A method, device and equipment for adaptive switching control of multi-scenario operation modes of an energy storage system

Abstract

The application provides a kind of energy storage system multi-scene operation mode adaptive switching control method, device and equipment, by collecting power grid state, load demand, economic index data multidimensional fusion analysis generates comprehensive state characteristic vector;Scene recognition classification is generated based on the characteristic vector Scene type identification, according to the demand priority ranking Energy space-time transfer analysis generates space-time energy distribution scheme;Distribution scheme is analyzed to identify potential conflict points Scene conflict, build scene coordination mechanism Fusion strategy is generated;Based on the debt risk parameter of the energy overdraft capacity and compensation timing Fusion strategy is obtained;Combined with debt risk parameter Multilevel fault-tolerant switching system is generated Main switching, standby switching and emergency switching scheme;Establish switching decision causal mapping relationship Dynamic adjustment generates adaptive switching instruction sequence;Through operation mode switching Execution data is obtained Reverse correction scene identification, realize energy storage system intelligent adaptive switching control.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to an adaptive switching control method, device and equipment for multi-scenario operation modes of an energy storage system. Background Technology

[0002] With the continuous expansion of new energy power generation and the deepening of power market reform, the strategic position of energy storage systems as a flexible resource for the power system is becoming increasingly prominent. Energy storage systems have technical advantages such as bidirectional power regulation, energy transfer in time and space, and multiple service functions. They can participate in various application scenarios such as grid frequency regulation and peak shaving, new energy consumption, load regulation, and emergency backup, and have become an essential component in building new power systems.

[0003] However, existing energy storage system operation mode switching control methods have significant technical shortcomings: insufficient accuracy in scenario identification, lack of comprehensive analysis capabilities for grid status, load demand, and economic indicators, leading to biased judgment of operation scenarios and inappropriate mode selection; lack of multi-scenario coordination mechanisms, resulting in spatiotemporal conflicts and resource competition among different operation scenarios, and a lack of effective conflict identification and coordination strategies, affecting the overall system operating efficiency; lack of intelligence and adaptability in switching decisions, often relying on fixed thresholds or empirical parameters, unable to adapt to complex and ever-changing operating environments and demand changes; weak risk assessment and fault-tolerant protection capabilities, lacking quantitative analysis of operational risks and hierarchical protection mechanisms, resulting in safety hazards and reliability issues. These technical limitations severely restrict the operational performance and economic benefits of energy storage systems in complex grid environments, urgently requiring the development of next-generation energy storage system operation control technologies. Summary of the Invention

[0004] This invention provides a method, device, and equipment for adaptive switching control of multi-scenario operation modes in energy storage systems, aiming to solve technical problems such as inaccurate scenario identification, insufficient multi-scenario coordination capabilities, and lack of adaptability in switching decisions. By integrating core technologies such as multi-dimensional data fusion, intelligent scenario identification, conflict coordination optimization, debt risk assessment, and fault-tolerant switching protection, an intelligent energy storage operation control system with environmental perception and adaptive learning capabilities is constructed. This enables collaborative operation and autonomous switching of energy storage systems in multiple scenarios, forming an intelligent operation control solution for energy storage systems in complex power grid environments.

[0005] The first aspect of this invention proposes an adaptive switching control method for multi-scenario operation modes of an energy storage system, comprising the following steps: Collect power grid status data, load demand data, and economic indicator data, and perform multi-dimensional fusion analysis on the power grid status data, load demand data, and economic indicator data to generate a comprehensive status feature vector; Based on the comprehensive state feature vector, scene identification and classification are performed to generate scene type identifiers. Based on the scene type identifiers, demand priority is sorted. Based on the demand priority sorting, spatiotemporal energy transfer analysis is performed to generate spatiotemporal energy distribution schemes. Scene conflict analysis is performed on the spatiotemporal energy distribution scheme to identify potential conflict points. A scene coordination mechanism is constructed based on the potential conflict points, and a scene fusion strategy is generated through the scene coordination mechanism. Based on the scenario fusion strategy, the energy overdraft capacity and compensation time sequence are obtained, and debt risk parameters are generated according to the energy overdraft capacity and the compensation time sequence. A multi-level fault-tolerant switching system is established based on the debt risk parameters and the comprehensive state feature vector. A primary switching scheme, a backup switching scheme, and an emergency switching scheme are generated based on the multi-level fault-tolerant switching system. A causal mapping relationship for handover decisions is established based on the primary handover scheme, the backup handover scheme, and the emergency handover scheme. An adaptive handover instruction sequence is generated by dynamically adjusting the handover decision causal mapping relationship. Based on the adaptive switching instruction sequence, the energy storage system is switched to different operating modes to obtain actual execution data. The actual execution data is then used to reverse-correct the scenario type identifier to obtain the corrected scenario type identifier, thus completing the adaptive switching control of multiple scenario operating modes.

[0006] A second aspect of this invention provides an adaptive switching control device for multi-scenario operation modes of an energy storage system, comprising: The data acquisition module is used to collect power grid status data, load demand data, and economic indicator data, and to perform multi-dimensional fusion analysis on the power grid status data, load demand data, and economic indicator data to generate a comprehensive status feature vector. The scene recognition module is used to identify and classify scenes based on the comprehensive state feature vector to generate scene type identifiers, sort the demand priorities according to the scene type identifiers, and perform spatiotemporal energy transfer analysis based on the demand priority sorting to generate spatiotemporal energy distribution schemes. The conflict coordination module is used to perform scene conflict analysis on the spatiotemporal energy distribution scheme, identify potential conflict points, construct a scene coordination mechanism based on the potential conflict points, and generate a scene fusion strategy through the scene coordination mechanism. The debt assessment module is used to obtain the energy overdraft capacity and compensation timeline based on the scenario fusion strategy, and generate debt risk parameters based on the energy overdraft capacity and the compensation timeline. The fault-tolerant switching module is used to establish a multi-level fault-tolerant switching system based on the debt risk parameters and the comprehensive state feature vector, and to generate a primary switching scheme, a backup switching scheme and an emergency switching scheme based on the multi-level fault-tolerant switching system. The decision adjustment module is used to establish a causal mapping relationship for handover decisions based on the primary handover plan, the backup handover plan, and the emergency handover plan, and to dynamically adjust and generate an adaptive handover instruction sequence based on the causal mapping relationship for handover decisions. The execution feedback module is used to switch the operating mode of the energy storage system based on the adaptive switching instruction sequence to obtain actual execution data, and use the actual execution data to reverse correct the scenario type identifier to obtain the corrected scenario type identifier, thereby completing the adaptive switching control of multiple scenario operating modes.

[0007] A third aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the adaptive switching control method for multi-scenario operation modes of an energy storage system disclosed in the first aspect.

[0008] The beneficial effects of this invention are reflected in the following points: First, by establishing multi-dimensional fusion analysis and intelligent scene recognition technology, it innovatively realizes the deep fusion processing of power grid status, load demand, and economic indicators, and the accurate identification of scene types based on comprehensive state feature vectors. Combined with demand priority ranking and spatiotemporal energy distribution analysis, it fundamentally solves the core technical problems of insufficient scene perception capability and low recognition accuracy of traditional energy storage control methods, significantly improves the environmental adaptability and scene response accuracy of energy storage systems, and enables energy storage systems to quickly and accurately identify operating scenarios and formulate optimal energy time and space scheduling strategies.

[0009] Secondly, through scenario conflict analysis and coordination mechanism construction technology, a complete coordination framework was established, including overlapping area identification, conflict degree assessment, potential conflict point marking, and scenario fusion strategy generation. This solved the key technical bottlenecks of spatiotemporal resource competition, difficulty in conflict identification, and lack of coordination mechanisms in multi-scenario operation. It achieved a technological leap from traditional single-scenario independent operation to multi-scenario collaborative optimization, significantly improved the resource utilization efficiency and operational stability of the energy storage system, and enhanced the coordination and cooperation capabilities of the energy storage system under complex constraints.

[0010] Finally, through quantitative assessment of debt risk parameters and multi-level fault-tolerant switching technology, the system innovatively achieves precise quantification of operational risks based on energy overdraft capacity and compensation timing, hierarchical fault-tolerant protection configuration, and intelligent switching decisions based on causal mapping relationships. Combined with an adaptive command sequence generation and execution data feedback correction mechanism, it completely solves the fundamental problems of weak risk assessment capabilities, single protection mechanisms, and rigid decision-making logic in traditional energy storage control methods. This results in an intelligent energy storage operation control scheme with autonomous learning and dynamic optimization characteristics, which greatly improves the safety, reliability, and environmental adaptability of the energy storage system.

[0011] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0012] The accompanying drawings illustrate specific examples of the technical solutions described in this invention and, together with the detailed embodiments, form part of the specification, serving to explain the technical solutions, principles, and effects of this invention.

[0013] Unless otherwise specified or defined, the same reference numerals in different figures represent the same or similar technical features, and different reference numerals may be used to represent the same or similar technical features.

[0014] Figure 1 This is a flowchart illustrating an adaptive switching control method for multi-scenario operation modes of an energy storage system according to the present invention.

[0015] Figure 2 This is a structural block diagram of an adaptive switching control device for multi-scenario operation modes of an energy storage system according to the present invention.

[0016] Figure 3 This is a schematic diagram of the structure of a computer device according to the present invention. Detailed Implementation

[0017] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0018] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0019] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0020] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0021] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0023] The technical solutions of the embodiments of this application will be described below.

[0024] like Figure 1 As shown, this embodiment of the invention provides an adaptive switching control method for multi-scenario operation modes of an energy storage system, including the following steps S110-S170: Step S110: Collect power grid status data, load demand data, and economic indicator data, and perform multi-dimensional fusion analysis on the power grid status data, load demand data, and economic indicator data to generate a comprehensive status feature vector.

