Power distribution network energy storage optimization scheduling method based on dynamic load prediction

Abstract

The invention relates to the technical field of smart power grids, in particular to a power distribution network energy storage optimization scheduling method based on dynamic load prediction, which comprises the following steps: acquiring power data, extracting a reversal node, calibrating a disturbance section, identifying a charge trend, screening a scheduling unit, constructing a scheduling link, identifying an intervention node by fusing voltage and current directions, and setting release limitation. And generating an energy storage release freezing instruction set. According to the method, the direction reversal point in the power data of the energy storage access node of the power distribution network is extracted, a trend sequence is constructed in combination with the change trend of the state of charge, a continuous descending unit is eliminated, synchronous matching of the energy storage trend and the disturbance behavior is realized in a time axis, and the power distribution network energy storage access node power data is obtained through combined screening of trend direction consistency and communication path conditions. An energy storage path map is constructed, the structural combination capability of energy storage resources is enhanced, the voltage fluctuation main vector and the current release auxiliary vector are fused, and the response capability of the energy storage resources of the power distribution network and the path regulation and control precision are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart grids, and particularly to a method for optimizing the dispatching of a distribution network energy storage based on dynamic load prediction. Background Art

[0002] The technical field of smart grids involves the deep integration of power systems and information and communication systems, aiming to achieve digital, automated, and visualized management of power systems. The core content of this technical field includes real-time monitoring and intelligent control of the entire process of power generation, power transmission, power distribution, and power consumption, and improves the flexibility and economy of grid operation through means such as data acquisition, two-way communication, state perception, load prediction, and resource coordination. Smart grids systematically integrate distributed energy access, energy storage regulation, power market mechanisms, and user-side interactive responses to form an energy Internet infrastructure for multi-agent, multi-energy, and multi-scenario collaborative operation.

[0003] Among them, the method for optimizing the dispatching of a distribution network energy storage based on dynamic load prediction refers to a power dispatching method that combines load change trend analysis and energy storage operation. This patent theme mainly aims at the problems of strong load volatility and unstable power supply and demand matching in the operation of the distribution network, covering matters such as load prediction modeling, analysis of the time series distribution of electricity, and formulation of charge and discharge strategies for energy storage devices. The method is to construct a load prediction model based on load data and time series characteristics to obtain the short-term dynamic load change trend, and then conduct power balance analysis and formulate a time-sharing energy storage plan according to the prediction results. Specifically, mathematical optimization methods are used to allocate the charge and discharge time sequence of the energy storage to meet the grid power regulation requirements and local power supply and demand balance requirements.

[0004] Although the existing technology can achieve short-term trend judgment through a load prediction model and formulate an energy storage plan, it lacks a node-level mutation recognition mechanism when dealing with sudden disturbances, which is likely to cause delays or misjudgments in abnormal load responses. The identification of the state of charge stays at the mean value or maximum capacity judgment within a cycle, and fails to use the state change trend as the preferred basis for dispatching, making it difficult to dynamically respond to the current energy storage characteristics. In terms of energy storage path organization, most use fixed node weight modeling, ignoring the joint evaluation of trend consistency and structural connection ability, and the path construction lacks systematicness and scalability. Voltage and current characteristics are only used as result monitoring indicators, and no active discrimination parameter combination is formed, making it difficult to screen intervention targets before dispatching. In actual operation, problems such as energy storage response mismatch, continuous voltage fluctuation, and ineffective current release occur, reducing the flexible regulation ability of the grid and the response efficiency of energy storage resources. Summary of the Invention

