achieve normalization and relative comparison.

[0014] Preferably, within the short-term monitoring window, the specific steps for generating the power flow reverse path compression factor after in-depth analysis of the power flow path aggregation degree after the reverse injection of the substation node are as follows: After the reverse power injection of the substation node, by analyzing all transmission lines connected to this node and their downstream nodes, the path of the reverse power flow injection is identified. The power aggregation degree of each line is quantified based on the power flow density, and the power flow density is expressed as the ratio of the power change rate along each line to the grid topological distance. The specific calculation formula is: , where: represents the power flow density from substation node i to substation node j, that is, the power transmission intensity per unit grid distance, is the power flow from substation node i to substation node j, represents the magnitude of the power flow, without considering the direction of the power (that is, not considering whether it is flowing forward or backward), only focusing on the intensity of the power; is the distance between substation node i and substation node j (measured by electrical distance or topological distance); After identifying the power flow path related to the substation node, by analyzing the compression effect of the reverse power injection path, the power flow reverse path compression factor is further generated. The power flow reverse path compression factor is calculated through the non-linear relationship between the weighted power flow density and the grid topology. The calculation expression is: , where: is the set of all reverse power paths connected to substation node j; is the sum of all power flow densities related to substation node j; k is the index representing each path in the reverse power path set ; n is the index representing the reverse power path set The number of all paths connected to the substation node j; is the reverse path compression factor of the power flow.

[0015] Preferably, within a short-term monitoring window, after deeply analyzing the asymmetry degree of the three-phase power reverse injection of the substation node, the specific steps for generating the node injection imbalance factor are as follows: Within a short-term monitoring window, extract the reverse active power injection values of phases A, B, and C of the substation node respectively. To eliminate the absolute amplitude interference caused by the dispatching scale, introduce a normalization ratio function to construct the maximum amplitude relative deviation between the three phases. The constructed expression is: , where: , , respectively represent the reverse active power injection values of the three phases of the substation node within the current monitoring window; is a small positive number (such as ) to prevent numerical instability caused by a zero denominator; represents the maximum amplitude relative deviation between the three phases (asymmetry amplitude measurement); In actual operation, the three-phase reverse injection power not only has amplitude differences, but may also cause phase difference expansion effects due to reasons such as equipment parameter differences, uneven line losses, and local voltage distortion. To capture such dynamic interference characteristics, introduce a phase shift superposition term to construct the node injection imbalance factor. The constructed expression is: , where: represents the main frequency phase difference of the reverse injection power waveforms between phase A and phase B; represents the main frequency phase difference of the reverse injection power waveforms between phase B and phase C; represents the main frequency phase difference of the reverse injection power waveforms between phase C and phase A; is the phase sensitivity factor, which controls the amplification effect of the phase shift on the imbalance degree (recommended 0.2 - 0.5); is the node injection imbalance factor, which combines the maximum amplitude asymmetry degree and the phase perturbation intensity, and is used to quantitatively characterize the structural sensitivity of the node to external injection.

[0016] In the above technical solution, the technical effects and advantages provided by the present invention are: By obtaining the reverse power flow data of the substation node in real time and combining feature engineering and the support vector machine model, the platform of the present invention can intelligently identify the power flow susceptibility degree of each node, and accordingly dynamically adjust the discharge ratio of the energy storage unit, avoiding the problems of abnormal voltage rise and protection misoperation caused by multiple energy storage stations concentrating on reverse injection into the same node from the source, effectively improving the safety, stability of the power grid operation and the economy of energy storage resource scheduling, and realizing the coordinated optimization between the intelligent scheduling of the energy storage system and the safe operation of the power grid. Description of the Drawings

[0017] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a module schematic diagram of a cluster energy storage collaborative control cloud service platform of the present invention. Detailed implementation manners

[0019] Now, example embodiments will be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the example embodiments to those skilled in the art.

[0020] The present invention provides a Figure 1 cluster energy storage collaborative control cloud service platform as shown in In the dynamic scheduling scenario of the energy storage cluster, in order to comprehensively perceive the operating states of each substation node, first, high-precision current sensors, voltage sensors, and active power sensors are deployed at each substation node to obtain the reverse power flow data of each substation node in real time, and the collected data is uploaded to the cloud service platform in real time through the edge gateway. These sensors should have a sampling rate and communication ability at the millisecond level, and be able to accurately obtain the reverse power flow behavior presented by the node during energy storage discharge, including core indicators such as power flow direction (forward or reverse), injection amplitude, voltage response, etc. The collected data is uploaded to the cloud service platform in real time through the edge gateway for subsequent data analysis and scheduling decision-making. The role of this step is to establish a "power flow data perception layer" to provide basic data support for the entire prediction and regulation process, and ensure that the system has an accurate and real-time view of the power grid state.