[0025] Specifically, power grid status data acquisition relies on a distributed intelligent monitoring network and is equipped with high-precision power parameter measurement devices to achieve comprehensive perception of the power grid's operating status. The power grid status data acquisition system adopts a three-layer architecture, including a field measurement layer, a communication aggregation layer, and a data processing layer. The field measurement layer deploys intelligent electronic equipment, configuring digital protection and control devices and synchronous phasor measurement units. The sampling frequency is set to 4000Hz, with voltage measurement accuracy reaching 0.2 level, current measurement accuracy reaching 0.2S level, and phase angle measurement accuracy reaching 0.01 degrees. The communication aggregation layer adopts a redundant transmission scheme combining fiber optic Ethernet and power line carrier communication. Data transmission follows the IEC 61850 communication protocol, with a transmission delay of less than 10ms. The data processing layer deploys a real-time database and a state estimation algorithm, using weighted least squares for state estimation, achieving an estimation accuracy better than 99.5% with an update cycle of 5 seconds. Load demand data acquisition is achieved through an intelligent electricity consumption information acquisition system, covering full-user load monitoring for high-voltage large users, dedicated transformer users, and low-voltage centralized meter reading users. High-voltage large users employ dedicated metering devices to monitor their active power, reactive power, and energy consumption in real time, with a data acquisition cycle of 1 minute. Dedicated transformer users collect load data through distribution transformer monitoring terminals, monitoring the real-time load rate, power factor, and three-phase current imbalance of the distribution transformer, with a data upload cycle of 15 minutes. Low-voltage centralized meter reading users remotely read meters using smart meters, recording their time-of-use consumption and maximum demand, with a data acquisition cycle of 1 hour. Economic indicator data collection covers multi-dimensional economic information such as electricity market prices, generation costs, equipment investment, and operation and maintenance costs. Electricity market price data is obtained through the electricity trading platform, including day-ahead time-of-use market prices, real-time market prices, ancillary service prices, and transmission and distribution prices, with price data released and updated in 15-minute increments. Generation cost data includes fuel costs, labor costs, maintenance costs, and environmental costs. Fuel costs are calculated based on fuel prices such as coal and natural gas and the unit's heat consumption rate, while environmental costs include carbon emission costs and pollutant treatment costs. Equipment investment data includes the initial investment, installation costs, and ancillary facility costs of energy storage equipment, which are depreciated and amortized over the equipment's service life. Operation and maintenance cost data includes equipment repair costs, spare parts costs, insurance costs, and management costs, which are statistically analyzed through the equipment management system and financial system. Through the collaborative work of the above data collection systems, grid status data reflecting the real-time operation of the power grid, load demand data reflecting user electricity consumption patterns, and economic indicator data reflecting the economic efficiency of system operation are obtained.

[0026] Multidimensional fusion analysis is performed on power grid status data, load demand data, and economic indicator data. Deep learning and data mining techniques are used to achieve unified processing and feature extraction of heterogeneous data. The multidimensional fusion analysis first preprocesses the three types of data, including data cleaning, outlier detection, missing value imputation, and data standardization, to ensure data quality and consistency. Data cleaning uses statistical analysis methods to identify and remove outlier data points; outlier detection uses the 3σ criterion and box plot method; and missing value imputation uses linear interpolation and regression analysis. Data standardization uses a combination of Z-score standardization and Min-Max standardization to convert data with different dimensions and numerical ranges into a unified standard format. Feature extraction uses a dimensionality reduction method combining principal component analysis and independent component analysis to extract key feature components from the high-dimensional raw data. Power grid status data feature extraction focuses on the safety and stability characteristics of the power grid, including safety indicators such as voltage stability margin, power angle stability margin, frequency stability margin, and transient stability margin. Load demand data feature extraction focuses on the time-varying and predictive characteristics of the load, including dynamic indicators such as load peak-to-valley difference, load change rate, load prediction error, and load uncertainty. The feature extraction of economic indicators focuses on economic benefit and cost-risk characteristics, including economic indicators such as payback period, net present value, internal rate of return, and cost sensitivity. Feature fusion employs a combination of weighted fusion and deep fusion methods. Weighted fusion uses the analytic hierarchy process (AHP) to determine the weight coefficients of each feature dimension, while deep fusion uses neural networks to learn the nonlinear mapping relationships between features. Through multidimensional fusion analysis, the original grid status data, load demand data, and economic indicator data are transformed into a unified feature representation format, generating a comprehensive state feature vector that fully reflects the operating environment of the energy storage system.

[0027] Step S120: Based on the comprehensive state feature vector, scene identification and classification are performed to generate scene type identifiers. Based on the scene type identifiers, the demand priority is sorted. Based on the demand priority sorting, energy spatiotemporal transfer analysis is performed to generate spatiotemporal energy distribution schemes.

[0028] Specifically, scene recognition and classification are performed based on comprehensive state feature vectors. A hybrid model architecture combining convolutional neural networks (CNNs) and recurrent neural networks (RNNs) is adopted to perform multi-level feature learning and pattern recognition on the comprehensive state feature vectors. CNNs are responsible for extracting spatial correlation features from the comprehensive state feature vectors, while RNNs are responsible for capturing temporal dependency features. The two types of features are organically combined through a feature fusion layer. The training dataset contains historical operating scene samples and labeled samples covering typical operating modes for model training. The classification algorithm employs an ensemble learning strategy using multiple machine learning methods such as support vector machines (SVMs), random forests, and gradient boosting decision trees. Weighted voting mechanisms and probabilistic fusion techniques are used to improve classification accuracy and robustness. Scene feature extraction focuses on the combination patterns of key factors such as power grid operating status, load change trends, economic indicators, and environmental constraints. The contribution weights of each dimension of features to scene classification are determined using feature importance evaluation methods such as information gain and Gini coefficient. The model training process employs k-fold cross-validation and grid search hyperparameter optimization techniques. Performance evaluation on the validation and test sets ensures the generalization ability and robustness of the classification model, achieving high-precision classification on the test dataset. Scene recognition results are output in the form of multi-class probability distributions, including the membership probability and confidence level of each scene type. The final scene category is determined by setting probability thresholds and lower confidence limits. For example, when the comprehensive state feature vector shows large fluctuations in grid frequency, drastic load changes, and peak electricity prices, the algorithm identifies it as a grid frequency regulation scenario; when the feature vector shows excess renewable energy output, low load demand, and significant environmental benefits, the algorithm identifies it as a renewable energy consumption scenario. Through scene recognition and classification processing, the abstract comprehensive state feature vector is transformed into a specific operational scenario description, generating a scene type identifier that includes the main scene type code, subtype code, and confidence level.

[0029] Demands are prioritized based on scenario type identifiers. A scenario-demand mapping matrix is ​​established, associating main type codes, subtype codes, and basic demand weight coefficients to construct a multi-level mapping table. Confidence level serves as a weight adjustment factor; when the confidence level is high, the standard weight configuration corresponding to the scenario is adopted; when the confidence level is low, the weight configuration is adjusted towards a more balanced distribution, reducing the magnitude of extreme weight allocation. A pairwise comparison judgment matrix is ​​established by combining expert knowledge and historical data, and the relative importance weight coefficients of each demand element are calculated using the eigenvector method and consistency testing techniques. The weight calculation process comprehensively considers characteristic parameters such as the urgency index, persistence index, and impact range index of the scenario, as well as the influence of the confidence level on the reliability of the weights. Fuzzy membership functions are constructed using triangular fuzzy numbers and trapezoidal fuzzy numbers, where the support set width of the fuzzy numbers is adjusted according to the confidence level; the higher the confidence level, the tighter the boundary of the fuzzy set, and the stronger the weight determinism. Fuzzy comprehensive evaluation transforms qualitative expert judgments and confidence information into quantitative weight values ​​through confidence-weighted fuzzy synthesis operations and defuzzification processing. For example, in peak load regulation scenarios, when the confidence level is high, the weight for economic benefit requirements is precisely set to 0.6, and the weight for safety requirements is 0.3. When the confidence level is medium, the weight allocation converges towards the central value, with the weight for economic benefit requirements adjusted to 0.5 and the weight for safety requirements adjusted to 0.35 to reduce the risk of misidentification. Through the precise mapping of the three components of the scenario type identifier to the requirement weights and the adjustment of the confidence level, a priority ranking of requirements reflecting the characteristics of the current operating scenario and the reliability of identification is formed.

[0030] In some embodiments, the step of generating a spatiotemporal energy distribution scheme by performing energy spatiotemporal transfer analysis based on the demand priority ranking includes: performing time-dimensional decomposition on the demand priority ranking to obtain a temporal demand distribution; performing spatial-dimensional mapping on the temporal demand distribution to generate a spatial energy demand matrix; analyzing the spatial energy demand matrix to obtain an optimal transmission path; and constructing a spatiotemporal energy distribution scheme based on the optimal transmission path.

[0031] First, the demand priority ranking is decomposed along a time dimension to obtain the time-series demand distribution. Time series analysis and demand decomposition techniques classify and map the weighted results of the demand priority ranking according to time characteristics, establishing a dynamic model of demand intensity changing over time. The decomposition process comprehensively considers time attribute parameters such as demand response time requirements, duration constraints, periodicity, and urgency levels, establishing a multi-dimensional demand time characteristic database. A piecewise linear function is used to describe the time trend of demand intensity, and an exponential decay function is used to describe the time decay characteristics of demand urgency. A composite time distribution model is established through function superposition and parameter fitting. The continuous time axis is discretized into time segments of equal or variable length, each time segment corresponding to a specific demand combination and intensity level. The demand allocation ratio and scheduling strategy for different time windows are determined through weighted allocation algorithms and constraint optimization methods. The time-series demand distribution is represented in vector form, with the vector dimension corresponding to the number of time segments, and the vector elements representing the demand intensity and type identifier of that time segment. For example, when safety requirements have a high weight, rapid response requirements account for 70% of the total demand, allocated to a 1-second response period, while maintenance requirements account for 20%, allocated to a 1-5 second period. When economic requirements have a high weight, peak-price discharge requirements account for 60%, off-peak-price charging requirements account for 30%, and flat-price standby requirements account for 10%. Through time-dimensional demand decomposition and optimization, a time-series demand distribution describing the distribution patterns and scheduling strategies of each priority demand on the time axis is obtained.

[0032] Next, a spatial dimension mapping is performed based on the temporal demand distribution to generate a spatial energy demand matrix. Power grid modeling and spatial analysis techniques realize the information conversion from temporal demand distribution to spatial dimensions, transforming time-dimensional information into spatial energy allocation requirements, comprehensively considering spatial constraints such as power grid topology, node electrical characteristics, and geographical distribution. A spatiotemporal mapping algorithm is established, using a power flow calculation engine and load flow analysis module, combined with node impedance matrices and network connection matrices, to determine a spatial power allocation scheme that meets the requirements of the temporal demand distribution. The mapping process divides the power grid into hierarchical partitions according to voltage levels and geographical regions, establishing a hierarchical spatial node system. Each spatial unit corresponds to specific load density indicators, generation capacity configurations, and energy storage demand levels, while considering network characteristic parameters such as electrical distance, transmission impedance, connection relationships, and capacity constraints of each node. The spatial energy demand matrix is ​​represented by a two-dimensional sparse matrix data structure. The row dimension represents the discrete time segment index, the column dimension represents the power grid spatial node number, and the matrix elements store information such as the energy demand value, priority identifier, and constraints of a specific node at a specific time. The matrix construction process employs iterative optimization algorithms and constraint satisfaction techniques to ensure that the spatial energy distribution satisfies both the temporal constraints of the time-series demand distribution and the spatial physical constraints of the power grid operation. For example, when the time-series demand distribution shows that rapid response demand dominates, the spatial mapping algorithm prioritizes high-voltage nodes with short electrical distances and low transmission delays, allocating 70% of the demand to 500kV main grid nodes. When economic optimization demand dominates, the spatial mapping algorithm prioritizes user-side nodes with high load density and strong price sensitivity, allocating 60% of the demand to industrial park nodes. Through spatial dimension transformation and optimized mapping of the time-series demand distribution, a spatial energy demand matrix reflecting the spatial configuration requirements and constraints of the energy storage system is generated.