[0005] In order to solve the technical problems existing in the prior art, such as the lack of a node-level mutation recognition mechanism when dealing with sudden disturbances, which easily leads to abnormal response delays or misjudgments of the load. The recognition of the state of charge stays at the average value within the cycle or the maximum capacity judgment, and fails to use the state change trend as the preferred basis for scheduling, making it difficult to dynamically respond to the current energy storage characteristics. In terms of the organization of the energy storage path, it mostly models with fixed node weights, ignoring the joint evaluation of trend consistency and structural connection ability, and the path construction lacks systematicness and scalability. The voltage and current characteristics are only used as result monitoring indicators, and no parameter combination for active discrimination is formed, making it difficult to screen intervention targets before scheduling. In actual operation, problems such as energy storage response mismatch, continuous voltage fluctuations, and ineffective current release occur, reducing the flexible regulation ability of the power grid and the response efficiency of energy storage resources. The present invention provides an optimized scheduling method for distribution network energy storage based on dynamic load prediction. The technical solution is as follows: On the one hand, an optimized scheduling method for distribution network energy storage based on dynamic load prediction is provided, and the method includes: S1: Collect the power data of consecutive time slices of the energy storage access nodes in the distribution network, detect the power change direction of adjacent time slices, extract the direction reversal points as mutation candidate nodes, calibrate the starting node after comparing and judging the mutation recognition reference direction, locate the end node of the reversal, and generate a load disturbance section sequence; S2: Based on the load disturbance section sequence, mark the time window, extract the state of charge records of the energy storage access nodes in the distribution network, compare the state of charge change directions of adjacent cycles in turn, judge whether the trend direction shows a continuous decline, and eliminate the continuously declining units to generate an energy storage unit list; S3: Call the energy storage unit list, map the state of charge trend sequence and the load disturbance section sequence on the time axis, according to the connection relationship of the energy storage nodes in the distribution network, retrieve the unit combinations with path connection ability and consistent trend directions, and sequentially construct a scheduling link in the connection order between nodes to generate an energy storage path map; S4: Based on the energy storage path map, calculate the main vector composed of the fluctuation direction and the fluctuation rate, combine the current release direction of the energy storage unit and the state of charge change direction to construct an auxiliary vector, and generate an intervention node trend set.

[0006] As a further solution of the present invention, the load disturbance section sequence includes the disturbance starting point position, the disturbance end point number, and the number of disturbance direction changes. The energy storage unit list includes the energy storage node number, the state of charge trend direction, and the average state of charge within the cycle. The energy storage path map includes the node connection order, the trend consistency identifier, and the path scheduling priority sequence. The intervention node trend set includes the main vector of voltage change, the auxiliary vector of current release, and the node trend aggregation label.

[0007] As a further solution of the present invention, the specific steps for obtaining the load disturbance section sequence are as follows: S101: Collect the power data of consecutive time slices at the energy storage access nodes of the distribution network, detect the power values between adjacent two time slices, construct a power change direction sequence according to the difference signs of the front and back power values, screen the points where the change direction changes from positive to negative or from negative to positive as the direction reversal points, extract them as mutation candidate nodes, and generate a mutation candidate node sequence; S102: Based on the node positions in the mutation candidate node sequence, call the power values of the two time slices before and after the node, calculate the forward change amount and the backward change amount respectively, compare the two with the node change direction, and determine the nodes whose change amplitude direction is consistent with the original change direction as the starting nodes for identifying the reference direction, and obtain the direction reference positioning node set; S103: According to the direction reference positioning node set, detect whether the power change directions of adjacent nodes show continuous reversals. If the number of continuous reversals exceeds the set reversal number, it is positioned as the disturbance end node, mark the time slices between the starting node and the disturbance end node, and obtain the load disturbance section sequence.

[0008] As a further solution of the present invention, the steps for obtaining the energy storage unit list are specifically as follows: S201: Based on the intervals marked in the load disturbance section sequence, extract the charge state records of the energy storage access nodes of the distribution network within the corresponding time periods, arrange the charge state values of the nodes in chronological order, and obtain the energy storage node charge state trend data; S202: Call the charge state values of adjacent cycles in the energy storage node charge state trend data, compare each group of values in sequence, record the change direction marks and connect them to form a trend change sequence, identify the sections where the charge state values change continuously, and obtain the rising trend section time slice set; S203: According to the time positions corresponding to the rising trend section time slice set, screen the energy storage access node numbers with a rising charge state trend, sort out the qualified nodes and perform deduplication and summarization to generate an energy storage unit list.

