Energy storage rolling optimization configuration method based on incremental distribution network

Abstract

The application discloses an energy storage rolling optimization configuration method based on an incremental power distribution network, relates to the technical field of energy storage potential prediction, and comprises the following steps: identifying trend mutation points of a historical load time sequence through a structural change point detection algorithm, and constructing an initial trend baseline; segmenting and decomposing the load sequence based on the initial trend baseline, extracting stable trend items and short-term disturbance items in each segment, and constructing a feature tensor; adaptively evaluating each segment in the feature tensor, and marking and shielding the trend segment; performing collaborative optimization matching on the high adaptability segment, and outputting a segmented energy storage configuration scheme; and rolling extraction of the feature tensor along the time sequence direction, generation of a dynamic adaptation curve, and feedback correction of a long-term energy storage deployment plan. The application selects high adaptability segments for matching optimization, and effectively improves the matching precision between the energy storage system and the load evolution characteristics.

Description

Technical Field

[0001] This invention relates to the field of energy storage potential prediction technology, and more specifically, to a method for rolling optimization configuration of energy storage based on incremental distribution networks. Background Technology

[0002] With the gradual integration of distributed renewable energy and large-scale energy storage systems, the load in the power supply area is showing greater volatility and uncertainty, placing higher demands on the dynamic regulation capabilities of energy storage in the power system.

[0003] In load regulation and energy storage configuration technologies, a common practice is to identify the amplitude and rate of change of load fluctuations based on historical load time series data, thereby estimating the regulation potential that can be addressed by the energy storage system. However, existing technologies generally treat all load disturbance segments as adjustable targets, ignoring whether the disturbances themselves have the value of energy storage intervention. In reality, many load disturbances originate from pseudo-fluctuations (such as data anomalies, temporary load shedding, and short-term weather disturbances) or unadjustable areas (such as sudden changes in industrial batch processing and emergency load switching). Although these segments appear as "changes" on the load curve, they are not objects that the energy storage system can effectively respond to or regulate.

[0004] Due to the lack of a mechanism to identify the "tunability" of disturbance behavior, existing methods cannot separate such structurally uncontrollable disturbance regions from load trends, leading to distorted assessments of regulation potential. As a result, energy storage devices often encounter problems such as regulation failure, strategy underutilization, and low energy utilization during regulation. In severe cases, this can even cause the energy storage system to intervene at the wrong time, resulting in reverse regulation or resource waste.

[0005] For example, the invention patent announcement CN119577582A discloses a source-load matching rate prediction method, system, device, medium, and product for energy storage scheduling decision-making. By constructing an initial source-load matching rate prediction model based on a deep learning algorithm, using a prediction error function and a decision evaluation function to form a hybrid loss function, and optimizing the parameters of the initial source-load matching rate prediction model based on a hybrid gradient descent learning method, a source-load matching rate prediction model is obtained, thereby accurately predicting the source-load matching rate and improving the decision-making accuracy under the predicted source-load matching rate.

[0006] For example, the invention patent announcement CN116231696B, which describes a load-prediction-based energy storage switching control method and system, includes an information acquisition module, a retrieval module, a prediction module, and a judgment module. The judgment module is used to determine whether the predicted electricity consumption for the day is greater than the storage capacity of each energy storage device, based on the predicted electricity consumption for the day and the storage capacity of each energy storage device. If so, it removes energy storage devices whose storage capacity is less than a preset threshold and switches the remaining energy storage devices to the corresponding load devices. Otherwise, it matches the energy storage device with the largest storage capacity among all energy storage devices and determines whether the actual difference between this storage capacity and the predicted electricity consumption for the day is greater than a preset difference. If so, it switches the energy storage device corresponding to the largest storage capacity to that load device. Otherwise, it matches the two energy storage devices with the largest storage capacity among all current energy storage devices and switches them to the corresponding load devices.

[0007] The aforementioned publicly disclosed technical solutions have at least the following technical problems: In practical applications, some load fluctuations are not caused by normal periodic demand, but rather by sudden events (such as cold start events or process mutations) or factors that are not adjustable by energy storage (such as human switching operations or extreme weather events). Because existing technologies cannot identify these "pseudo-fluctuations" or "unadjustable areas," they are incorrectly included in the potential calculation of the energy storage system, causing subsequent matching failures, strategy execution deviations, or even system overload.