[0033] Then, the spatial energy demand matrix is ​​analyzed to obtain the optimal transmission path. Graph theory modeling and network optimization algorithms abstract the power grid topology into a weighted directed graph mathematical model, with substations, power plants, and important load nodes as the vertex set and transmission lines, distribution lines, and tie lines as the edge set. The edge weight parameters comprehensively represent multiple attributes such as transmission impedance, thermal stability limit, capacity constraint, and transmission cost. Based on the spatial energy demand matrix, a multi-objective optimization problem is established, and Pareto optimality theory and a non-dominated sorting genetic algorithm are used to solve for the optimal transmission path, simultaneously optimizing multiple objective functions such as shortest transmission distance, minimum transmission loss, balanced line load rate, and lowest transmission economic cost. The path optimization algorithm integrates improved versions and combined applications of classic graph theory algorithms such as Dijkstra's shortest path algorithm, Floyd-Warshall full-source shortest path algorithm, and Ford-Fulkerson maximum flow minimum cut algorithm. The algorithm fusion and parameter adaptation techniques improve the solution efficiency and solution quality. The path analysis process rigorously considers grid safety constraints such as line thermal stability limitations, static voltage stability constraints, transient power angle stability constraints, and the N-1 safety criterion. Constraint processing techniques and penalty function methods ensure that the selected transmission paths meet grid safety and reliability requirements under both normal operation and anticipated fault conditions. The transmission paths also possess dynamic reconfiguration and fault self-healing capabilities, establishing a redundant configuration system of primary transmission paths and multi-level backup transmission paths. Path switching logic and load transfer algorithms address line faults, equipment maintenance, and network topology changes. For example, when the spatial matrix shows that energy storage demand is mainly concentrated at high-voltage main grid nodes, the transmission path analysis algorithm prioritizes the 500kV UHV transmission channel as the primary transmission path, configuring 220kV and 110kV networks as hierarchical backup paths. When energy storage demand is mainly distributed at user-side nodes, the algorithm prioritizes the 110kV and 35kV distribution networks as primary transmission paths, utilizing local balancing and short-distance transmission strategies. Through in-depth analysis of the spatial energy demand matrix and multi-objective path optimization, the optimal transmission path that meets spatial power configuration requirements and safety constraints is obtained.

[0034] Finally, a spatiotemporal energy distribution scheme is constructed based on the optimal transmission path. A spatiotemporal collaborative optimization method is adopted, using the optimal transmission path as the backbone framework. Combining temporal demand distribution and spatial energy demand matrix, a three-dimensional optimization model of time, space, and energy is established to comprehensively consider the charging and discharging timing of the energy storage system, power spatial allocation, and the coordination of energy transmission paths. The scheme construction process is guided by the optimal transmission path, integrating the system with temporal demand distribution and spatial energy demand matrix to determine the operating strategy and control parameters of the energy storage system. Time-dimensional optimization determines the charging and discharging plan of the energy storage system based on the capacity characteristics of the transmission path; spatial-dimensional optimization determines the distribution ratio of energy storage power among nodes based on the node connections of the transmission path; and transmission path optimization determines the specific path selection and switching logic for energy transmission. The scheme also includes dynamic adjustment and real-time optimization mechanisms, allowing online correction of the distribution scheme based on changes in the state of the transmission path. For example, when the optimal transmission path is a single main path, the spatiotemporal energy distribution scheme adopts a centralized configuration, concentrating energy storage power at key nodes of the main path; when the optimal transmission path is a multi-branch parallel path, the scheme adopts a distributed configuration, distributing energy storage power to each branch node according to the path capacity ratio. By integrating and coordinating the optimal transmission path, a spatiotemporal energy distribution scheme adapted to the current operating scenario is constructed.

[0035] Step S130: Perform scene conflict analysis on the spatiotemporal energy distribution scheme to identify potential conflict points, construct a scene coordination mechanism based on the potential conflict points, and generate a scene fusion strategy through the scene coordination mechanism.

[0036] Specifically, scenario conflict analysis is conducted on spatiotemporal energy distribution schemes to identify potential conflict points. Multi-dimensional analysis techniques are used to detect contradictions and conflicts between spatiotemporal energy distribution schemes under different operating scenarios, focusing on the cross-influence and competition of multiple dimensions such as time conflicts, spatial conflicts, and resource conflicts. This allows for the early detection of potential contradictions in multi-scenario operation, avoiding reduced operating efficiency of energy storage devices and grid stability risks caused by conflicts.

[0037] In some embodiments, the step of performing scene conflict analysis on the spatiotemporal energy distribution scheme to identify potential conflict points includes: identifying overlapping regions of multiple scenes in the spatiotemporal energy distribution scheme to obtain intersecting regions; analyzing the intersecting regions to generate a conflict degree assessment result; determining the conflict level based on the conflict degree assessment result; and marking potential conflict points according to the conflict level.

[0038] First, overlapping regions across multiple scenarios are identified to obtain intersection areas for the spatiotemporal energy distribution scheme. Through time axis projection and interval overlap calculation, the intersection periods of charging and discharging time sequences in different scenarios are determined, and a time overlap matrix is ​​established to record the temporal intersection relationships between scenarios. Spatial dimension overlap identification uses spatial node mapping and power demand superposition to determine common nodes and regions for power allocation in different scenarios, and a spatial overlap matrix is ​​established to record the spatial intersection relationships between scenarios. Resource dimension overlap identification uses capacity demand accumulation and power demand superposition to determine the common demand and competition relationships for energy storage resources in different scenarios, and a resource overlap matrix is ​​established to record the resource intersection relationships between scenarios. The overlapping region identification algorithm employs set theory and graph theory methods, abstracting the spatiotemporal energy distribution scheme into a set representation of spatiotemporal resource occupancy. Multi-scenario overlapping regions are identified through set intersection operations and graph overlap analysis. The identification process considers the boundary definition and fuzzy handling of overlapping regions. Fuzzy set theory is used to handle overlapping regions with ambiguous boundaries, and the continuity characteristics of the degree of overlap are quantified through membership functions. For example, during weekday evening peak hours, both peak load regulation and grid support scenarios require energy storage devices to output power at major load centers. The overlap identification algorithm determines the time overlap area to be 18:00-22:00, the spatial overlap area to be multiple 110kV substation nodes in the city center, and the resource overlap demand to be more than 80% of the total energy storage capacity. Through multi-dimensional overlap analysis and intersection calculation, the overlap area reflecting the cross-impact range of different scenario solutions is finally obtained.

[0039] Next, the cross-regional analysis generates conflict severity assessment results. Based on the specific characteristics and scope information of the cross-regional areas, the conflict severity assessment algorithm quantitatively analyzes the intensity of resource competition and scheme compatibility within the overlapping areas. The assessment process establishes a multi-index evaluation system, including assessment dimensions such as temporal conflict intensity, spatial conflict density, resource conflict severity, and comprehensive conflict impact. Temporal conflict intensity is quantified by indicators such as the ratio of overlapping time length to total time, time window overlap, and degree of temporal constraint violation, reflecting the severity of temporal conflict. Spatial conflict density is quantified by indicators such as the ratio of overlapping nodes to total nodes, power demand superposition multiple, and degree of spatial constraint violation, reflecting the concentration of spatial conflict. Resource conflict severity is quantified by indicators such as the excess capacity demand ratio, power demand overrun multiple, and degree of resource constraint violation, reflecting the intensity of resource conflict. Comprehensive conflict impact is assessed by using a weighted comprehensive evaluation method to fuse multi-dimensional conflict indicators, forming a comprehensive assessment index reflecting the overall severity of conflict. The assessment algorithm uses the analytic hierarchy process (AHP) to determine the weight coefficients of each assessment indicator, and the weight parameters are determined through expert experience and historical data training. The conflict severity quantification employs a combination of continuous scoring and discrete grading. Continuous scoring reflects the precise numerical value of the conflict severity, while discrete grading facilitates conflict classification and the selection of handling strategies. For example, when the total power demand of multiple scenarios at the same node in an overlapping area exceeds 150% of the node's capacity limit during the same time period, the conflict severity assessment algorithm calculates a resource conflict severity index of 0.8, a spatial conflict density index of 0.7, and a comprehensive conflict impact index of 0.75, indicating the presence of high-intensity conflict in this overlapping area. Through in-depth analysis of the overlapping area and multi-index assessment, a conflict severity assessment result is generated that quantitatively describes the severity and scope of the conflict.

[0040] Then, the conflict level is determined based on the conflict severity assessment results. The level classification uses a combination of threshold segmentation and cluster analysis, dividing the assessment results into different levels such as minor conflict, moderate conflict, severe conflict, and extremely severe conflict by setting conflict severity thresholds. Minor conflict corresponds to a low conflict severity assessment value, indicating mild competition between scenarios that can be resolved through simple adjustments; moderate conflict corresponds to a moderate assessment value, indicating significant competition between scenarios requiring coordination; severe conflict corresponds to a high assessment value, indicating intense competition between scenarios requiring focused attention and priority; extremely severe conflict corresponds to the highest assessment value, indicating fundamental contradictions between scenarios requiring redesign and in-depth coordination. The level determination process considers the multi-dimensional characteristics and comprehensive impact of conflict, employing a multi-criteria decision analysis method to comprehensively consider the conflict severity across time, space, resources, and other dimensions, determining the final conflict level through weighted assessment and comprehensive judgment. The level classification also considers the dynamic and time-varying characteristics of conflict, establishing a dynamic level adjustment mechanism to update and adjust the conflict level in real time according to changes in operating conditions and conflict evolution trends. The conflict severity assessment results are represented using a standardized coding method, including a severity code, a numerical value for severity, a description of the scope of impact, and a processing priority. For example, if the conflict severity assessment results show an overall conflict impact index of 0.75 and a resource conflict severity index of 0.8, the grading algorithm determines the conflict level as severe conflict, with a severity code of "L3" and a processing priority of "high priority," requiring the initiation of a focused coordination strategy. Through the level conversion and classification management of the assessment results, the conflict level reflecting the severity and urgency of the conflict is determined.