[0009] As a further solution of the present invention, the steps for obtaining the energy storage path map are specifically as follows: S301: Call the energy storage unit list, correspond the corresponding charge trend sequence with the load disturbance section sequence in the time axis position, calculate the peak interval state value of the energy storage unit, screen the energy storage units whose charge state shows an upward trend and the charge state value is in the peak interval within the node time period during the disturbance time slice, and obtain the peak interval trend energy storage node set; S302: Based on the charge state trend sequence of the nodes in the peak interval trend energy storage node set, count the number of time slices of the continuous upward trend of the nodes, and number and mark the nodes in descending order according to the trend duration to obtain the trend sorting energy storage node sequence; S303: Sort each pair of nodes in the energy storage node sequence according to the said trend, retrieve the node pairs with a path connection relationship in the distribution network structure, determine whether the trend directions of the nodes on the connection path are consistent, and if they are consistent, arrange them in sequence according to the connection order in the network path to generate an energy storage path map.

[0010] As a further solution of the present invention, the calculation of the peak interval state value of the energy storage unit adopts the formula: ; wherein, represents the state value of the energy storage unit in the peak interval, represents the state of charge value of the th time slice, represents the state of charge value of the previous time slice, represents the time interval, represents the load disturbance value of the th time slice, is the minimum value, is the total number of time slices.

[0011] As a further solution of the present invention, the specific steps for obtaining the intervention node trend set are as follows: S401: Based on the timing information corresponding to the node paths in the energy storage path map, combined with the real-time voltage values of the nodes within the time period, calculate the ratio of the voltage value difference to the time interval in consecutive time slices, and combine it with the voltage change direction in the adjacent time period to construct a main vector, and obtain a set of node main vectors; S402: According to the data of the current release direction and the state of charge change direction of the nodes in the same time period in the set of node main vectors, construct an auxiliary vector, and screen the nodes that simultaneously meet the conditions that the voltage rate is in an increasing state, the voltage value is continuously in a descending section, and the current continuous release direction is consistent to generate an intervention node trend set.

[0012] As a further solution of the present invention, the calculation of the ratio of the voltage value difference to the time interval in consecutive time slices adopts the formula: ; wherein, represents the ratio of the voltage difference between node and node to the time interval within the time period, and respectively represent the voltage values of node and node at time points and , represents the time interval between node and node , represents the voltage change direction within the th time period, and represents the total number of voltage change directions.

[0013] As a further aspect of the present invention, the method further includes step S5: S5: Invoke the intervention node trend set, adjust the standard of the release curve to set the release change limit, perform amplitude limiting processing on the current release rate, and at the same time perform a scheduling behavior direction locking operation on the energy storage units that form a connected relationship with the nodes in the energy storage path, generating an energy storage release freeze instruction set; The energy storage release freeze instruction set includes a release rate limit value, a path direction locking number, and a freeze response instruction code.

[0014] As a further aspect of the present invention, the obtaining step of the energy storage release freeze instruction set is specifically as follows: S501: Based on the intervention node trend set, extract the release rate change value of the node within the time period, and invoke the release change limit interval in the release curve adjustment setting. Compare the node change value with the upper and lower limits of the interval, and analyze the adjustment range of the release curve by determining whether it exceeds the upper and lower boundaries, generating a release change determination trend value; S502: According to the release change determination trend value, perform amplitude limiting processing on the real-time node release current rate, extract the release amplitude parameter corresponding to the rate value, and perform interval clipping operation with the boundary values of the release change limit interval to update the rate value of the release current, generating a node release limit rate; S503: Invoke the node release limit rate, match the energy storage units that form a connected relationship with the real-time intervention node in the energy storage path, compare the scheduling direction value of the energy storage unit with the direction parameter in the constraint release rate set, lock the scheduling direction and write the direction value into the freeze instruction field, generating an energy storage release freeze instruction set.

[0015] The beneficial effects brought by the technical solution provided by the embodiments of the present invention at least include: By extracting the direction reversal points in the power data of the energy storage access nodes in the distribution network and locating the load disturbance boundary based on amplitude judgment and reversal continuity, the boundary recognition and dynamic modeling of the load mutation behavior are realized, the positioning accuracy of the disturbance response is improved, a trend sequence is constructed in combination with the change trend of the state of charge, the continuously decreasing units are excluded, and only the rising trend section is retained as the scheduling object to achieve the synchronous matching of the energy storage trend and the disturbance behavior within the time axis. Through the joint screening of the trend direction consistency and the connected path conditions, an energy storage path map is constructed to enhance the structural combination ability of the energy storage resources. By fusing the main vector of voltage fluctuation and the auxiliary vector of current release, the energy storage nodes that meet the three characteristic conditions are identified, and the key regulation positions are further screened. Combining the release rate limit and the behavior direction locking mechanism, the behavior boundary and intensity adjustment criteria within the path are set, and the energy storage state, power disturbance, voltage feedback and scheduling link are connected through the trend behavior as a clue, and a linkage regulation mechanism is constructed from the dynamic behavior evolution to significantly improve the response ability of the distribution network energy storage resources and the accuracy of path regulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a schematic diagram of the working process of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0017] The technical solutions in the present invention will be described below with reference to the accompanying drawings.