[0008] To address the above problems, this invention proposes a solution. Summary of the Invention

[0009] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a rolling optimization configuration method for energy storage based on incremental distribution networks, which improves the accurate matching capability between energy storage strategies and load behavior, thereby solving the core problems of energy storage configuration mismatch, strategy ineffectiveness, or regulation failure in the prior art.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] The rolling optimization configuration method for energy storage based on incremental distribution networks includes the following steps: identifying trend abrupt change points in historical load time series using a structural change point detection algorithm to construct an initial trend baseline; segmenting the load series based on the initial trend baseline, extracting stable trend terms and short-term disturbance terms within each segment, and constructing a feature tensor; for each segment in the feature tensor, conducting an adaptability assessment based on key technical constraints limiting the adjustable potential of the energy storage system, and marking and masking trend segments; for highly adaptable segments, performing collaborative optimization matching with multiple objectives of minimizing the total life cycle cost and maximizing the grid regulation demand satisfaction rate, and outputting segmented energy storage configuration schemes; rolling the feature tensor along the time series direction to generate dynamic adaptation curves and providing feedback to correct long-term energy storage deployment plans.

[0012] In a preferred embodiment, the step of identifying trend abrupt changes in historical load time series using a structural change point detection algorithm and constructing a dynamic initial trend baseline specifically involves: acquiring load data of the power supply area within a historical window to form an original time series; constructing a sliding window of length t to traverse all original time series, fitting an autoregressive prediction model in each window and calculating the residual sequence; identifying structural change points in the residual sequence using a cumulative sum change point detection algorithm; dividing the original load series into several sub-segments based on the change points, independently fitting a trend function within each sub-segment, and splicing them together to form a segmented continuous trend baseline.

[0013] In a preferred embodiment, the method of identifying structural change points in the residual sequence through a cumulative and change point detection algorithm specifically involves: calculating the average value of the residuals based on the residual sequence and constructing a cumulative deviation sequence function; detecting the inflection point of the cumulative deviation sequence function; and determining a structural change point when the change in function value at the inflection point exceeds a preset fluctuation threshold.

[0014] In a preferred embodiment, the step of segmenting the load sequence based on the initial trend baseline, extracting the stable trend term and short-term disturbance term within each segment, and constructing a feature tensor specifically involves:

[0015] Based on the piecewise trend function fitted in the trend baseline, the load time series is divided into multiple trend stable segments. Within each trend stable segment, a stable trend term is extracted based on the fitted trend function, and a disturbance sequence is generated by the difference between the original load sequence and the trend term. The disturbance sequence is decomposed into a multi-scale sliding window to extract disturbance behavior features within the window, including the dominant disturbance direction, disturbance density spectrum, disturbance energy intensity, and disturbance evolution path. The disturbance behavior features are fused with the trend slope and trend drift rate to construct a multidimensional feature vector for a single trend segment. The multidimensional feature vectors corresponding to each trend segment are stacked in temporal order to form a feature tensor with a multidimensional nested structure.

[0016] In a preferred embodiment, the adaptation assessment of the key technical constraints limiting the adjustable potential of the energy storage system for each segment of the feature tensor includes: extracting the trend slope, drift rate, disturbance amplitude, and disturbance frequency of each trend segment in the feature tensor to construct a dynamic demand vector; synchronously acquiring the operating state parameters of the energy storage system within that time period based on the temporal position of the trend segment, including energy storage charge / discharge power margin, state of charge, regulation direction margin, response delay, and operating aging factor; constructing an adaptation matching model based on the dynamic demand vector and the set of operating state parameters; identifying the main technical constraints that limit the energy storage's adjustment capability in that trend segment through the adaptation matching model; and binding the adjustment demand of each trend segment with the constraints to generate a trend segment-level adjustable potential limiting factor map.

[0017] In a preferred embodiment, the step of marking and masking trend segments specifically involves: calculating a suitability score based on the adjustable potential limiting factor map of the trend segments, wherein the suitability score is generated by weighting the degree of matching between energy storage response capacity and trend disturbance demand; marking trend segments with suitability scores higher than a preset suitability threshold as high-suitability segments and writing them into the label field of the trend tensor; assigning priority matching weights to high-suitability segments during the strategy generation stage and generating priority adjustment control instructions; marking trend segments with scores lower than the preset suitability threshold as low-suitability segments; implementing differentiated masking strategies for low-suitability segments based on their limiting factor types; and finally writing the marking information of high- and low-suitability segments into the trend tensor structure.