[0041] Finally, potential conflict points are marked according to their conflict levels. Based on the conflict level, the conflict point marking algorithm accurately locates and identifies specific conflict positions in the spatiotemporal energy distribution scheme, establishing a complete description of the spatial coordinates, temporal coordinates, and attribute information of the conflict points. The marking process maps conflict level information with the specific location information of the intersection area to determine the precise location, conflict type, severity, and impact range of each potential conflict point. Conflict point marking adopts a multi-level marking system, including coarse-grained area marking and fine-grained node marking. Coarse-grained marking is used for macro-level conflict distribution analysis, while fine-grained marking is used for precise conflict location and handling. The marked information is stored in a structured data format, containing fields such as conflict point number, spatiotemporal coordinates, conflict level, conflict type, impact assessment, and handling status. The marking algorithm also establishes a correlation analysis and cluster detection function for conflict points, identifying the correlation between adjacent conflict points and conflict clustering areas, providing spatial clustering information for subsequent coordination strategy design. Conflict point visualization uses graphical marking and color coding technology, using different colors and symbols to mark conflict points of different levels on the spatiotemporal energy distribution map, facilitating intuitive analysis of conflict distribution and management decisions. The marking results also support dynamic updates and historical tracking, recording the evolution and handling process of conflict points, providing data support for conflict prediction and prevention. For example, when the conflict level is severe, the marking algorithm marks a red warning symbol at the corresponding spatiotemporal coordinates, with the conflict point number "CP-001", the time coordinate "19:30-20:30", the spatial coordinate "220kV-Station-A", the conflict type "time-resource conflict", and the impact assessment "high impact". Based on the precise location and structured marking of the conflict level, potential conflict points that can accurately indicate the location and characteristics of the conflict are obtained.

[0042] A scenario coordination mechanism is constructed based on potential conflict points. Differentiated strategies are formulated according to the specific attribute information of potential conflict points, including key parameters such as conflict point number, spatiotemporal coordinates, conflict level, conflict type, and impact assessment. For conflict points marked as "time-resource conflict" and at level "L3," the coordination mechanism designs a combination of time slicing and resource reservation strategies to finely divide conflict periods and allocate dedicated time windows and capacity quotas for different scenarios. For conflict points marked as "space-power conflict" and located at specific substation nodes, the coordination mechanism designs spatial diversion and node switching strategies to redistribute power demand to adjacent or backup nodes, alleviating pressure on single points. The selection of coordination strategies and parameter settings directly correspond to the conflict point's level code and impact assessment results. High-level conflict points employ mandatory coordination measures, while medium- and low-level conflict points employ flexible coordination measures. The mechanism also establishes inter-conflict coordination processing; when multiple adjacent conflict points form a cluster, a regional overall coordination strategy is adopted to avoid overall conflict transfer caused by local optimization. For example, for a cluster of consecutive conflict points numbered "CP-001" to "CP-005", a regional spatiotemporal resource reallocation scheme is designed to uniformly adjust the runtime sequence and power configuration of each scenario within the region. Through conflict characteristic analysis and multi-strategy coordination design, a scenario coordination mechanism that can effectively resolve potential conflicts is finally constructed.

[0043] A scenario fusion strategy is generated through a scenario coordination mechanism. The handling schemes for each conflict point in the coordination mechanism are extracted, including time adjustment schemes, spatial reallocation schemes, and resource allocation schemes, forming coordination constraints and optimization objectives between scenarios. The time fusion strategy establishes a unified time-series scheduling table for multiple scenarios based on the time-slicing results of the coordination mechanism, staggering originally conflicting time periods according to the allocation scheme of the coordination mechanism. The spatial fusion strategy establishes a unified spatial configuration map for multiple scenarios based on the spatial distribution results of the coordination mechanism, distributing the load of originally competing nodes according to the reallocation scheme of the coordination mechanism. The resource fusion strategy establishes a unified resource management table for multiple scenarios based on the capacity allocation results of the coordination mechanism, dynamically allocating originally contested energy storage resources according to the reservation scheme of the coordination mechanism. The fusion strategy also integrates the priority scheduling logic and compensation strategy design from the coordination mechanism to ensure that multi-scenario switching can be executed according to the predetermined coordination scheme in actual operation. Strategy verification verifies the effectiveness and feasibility of the fusion strategy in eliminating the original conflict points by simulating conflict scenarios handled by the coordination mechanism. For example, when the coordination mechanism adjusts the peak load regulation scenario to the 17:00-19:00 time period and the power grid support scenario to the 19:00-21:00 time period, the fusion strategy establishes a continuous scheduling scheme for 17:00-21:00 accordingly, ensuring seamless time connection between the two scenarios while avoiding resource conflicts. Through the comprehensive application and multi-level fusion optimization of the coordination mechanism, a scenario fusion strategy capable of coordinating the collaborative operation of multiple scenarios is obtained.

[0044] Step S140: Obtain the energy overdraft capacity and compensation time series based on the scenario fusion strategy, and generate debt risk parameters based on the energy overdraft capacity and compensation time series.

[0045] Specifically, while the scenario fusion strategy resolves conflicts between scenarios by coordinating the spatiotemporal energy distribution of multiple operating scenarios, short-term energy supply and demand imbalances may still occur in actual operation, requiring energy storage devices to bear a certain risk of energy overdraft. To quantify this overdraft risk, this invention introduces a debt risk parameter as a quantitative indicator of the energy overdraft risk of the energy storage system. Precise calculation of overdraft capacity and compensation timing can quantify the risk tolerance and compensation arrangements of the energy storage device during the execution of the fusion strategy, providing a safety boundary and risk control basis for multi-scenario collaborative operation.

[0046] In some embodiments, obtaining the energy overdraft capacity and compensation timing based on the scenario fusion strategy includes: analyzing the scenario fusion strategy to obtain the energy supply-demand difference; establishing a debt capacity assessment mechanism based on the energy supply-demand difference; obtaining the energy overdraft capacity using the debt capacity assessment mechanism; and determining the compensation timing based on the energy overdraft capacity.

[0047] First, the scenario fusion strategy is analyzed to obtain the energy supply and demand difference. The energy supply and demand analysis algorithm performs detailed calculations and comparative analyses of the energy demand and supply capacity of each operating scenario within the scenario fusion strategy, identifying time periods and regions where energy supply and demand are unbalanced. The analysis process establishes an energy supply and demand balance equation, taking the charging capacity of the energy storage device as the energy supply side and the discharging demand of each scenario as the energy demand side. Energy surplus or deficit for each time period is determined through supply and demand comparison calculations. The supply and demand difference calculation considers technical limitations such as capacity constraints, power constraints, and efficiency losses of the energy storage device, as well as external constraints such as grid dispatch requirements, market price signals, and changes in user demand. The calculation process employs time-series analysis and rolling calculation methods, dynamically calculating the supply and demand difference with a 15-minute time granularity, forming an energy supply and demand difference sequence covering the next 24-72 hours. The difference sequence is represented by positive and negative values; positive values ​​indicate energy surplus, and negative values ​​indicate energy deficit, with the magnitude reflecting the degree of surplus or deficit. The analysis algorithm also establishes an uncertainty assessment of the supply-demand gap. Through Monte Carlo simulation and probabilistic analysis, it quantifies the impact of load forecasting errors, fluctuations in renewable energy output, and equipment failure risks on the supply-demand gap. For example, in the analysis of a certain workday, the energy surplus is shown to be 200 MWh from 10:00 AM to 12:00 PM, the energy deficit is shown to be 150 MWh from 6:00 PM to 8:00 PM, and the energy surplus is shown to be 300 MWh from 10:00 PM to 6:00 AM the next day. After comprehensive analysis and calculation using a scenario fusion strategy, the final energy supply-demand gap reflecting the energy supply-demand balance during the operation of the energy storage device is obtained.

[0048] Next, a debt capacity assessment mechanism is established based on the energy supply-demand gap. First, a characteristic analysis of the energy supply-demand gap is conducted, including key characteristic parameters such as the distribution density of gap periods, the statistical characteristics of gap values, the variation pattern of gap duration, and the distribution pattern of gap intervals. When the supply-demand gap shows a continuous large gap pattern, the mechanism design adopts a large-capacity long-term debt management configuration, setting the maximum debt capacity to 80% of the installed capacity, extending the longest debt term to 36 hours, and setting five debt tiers. When the supply-demand gap shows an intermittent small gap pattern, the mechanism design adopts a small-capacity short-term debt management configuration, limiting the maximum debt capacity to 60% of the installed capacity, shortening the longest debt term to 12 hours, and setting three debt tiers. The mechanism's constraint parameters are adaptively adjusted according to the extreme value characteristics of the supply-demand gap. A strict constraint mode is activated when the maximum difference exceeds 50% of the installed capacity, and a relaxed constraint mode is used when the maximum difference is less than 30% of the installed capacity. The mechanism's risk assessment module sets different risk coefficients based on the volatility characteristics of the supply-demand gap, increasing the risk weight when the difference fluctuates sharply and decreasing the risk weight when the difference is relatively stable. The mechanism also designs debt maturity management strategies based on the time concentration characteristics of the supply-demand gap. When the gap is concentrated in a short period, a rapid debt repayment strategy is adopted; when the gap is dispersed over a long period, a gradual debt management strategy is adopted. For example, for the aforementioned 50MWh concentrated gap in terms of supply and demand, the mechanism establishes a medium-capacity short-term debt allocation, sets the maximum debt capacity at 70% of the installed capacity, the maximum debt term at 18 hours, establishes a three-tier debt management system, sets the risk coefficient at 0.6, and adopts a rapid repayment model. By utilizing the characteristics of the energy supply-demand gap to guide and optimize allocation, a targeted debt capacity assessment mechanism is generated.

[0049] Then, the overdraft capacity is obtained using a debt capacity assessment mechanism. First, the mechanism's debt status analysis module is invoked to assess the current debt balance, debt structure, and repayment capacity, establishing initial conditions for overdraft capacity calculation. The mechanism's capacity optimization module employs a multi-constraint optimization algorithm, comprehensively considering multiple constraints such as debt balance constraints, debt maturity constraints, safety reserve constraints, and repayment capacity constraints, to determine the optimal overdraft capacity configuration through constraint optimization. The calculation process integrates the mechanism's risk control module, using scenario analysis and stress testing methods to assess the safety boundary of the overdraft capacity under normal, adverse, and extreme scenarios, ensuring the controllability of overdraft decisions under various operating conditions. The mechanism's dynamic adjustment module updates and revises the overdraft capacity online based on real-time changes in debt status and external conditions, maintaining the timeliness and accuracy of overdraft authorization. The overdraft capacity acquisition also invokes the mechanism's hierarchical management module, setting different levels of overdraft capacity based on debt risk level and urgency, establishing a hierarchical authorization system for regular overdraft, priority overdraft, and emergency overdraft. For example, with a debt balance of 30% of installed capacity and an expected repayment time of 12 hours, the debt capacity assessment mechanism calculates that the current available overdraft capacity is 40MWh, the sustainable overdraft time is 2 hours, the safety margin is 80%, and the overdraft level is conventional overdraft. Through comprehensive assessment and optimization by the debt capacity assessment mechanism, a reliable energy overdraft capacity is determined.