[0018] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "example" in the present invention should not be construed as being more preferred or more advantageous than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" may be both, or either one of the two.

[0019] To make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0020] Please refer to Figure 1 , the embodiments of the present invention provide an optimized scheduling method for distribution network energy storage based on dynamic load prediction, and the processing flow of the method may include the following steps: S1: Collect the power data of consecutive time slices of the energy storage access nodes in the distribution network, detect the power change direction of adjacent time slices, extract the direction reversal points as the mutation candidate nodes, call the power change amplitude within the time slices before and after the candidate nodes, calibrate the starting node after comparing and judging the mutation recognition reference direction, detect whether the change direction appears continuous reversal, locate the reversal end node, and generate a load disturbance section sequence; S2: Based on the load disturbance section sequence identification time window, extract the state of charge records of the energy storage access nodes in the distribution network, compare the state of charge change directions in adjacent cycles in sequence and form a trend sequence, determine whether the trend direction shows continuous decline and eliminate the continuously declining units, retain the units with rising trends as the objects participating in scheduling, and generate an energy storage unit list; S3: Call the energy storage unit list, perform time-axis mapping on the charge trend sequence and the load disturbance section sequence, screen the energy storage units that show an upward trend and whose state of charge is in the peak section within the disturbance section, mark them according to the trend persistence, and retrieve the unit combinations with path connection capabilities and the same trend direction according to the connectivity relationship of the energy storage nodes in the distribution network. Build a scheduling link in sequence according to the connection order between nodes to generate an energy storage path map; S4: Based on the energy storage path map, combine the real-time voltage change data, calculate the main vector composed of the fluctuation direction and the fluctuation rate, and construct an auxiliary vector by combining the current release direction of the energy storage unit and the state of charge change direction. Identify the nodes that meet the conditions of all three items of voltage rate increase, continuous voltage drop, and continuous current release to generate an intervention node trend set; S5: Call the intervention node trend set, set the release change limit according to the release curve adjustment standard, perform amplitude limiting processing on the current release rate, and at the same time perform a locking operation on the scheduling behavior direction of the energy storage units that form a connectivity relationship with the nodes in the energy storage path to generate an energy storage release freeze instruction set; The load disturbance section sequence includes the disturbance starting point position, the disturbance ending point number, and the number of disturbance direction changes. The energy storage unit list includes the energy storage node number, the charge trend direction, and the average charge within the cycle. The energy storage path map includes the node connection order, the trend consistency identifier, and the path scheduling priority sequence. The intervention node trend set includes the voltage change main vector, the current release auxiliary vector, and the node trend aggregation label. The energy storage release freeze instruction set includes the release rate limit value, the path direction locking number, and the freeze response instruction code.

[0021] The specific steps for obtaining the load disturbance section sequence are as follows: S101: Collect the continuous time-slice power data of the energy storage access nodes in the distribution network, detect the power values between adjacent two time slices, and construct a power change direction sequence according to the difference signs of the front and back power values. Screen the points where the change direction changes from positive to negative or from negative to positive as the direction reversal points, and extract them as mutation candidate nodes to generate a mutation candidate node sequence; With the help of automated monitoring, such as distribution automation terminals (DTUs) or edge computing units, the node power is sampled and recorded in real time, and the data is stored in a structured manner at fixed intervals. The sampling interval is set to every 5 minutes. The power of a certain node within 24 hours is recorded as 288 consecutive time slices, forming a complete power time series. The power values of adjacent time slices are extracted in sequence and the difference is judged. Starting from the first time slice, the power change between every two consecutive time slices is calculated one by one. For each pair of time slices, set the 10th and 11th slices, read their power values respectively, and calculate the magnitude and sign of the difference, and judge whether it is rising (positive), falling (negative) or flat (zero). Traverse the entire power sequence continuously in this way, record the power change direction at each time point, generate a direction sequence representing the change trend, identify the node positions in the generated direction sequence where the previous time slice is positive and the subsequent time slice is negative, or the previous is negative and the subsequent is positive, and extract the node positions where such change direction turns. For example, if the power of a certain node rises from 2 kW to 2.5 kW at 2 pm and then drops to 2.2 kW at 2:05 pm, then 2:05 pm is an obvious power change reversal point, which is used as a mutation candidate node, and the mutation candidate node sequence is obtained by summarizing them one by one.