[0021] The specific steps for deploying high-precision current sensors, voltage sensors, and active power sensors at each substation node to obtain reverse power flow data in real time are as follows: First, according to the primary wiring diagram of the substation and the energy storage access topology, determine the appropriate sensor installation locations, and preferentially select the bus outlet, energy storage grid connection node, or main transformer side bus; secondly, install current transformers (CTs), voltage transformers (PTs), and digital power sensors to ensure that they have high sampling accuracy (such as kHz level) and wide dynamic range; thirdly, connect the sensor data to the edge gateway device in real time through a communication unit (such as RS485, CAN, Ethernet); fourthly, perform preliminary screening, caching, and timestamp calibration on the collected sensor data in the edge gateway to form a structured power flow data stream.

[0022] This step realizes the continuous monitoring and perception of the energy storage injection behavior at the physical level and is the basis for subsequent power flow characteristic identification and control decision-making.

[0023] Perform preprocessing operations on the reverse power flow data through the cloud service platform to improve data quality; Perform preprocessing operations on the reverse power flow data through the cloud service platform, which mainly includes the following aspects: First, perform outlier detection and removal to identify and remove abnormal fluctuation data caused by communication failures, electromagnetic interference, or sensor drift; secondly, perform data denoising and smoothing processing. Common methods include moving average, wavelet transform, or median filtering to eliminate short-term spikes and high-frequency disturbances; then perform data alignment and time synchronization to unify the timestamps of the data collected at different nodes to ensure the accuracy of horizontal comparison and time series analysis; finally, perform missing data filling and interpolation processing, and use linear interpolation, Lagrange interpolation, or a time series prediction model to repair local data gaps.

[0024] Through the above preprocessing steps, the continuity, accuracy, and availability of the reverse power flow data can be significantly improved, providing reliable support for subsequent power flow feature extraction and risk prediction.

[0025] Use feature engineering techniques to extract sensitivity features from the preprocessed data that can reflect the response intensity of the substation node to the external power injection. After in-depth analysis of the extracted features, construct the analyzed power flow sensitivity features into a feature vector to characterize the response intensity of each substation node when facing external power shocks; Use feature engineering techniques to extract sensitivity features from the preprocessed data that can reflect the response intensity of the substation node to the external power injection. Among them, the extracted features include the aggregation degree of the power flow path after reverse injection and the asymmetry degree of the three-phase power reverse injection. Within a short-term monitoring window, after in-depth analysis of the extracted features, generate the power flow reverse path compression factor and the node injection imbalance factor respectively, and construct the analyzed power flow reverse path compression factor and the node injection imbalance factor into a feature vector to characterize the response intensity of each substation node when facing external power shocks.

[0026] When the power transformation node shows a high degree of concentration of power flow paths after reverse injection, it usually indicates that the node is relatively sensitive to external power shocks and is prone to local anomalies. The fundamental reason is that: a high degree of concentration of power flow paths means that a large amount of reverse electric energy quickly converges within a short path, a small number of branches, or a concentrated topological area, and cannot be effectively diffused or shunted. This concentrated injection will cause a rapid increase in local voltage and a sharp surge in current density, further amplifying electrical disturbances and triggering phenomena such as harmonics, frequency fluctuations, or protection malfunctions. At the same time, the equipment thermal load and electrical stress in the power flow concentration area suddenly increase, further exacerbating the system's overreaction to the injection disturbance. Therefore, the concentration of power flow paths is essentially an indication of insufficient injection redundancy space, which is highly correlated with the electrical stability margin of the node and is one of the key indicators for judging its power flow sensitivity.