[0050] Finally, the compensation sequence is determined based on the energy overdraft capacity. A multi-objective time-series optimization model is established, simultaneously considering multiple optimization objectives such as compensation timeliness, compensation economy, compensation smoothness, and compensation reliability. A Pareto-optimal compensation sequence scheme is solved using a multi-objective evolutionary algorithm. The compensation sequence design adopts a hierarchical planning strategy. First, the overall time frame and stage division of compensation are determined. Then, the specific time period arrangements and power configurations for each stage are refined. Finally, the time coordination and power smoothing of the compensation process are optimized. The timing determination considers the urgency and priority ranking of compensation. The priority order of compensation is determined according to the risk level and expiration date of the overdraft capacity, prioritizing compensation periods for high-risk overdrafts and those nearing expiration. The compensation sequence also considers the technical constraints of energy storage devices and grid dispatch constraints to ensure that the compensation scheme is technically feasible and dispatch-acceptable. An emergency adjustment mechanism is established in the time-series optimization process. When prediction deviations or emergencies occur, the compensation sequence can be quickly replanned to maintain the flexibility and adaptability of the compensation plan. The robustness verification of the compensation sequence is conducted through uncertainty analysis and sensitivity testing to evaluate the stability and reliability of the time-series scheme under parameter disturbances and environmental changes. For example, for a 40MWh overdraft capacity, the compensation sequence is divided into two phases: the first phase, from 21:30 to 22:30, involves 25MWh of compensation, taking advantage of the charging opportunity during the nighttime load reduction; the second phase, from 09:30 to 10:30 the following day, involves 15MWh of compensation, utilizing the surplus period of new energy power generation growth. This phased compensation approach ensures timely debt repayment while optimizing the economics and feasibility of the compensation. Through coordinated planning and consideration of overdraft capacity limitations and multi-objective planning, a feasible compensation sequence is obtained.

[0051] In some embodiments, generating debt risk parameters based on the energy overdraft capacity and the compensation time series includes: assessing the risk exposure of the energy overdraft capacity to obtain an overdraft risk value; performing time risk analysis based on the compensation time series to generate a time series risk coefficient; establishing a risk weight allocation mechanism based on the overdraft risk value and the time series risk coefficient; and using the risk weight allocation mechanism to perform weighted fusion to generate debt risk parameters.

[0052] First, a risk exposure assessment is conducted on the overdraft capacity to obtain the overdraft risk value. A risk exposure calculation model is established, employing risk measurement methods such as Value at Risk (VaR) and Conditional Value at Risk (CVaR) to calculate the potential loss scale and probability of occurrence at different confidence levels. The overdraft risk assessment comprehensively considers the combined impact of multiple risk factors, including market risk, technological risk, operational risk, and policy risk. Market risk assessment considers the impact of electricity price fluctuations and ancillary service price changes on overdraft costs; technological risk assessment considers the impact of equipment failures and capacity decay on overdraft capacity; operational risk assessment considers the impact of dispatch changes and load fluctuations on overdraft demand; and policy risk assessment considers the impact of regulatory policies and market rule changes on overdraft compliance. Risk quantification employs a combination of historical simulation and Monte Carlo simulation, using numerous random scenarios to statistically analyze the loss distribution characteristics of overdraft capacity under various risk scenarios. The assessment algorithm also establishes a dynamic monitoring mechanism for risk exposure, continuously updating the risk exposure assessment results based on real-time changes in overdraft capacity and the evolution of the external risk environment. The risk exposure assessment results are expressed using a standardized risk value, ranging from 0 to 1. A higher value indicates a higher risk exposure and a more severe potential loss from overdraft behavior. For example, when the overdraft capacity is 45 MWh, accounting for 45% of the installed capacity, the overdraft risk value calculated through the risk exposure assessment at a 95% confidence level is 0.25, indicating a 5% probability of suffering a loss equivalent to 25% of the overdraft value. Based on in-depth analysis and multi-dimensional assessment of the risk exposure, an accurate overdraft risk value is obtained.

[0053] Next, a time risk analysis is conducted based on the compensation timeline to generate a time risk coefficient. A time risk assessment model is established to evaluate the time risk level from dimensions such as compensation time adequacy, timeline distribution uniformity, and execution certainty. Compensation time adequacy is assessed by calculating the ratio of overdraft due time to available compensation time, evaluating the time pressure and urgency of the compensation timeline; a smaller ratio indicates greater time pressure and higher risk. Timeline distribution uniformity is assessed by analyzing the distribution density and interval rationality of compensation periods, evaluating the smoothness and executability of the timeline arrangement; a more uneven distribution indicates more concentrated compensation pressure and greater risk. Execution certainty is assessed by analyzing the prediction accuracy and scheduling stability of compensation periods, evaluating the reliability and anti-interference capability of the timeline execution; higher uncertainty indicates greater execution risk. Time risk quantification employs a combination of fuzzy comprehensive evaluation and hierarchical analysis, transforming qualitative time characteristic assessments into quantitative risk coefficients. The risk coefficient calculation also considers the flexibility and emergency adjustment capability of the compensation timeline; timeline schemes with strong adaptability correspond to lower time risk coefficients. The analysis algorithm establishes a sensitivity analysis function for time risk, assesses the impact of changes in key time parameters on time-series risk, and identifies key risk control points in the time-series scheme. For example, the analysis results of a certain compensation time-series scheme show that the compensation time adequacy index is 0.75, the time-series distribution uniformity index is 0.85, and the execution certainty index is 0.70. Through comprehensive evaluation, the time-series risk coefficient is calculated to be 0.30, indicating that this time-series scheme has a moderately low level of time risk. A reliable time-series risk coefficient is formed through comprehensive analysis and risk quantification of the compensation time series.

[0054] Then, a risk weight allocation mechanism is established based on the overdraft risk value and the time-series risk coefficient. A weight determination method combining expert judgment and data-driven approaches is adopted. The basic weight coefficients for overdraft risk and time-series risk are determined through expert experience surveys and historical data statistical analysis. Weight allocation considers the scenario dependence and dynamic change characteristics of risks, employing differentiated weight configuration strategies under different operating scenarios and risk environments. When the overdraft capacity is relatively large, the weight of overdraft risk is increased accordingly, focusing on the impact of overdraft scale on risk level; when the compensation timeline is relatively tight, the weight of time-series risk is increased accordingly, focusing on the impact of time constraints on execution risk. The mechanism design establishes a risk correlation analysis function, assessing the degree of mutual influence and coupling relationship between overdraft risk and time-series risk through correlation coefficient calculation and independence tests, considering the synergistic amplification effect and mutual offsetting effect between risks in weight allocation. An adaptive optimization algorithm is used for weight allocation, dynamically adjusting weight parameters based on the accuracy and predictive effect of historical risk assessments to continuously improve the accuracy and reliability of comprehensive risk assessment. The mechanism also establishes a stability testing function for weight allocation, verifying the stability and applicability of the weight allocation scheme under different parameter perturbations through sensitivity analysis and robustness testing. For example, under normal operating conditions, the overdraft risk weight is set to 0.65, and the time-series risk weight is set to 0.35; under emergency dispatch conditions, the overdraft risk weight is adjusted to 0.45, and the time-series risk weight is adjusted to 0.55. This scenario-based weight adjustment better reflects the risk focus under different conditions. Through the fusion analysis and mechanism construction of overdraft risk value and time-series risk coefficient, an applicable risk weight allocation mechanism is generated.

[0055] Finally, a risk weighting mechanism is used for weighted fusion to generate debt risk parameters. Using the weighting coefficients determined by the weighting mechanism, a weighted average is calculated on the overdraft risk value and the time-series risk coefficient to form a quantitative parameter comprehensively reflecting the debt risk level of the energy storage device. The fusion calculation employs a hybrid method combining linear and nonlinear weighting. Linear weighting is used when the two types of risks are relatively independent, while nonlinear weighting is used when there is a significant interaction between risks. The parameter generation process integrates the corrective effects of risk amplification and risk mitigation factors. The risk amplification factor considers the risk superposition and amplification effects under extreme conditions, while the risk mitigation factor considers the mitigation effects of risk control measures and protection mechanisms. The debt risk parameter is represented by continuous numerical values ​​ranging from 0 to 1. A higher value indicates a higher debt risk, requiring stricter risk control and regulatory measures; a lower value indicates a lower debt risk, allowing the energy storage device more operational flexibility and overdraft capacity. The parameter generation includes a confidence interval calculation function. Through uncertainty propagation analysis and probabilistic reasoning methods, the confidence interval and credibility level of the debt risk parameter are provided, offering uncertainty information for risk decision-making. The risk parameters are updated in real time and dynamically corrected. Based on changes in overdraft capacity and compensation timing, as well as changes in the external risk environment and market conditions, the debt risk parameters are adjusted and updated online and on a rolling basis. For example, when the overdraft risk value is 0.25, the timing risk coefficient is 0.30, and the corresponding weights are 0.65 and 0.35 respectively, the weighted fusion calculation yields a debt risk parameter of 0.27, indicating that the current debt status is at a low to medium risk level, and the energy storage device can undertake the corresponding overdraft tasks under appropriate risk control. Reasonable debt risk parameters are generated through system fusion and parameter optimization via the risk weight allocation mechanism.

[0056] Step S150: Establish a multi-level fault-tolerant switching system based on debt risk parameters and comprehensive status feature vectors, and generate a primary switching scheme, a backup switching scheme, and an emergency switching scheme based on the multi-level fault-tolerant switching system.

[0057] Specifically, debt risk parameters reflect the risk tolerance level and safety boundary of energy storage devices, while the comprehensive state feature vector reflects the complexity and dynamic characteristics of the power grid operating environment. The organic combination of these two provides a scientific basis for establishing a hierarchical fault-tolerant switching system. This system can automatically select appropriate switching protection strategies based on different risk levels and operating states, significantly improving the reliability and safety of energy storage devices in complex operating environments.

[0058] In some embodiments, establishing a multi-level fault-tolerant switching system based on the debt risk parameters and the comprehensive state feature vector includes: dividing risk levels based on the debt risk parameters to generate risk stratification results; constructing corresponding switching protection levels according to the risk stratification results and the comprehensive state feature vector; analyzing the switching protection levels to generate fault-tolerant mechanism configurations; and establishing a multi-level fault-tolerant switching system based on the fault-tolerant mechanism configurations.