[0022] S102: Based on the node positions in the mutation candidate node sequence, call the power values of the two time slices before and after the node, and calculate the forward change amount and the backward change amount respectively. Compare the two with the node change direction, and judge the nodes whose change amplitude direction is consistent with the original change direction as the starting nodes for identifying the reference direction, and obtain the set of direction reference positioning nodes; Call the power values of the two time slices before and after each candidate node in turn, and calculate the differences from the power of the current node respectively, forming the forward change amount and the backward change amount, to compare whether their change trends are consistent. Read the power P(t) of each candidate node at time t, and the power P(t - 1) at the previous moment and the power P(t + 1) at the next moment of time t, calculate the differences between P(t) and the values on both sides respectively, and judge the change direction according to the positive and negative of the difference. If it is found that the forward and backward powers of the node are both increasing or decreasing trends, it means that the change direction at the node has the characteristic of continuity and can be further recognized as the starting reference point for direction recognition. For example, the power of a certain energy storage node is 4.0 kW at 3 pm, 3.5 kW at 2:55 pm, and 4.3 kW at 3:05 pm. It is found that both the forward and backward are positive changes, and it is judged that this node is a key starting node in the stable rising trend. Conduct similar analyses on the entire mutation candidate sequence, judge the consistency of the change trend of each node one by one, eliminate the nodes with discontinuous change directions or excessive fluctuations, and summarize and screen out a set of representative direction reference positioning node sets.

[0023] S103: Locate the node set according to the direction reference, and detect whether the power change direction of adjacent nodes shows continuous reversal. If the number of continuous reversals exceeds the set reversal times, the node is located as the disturbance end node, mark the time slices between the starting node and the disturbance end node, and obtain the load disturbance section sequence; Continuously judge the power change direction between any two adjacent nodes, mainly checking whether there is a frequent reversal phenomenon of the power direction. Traverse the direction reference nodes one by one. Starting from the first node, detect whether the power change trends of each pair of adjacent nodes are opposite backward. If it is found that the number of continuous alternating changes in direction reaches or exceeds the preset number threshold, such as 3 times, it is considered that the time when the current node is located no longer continues the initial disturbance trend, and mark it as the disturbance end node. Combine and judge the time slices between the starting node and the end node, and uniformly mark them as a complete disturbance section. If the power continuously rises from 1 pm and turns down at 1:20 pm, rises again at 1:25 pm, and turns down again at 1:30 pm, it can be recorded as 3 consecutive direction reversals. Combining the set reversal threshold of 3 times, it is determined that 1:30 pm is the disturbance end point, and the power records between the starting 1 pm and the end 1:30 pm will be uniformly identified as a disturbance section. During the whole process, the maximum allowable interval between each reversal can also be set, such as the time difference between reversal points shall not exceed 10 minutes, to avoid misidentifying fluctuations with too large intervals as continuous disturbance phenomena, ensure the accuracy and continuity of the disturbance section division, and obtain the load disturbance section sequence.