[0027] The specific steps for generating the power flow reverse path compression factor after in-depth analysis of the concentration of power flow paths after reverse injection of the power transformation node within a short-term monitoring window are as follows: After reverse power injection at the power transformation node, by analyzing all transmission lines connected to the node and their downstream nodes, identify the paths of reverse power flow injection. Quantify the power concentration degree of each line based on the power flow density, which is expressed as the ratio of the power change rate along each line to the grid topological distance. The specific calculation formula is: , where: represents the power flow density from power transformation node i to power transformation node j, that is, the power transmission intensity per unit grid distance, is the power flow from power transformation node i to power transformation node j, represents the magnitude of the power flow, without considering the direction of the power (that is, not considering whether it is flowing forward or backward), only focusing on the intensity of the power; is the distance between power transformation node i and power transformation node j (measured by electrical distance or topological distance); After reverse power injection occurs at the power transformation node, the identification of the reverse injection path can be achieved by combining the grid topological structure and real-time power flow measurement data. The specific approach is: first, based on the SCADA system or the GIS grid topological map, construct the mapping relationship between the adjacent nodes and transmission lines of the power transformation node to determine the set of all branches and downstream nodes directly or indirectly connected to the node; subsequently, combine the real-time data collected by current, voltage, and power sensors deployed on each line to determine the power flow direction on each branch; when the power flow direction of a certain branch is from the power transformation node to other nodes, it can be determined that this line is part of the reverse injection path; finally, through a recursive path expansion algorithm, gradually trace the continuous reverse power flow paths starting from the power transformation node to identify the complete path set of the reverse power flow. This method can dynamically depict the transmission chain during the power back-injection process and provide accurate support for subsequent power flow compression analysis and sensitive node identification.

[0028] This step identifies the concentrated paths of reverse power flow by evaluating the power distribution on each line, and measures the concentration of reverse power near the substation node through the power flow density of this path. This step helps to identify and quantify the power flow concentration paths, serving as the basis for subsequent calculation of the power flow reverse path compression factor.

[0029] After identifying the power flow paths related to the substation node, the power flow reverse path compression factor is further generated by analyzing the compression effect on the reverse power injection path. The power flow reverse path compression factor is calculated through the non-linear relationship between the weighted power flow density and the power grid topology, and the calculation formula is: , where: is the set of all reverse power paths connected to substation node j; is the sum of the power flow densities of all paths related to substation node j; k is the index representing each path in the reverse power path set ; n is the number of all paths connected to substation node j in the reverse power path set ; is the power flow reverse path compression factor; This formula comprehensively evaluates the concentration of reverse power flow and the sensitivity of node response through the weighted power flow density and the path compression ratio. The higher the power flow reverse path compression factor, the more concentrated the power flow path of this substation node, and the more sensitive its voltage response and system stability are, and it is prone to fluctuations or instability due to external power injection.

[0030] It can be seen from the reverse power flow path compression factor that within a short-term monitoring window, the larger the performance value of the reverse power flow path compression factor generated after in-depth analysis of the power flow path aggregation degree after reverse injection, the higher the power flow sensitivity of the substation node when facing external power shocks. Conversely, it indicates that the power flow sensitivity of the substation node is lower when facing external power shocks. The reason is that the reverse power flow path compression factor reflects the aggregation degree of power flow in the space around a certain substation node after external power injection. The larger its performance value, the more the reverse power flow is concentrated on a few paths and quickly returns, forming a highly compressed power flow channel, and the electrical energy lacks dispersion and buffering in spatial distribution. This aggregation effect will cause the voltage of local nodes to rise rapidly and the electrical shock to increase significantly, thus triggering phenomena such as overvoltage, frequency disturbance, and even misoperation of relay protection, indicating that the node has a large response intensity to power injection, a low stability margin, and a high power flow sensitivity. Conversely, if the reverse power flow path compression factor is small, it means that the injected power is relatively evenly distributed in multiple paths, and the node has good energy absorption and power flow buffering capabilities, and it shows characteristics such as small voltage fluctuations, strong system stability, and low power flow sensitivity under external power disturbances. Therefore, this reverse power flow path compression factor can be used as an effective indicator to measure the power flow sensitivity of nodes.

[0031] If the degree of asymmetry of the three-phase power reverse injection of the substation node is relatively high, it usually indicates that the substation node has a high response sensitivity when facing external power shocks. The reason is that the three-phase asymmetry reflects the uneven distribution of the injected power in space, and this unevenness will cause negative sequence current and zero sequence current to appear in the power grid, which will in turn trigger phenomena such as voltage fluctuations, harmonic amplification, electromagnetic force pulsation, and local overheating, especially more obvious in scenarios where the grid access capacity is limited or the transformer neutral point is unstable. Asymmetric injection is also likely to induce problems such as misjudgment of protection devices, expansion of voltage imbalance, and aging of equipment insulation, thus exacerbating the electrical sensitivity of the node to power disturbances. Therefore, the degree of asymmetry of the three-phase power reverse injection of the substation node is one of the important sensitivity characteristics to measure the dynamic stability and anti-interference ability of the node.