[0018] In a preferred embodiment, the high-adaptability segmentation is optimized and matched collaboratively with multiple objectives, including minimizing the total lifecycle cost and maximizing the grid regulation demand satisfaction rate, to output a segmented energy storage configuration scheme. Specifically, for each high-adaptability segment, its trend disturbance characteristics, adjustable capacity demand, regulation direction, and regulation timeliness are extracted to construct a target vector to be matched. Based on a context-aware mechanism, the weight coefficients of response rate, regulation cost, energy utilization efficiency, and operational risk in the matching evaluation index are dynamically adjusted by combining the pre-sequence and post-sequence characteristics of the trend segment. Using the target vector to be matched as input, a multi-objective matching score is performed with the operational capability vectors of each energy storage unit in a preset energy storage type parameter library to select a set of candidate energy storage units that meet the regulation requirements. In a preset set of operational strategies, regulation strategy templates compatible with the candidate energy storage unit set are retrieved and jointly optimized. The optimal combination of energy storage unit and control strategy is output as the matching result and written into the result domain of the trend tensor.

[0019] In a preferred embodiment, the step of retrieving a regulation strategy template compatible with the candidate energy storage unit set from the preset set of operating strategies and performing joint optimization specifically involves: extracting the dynamic capability constraint vector of each candidate energy storage unit, retrieving a strategy template that satisfies the capability constraint conditions of the energy storage unit from the preset set of operating strategies, and eliminating invalid strategies, wherein the elimination conditions include directional conflict, insufficient regulation bandwidth, or mismatch in scheduling granularity; constructing a joint benefit function for each combination of energy storage units and feasible strategies, and performing multi-objective optimization to solve for the optimal regulation template.

[0020] In a preferred embodiment, the step of extracting feature tensors along the time series direction, generating dynamic adaptation curves, and providing feedback to correct long-term energy storage deployment plans specifically involves: using the time series as an index, extracting trend segments marked as highly adaptable and already matched, and splicing them according to a time segment sliding window to form candidate rolling adjustment paths; based on the optimal energy storage-strategy combination of the candidate rolling adjustment paths, obtaining their resource consumption intensity and performance degradation predictions in future adjustment cycles, and constructing a feasibility assessment map; for paths composed of multiple consecutive highly adaptable segments, evaluating their cumulative response load, resource consumption intensity, expected revenue, and strategy switching frequency based on the feasibility assessment map, and eliminating unsustainable paths according to preset constraints; splicing the remaining feasible paths along the time axis to form a complete dynamic rolling adaptation curve; dynamically adjusting the medium- and long-term deployment plan based on the structural characteristics of the rolling adaptation curve, combined with the overall load evolution trend of the power supply area and energy storage asset life cycle indicators; and writing the finally determined rolling adaptation curve and its deployment recommendations into the system control task pool.

[0021] The technical effects and advantages of the energy storage rolling optimization configuration method based on incremental distribution networks in this invention are as follows:

[0022] 1. This invention constructs a feature tensor containing multi-dimensional features such as trend slope, disturbance intensity, and evolution path to achieve unified deconstruction and in-depth characterization of long-term trends and short-term disturbances in load sequences. Based on this, an adaptability scoring model is established by integrating energy storage operation state parameters to accurately identify technical constraints that limit the adjustment capability in trend segments and select highly adaptable segments for matching optimization. This effectively improves the matching accuracy between the energy storage system and load evolution characteristics, significantly outperforming the traditional static planning method based solely on load average or maximum power indicators.

[0023] 2. This invention extracts high-fitting segments from trend tensors based on time continuity, constructs rolling fit curves, and generates combined matching results of "energy storage unit + regulation strategy" through feasibility assessment and joint optimization mechanisms. This allows for dynamic adjustment of equipment selection priorities, capacity configuration schemes, and strategy deployment time plans. This mechanism not only achieves closed-loop adaptive matching between real-time load changes and energy storage response but also supports intelligent deployment decisions across multiple periods and cycles, demonstrating good scalability and engineering practicality. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the energy storage rolling optimization configuration method based on incremental distribution networks according to the present invention.

[0025] Figure 2 A feasibility assessment diagram for the scrolling adaptation path. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1, Figure 1 The present invention provides a rolling optimization configuration method for energy storage based on incremental distribution networks, comprising the following steps:

[0028] S1, identify trend abrupt change points in historical load time series using a structural change point detection algorithm, and construct an initial trend baseline;

[0029] In this embodiment, the initial trend baseline is used to identify non-stationary structural turning points caused by changes in user behavior cycles, fluctuations in operating rhythm, or non-periodic disturbances.

[0030] The initial trend baseline is constructed based on the structural change point detection algorithm, specifically as follows:

[0031] Obtain load data for the power supply area within a historical window to form the original time series;

[0032] Construct a sliding window of length t to traverse all the original time series, and fit an autoregressive prediction model in each window. And calculate the residual sequence ;in, , The fitted value of the autoregressive prediction model. These are the coefficients of the autoregressive prediction model. For white noise, The original load value at time t. This represents the order of the autoregressive prediction model.