[0024] The specific steps for obtaining the energy storage unit list are as follows: S201: Based on the intervals marked by the load disturbance section sequence, extract the state of charge records of the distribution network energy storage access nodes within the corresponding time period, arrange the state of charge values of the nodes in chronological order, and obtain the state of charge trend data of the energy storage nodes; It is necessary to extract the state of charge data within the corresponding time period of the section from the real-time operation records of the energy storage access node, retrieve the time boundary information within the disturbance section. If a certain disturbance section is defined as from 8:00 to 8:45, then it is necessary to query the SOC (State of Charge) data records of relevant energy storage nodes within this time range. When extracting, it is required that the data has continuity and timestamp integrity to ensure that it can be accurately corresponding to each time slice within the disturbance section. After that, arrange the state of charge data of each node in ascending order of timestamp to form a state of charge time series indexed by time. The SOC value at each time point constitutes the energy storage data of the node. If the SOC values recorded by energy storage node A at 8:00, 8:05, and 8:10 are 40%, 42%, and 43% respectively, then the state of charge trend of this node within the disturbance section is 40%-42%-43%. During the whole process, it is necessary to ensure the synchronous processing of the time slices of the nodes and not disrupt the continuity of the trend sequence due to missing values. If there are data gaps, they can be repaired by linear interpolation or time window filling methods. After processing the disturbance section and the corresponding nodes, store the data of each node within each disturbance section independently and mark the corresponding node number and time period number to obtain the state of charge trend data of the energy storage node.

[0025] S202: Call the state of charge values of adjacent cycles in the state of charge trend data of the energy storage node, compare each group of values successively, record the change direction markers and connect them to form a trend change sequence, identify the sections where the state of charge values change continuously, and obtain the set of time slices of the upward trend section; Read the SOC values of adjacent cycles in each trend sequence one by one, and perform a sequential comparison operation on each pair of values. Read the SOC values of every two consecutive time slices. Set the SOC value at time t1 to 41% and at time t2 to 42%. Determine that its change direction is upward and record it as "+1". If t2 is lower than t1, record it as "-1", and if they are equal, record it as "0". Such marks are concatenated item by item to form a trend change sequence, which is used to reflect the change pattern of SOC during the perturbation. Set the SOC change of node B within a certain perturbation segment to 41% - 42% - 43% - 42%, then its trend change sequence is "+1, +1, -1". Traverse the trend change sequence to identify the segments that are continuously "+1", indicating that the SOC of this node is in a continuous upward state. The time indices corresponding to the segments will be extracted as the time slice set of the upward trend section. To improve the recognition accuracy, a minimum length threshold for continuous segments can be set, such as the continuous upward time should not be less than two cycles (i.e., at least 10 consecutive minutes), to filter out false upward trends caused by single charging fluctuations. Set that if the SOC of node C is 45%, 46%, 47%, 48% between 8:00 and 8:30, then this segment can be used as an upward trend section, and the time points 8:00, 8:05, 8:10, 8:15 will be added to the time slice set for subsequent node screening and determination, obtaining the time slice set of the upward trend section.

[0026] S203: According to the time positions corresponding to the time slice set of the upward trend section, screen the numbers of energy storage access nodes with an upward state of charge trend, organize the qualified nodes and perform deduplication and summarization to generate a list of energy storage units; Screen out the numbers of energy storage access nodes that show a continuous upward trend in the state of charge within this segment, establish a mapping relationship between time slices and nodes, mark the nodes that show continuous charging behavior within each upward trend time period, traverse the time slice set, retrieve the energy storage nodes with a complete upward sequence correspondingly, confirm that there is continuous charging behavior during the entire perturbation period, and add the node numbers to the candidate list. To avoid redundant records, the energy storage node numbers will be deduplicated after the section analysis is completed to ensure that each node only appears once in the list. If node D has an upward SOC trend at multiple time points within the perturbation segment from 9:00 am to 9:20 am and meets the upward determination criteria after verification, then record the number of node D and output it in a unified format of node numbers for subsequent control strategies, charge and discharge scheduling, or energy optimization, ensuring the integrity and operability of the energy storage unit identification process, and generating a list of energy storage units.