[0032] Within a short-term monitoring window, the specific steps to generate the node injection imbalance factor after in-depth analysis of the degree of asymmetry of the three-phase power reverse injection of the substation node are as follows: Within a short-term monitoring window, extract the reverse active power injection values of phases A, B, and C of the substation node respectively. To eliminate the absolute amplitude interference brought by the dispatching scale, introduce a normalization ratio function to construct the maximum amplitude relative deviation between the three phases. The constructed expression is: , where: 、 、 respectively represent the reverse active power injection values of the three phases of the substation node within the current monitoring window; To prevent the micro-positive number that causes numerical instability due to a zero denominator (such as ); represents the maximum amplitude relative deviation between three phases (asymmetric amplitude measurement); This step is used to measure the maximum relative amplitude difference in the three-phase power reverse injection, highlighting the response of the "weakest point" of local imbalance.

[0033] In actual operation, the reverse injection power of three phases not only has amplitude differences, but may also cause phase difference expansion effects due to equipment parameter differences, uneven line losses, local voltage distortion, etc. To capture such dynamic interference characteristics, a phase shift superposition term is introduced to construct a node injection imbalance factor, and the constructed expression is: , where: represents the main frequency phase difference of the reverse injection power waveforms between phase A and phase B; represents the main frequency phase difference of the reverse injection power waveforms between phase B and phase C; represents the main frequency phase difference of the reverse injection power waveforms between phase C and phase A; is the phase sensitivity factor, which controls the amplification effect of phase shift on the imbalance degree (recommended 0.2 - 0.5); is the node injection imbalance factor, which integrates the maximum amplitude asymmetry and the phase perturbation intensity, and is used to quantitatively characterize the structural sensitivity of the node to external injection; This step introduces the structural interference (phase misalignment) between three phases into the index evaluation logic, strengthening the description of the complexity of electrical interference from the "amplitude + phase" two dimensions, so that the node injection imbalance factor can not only measure the conventional voltage skew, but also sense advanced instability characteristics such as potential resonance and inter-phase disturbance propagation.

[0034] From the node injection imbalance factor, it can be seen that within a short-term monitoring window, the larger the value of the node injection imbalance factor generated after in-depth analysis of the asymmetry degree of the three-phase power reverse injection of the substation node, the higher the sensitivity of the substation node to the power flow when facing external power shocks. On the contrary, it indicates that the substation node has a lower sensitivity to the power flow when facing external power shocks. The reason is that when a substation node shows strong asymmetry (including amplitude differences and phase misalignments) during the three-phase reverse power injection process, it means that there are large imbalances or local impedance differences in the electrical structure of the substation node, which will make it more likely to cause electrical instability phenomena such as voltage fluctuations, harmonic distortions, enhanced negative sequence currents, and malfunctions of relay protection when receiving sudden power disturbances. This structural asymmetry reflects the inadaptability of the substation node to disturbances. Therefore, the larger the node injection imbalance factor, the more it can quantify its vulnerability to power flow disturbances and the higher the sensitivity; on the contrary, a smaller node injection imbalance factor indicates that the node has a stronger power symmetry acceptance ability, a more balanced response, and a more controllable impact of power flow disturbances, and the sensitivity is relatively low.

[0035] Taking the generated sensitivity feature vector as input, it is fed into a pre-trained support vector machine model for real-time inference, and the power flow susceptibility score of each substation node within the current short-term monitoring window is output, for the intelligent prediction of the power flow risk state of each substation node; The sensitivity feature vector composed of the power flow reverse path compression factor and the node injection imbalance factor is input into a pre-trained support vector machine model for real-time inference. Through the support vector machine model, the power flow susceptibility score of each substation node within the current short-term monitoring window is output, and the intelligent prediction of the power flow risk state of each substation node is carried out through the power flow susceptibility score.

[0036] The pre-trained support vector machine model refers to a supervised machine learning model with classification or regression discrimination ability constructed through offline training based on a large amount of historical power grid operation data and real power flow response results before being formally applied to the power flow susceptibility prediction task of substation nodes. The core task of this model is to learn the "mapping relationship between input features (such as power flow reverse path compression factor, node injection imbalance factor) and output labels (such as power flow susceptibility level or power flow disturbance response intensity)", so that after actual deployment, it can quickly make a prediction output of risk level or numerical score for the newly input real-time feature vector.