[0033] Structural change points are identified in the residual sequence using a cumulative sum change point detection algorithm;

[0034] Based on the point of change, the original load sequence is divided into several segments. Linear or nonlinear trend functions are fitted to each segment, and the segments are spliced ​​together to form a segmented continuous trend baseline.

[0035] Furthermore, whether or not to use a nonlinear trend function is determined by the following conditions: when the root mean square of the fitting residual of the nonlinear model is reduced by more than a set threshold compared to the linear model, or when there are multiple trend reversal points within the trend segment, a nonlinear function is used to fit the trend line.

[0036] The method for identifying structural change points based on residual sequences using a cumulative sum change point detection algorithm specifically includes:

[0037] Calculating the average value of residuals based on residual sequences And construct the cumulative deviation sequence function;

[0038] The inflection point of the cumulative deviation sequence function is detected. When the change in function value at the inflection point exceeds the preset fluctuation threshold, it is determined to be a structural change point.

[0039] The cumulative deviation sequence function is specifically as follows:

[0040]

[0041]

[0042] in, For cumulative deviation, Let be the prediction residual at time i. For the current moment, The average value of the residuals. This is the actual load value. To predict load values.

[0043] S2, based on the initial trend baseline, segments the load sequence, extracts the stable trend term and short-term disturbance term within each segment, and constructs a feature tensor, specifically:

[0044] Based on the structural change points of the initial trend baseline, the load time series is divided into multiple trend-stable segments;

[0045] In each trend stabilization segment, a stable trend term is extracted based on the fitted trend function, and then processed using the original load sequence. With trend items The difference is used to generate a perturbation sequence. , Let be the disturbance value at time t in the j-th trend segment;

[0046] The perturbation sequence is decomposed into a multi-scale sliding window to extract the perturbation behavior features within each window. These perturbation behavior features include the dominant perturbation direction (e.g., rising, fluctuating, or falling), perturbation density spectrum (reflecting the degree of perturbation concentration), perturbation energy intensity (estimated by the coefficient of variation or Fourier frequency domain energy distribution), and perturbation evolution path (e.g., the start and end positions of the perturbation and the trajectory of the fluctuation amplitude).

[0047] By integrating disturbance behavior features with trend slope and trend drift rate, a multi-dimensional feature vector for a single trend segment is constructed.

[0048] The multidimensional feature vectors corresponding to each trend segment are stacked in temporal order to form a feature tensor with a multidimensional nested structure.

[0049] The feature tensor's dimensional structure includes a time series axis, a trend segment index axis, a feature dimension axis, and an optional perturbation scale hierarchy axis, providing good structural visibility and information compression capabilities.

[0050] In this embodiment, by constructing a feature tensor, long-term trends (such as linear load growth and steady decline) and short-term disturbances (such as peak fluctuations and intraday jitter) are uniformly encoded into structured vectors, achieving integrated trend-disturbance modeling. This also provides multi-dimensional inputs for energy storage regulation strategies, supports time-series rolling forecasting and continuous strategy control, and facilitates multi-objective collaborative evaluation and model learning, thereby enhancing intelligent decision-making capabilities.

[0051] S3, for each segment in the feature tensor, performs an adaptability assessment based on the key technical constraints limiting the adjustable potential of the energy storage system, and marks and masks trend segments, specifically:

[0052] Based on each trend segment in the feature tensor, we extract its trend slope (representing the growth rate of the trend function in the trend segment), drift rate (representing the rate of change of the trend slope within the trend segment, used to characterize trend stability), disturbance amplitude (representing the fluctuation intensity of the disturbance sequence within the trend segment), and frequency index (representing the rate of change of the disturbance), and construct a dynamic demand vector to characterize the load behavior adjustment needs of that segment.

[0053] Based on the time sequence position of the trend segment, the operating status parameters of the energy storage system within that time period are acquired synchronously. These parameters include energy storage charge and discharge power margin, state of charge, adjustment direction margin, response delay parameters, and operating aging factor.

[0054] Construct an adaptive matching model between dynamic demand vectors and operating state parameters, and based on the adaptive matching model, identify the main technical constraints that limit the energy storage's adjustment capability in this trend segment;

[0055] By binding the adjustment needs of each trend segment with the limiting factors, a map of adjustable potential limiting factors at the trend segment level is constructed.

[0056] Furthermore, the process involves constructing an adaptive matching model between the dynamic demand vector and the operating state parameters, and based on this model, identifying the main technical constraints that limit the energy storage's ability to regulate during this trend period. Specifically:

[0057] The matching score between the dynamic demand vector and the operating status parameters of the trend segment is calculated using an adaptive matching model.