[0059] First, risk levels are classified based on debt risk parameters, generating risk stratification results. A combination of threshold segmentation and cluster analysis is used to classify debt risk parameters into five basic levels: low risk, low-to-medium risk, medium risk, medium-to-high risk, and high risk, by setting risk thresholds. Low risk corresponds to a debt risk parameter range of 0-0.2, indicating relatively low operational risk for energy storage devices, allowing for conventional protection strategies; low-to-medium risk corresponds to 0.2-0.4, indicating some risk requiring moderate attention; medium risk corresponds to 0.4-0.6, indicating moderate risk requiring close monitoring; medium-to-high risk corresponds to 0.6-0.8, indicating higher risk requiring enhanced protection measures; and high risk corresponds to 0.8-1.0, indicating severe risk requiring the activation of emergency protection mechanisms. The risk level classification also considers the dynamic characteristics and trend analysis of risk changes, dynamically adjusting the level classification based on risk change rates and risk development trends. The classification results are represented using a structured coding method, including level codes, risk values, safety margins, and recommended measures, providing hierarchical guidance for subsequent protection level design. For example, when the debt risk parameter is 0.35, the risk level is determined to be low to medium risk, with a level code of "R2", a safety margin of 65%, and the recommended measure is "moderate protection". Through scientific analysis and level conversion of the debt risk parameters, the risk stratification results are generated.

[0060] Next, based on the risk stratification results and the comprehensive state feature vector, a corresponding switching protection level is constructed. Data fusion employs a layered fusion mechanism. First, a basic risk level is determined based on debt risk parameters as the dominant factor for the protection level. Then, key feature components in the comprehensive state feature vector are analyzed, including dimensions such as grid security characteristics, load dynamic characteristics, economic benefit characteristics, and environmental constraint characteristics. The basic risk level is then refined and adjusted through feature weight analysis and state assessment. Specifically, when the comprehensive state feature vector indicates a deterioration in grid operation, the current risk level is increased by one level; when the feature vector indicates stable system operation and sufficient safety margin, the current risk level is decreased by one level. However, the adjusted risk level does not exceed ±1 of the original level, ensuring the rationality and stability of the adjustment. The protection hierarchy design adopts a hierarchical structure, establishing four main levels: basic protection layer, standard protection layer, enhanced protection layer, and emergency protection layer. The mapping relationship to the five risk levels is as follows: The basic protection layer corresponds to low risk level (0-0.2), employing a minimum protection strategy to prioritize basic safety functions; the standard protection layer corresponds to low-to-medium risk level (0.2-0.4), employing a conventional protection strategy to balance safety and economic requirements; the enhanced protection layer corresponds to medium risk level (0.4-0.6), employing a strengthened protection strategy to improve safety margin and reliability; the emergency protection layer corresponds to medium-to-high risk level and high risk level (0.6-1.0), employing a maximum protection strategy to prioritize the safety of the device and the power grid. When the risk level is medium-to-high (0.6-0.8), the emergency protection layer adopts the standard emergency mode; when the risk level reaches high risk (0.8-1.0), the emergency protection layer upgrades to the highest emergency mode, further shortening response time and increasing protection strength. The hierarchical construction also considers the dynamic changes in the comprehensive state feature vector, adjusting the response sensitivity and switching threshold of the protection level according to the changing trend and fluctuation degree of the feature vector. For example, when the risk stratification result is low to medium risk and the comprehensive state feature vector shows that the power grid operation is relatively stable, a standard protection level is constructed, and a medium response sensitivity and a regular switching threshold are set. Based on the comprehensive analysis of risk stratification results and state characteristics, the switching protection level is constructed.

[0061] For example, the analysis of the switching protection level to generate a fault-tolerant mechanism configuration includes: performing an inter-level coordination assessment on the switching protection level to obtain coordination indicators, the coordination indicators including level response time and level conflict frequency; determining a protection priority sequence based on the coordination indicators; allocating fault-tolerant resources according to the protection priority sequence to generate a resource configuration scheme; and generating a fault-tolerant mechanism configuration based on the resource configuration scheme.

[0062] Then, the switching protection levels are analyzed to generate a fault-tolerant mechanism configuration. First, the inter-level coordination of the switching protection levels is assessed to obtain coordination indicators, including two core parameters: level response time and level conflict frequency. Level response time is evaluated by analyzing the activation time, processing time, and switching time of each protection level; a shorter response time indicates better coordination. Level conflict frequency is evaluated by statistically analyzing the frequency of simultaneous activation or mutual interference of different protection levels; a lower conflict frequency indicates higher coordination. Next, a protection priority sequence is determined based on the coordination indicators. According to the speed of level response time and the frequency of conflict, a priority ranking is established for each protection level. Levels with shorter response times and lower conflict frequencies receive higher priority, while levels with longer response times or higher conflict frequencies have lower priority. Then, fault-tolerant resources are allocated according to the protection priority sequence to generate a resource configuration scheme. Limited fault-tolerant resources are allocated according to priority, prioritizing the resource needs of high-priority levels to ensure resource guarantees for critical protection functions, while also comprehensively considering the basic resource needs and coordination requirements of each level. Finally, a fault-tolerant mechanism configuration is generated based on the resource allocation scheme. This transforms the resource allocation results into specific fault-tolerant strategies and protection parameters, including fault-tolerant algorithm selection, protection threshold setting, response time configuration, and coordination logic design, forming a complete fault-tolerant mechanism configuration scheme. The fault-tolerant mechanism configuration also establishes a dynamic adjustment function, optimizing and adaptively adjusting the configuration parameters online based on changes in operating status and protection effect feedback. The fault-tolerant mechanism configuration is generated through system analysis and optimization configuration by switching protection levels.

[0063] Finally, a multi-level fault-tolerant switching system is established based on the fault-tolerant mechanism. A hierarchical architecture is adopted, organically organizing and coordinating each protection level according to priority and functional characteristics. The system architecture includes four functional layers: perception layer, decision layer, execution layer, and monitoring layer. The perception layer is responsible for real-time monitoring of the operating status of energy storage devices and the power grid, collecting various status information and fault signals; the decision layer analyzes and processes the monitoring information according to the strategies and algorithms configured by the fault-tolerant mechanism, determining the type and level of switching protection; the execution layer implements specific switching protection actions and fault-tolerant measures according to the instructions of the decision layer; the monitoring layer supervises and evaluates the execution process and effect of switching protection, providing feedback information and optimization suggestions. The system also integrates auxiliary functions such as fault diagnosis, early warning alarm, emergency response, and recovery control, forming a comprehensive fault-tolerant protection capability. The system's reliability design employs redundant configuration and backup mechanisms to ensure that basic switching protection functions are maintained even when some protection levels fail. The system also establishes a self-learning and self-optimization mechanism, continuously improving fault-tolerant strategies and protection parameters through operational experience accumulation and effect evaluation feedback. For example, the established multi-level fault-tolerant switching system includes four protection levels, three types of fault-tolerant strategies, five switching modes, and two backup mechanisms, capable of meeting switching protection needs under different risk levels and complex operating environments. The multi-level fault-tolerant switching system is established through system integration and architecture design configured with fault-tolerant mechanisms.

[0064] Based on a multi-level fault-tolerant switching system, primary switching schemes, backup switching schemes, and emergency switching schemes are generated. According to the hierarchical structure and fault-tolerant mechanism configuration of the multi-level fault-tolerant switching system, corresponding switching implementation strategies are formulated for different risk levels and operating states. The primary switching scheme corresponds to the basic protection level and standard protection level (risk level 0-0.4), employing conventional switching strategies and optimized switching algorithms. It is suitable for switching requirements under normal operating conditions, offering high economic efficiency and a good user experience. The switching response time is set at 100-200ms, and the switching success rate is required to reach over 99%. The backup switching scheme corresponds to the enhanced protection level (risk level 0.4-0.6), employing reinforced switching strategies and redundant protection mechanisms. It is suitable for situations where the primary switching scheme cannot be implemented or the operating environment deteriorates. It provides switching functionality while ensuring safety and reliability, with a switching response time set at 50-100ms and a switching success rate required to reach over 98%. The emergency switching scheme corresponds to the emergency protection level (risk level 0.6-1.0), employing a rapid switching strategy and a minimal protection mechanism. It is suitable for emergency switching needs under severe faults, extreme conditions, or critical situations, prioritizing the basic safety of energy storage devices and the power grid. The switching response time is set at 20-50ms, and the switching success rate is required to reach over 95%. Three switching schemes establish a hierarchical triggering mechanism and switching logic, automatically selecting the appropriate switching scheme based on real-time changes in debt risk parameters and comprehensive state characteristic vectors, and establishing seamless switching and fault recovery mechanisms between schemes. Through the system design and strategy optimization of a multi-level fault-tolerant switching architecture, the primary switching scheme, backup switching scheme, and emergency switching scheme are ultimately generated.

[0065] Step S160: Establish a causal mapping relationship for handover decisions based on the primary handover scheme, the backup handover scheme, and the emergency handover scheme; and dynamically adjust and generate an adaptive handover instruction sequence based on the causal mapping relationship for handover decisions.

[0066] Specifically, the three switching schemes are causally related and mutually influential in their decision-making logic and triggering conditions. By establishing a causal mapping relationship for switching decisions, the decision-making patterns and conversion logic among different switching schemes are deeply explored, enabling the intelligent generation of adaptive switching instructions.

[0067] In some embodiments, establishing a causal mapping relationship for handover decisions based on the primary handover scheme, the backup handover scheme, and the emergency handover scheme includes: extracting decision conditions for the primary handover scheme, the backup handover scheme, and the emergency handover scheme; establishing a condition-result mapping table based on the decision conditions; analyzing the condition-result mapping table to obtain the causal association strength; constructing a dynamic weighting mechanism based on the causal association strength; and obtaining the causal mapping relationship for handover decisions based on the dynamic weighting mechanism.

[0068] First, decision conditions were extracted for the primary switchover, backup switchover, and emergency switchover schemes. Knowledge extraction and rule parsing techniques were used to extract key decision condition elements from the strategy descriptions, parameter configurations, and triggering logic of the three switchover schemes. The decision condition extraction focused on core elements such as state judgment conditions, threshold setting conditions, timing constraints, and safety boundary conditions, establishing a structured decision condition database. The primary switchover scheme's decision conditions extraction included elements such as normal operation state judgment, economic optimization target setting, response time requirements, and switchover success rate thresholds, forming a set of decision conditions oriented towards efficiency optimization. The backup switchover scheme's decision conditions extraction included elements such as abnormal state identification, safety margin assurance, redundancy protection activation, and reliability requirements, forming a set of decision conditions oriented towards safety assurance. The emergency switchover scheme's decision conditions extraction included elements such as hazardous state detection, rapid response triggering, minimum function maintenance, and emergency protection activation, forming a set of decision conditions oriented towards emergency response. The extraction process employed natural language processing and semantic analysis techniques to convert the textually described decision logic into structured conditional expressions and logical relationships. The extracted conditions were categorized and labeled according to functional attributes, importance, and triggering frequency. For example, key decision conditions such as "debt risk parameter ≤ 0.4", "grid frequency deviation ≤ ±0.2Hz", and "load change rate ≤ 10% / min" are extracted from the main switching scheme, while emergency decision conditions such as "equipment fault detection signal = TRUE", "grid frequency deviation ≥ ±0.5Hz", and "response time ≤ 50ms" are extracted from the emergency switching scheme. Through system analysis and element extraction of the three switching schemes, a complete set of decision conditions is obtained.