[0027] The specific steps for obtaining the energy storage path map are as follows: S301: Call the energy storage unit list, align the corresponding state-of-charge trend sequence with the load disturbance section sequence on the time axis, calculate the state value of the energy storage unit in the peak interval, filter out the energy storage units whose state of charge shows an upward trend and the state-of-charge value is within the peak interval during the disturbance time slice, and obtain the peak interval trend energy storage node set; Calculate the state value of the energy storage unit in the peak interval using the formula: ; where, represents the state value of the energy storage unit in the peak interval, represents the state-of-charge value at the th time slice, represents the state-of-charge value of the previous time slice, represents the time interval, represents the th load disturbance value at the time slice, is a minimum value, is the total number of time slices; Parameter meaning and formula calculation derivation process: State-of-charge change : Represents the absolute value of the difference between the state-of-charge value at the th time slice and the state-of-charge value of the previous time slice; Obtaining method: By real-time monitoring the state of charge of the energy storage unit, obtain the state-of-charge value of each time slice ; Set the state-of-charge value at the th time slice as , and the state-of-charge value of the previous time slice as , then the state-of-charge change is: ; Time interval : Represents the time interval between two consecutive time slices; Obtaining method: Determine the time interval according to the sampling frequency. If sampling is done once per minute, the time interval is 1 minute, minute Load disturbance value : Represents the load disturbance value at the th time slice; Obtaining method: By real-time monitoring the load change of the distribution network, obtain the load disturbance value of each time slice ; Set the load disturbance value at the th time slice as ; Minimum value to prevent division by zero : Represents a very small constant used to prevent division by zero in calculations; Obtaining method: Set it as a very small constant, such as ; Total number of time slices : Represents the total number of time slices in the time series; Obtaining method: Determine the total number of time slices according to the monitored time range and time interval. If the monitoring time is 24 hours and the sampling is done once per minute, then the total number of time slices is: ; Set to sample once per minute within a certain time period, with a total of 5 time slices. The state of charge values and load disturbance values are shown in the following table: Table 1 State of charge values and load disturbance value data ; As shown in Table 1, the state of charge values and load disturbance values for each time slice are given; Based on the above data, calculate the state of charge change amount, time interval, and load disturbance value for each time slice: Time slice ; Time slice ; Time slice ; Time slice ; Time slice ; Substitute the above data into the formula to calculate the contribution value for each time slice: Time slice 1: ; Time slice 2: ; Time slice 3: ; Time slice 4: ; Time slice 5: ; Add up the contribution values of all time slices: ; This result shows that the state value of the energy storage unit in the peak interval is 0.5706, indicating that during this time period, the state of charge change trend of the energy storage unit is relatively obvious, and the influence of load disturbance is small, which is suitable for energy storage optimal scheduling.

[0028] S302: Based on the state-of-charge trend sequence of the concentrated nodes of the energy storage nodes in the peak interval trend, count the number of time slices of the continuous upward trend of the nodes, and number and mark the nodes in descending order according to the trend duration to obtain the trend-sorted energy storage node sequence; Further analyze the state-of-charge trend sequence of each node, count the number of time slices covered by the continuous upward trend therein, traverse the trend sequence of each node, record the start and end positions of the continuous upward segment, and count the total number of time slices with continuous "upward" marks. Set that if a certain node has 4 time slices showing continuous upward between 10:00 and 10:30, it is considered that its trend duration is 4 slices. After completing the node statistics, sort the nodes according to the duration of their upward trends. The sorting rule is that the longer the duration, the higher the ranking. Use the numbering method to assign sorting marks to the nodes. If node B has 6 continuous time slices and node C has 4 continuous time slices, then node B is numbered 1 and node C is numbered 2, and so on. After sorting, the information such as the number, node ID, and trend length of each node will be formed into a structured record for further network analysis. To ensure the fairness of sorting, unify the time slice interval and continuous judgment criteria of each node before sorting. Set that at least two continuous time slices are counted into the trend length to exclude short-term jitters or non-continuous upward trends, and ensure the stability and reliability of the statistical results, and obtain the trend-sorted energy storage node sequence.

[0029] S303: For each pair of nodes in the trend-sorted energy storage node sequence, retrieve the node pairs with path connection relationships in the distribution network structure, and judge whether the trend directions of the nodes on the connection path are the same. If they are the same, arrange them in sequence according to the connection order in the network path to generate the energy storage path map; Retrieve the physical connection relationships of each energy storage node in the network through the distribution network GIS or the master station SCADA, establish the path information on whether there is a direct or indirect electrical connection between the nodes. For the node pairs with path connections, retrieve the intermediate nodes included in the connection path one by one to obtain the SOC trend direction information of the nodes in the disturbance section. The judgment criterion is whether the state-of-charge change direction of the nodes on the path in the corresponding time slice is the same as that of the start and end nodes of the path. Set that if both the start and end points show an upward trend, and each node on the path also shows an upward SOC trend in the corresponding time slice, it is considered that the path trend is the same. At this time, arrange the node numbers in the order of the power grid path to form a trend-consistent path chain. For example, if nodes 1, 2, 3, and 4 form a continuous path in physical connection, and these four nodes all show continuous upward SOC trends between 10:00 and 10:30, then record the path as "1→2→3→4", which is used as a trend link. The entire map consists of multiple trend-consistent paths, and each path represents the energy transfer or charging synchronization chain formed by the energy storage nodes during the disturbance period, providing a structural basis for subsequent scheduling or coordinated control strategies, and generating the energy storage path map.