[0058] When the score of a certain trend segment is lower than the preset score threshold, the system triggers the cause-finding module, which perturbs each dimension of the dynamic demand vector in turn to construct a differential score.

[0059] The score changes caused by disturbances in each dimension are normalized into a relative sensitivity index;

[0060] Based on the sensitivity ranking of each dimension, the dominant influencing factors for the decline in matching scores are identified, such as insufficient SOC, maximum power limitation, or excessively long response latency. This factor will be marked as a key technical constraint limiting the adjustment capability of this trend segment, forming a "trend-constraint factor" comparison label for subsequent shielding judgment and strategy fine-tuning.

[0061] The dominant factor is used as a technical constraint label for this trend segment and written into the adjustable potential limited factor map.

[0062] Dynamically adjust demand vector , , , , , These represent the required adjustment energy, adjustment time, response rate, disturbance frequency, and adjustment direction, respectively (+1 for discharging and −1 for charging).

[0063] Current operating state vector of energy storage device , , , , , These represent the current remaining adjustable capacity, allowable adjustment time, maximum response rate, allowable disturbance frequency range, and current adjustment direction capability of the energy storage device, respectively.

[0064] The adaptive matching model is specifically as follows:

[0065]

[0066] in, To match the rating, The importance weight of the j-th indicator is preset. The total number of dimensions. Let j be the residual function in the j-th target dimension. The normalization coefficients are preset for each dimension.

[0067] The process of marking and masking trend segments specifically involves:

[0068] Based on the adjustable potential limiting factor map of the trend segment, the adaptability score of each trend segment is calculated. The adaptability score is generated by weighting the degree of matching between the energy storage response capability and the trend disturbance demand.

[0069] Trend segments with fitness scores higher than a preset fitness threshold are marked as high fitness segments and written into the label field of the trend tensor.

[0070] For highly adaptable segments, priority matching weights are assigned during the strategy generation stage to generate priority adjustment and control instructions for that trend segment.

[0071] Trend segments with scores below a preset fit threshold are marked as low fit segments;

[0072] For low-fit fragments, implement a differentiated masking strategy based on their constraint factor type;

[0073] Finally, the labeling information of high and low adaptability segments is written into the trend tensor structure to support weight allocation and path avoidance processing in the subsequent rolling strategy selection and time series optimization process.

[0074] The priority matching weight is used by the control strategy generation module to assign resource allocation priority to the adjustment instructions of highly adaptable trend segments. Its weight value is generated by weighting the following factors: the adaptability score of the trend segment, which reflects the degree of matching of its adjustment capability; the adjustment urgency of the trend segment, which reflects the rate of change of disturbance and the magnitude of its impact; and the time proximity of the trend segment, which indicates the position of the trend segment relative to the current time.

[0075] The shielding strategy includes:

[0076] If the limiting factor is the energy storage SOC limit or insufficient power margin, then control commands will not be generated temporarily and the segment will be removed in the strategy optimization.

[0077] If the limiting factor is excessive response delay or conflicting adjustment directions, then a virtual load curve is used to compensate for the disturbance in this segment during simulation, and the segment structure is retained to participate in strategy optimization.

[0078] The compatibility score is specifically as follows:

[0079]

[0080]

[0081]

[0082]

[0083]

[0084] in, For fit rating, , , These are preset weighting coefficients for available capacity ratio, response matching degree, and response latency risk, respectively. This is the ratio of available capacity. In response to the matching degree, In response to the risk of delay, For maximum available charge / discharge power, Let i be the duration of the i-th segment. This represents the equivalent regulating energy caused by load disturbances within this trend segment. The value of the original load time series at time point t. The trend baseline value at time point t. The i-th segment is the principal component of the perturbation frequency. For the actual frequency response bandwidth of energy storage, This represents the system's maximum attention frequency. This represents the typical adjustment response time of an energy storage device under current conditions. This represents the duration of the disturbance within the trend segment.

[0085] S4, for highly adaptable segments, performs collaborative optimization matching with multiple objectives of minimizing the total life cycle cost and maximizing the grid regulation demand satisfaction rate, outputting a segmented energy storage configuration scheme, specifically:

[0086] For each highly fit segment, its trend perturbation features, adjustable capacity requirements, adjustment direction, and adjustment timeliness are extracted and encoded in multiple dimensions to form the target vector to be matched. ,in As a trend strength indicator, As an indicator of disturbance intensity, Adjust the capacity as needed. To adjust the direction (+1 for discharging, −1 for charging). To adjust for timeliness;

[0087] Based on the context-aware mechanism, the weight coefficients of response rate, regulation cost, energy utilization efficiency and operational risk in the matching evaluation indicators are dynamically adjusted by combining the pre-sequence and post-sequence features of the trend segment.