[0069] Next, a condition-result mapping table is established based on the decision conditions. Using association rule mining and decision tree analysis, the extracted decision conditions are matched with their corresponding switching results to establish an "IF-THEN" form condition-result correspondence. The mapping table structure includes fields such as condition number, condition description, condition type, condition value range, result number, result description, result type, and confidence level. The condition field records the specific judgment conditions and threshold settings, the result field records the corresponding switching actions and execution parameters, and the confidence level field records the reliability of the condition-result relationship. The mapping table establishment process comprehensively considers the combined effects and interactive influences of conditions, establishing not only direct mappings between single conditions and results but also complex mapping relationships between multiple condition combinations and composite results. The table construction integrates special mapping rules for anomaly handling and boundary cases to ensure the integrity and robustness of the mapping table. For example, the established mapping table contains key mapping relationships such as "debt risk parameter 0.2-0.4 → main switching scheme activated, confidence level 0.95" and "grid frequency deviation ≥ 0.5Hz → emergency switching scheme initiated, confidence level 0.98", covering the main decision-making scenarios and switching logic. Based on in-depth analysis and correlation mining of decision conditions, a complete condition-result mapping table is formed.

[0070] Then, the condition-outcome mapping table is analyzed to obtain the causal association strength. Statistical analysis and machine learning methods are used to quantitatively evaluate the condition-outcome correspondences in the mapping table, calculating the influence and predictive ability of each condition element on the switching outcome. Association strength analysis employs multiple metrics, including association rule evaluation indicators such as support, confidence, lift, and chi-square statistic, as well as feature importance evaluation indicators such as information gain, Gini coefficient, and correlation coefficient. Support assesses the frequency of occurrence of condition-outcome combinations in the overall sample, reflecting the universality and representativeness of the mapping relationship; confidence assesses the probability of the outcome occurring when the conditions are met, reflecting the reliability and accuracy of the mapping relationship; lift assesses the degree to which the conditions improve the outcome prediction, reflecting the effectiveness and gain effect of the mapping relationship. Causal association strength also considers the time dimension's delay and cumulative effects, identifying the time propagation characteristics and delayed response patterns of the impact of condition changes on the outcome through time series analysis and lag correlation analysis. The stability and consistency of association strength are evaluated under different time windows and condition granularities. The analysis results are represented by standardized scores, ranging from 0 to 1. Higher values ​​indicate stronger causal relationships and more significant influence of the conditions on the outcomes. For example, the analysis found a causal relationship strength of 0.85 between "debt risk parameters" and "primary switching scheme selection," and 0.92 between "grid frequency deviation" and "emergency switching activation," indicating that these conditions have a strong decisive influence on the corresponding results. Accurate causal relationship strengths are obtained through in-depth mining and correlation analysis of the condition-outcome mapping table.

[0071] Next, a dynamic weighting mechanism is constructed based on the strength of causal relationships. A multi-level weighting allocation strategy is adopted, including three levels: intra-condition weights, inter-option weights, and comprehensive decision weights. Intra-condition weights are allocated according to the strength of the causal relationship between each decision condition; conditions with high correlation strength receive greater weights, while those with low correlation strength receive correspondingly lower weights, ensuring that important conditions play a leading role in the decision-making process. Inter-option weights are allocated based on the applicability and execution frequency of the three switching options; options with broad applicability and high usage frequency receive a larger base weight, while those with strong specialization and low usage frequency receive a smaller base weight. The comprehensive decision weight is calculated by fusing condition weights and option weights, using a combination of weighted averaging and nonlinear fusion methods, considering the interaction and synergistic effects between weights. The dynamic adjustment mechanism updates and optimizes the weight allocation in real time based on changes in operational status and feedback on decision-making effects, improving the adaptability and accuracy of weight allocation. The weighting mechanism sets constraints and boundary limits to ensure the rationality and stability of weight allocation and avoid decision-making biases caused by extreme weight allocations. For example, the constructed dynamic weighting mechanism assigns a weight of 0.35 to "debt risk parameters," 0.28 to "power grid status indicators," 0.22 to "load change characteristics," and 0.15 to "economic indicators," with dynamic adjustments of ±10% based on operational conditions. A reliable dynamic weighting mechanism is formed based on scientific analysis of causal correlation strength and weight optimization.

[0072] Finally, a dynamic weighting mechanism is used to obtain the causal mapping relationship for switching decisions. A combination of weighted decision trees and causal inference networks is employed to systematically integrate the weight allocation results of the dynamic weighting mechanism with the association information of the condition-outcome mapping table, forming a complete causal mapping relationship model. The mapping relationship is represented in the form of a directed graph, where nodes represent decision conditions and switching results, edges represent causal relationships, and the weights of the edges reflect the strength and degree of influence. The relationship model establishes a multi-level mapping structure, including direct causal relationships, indirect causal relationships, and compound causal relationships. Direct relationships represent the direct impact of conditions on results, indirect relationships represent the impact transmitted through intermediate variables, and compound relationships represent the comprehensive impact of multiple conditions combined on results. The mapping relationship integrates the causal transmission characteristics of the time dimension, establishing a temporal causal network to reflect the time delay and propagation path of the impact of condition changes on results. The relationship model has dynamic updating and learning optimization functions, continuously improving and optimizing the causal mapping relationship based on new operational data and decision-making experience, thereby improving the accuracy and adaptability of decision-making. For example, the established causal mapping relationships show key causal transmission chains such as: "Debt risk parameter increases by 0.1 → Primary switching scheme activation probability decreases by 15% → Backup switching scheme activation probability increases by 12%", and "Grid frequency deviation exceeds 0.3Hz → Emergency switching scheme activation probability increases by 80% → Response time shortened to within 30ms". Through the systematic application of dynamic weighting mechanisms and causal reasoning, a scientifically reliable causal mapping relationship for switching decisions is ultimately obtained.

[0073] An adaptive switching command sequence is dynamically generated based on the causal mapping relationship of switching decisions. The command sequence generation employs an intelligent decision-making algorithm, which reads the real-time operating status information of the energy storage system and the power grid. Through reasoning analysis using a causal mapping model, it automatically determines the most suitable switching scheme and execution parameters. The generation process establishes a multi-step decision-making flow. First, state assessment and condition matching are performed, comparing the current operating state with the decision conditions in the causal mapping relationship to identify satisfied condition combinations and triggered causal relationships. Next, scheme selection and parameter determination are performed, selecting the optimal switching scheme and determining specific execution parameters based on the causal reasoning results and dynamic weight allocation. Finally, command generation and sequence optimization are performed, converting the decision results into standardized switching commands and optimizing the command sequence according to execution constraints and timing requirements. The adaptive adjustment mechanism continuously corrects and updates the command sequence in real time based on execution effect feedback and environmental changes, ensuring that the switching commands always adapt to current operating conditions and changing demands. The command sequence also includes fault tolerance and anomaly recovery functions. When anomalies or deviations in command execution are detected, backup commands or emergency handling procedures are automatically activated to ensure the continuity and reliability of the switching process. Sequence optimization considers the coordination and timing constraints between instructions. Through timing optimization algorithms and constraint satisfaction techniques, it ensures that the instruction sequence is reasonable in time, consistent in logic, and feasible in execution. For example, the generated adaptive switching instruction sequence includes a complete process: "status monitoring → condition judgment → scheme selection → parameter setting → instruction issuance → execution monitoring → effect evaluation → dynamic adjustment." The instruction response time is 50-200ms, and the instruction adaptive adjustment cycle is 1-5 seconds, ensuring the intelligent and adaptive characteristics of the switching process. Through the intelligent application and dynamic optimization of causal mapping relationships, a highly efficient and reliable adaptive switching instruction sequence is ultimately generated.

[0074] Step S170: Based on the adaptive switching instruction sequence, the energy storage system is switched to the operating mode to obtain actual execution data. The actual execution data is used to reverse correct the scenario type identifier to obtain the corrected scenario type identifier, thus completing the adaptive switching control of multiple scenario operating modes.

[0075] Specifically, the system acquires actual execution data by switching the operating mode of the energy storage device based on an adaptive switching command sequence. The command execution process follows the command sequence to perform the actual switching operation of the energy storage device's operating mode, monitoring key execution parameters in real time, such as changes in charging and discharging power, capacity status, response time, switching success rate, and operational stability. Data acquisition uses high-frequency sampling to record detailed operational data during the switching process, including voltage and current waveforms, power change curves, temperature change trends, equipment status signals, and protection action information. Execution data also includes impact data such as comparisons of grid status before and after the switching, load response effects, economic benefits, and changes in safety indicators. Data acquisition employs multi-sensor fusion technology to ensure data integrity and accuracy, covering the entire process and all parameters of the energy storage device's operation. Execution effect evaluation calculates key performance indicators such as switching accuracy, response time deviation, power tracking error, and economic benefit deviation through comparative analysis of preset targets and actual results. Through the execution of the adaptive switching command sequence and full-process monitoring, actual execution data reflecting the actual switching performance of the energy storage device is obtained.

[0076] The algorithm uses actual execution data to reverse-correct scene type identifiers, obtaining the corrected scene type identifiers to achieve adaptive switching control of multiple scene operation modes. The reverse correction algorithm employs machine learning and data mining techniques to compare and analyze actual execution data with the original scene type identifiers, identifying deviations and misjudgments in scene identification. The correction process first establishes a mapping between execution effect and scene type, building a standard pattern library of scene-execution data by analyzing typical execution data characteristics under different scene types. Then, pattern matching and similarity calculation methods are used to compare the current actual execution data with the standard pattern library, calculating the matching degree and similarity score for each scene type. When the matching degree between the actual execution data and the standard pattern corresponding to the original scene type identifier is lower than a set threshold, the scene re-identification process is initiated, re-determining the most matching scene type based on the execution data characteristics. The correction algorithm also considers the impact of changes in the execution environment and external interference factors, using environmental compensation and interference filtering techniques to eliminate execution deviations caused by non-scene factors, ensuring the accuracy of the correction results. The correction of scene type identifiers employs a confidence-weighted and probability-updating mechanism, dynamically adjusting the confidence level of the scene type identifiers based on the reliability and representativeness of the actual execution data. The revised scene type identifier includes information such as the main type code, subtype code, revised confidence level, and revision basis explanation. The revised scene type identifier provides feedback to optimize the parameter configuration and judgment logic of the scene recognition algorithm, while also using execution effect data to optimize the causal mapping relationship and weight allocation mechanism, and improving the generation strategy of switching instruction sequences. Adaptive switching of multiple scene operation modes enables intelligent management and control of energy storage devices in complex operating environments. It can automatically select the optimal operation mode and switching strategy based on dynamic changes in grid status, load demand, and economic requirements. For example, during weekday evening peak hours, when grid frequency fluctuates and load increases rapidly, it automatically identifies the scenario as a grid support scenario and initiates a fast response mode; during weekend night hours, when there is a surplus of renewable energy generation and electricity prices are low, it automatically switches to a renewable energy consumption scenario and initiates a charging mode. Through feedback from the revised scene type identifier and full-process optimization, adaptive switching control of multiple scene operation modes is ultimately achieved.