[0037] As a classic supervised learning algorithm, the support vector machine model has good small-sample modeling ability, strong generalization ability and non-linear classification ability, and is especially suitable for problems such as power flow susceptibility with high-dimensional sparsity and unclear non-linear boundaries. In the model training stage, the platform will first construct a training data set based on the power flow historical data (including eigenvalue and actual response performance) of a large number of substation nodes. Features such as "power flow reverse path compression factor" and "node injection imbalance factor" respectively describe the power aggregation degree and the three-phase electric energy asymmetry degree, and then the corresponding power flow sensitivity level or index score is marked through manual or rule labels to form an "input-output" training pair. During the training process, the support vector machine model constructs an optimal classification hyperplane (or regression fitting function) to learn the boundary structure that can maximize the inter-class interval and minimize the classification error, so as to form a decision function that can distinguish different power flow risk levels. For complex non-linear problems, the support vector machine model can also use kernel function technology (such as Gaussian radial basis kernel, polynomial kernel, etc.) to map the input features to a higher-dimensional space, so that the originally inseparable data can be linearly separable in the higher-dimensional space, improving the model accuracy and adaptability.

[0038] The support vector machine model after training is the "pre-trained support vector machine model". This model has the ability to "understand the complex relationship between input features and power flow risks" and can be deployed to the platform for real-time inference and prediction. In the application stage, when the platform extracts the "power flow reverse path compression factor" and "node injection imbalance factor" within the current short-term monitoring window of each substation node in real time, these two high-dimensional electrical response features will form a new two-dimensional feature vector and be input into the pre-trained support vector machine model. Based on its internal parameters and the decision boundary obtained through training, the model quickly determines whether the current node is in a high-risk, medium-risk, or low-risk interval, or directly outputs a continuous power flow susceptibility score, which is the "power flow susceptibility score".

[0039] Through the power flow susceptibility score, the platform can realize the intelligent dynamic identification and quantitative expression of the power flow risk status of each substation node. In addition, since the training samples of the support vector machine model can be continuously updated, it has good scalability and online re-training ability, and can adapt to changes in grid operating conditions, types of connected devices, etc., and maintain the reliability and accuracy of the prediction effect in the long term. Therefore, the "pre-trained support vector machine model" is not only the core intelligent module for power flow susceptibility identification in this solution, but also the key foundation for the energy storage scheduling system to achieve adaptive power flow management and risk closed-loop control.

[0040] The support vector machine model is not specifically limited here. Any support vector machine model that can comprehensively analyze the power flow reverse path compression factor and the node injection imbalance factor to generate the power flow susceptibility score is acceptable. To implement the technical solution of the present invention, the present invention provides a specific implementation method; the expression for generating the power flow susceptibility score is: , where , are the preset proportionality coefficients of the power flow reverse path compression factor and the node injection imbalance factor respectively, and , are both greater than 0.

[0041] In this formula expression, the so-called preset proportionality coefficients (i.e., and in the formula) refer to the weighted coefficients artificially set by the system or obtained by fitting historical data when generating the power flow susceptibility score ( ), which are used to weight the two input features "power flow reverse path compression factor and node injection imbalance factor Adjust the relative importance in the comprehensive analysis.

[0042] Specifically, these two proportionality coefficients determine the contribution weights of the two parameters to the final output result when generating the comprehensive sensitivity evaluation result.

[0043] For example: If is relatively large, it indicates that the system emphasizes more on the impact of the "reverse power flow path aggregation effect" on the node susceptibility; If is even larger, it means that the impact of the unbalanced three-phase injection of the node on the power grid fluctuation is given priority consideration.

[0044] Such preset coefficients are usually obtained through the following methods: Set by expert experience: Assign fixed weights to the sensitivities of electrical quantities based on engineering experience; Fitted from historical data: Optimize with minimum error (such as regression analysis, support vector regression) through the power flow response effect in the training data; Dynamically corrected by system tuning: Continuously fine-tune the coefficient values according to the prediction error feedback during the system operation.

[0045] Its role is to give different characteristic parameters differentiability, so that the power flow susceptibility score not only reflects the key influencing factors, but also has strong adaptability and interpretability. The generated power flow susceptibility score can better reflect the comprehensive response intensity of the substation node to the power disturbance in the actual operation.