[0088] Using the target vector to be matched as input, multi-target matching and scoring are performed with the operating capacity vectors of multiple energy storage units in the preset energy storage type parameter library, and a set of candidate energy storage units that meet the regulation requirements is selected.

[0089] Based on the candidate energy storage unit set, a compatible regulation strategy template is retrieved from the preset set of operation strategies, and joint optimization is performed.

[0090] Finally, the optimal combination of energy storage unit and control strategy corresponding to each highly adaptable segment is output and written into the result domain of the trend tensor as the adaptation matching result of that trend segment.

[0091] The step of retrieving compatible adjustment strategy templates from a preset set of operating strategies based on a candidate energy storage unit set and performing joint optimization specifically involves:

[0092] Extract the dynamic capability constraint vector for each candidate energy storage unit, including the current state of charge, maximum generation / charging power, remaining adjustable capacity, minimum response time, and operating mode limitations;

[0093] Search the preset set of operating strategies for strategy templates that meet the capacity constraints of the energy storage unit, and remove invalid strategies due to directional conflicts, insufficient adjustment bandwidth or mismatched scheduling granularity.

[0094] For each combination of "energy storage unit + feasible strategy", a joint benefit function of adjustment benefits and costs is constructed. The function comprehensively considers the adjustment amplitude response quality, energy efficiency, strategy execution cost and cycle life impact factor.

[0095] Under this joint benefit function, multi-objective optimization is performed to select the strategy that maximizes the objective function value or minimizes the cost function as the optimal adjustment template for the energy storage unit.

[0096] Finally, all valid combinations in the candidate unit set are scored and ranked, and the globally optimal energy storage unit and its matching strategy are selected as the final adaptation output for this trend segment.

[0097] The joint benefit function is specifically as follows:

[0098]

[0099]

[0100]

[0101]

[0102]

[0103] in, For the joint benefit function, , , , The weighting coefficients are dynamically set according to the operational objectives to adjust them. To adjust the amplitude, rate, and target tracking quality of the response, To optimize the energy utilization efficiency of this energy storage unit, The cost of executing the strategy on this unit. This strategy may lead to potential losses in the cycle life of energy storage units. This represents the actual output power of the energy storage system under the execution of the strategy. The expected power value given by the forecast / demand side. To assess the total duration of the time period, To provide energy for regulation to the energy storage system. In this round of strategy, energy is injected into the power grid by energy storage systems. The startup costs incurred during the strategy preparation phase, This is due to the control resource consumption caused by strategy switching. This refers to the power change rate loss during strategy execution. The total number of charge-discharge cycles. Let be the magnitude of the state of charge change during the i-th charge-discharge cycle. The rated SOC range for the energy storage unit. This is a preset correction function for temperature and magnification factors.

[0104] S5 extracts feature tensors along the time series direction, generates dynamic adaptation curves, and provides feedback to correct long-term energy storage deployment plans. Specifically:

[0105] Using time series as an index, trend segments marked as highly fit and already matched in the feature tensor are extracted sequentially and then spliced ​​according to the time segment sliding window to form candidate rolling adjustment paths;

[0106] Based on the optimal energy storage-strategy combination of candidate rolling adjustment paths, the resource consumption intensity and performance degradation prediction in future adjustment cycles are extracted to construct a feasibility assessment map.

[0107] For a path consisting of multiple consecutive highly adaptable segments, its cumulative response load, resource consumption intensity, expected revenue and strategy switching frequency are evaluated based on the feasibility assessment map, and unsustainable paths are eliminated according to preset constraints.

[0108] The remaining feasible paths are pieced together along the timeline to form a complete dynamic rolling adaptation curve, reflecting the adjustment capability boundary, resource consumption trajectory and strategy response sequence of the highly adaptable adjustment segment in actual deployment.

[0109] Based on the structural characteristics of the rolling adaptation curve, and combined with the overall load evolution trend of the power supply area and the life cycle indicators of energy storage assets, the medium- and long-term deployment plan is dynamically adjusted. The deployment plan includes the priority of energy storage equipment selection, capacity configuration recommendations and strategy deployment plan.

[0110] The finalized rolling adaptation curve and its deployment recommendations are written into the system control task pool for subsequent execution modules to call and update.

[0111] The deployment plan includes energy storage equipment selection priorities, capacity configuration recommendations, and strategic deployment plans, specifically:

[0112] Prioritization of energy storage equipment selection:

[0113] Based on the distribution of adjustment characteristics of each trend segment in the rolling adaptation curve and the identified adjustment capability constraints, a selection requirement matrix is ​​constructed, including requirements for response speed priority, energy capacity priority, and bidirectional adjustment.