[0077] To implement the adaptive switching control method for multi-scenario operation modes of the energy storage system corresponding to the above method embodiments, and to achieve the corresponding functions and technical effects. See also Figure 2 , Figure 2 This diagram illustrates a structural block diagram of an adaptive switching control device 200 for multi-scenario operation modes of an energy storage system according to an embodiment of this application. For ease of explanation, only the parts relevant to this embodiment are shown. The adaptive switching control device 200 for multi-scenario operation modes of an energy storage system provided in this embodiment includes: The data acquisition module 201 is used to collect power grid status data, load demand data and economic indicator data, and to perform multi-dimensional fusion analysis on the power grid status data, the load demand data and the economic indicator data to generate a comprehensive status feature vector; Scene recognition module 202 is used to perform scene recognition and classification based on the comprehensive state feature vector to generate scene type identifiers, sort the demand priority according to the scene type identifiers, and perform energy spatiotemporal transfer analysis based on the demand priority sorting to generate spatiotemporal energy distribution schemes. Conflict coordination module 203 is used to perform scene conflict analysis on the spatiotemporal energy distribution scheme to identify potential conflict points, construct a scene coordination mechanism based on the potential conflict points, and generate a scene fusion strategy through the scene coordination mechanism. Debt assessment module 204 is used to obtain energy overdraft capacity and compensation time sequence based on the scenario fusion strategy, and generate debt risk parameters based on the energy overdraft capacity and the compensation time sequence; The fault-tolerant switching module 205 is used to establish a multi-level fault-tolerant switching system based on the debt risk parameters and the comprehensive state feature vector, and to generate a main switching scheme, a backup switching scheme and an emergency switching scheme based on the multi-level fault-tolerant switching system. Decision adjustment module 206 is used to establish a causal mapping relationship for switching decisions based on the primary switching plan, the backup switching plan and the emergency switching plan, and to dynamically adjust and generate an adaptive switching instruction sequence based on the causal mapping relationship for switching decisions. The execution feedback module 207 is used to switch the operating mode of the energy storage system based on the adaptive switching instruction sequence to obtain actual execution data, and use the actual execution data to reverse correct the scene type identifier to obtain the corrected scene type identifier, thereby completing the adaptive switching control of multiple scene operating modes.

[0078] The aforementioned adaptive switching control device 200 for multi-scenario operation modes of an energy storage system can implement the adaptive switching control method for multi-scenario operation modes of an energy storage system as described in the above method embodiments. The options in the above method embodiments are also applicable to this embodiment, and will not be detailed here. The remaining content of this application embodiment can be referred to the content of the above method embodiments, and will not be repeated in this embodiment.

[0079] like Figure 3 As shown, the third embodiment of the present invention also provides a computer device, including a memory 301, a processor 302, and a computer program stored in the memory 301 and executable on the processor 302, characterized in that the processor 302 executes the program to implement the steps of the adaptive switching control method for multi-scenario operation modes of an energy storage system as described in the first embodiment of the present invention.

[0080] The purpose of the above embodiments is to reproduce and derive the technical solution of the present invention by way of example, and to fully describe the technical solution, purpose and effect of the present invention. The purpose is to enable the public to have a more thorough and comprehensive understanding of the disclosure of the present invention, and not to limit the scope of protection of the present invention.

[0081] The above embodiments are not an exhaustive list based on the present invention, and there may be many other embodiments not listed. Any substitutions and improvements made without departing from the concept of the present invention are within the protection scope of the present invention.

Claims

1. A method for adaptive switching control of multiple operating modes in an energy storage system, characterized in that, include: Collect power grid status data, load demand data, and economic indicator data, and perform multi-dimensional fusion analysis on the power grid status data, load demand data, and economic indicator data to generate a comprehensive status feature vector; Based on the comprehensive state feature vector, scene identification and classification are performed to generate scene type identifiers. Based on the scene type identifiers, demand priority is sorted. Based on the demand priority sorting, spatiotemporal energy transfer analysis is performed to generate spatiotemporal energy distribution schemes. Scene conflict analysis is performed on the spatiotemporal energy distribution scheme to identify potential conflict points. A scene coordination mechanism is constructed based on the potential conflict points, and a scene fusion strategy is generated through the scene coordination mechanism. Based on the scenario fusion strategy, the energy overdraft capacity and compensation time sequence are obtained, and debt risk parameters are generated according to the energy overdraft capacity and the compensation time sequence. A multi-level fault-tolerant switching system is established based on the debt risk parameters and the comprehensive state feature vector. A primary switching scheme, a backup switching scheme, and an emergency switching scheme are generated based on the multi-level fault-tolerant switching system. A causal mapping relationship for handover decisions is established based on the primary handover scheme, the backup handover scheme, and the emergency handover scheme. An adaptive handover instruction sequence is generated by dynamically adjusting the handover decision causal mapping relationship. Based on the adaptive switching instruction sequence, the energy storage system is switched to different operating modes to obtain actual execution data. The actual execution data is then used to reverse-correct the scenario type identifier to obtain the corrected scenario type identifier, thus completing the adaptive switching control of multiple scenario operating modes.

2. The method according to claim 1, characterized in that, The step of generating a spatiotemporal energy distribution scheme based on the demand priority ranking and spatiotemporal transfer analysis includes: The time-series demand distribution is obtained by decomposing the demand priority ranking into a time dimension. A spatial energy demand matrix is ​​generated by spatial dimension mapping based on the aforementioned temporal demand distribution. The optimal transmission path is obtained by analyzing the space energy demand matrix. A spatiotemporal energy distribution scheme is constructed based on the optimal transmission path.

3. The method according to claim 1, characterized in that, The spatiotemporal energy distribution scheme is analyzed to identify potential conflict points, including: The spatiotemporal energy distribution scheme is subjected to multi-scene overlapping region identification to obtain the intersection region; The analysis of the intersecting regions generates a conflict level assessment result; The conflict level is determined based on the conflict severity assessment results; Potential conflict points are marked according to the conflict level.

4. The method according to claim 1, characterized in that, The process of obtaining the energy overdraft capacity and compensation timing based on the scene fusion strategy includes: The energy supply and demand difference is obtained by analyzing the scenario fusion strategy. A debt capacity assessment mechanism shall be established based on the aforementioned energy supply and demand difference. The energy overdraft capacity is obtained using the aforementioned debt capacity assessment mechanism; The compensation timing is determined based on the energy overdraft capacity.

5. The method according to claim 1, characterized in that, The generation of debt risk parameters based on the energy overdraft capacity and the compensation timing includes: The overdraft capacity is assessed for risk exposure to obtain the overdraft risk value; Based on the compensation time series, a time series risk coefficient is generated through time risk analysis. A risk weight allocation mechanism is established based on the overdraft risk value and the time-series risk coefficient; The risk weight allocation mechanism is used to generate debt risk parameters through weighted fusion.

6. The method according to claim 1, characterized in that, The establishment of a multi-level fault-tolerant switching system based on the debt risk parameters and the comprehensive state feature vector includes: Risk stratification results are generated by classifying risk levels based on the aforementioned debt risk parameters; Construct a corresponding switching protection level based on the risk stratification results and the comprehensive state feature vector; The switching protection level is analyzed to generate a fault tolerance mechanism configuration; A multi-level fault-tolerant switching system is established based on the aforementioned fault-tolerant mechanism.

7. The method according to claim 1, characterized in that, The step of establishing a causal mapping relationship for handover decisions based on the primary handover scheme, the backup handover scheme, and the emergency handover scheme includes: Decision conditions are extracted for the primary handover scheme, the backup handover scheme, and the emergency handover scheme; Establish a condition-outcome mapping table based on the aforementioned decision conditions; The condition-result mapping table is analyzed to obtain the strength of causal association. A dynamic weighting mechanism is constructed based on the strength of the causal relationship. The causal mapping relationship of the switching decision is obtained based on the dynamic weighting mechanism.

8. The method according to claim 6, characterized in that, The analysis of the switching protection level and the generation of the fault tolerance mechanism configuration include: The coordination indicators are obtained by performing an inter-level coordination assessment on the switching protection level, and the coordination indicators include level response time and level conflict frequency. The protection priority sequence is determined based on the aforementioned coordination indicators; Based on the protection priority sequence, a fault-tolerant resource allocation scheme is generated to produce a resource configuration scheme. A fault tolerance mechanism configuration is generated based on the resource configuration scheme.

9. A multi-scenario operation mode adaptive switching control device for an energy storage system, characterized in that, include: The data acquisition module is used to collect power grid status data, load demand data, and economic indicator data, and to perform multi-dimensional fusion analysis on the power grid status data, load demand data, and economic indicator data to generate a comprehensive status feature vector. The scene recognition module is used to identify and classify scenes based on the comprehensive state feature vector to generate scene type identifiers, sort the demand priorities according to the scene type identifiers, and perform spatiotemporal energy transfer analysis based on the demand priority sorting to generate spatiotemporal energy distribution schemes. The conflict coordination module is used to perform scene conflict analysis on the spatiotemporal energy distribution scheme, identify potential conflict points, construct a scene coordination mechanism based on the potential conflict points, and generate a scene fusion strategy through the scene coordination mechanism. The debt assessment module is used to obtain the energy overdraft capacity and compensation timeline based on the scenario fusion strategy, and generate debt risk parameters based on the energy overdraft capacity and the compensation timeline. The fault-tolerant switching module is used to establish a multi-level fault-tolerant switching system based on the debt risk parameters and the comprehensive state feature vector, and to generate a primary switching scheme, a backup switching scheme and an emergency switching scheme based on the multi-level fault-tolerant switching system. The decision adjustment module is used to establish a causal mapping relationship for handover decisions based on the primary handover plan, the backup handover plan, and the emergency handover plan, and to dynamically adjust and generate an adaptive handover instruction sequence based on the causal mapping relationship for handover decisions. The execution feedback module is used to switch the operating mode of the energy storage system based on the adaptive switching instruction sequence to obtain actual execution data, and use the actual execution data to reverse correct the scenario type identifier to obtain the corrected scenario type identifier, thereby completing the adaptive switching control of multiple scenario operating modes.

10. A computer device, characterized in that, It includes a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the method as described in any one of claims 1 to 8.

Images