[0046] From the power flow susceptibility score, it can be seen that within the short-term monitoring window, the larger the performance value of the power flow reverse path compression factor generated by in-depth analysis of the power flow path aggregation degree after reverse injection, and the larger the performance value of the node injection imbalance factor generated by in-depth analysis of the asymmetry degree of the three-phase power reverse injection of the substation node. That is, when using the support vector machine model to intelligently predict the power flow risk state of each substation node, the larger the performance value of the generated power flow susceptibility score, the higher the power flow sensitivity of the substation node when facing external power impact, and vice versa, it indicates that the power flow sensitivity of the substation node when facing external power impact is lower.

[0047] According to the power flow susceptibility score output by the support vector machine model, different discharge weights are assigned to the energy storage units connected to each substation node. A lower weight is assigned to the high-susceptibility nodes to limit their discharge ratio; a higher weight is assigned to the low-susceptibility nodes to enhance their discharge capacity, realizing the intelligent spatial distribution of energy storage resources, reducing the injection intensity in the high-risk area of the system, and avoiding voltage impact and protection maloperation from the source; According to the power flow susceptibility scores output by the support vector machine model, different discharge weights are assigned to the energy storage units connected to each substation node. The specific steps are as follows: After obtaining the power flow susceptibility scores of each substation node, an exponential decay function based on the risk suppression factor is constructed to map the power flow sensitivity to the discharge limit factor of the energy storage system, and the dynamic discharge weight is obtained. The calculation formula of the dynamic discharge weight is as follows: , where: is the dynamic discharge weight of the energy storage unit connected to the th substation node; is the currently available discharge power, representing the maximum power output supported by the energy storage SOC level; is the power flow susceptibility score predicted by the support vector machine model. The larger the value, the higher the power flow sensitivity; is the risk-sensitive adjustment coefficient, which controls the intensity of the effect of the power flow susceptibility score on the attenuation of the dynamic discharge weight; The design of this exponential decay function enables the dynamic discharge weight to be exponentially compressed at substation nodes with higher power flow risks, forming a risk-adaptive energy storage control logic.

[0048] To ensure that the system meets the overall power demand while maintaining the spatial balance of the distribution of substation nodes, the above-generated dynamic discharge weights are normalized to form the final allocation ratio vector. The normalization process uses an adaptive energy constraint model for weighted adjustment. The specific expression is: , where: is the final discharge ratio (normalized weight) of the th energy storage unit; N is the total number of energy storage substation nodes currently participating in discharge scheduling; the sub-multiplier term is the power flow susceptibility correction factor, which further compresses the discharge ratio of substation nodes with higher power flow sensitivity; q is a variable used to index all other substation nodes in the entire energy storage substation node set, used to complete the sum calculation, achieve normalization and relative comparison; This normalization step not only realizes the total control of the spatial allocation of energy storage resources, but also forms a dynamic suppression mechanism through the power flow risk weight factor, so that the final discharge decision reflects the spatial intelligent cooperation characteristics of "avoiding high-voltage sensitive substation nodes and giving priority to low-risk areas", thereby systematically avoiding voltage disturbances, relay malfunctions and power flow reconstruction risks.

[0049] The core role of this discharge weight allocation mechanism lies in: Based on the power flow susceptibility scores output by the support vector machine model, accurately identify the sensitivity of each substation node to external power disturbances during the current dispatching cycle. Based on this, dynamically adjust the discharge ratios of each energy storage unit to implement a "risk-driven" spatial allocation strategy. For nodes with high power flow sensitivity, the system effectively limits their reverse injection power by allocating lower discharge weights, reducing risks such as local voltage surges, grid structure impacts, and relay protection misoperation at the source; while for low-sensitivity nodes with strong power flow tolerance, higher discharge weights are allocated to enhance their regulation effect and stabilize the overall power distribution as a power flow buffer zone. Through this mechanism, not only is the regional coordination and safety optimization of energy storage resources achieved, but also the adaptive regulation ability of the power grid under high-fluctuation load conditions is improved, ensuring the safety, stability, and economy of system operation.