[0114] Then, the technical attribute vector matching is performed on the preset energy storage equipment type library, and a priority list of equipment types for adjustment tasks in different trend segments is generated by weighting the satisfaction rate and operating cost.

[0115] Capacity configuration recommendations generated:

[0116] Based on the cumulative adjustment energy, maximum load span, and typical disturbance amplitude of the historical high adaptability trend segment covered by the rolling adaptability curve, calculate the minimum available capacity range, recommended capacity range, and reserve margin factor for each type of equipment.

[0117] Simultaneously, taking into account the long-term load growth trend and planning scenarios of the power supply area, a capacity configuration suggestion table is output to support multi-phase rolling deployment;

[0118] Strategic deployment plan development:

[0119] Based on the adjustment cycle distribution of each trend segment, the frequency of strategy switching, and the current device lifecycle status, the execution frequency and time period adaptation of the matching strategy are evaluated.

[0120] Construct a Gantt chart for strategy implementation and set dynamic adjustment rules (such as strategy coverage overlap conflict resolution mechanism and strategy degradation replacement threshold), and output a strategy deployment plan table containing time, strategy ID, target energy storage unit and adjustment target.

[0121] Example 2, Figure 2A feasibility assessment diagram for rolling adaptation paths is presented. This diagram uses resource consumption intensity as the horizontal axis and response load as the vertical axis to present the core assessment dimensions of rolling adaptation paths. Green dots represent feasible paths (meeting constraints such as resource consumption and revenue), while red crosses indicate infeasible paths. The resource consumption threshold (0.8) and response load reference line are marked with dashed lines. Each path label simultaneously displays revenue and strategy switching frequency, intuitively reflecting the path's comprehensive performance in resource utilization, regulation efficiency, and economy. This provides visual support for energy storage strategy selection and capacity configuration, and assists in the accurate matching of energy storage potential based on time series and load characteristics.

[0122] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0123] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0124] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0125] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0126] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0127] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A rolling optimization configuration method for energy storage based on incremental distribution networks, characterized in that, Includes the following steps: The structural change point detection algorithm is used to identify trend abrupt change points in historical load time series and to construct an initial trend baseline. The load sequence is segmented based on the initial trend baseline, and the stable trend term and short-term disturbance term in each segment are extracted to construct a feature tensor. For each segment in the feature tensor, an adaptability assessment is conducted based on the key technical constraints limiting the adjustable potential of the energy storage system. Specifically, for each trend segment in the feature tensor, the trend slope, drift rate, disturbance amplitude, and disturbance frequency are extracted to construct a dynamic demand vector. Based on the temporal position of the trend segment, the operating status parameters of the energy storage system in the corresponding time period are obtained synchronously. These parameters include the energy storage charging and discharging power margin, state of charge, adjustment direction margin, response delay, and operating aging factor. Based on the dynamic demand vector and the set of operating status parameters, an adaptive matching model is constructed; the main technical constraints that limit the energy storage's adjustment capability in this trend segment are identified through the adaptive matching model; the adjustment demand of each trend segment is bound to the constraints to generate a trend segment-level map of adjustable potential constraints. And mark and mask the trend segments; For highly adaptable segments, a collaborative optimization matching is performed with multiple objectives of minimizing the total life cycle cost and maximizing the grid regulation demand satisfaction rate, and a segmented energy storage configuration scheme is output. Feature tensors are extracted by rolling along the time series direction, generating dynamic adaptation curves and providing feedback to correct long-term energy storage deployment plans.

2. The method for rolling optimization allocation of energy storage based on incremental distribution networks according to claim 1, characterized in that, The method involves identifying trend abrupt changes in historical load time series using a structural change point detection algorithm to construct an initial trend baseline. Specifically: Obtain load data for the power supply area within a historical window to form the original time series; Construct a sliding window of length t to traverse all the original time series, fit an autoregressive prediction model in each window and calculate the residual sequence; Structural change points are identified in the residual sequence using a cumulative sum change point detection algorithm; The original load sequence is divided into several segments based on the points of change. A trend function is independently fitted within each segment, and the segments are spliced ​​together to form a segmented continuous trend baseline.

3. The method for rolling optimization allocation of energy storage based on incremental distribution networks according to claim 2, characterized in that, The residual sequence is analyzed using an accumulation and change point detection algorithm to identify structural change points, specifically: The average value of the residuals is calculated based on the residual sequence, and the cumulative deviation sequence function is constructed. The inflection point of the cumulative deviation sequence function is detected. When the change in function value at the inflection point exceeds the preset fluctuation threshold, it is determined to be a structural change point.