[0050] Based on the above allocation results, during the dispatching execution process, the platform restricts the energy storage units connected to highly susceptible substation nodes, such as reducing their discharge power upper limits, delaying their response times, or setting them as standby response units. When a substation node is in a high-voltage or frequently disturbed area, automatically shield the substation node from participating in large-scale regulation tasks. This mechanism is self-adaptive and can dynamically adjust the response strategy to ensure that the power grid operates in a "low-impact, high-stability" state. By accurately controlling the injection behavior in high-sensitivity areas, ensure the safe operation of electrical equipment and effectively reduce systematic faults such as tripping and disconnection caused by over-injection.

[0051] The platform preferentially schedules the energy storage units connected to low-sensitivity substation nodes to participate in the current round of power release tasks, enabling them to undertake a greater proportion of the regulation work. At the same time, combined with the power grid power flow path optimization strategy and flexible switching equipment, guide the reverse power flow to these stable substation nodes with the ability of "voltage buffer pool" to form a flexible injection area. This mechanism can be superimposed with strategies such as energy storage state and electricity price prediction to ensure the dual goals of economy and stability. Use the substation node with the strongest pressure-bearing capacity in the system as the main regulation channel to effectively relieve the load of the main substation nodes and improve the overall optimization level of the dispatching robustness and power grid power flow distribution.

[0052] Through the above solutions, it is possible to achieve accurate prediction and active avoidance of power grid power flow risks during the dynamic dispatching of the energy storage cluster; by leveraging the real-time obtained reverse power flow data of substation nodes and combining feature engineering and the support vector machine model, the platform can intelligently identify the power flow susceptibility of each node and dynamically adjust the discharge ratios of energy storage units accordingly, avoiding problems such as abnormal voltage increases and protection misoperations caused by multiple energy storage stations injecting in reverse into the same node at the source, effectively improving the safety and stability of power grid operation and the economy of energy storage resource dispatching, and achieving the coordinated optimization between the intelligent dispatching of the energy storage system and the safe operation of the power grid.

[0053] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula that is closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0054] Only some exemplary embodiments of the present invention have been described above by way of illustration. Undoubtedly, for those of ordinary skill in the art, without departing from the spirit and scope of the present invention, the described embodiments can be modified in various different ways. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0055] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A cluster energy storage collaborative control cloud service platform, characterized in that: It includes data collection and edge upload module, data preprocessing module, sensitivity vector construction module, flow risk prediction module and energy storage weight control module; The data collection and edge upload module uses high-precision current sensors, voltage sensors, and active power sensors to obtain the reverse power flow data of each substation node in real time, and uploads the collected data to the cloud service platform through the edge gateway in real time; Data preprocessing module, which performs preprocessing operations on reverse power flow data through the cloud service platform; The sensitivity vector construction module uses feature engineering technology to extract sensitivity features that can reflect the response strength of substation nodes to external power injection from preprocessed data. After in-depth analysis of the extracted features, the analyzed power flow sensitivity features are constructed into feature vectors to represent the response strength of each substation node when facing external power impact. The power flow risk prediction module takes the generated sensitivity feature vector as input and sends it to the pre-trained support vector machine model for real-time reasoning. It outputs the power flow susceptibility score of each substation node in the current short-term monitoring window and makes intelligent predictions on the power flow risk status of each substation node. The energy storage weight control module assigns different discharge weights to the energy storage units connected to each substation node according to the power flow susceptibility score output by the support vector machine model.

2. A cluster energy storage collaborative control cloud service platform according to claim 1, characterized in that: The specific steps for deploying high-precision current sensors, voltage sensors, and active power sensors at each substation to obtain reverse power flow data in real time are as follows: Determine the sensor installation location based on the substation primary wiring diagram and energy storage access topology; Install current transformers, voltage transformers and digital power sensors to ensure they have high sampling accuracy and wide dynamic range; The sensor data is connected to the edge gateway device in real time through the communication unit; In the edge gateway, the collected sensor data is preliminarily screened, cached and timestamped to form a structured current data stream.

3. A cluster energy storage collaborative control cloud service platform according to claim 1, characterized in that: Feature engineering technology is used to extract sensitivity features that can reflect the response intensity of substation nodes to external power injection from the preprocessed data. The extracted features include the concentration of power flow path after reverse injection and the asymmetry of three-phase power reverse injection. Within the short-term monitoring window, the extracted features are deeply analyzed to generate power flow reverse path compression factor and node injection imbalance factor respectively. The analyzed power flow reverse path compression factor and node injection imbalance factor are constructed as feature vectors to characterize the response intensity of each substation node when facing external power impact.

4. A cluster energy storage collaborative control cloud service platform according to claim 3, characterized in that: The sensitivity feature vector composed of the power flow reverse path compression factor and the node injection imbalance factor is input into the pre-trained support vector machine model for real-time reasoning. The support vector machine model outputs the power flow susceptibility score of each substation node in the current short-term monitoring window, and the power flow risk status of each substation node is intelligently predicted based on the power flow susceptibility score.

5. A cluster energy storage collaborative control cloud service platform according to claim 4, characterized in that: According to the power flow susceptibility score output by the support vector machine model, different discharge weights are assigned to the energy storage unit connected to each substation node. The specific steps are as follows: After obtaining the power flow susceptibility score of each substation node, an exponential decay function based on the risk suppression factor is constructed to map the power flow sensitivity to the discharge limiting factor of the energy storage system, and obtain the dynamic discharge weight. The calculation formula of the dynamic discharge weight is as follows: ,in: For the The dynamic discharge weight of the energy storage unit connected to each substation node; is the currently available discharge power, indicating the maximum power output supported by the energy storage SOC level; Scoring the susceptibility of the tides predicted by the support vector machine model; is the risk-sensitive adjustment coefficient, which controls the effect of the power flow susceptibility score on the dynamic discharge weight attenuation.

6. A cluster energy storage collaborative control cloud service platform according to claim 5, characterized in that: The generated dynamic discharge weights are normalized to form the final allocation ratio vector. The normalization process uses an adaptive energy constraint model for weighted adjustment. The specific expression is: ,in: For the The final discharge ratio of each energy storage unit; N is the total number of energy storage substation nodes currently participating in the discharge scheduling; is the power flow susceptibility correction factor; q is a variable used to index all other substation nodes in the entire energy storage substation node set to complete the total calculation.

7. The cluster energy storage collaborative control cloud service platform according to claim 3, characterized in that: In the short-term monitoring window, the specific steps of generating the reverse flow path compression factor after in-depth analysis of the flow path concentration after reverse injection of the substation node are as follows: After the reverse power injection at the substation node, the reverse power injection path is identified by analyzing all transmission lines connected to the node and their downstream nodes. The power concentration of each line is quantified based on the power flow density. The power flow density is expressed as the ratio of the power change rate along each line to the grid topological distance. The specific calculation formula is: ,in: It represents the power flow density from substation node i to substation node j, that is, the power transmission intensity per unit grid distance. is the power flow from substation node i to substation node j, Indicates the magnitude of power flow, regardless of the direction of the power, only focusing on the intensity of the power; is the distance between substation node i and substation node j; After identifying the power flow path related to the substation node, the compression effect of the reverse power injection path is analyzed to further generate the power flow reverse path compression factor. The power flow reverse path compression factor is calculated by the nonlinear relationship between the weighted power flow density and the grid topology. The calculation expression is: ,in: is the set of all reverse power paths connected to substation j; is the sum of all power flow densities associated with substation node j; k is the set of reverse power paths The index of each path in; n represents the reverse power path set The number of all paths connected to substation node j in; is the compression factor of the reverse flow path.

8. The cluster energy storage collaborative control cloud service platform according to claim 3, characterized in that: In the short-term monitoring window, the specific steps of generating the node injection imbalance factor after in-depth analysis of the asymmetry degree of three-phase power reverse injection at the substation node are as follows: In the short-term monitoring window, the reverse active power injection values ​​of phase A, phase B, and phase C of the substation node are extracted respectively, and the normalized ratio function is introduced to construct the maximum relative deviation between the three phases. The constructed expression is: ,in: , , They respectively represent the three-phase reverse active power injection values ​​of the substation node in the current monitoring window; To prevent the denominator from being zero, which may cause numerical instability in the form of a slightly positive number; Indicates the maximum relative deviation between the three phases; The phase offset superposition term is introduced to construct the node injection imbalance factor, and the constructed expression is: ,in: Indicates the main frequency phase difference of the reverse power waveform between phase A and phase B; Indicates the main frequency phase difference of the reverse power waveform between phase B and phase C; Indicates the main frequency phase difference of the reverse power waveform between phase C and phase A; is the phase sensitivity factor, which controls the amplification effect of phase offset on imbalance; The imbalance factor is injected into the node, and the maximum amplitude asymmetry and phase perturbation intensity are integrated to quantitatively characterize the structural sensitivity of the node to external injection.

Images