4. The method for rolling optimization allocation of energy storage based on incremental distribution networks according to claim 3, characterized in that, The process involves segmenting the load sequence based on the initial trend baseline, extracting the stable trend term and short-term disturbance term within each segment, and constructing a feature tensor, specifically as follows: Based on the piecewise trend function fitted in the trend baseline, the load time series is divided into multiple trend-stable segments; Within each trend stabilization segment, a stable trend term is extracted based on the fitted trend function, and a perturbation sequence is generated by the difference between the original load sequence and the trend term. The perturbation sequence is decomposed into a multi-scale sliding window to extract the perturbation behavior features within the window. The perturbation behavior features include the perturbation dominance direction, perturbation density spectrum, perturbation energy intensity, and perturbation evolution path. By integrating disturbance behavior features with trend slope and trend drift rate, a multi-dimensional feature vector for a single trend segment is constructed. The multidimensional feature vectors corresponding to each trend segment are stacked in temporal order to form a feature tensor with a multidimensional nested structure.

5. The method for rolling optimization allocation of energy storage based on incremental distribution networks according to claim 4, characterized in that, The process of marking and masking trend segments specifically involves: Based on the adjustable potential limiting factor map of the trend segment, the adaptability score is calculated. The adaptability score is generated by weighting the degree of matching between the energy storage response capability and the trend disturbance demand. Trend segments with fitness scores higher than a preset fitness threshold are marked as high fitness segments and written into the label field of the trend tensor. For highly adaptable segments, priority matching weights are assigned during the strategy generation stage to generate priority adjustment control instructions; Trend segments with scores below the preset fit threshold are marked as low fit segments; For low-fit fragments, implement a differentiated masking strategy based on their constraint factor type; Finally, the labeling information of high and low fit segments is written into the trend tensor structure.

6. The method for rolling optimization allocation of energy storage based on incremental distribution networks according to claim 5, characterized in that, The highly adaptable segments are then collaboratively optimized and matched with multiple objectives, including minimizing the total lifecycle cost and maximizing the grid regulation demand satisfaction rate, to output a segmented energy storage configuration scheme, specifically: For each highly adaptable segment, extract its trend perturbation features, adjustable capacity requirements, adjustment direction and adjustment timeliness, and construct the target vector to be matched; Based on the context-aware mechanism, the weight coefficients of the matching evaluation indicators are dynamically adjusted by combining the features of the preceding and following sequences of the trend segment. Using the target vector to be matched as input, multi-objective matching and scoring are performed with the operating capability vectors of each energy storage unit in the preset energy storage type parameter library to filter the set of candidate energy storage units that meet the regulation requirements; Retrieve adjustment strategy templates that are compatible with the preset set of candidate energy storage units, perform joint optimization, and output the optimal combination of energy storage units and control strategies as the adaptation matching result, which is written into the result domain of the trend tensor.

7. The method for rolling optimization allocation of energy storage based on incremental distribution networks according to claim 6, characterized in that, The process involves retrieving adjustment strategy templates compatible with a preset set of candidate energy storage units and performing joint optimization, specifically: Extract the dynamic capability constraint vector of each candidate energy storage unit, and retrieve the strategy that satisfies the capability constraint of the energy storage unit from the preset set of operating strategies. Eliminate invalid strategies, including directional conflict, insufficient adjustment bandwidth or mismatch of scheduling granularity. For each combination of energy storage units and feasible strategies, a joint benefit function is constructed, and multi-objective optimization is performed to solve for the optimal regulation strategy.

8. The method for rolling optimization allocation of energy storage based on incremental distribution networks according to claim 7, characterized in that, The process of extracting feature tensors along the time series direction, generating dynamic adaptation curves, and feeding back to correct long-term energy storage deployment plans specifically involves: Using time series as an index, extract trend segments marked as highly adaptable and already matched, and stitch them together according to the time segment sliding window to form candidate rolling adjustment paths; Based on the optimal energy storage-strategy combination of candidate rolling adjustment paths, we obtain the predicted resource consumption intensity and performance degradation in future adjustment cycles and construct a feasibility assessment map. For a path consisting of multiple consecutive highly adaptable segments, its cumulative response load, resource consumption intensity, expected revenue and strategy switching frequency are evaluated based on the feasibility assessment map, and unsustainable paths are eliminated according to preset constraints. The remaining feasible paths are pieced together along the timeline to form a complete dynamic scrolling adaptation curve; Based on the structural characteristics of the rolling adaptation curve, and combined with the overall load evolution trend of the power supply area and the life cycle indicators of energy storage assets, the medium- and long-term deployment plan is dynamically adjusted. The finalized rolling adaptation curve and its deployment recommendations are written into the system control task pool.

Images