[0030] The steps for obtaining the intervention node trend set are specifically as follows: S401: Based on the timing information corresponding to the node paths in the energy storage path map, combined with the real-time voltage values of the nodes within a time period, calculate the ratio of the voltage value difference to the time interval in consecutive time slices, and combine it with the voltage change direction in the adjacent time period to construct a main vector, and obtain the node main vector set; Calculate the ratio of the voltage value difference to the time interval in consecutive time slices using the formula: ; where, represents the node and the node the ratio of the voltage difference to the time interval between them within the time period, and respectively represent the nodes and the node at the time point and the voltage values, represents the node and the node the time interval between them, represents the th time period of the voltage change direction, represents the total number of voltage change directions; Parameter meaning and formula calculation derivation process: represents the node and the node the ratio of the voltage difference to the time interval between them, with the unit of volts per second, and this parameter is obtained by monitoring the voltage changes of the node in different time periods; and respectively represent the nodes and the node at the moment and the voltage values, with the unit of volts, and the voltage values are obtained through real-time monitoring by voltage sensors; represents the node and the node the time interval between them, with the unit of seconds, and this time interval is obtained by synchronizing the clocks to record the difference between two time points; represents the th time period of the voltage change direction, with the unit of dimensionless, and the voltage change direction is judged by calculating the sign of the voltage change within this time period. If the voltage changes from positive to negative, then , otherwise ; is the number of time periods, representing the total number of voltage change directions, with the unit of "number". This value is quantified according to the analysis time window and actual monitoring data; Obtaining the specific value: , the voltage value is read through a voltage sensor; seconds, the time interval is obtained through a synchronous clock system; , set to select 3 time periods to calculate the voltage change direction; , deduced according to the voltage change direction in different time periods; Formula calculation and derivation: Calculate the voltage difference: ; Calculate the time interval: ; Calculate the voltage change direction and weighted summation: ; Substitute into the formula for calculation: ; This result indicates that between node and node , the voltage change rate is 0.48 volts per second, indicating that within 5 seconds, the change rate of the voltage difference relative to the time interval is 0.48 volts per second. This value is the key calculation result of this step and will be used as the basis for constructing the node main vector set later.

[0031] S402: According to the current release direction and charge state change direction data of nodes in the same time period in the node main vector set, construct an auxiliary vector, screen the nodes that simultaneously meet the conditions of the voltage rate being in an increasing state, the voltage value continuously being in a decreasing section, and the current continuously releasing in the same direction, and generate an intervention node trend set; Analyze the current release direction and the change direction of the state of charge of the analysis node within the same time period, construct an auxiliary vector based on this, extract the current flow data of the node within the time period corresponding to the main vector, judge whether the current release direction is outward (discharging) or inward (charging), and combine the change trend of the state of charge (SOC). Set that a decrease in SOC indicates that electrical energy is being released. Integrate the voltage change direction, current release direction, and SOC change trend into an auxiliary vector. Among them, the voltage needs to continuously be in the descending section, and the voltage change rate is in an increasing state (that is, the descending rate accelerates or remains the same), the current release direction remains consistent, and when the SOC shows a downward trend, this node is considered to meet the intervention condition. If the voltage of node B drops from 222V to 216V from 10:00 to 10:30, and the change rate gradually accelerates; the current always flows to the load end and the value is stably released, and the SOC drops from 90% to 75%, then it is considered that node B has typical intervention trend characteristics, and record its auxiliary vector as (↓ accelerating, out-flow, SOC decreasing). Traverse the set of node auxiliary vectors, and screen the nodes that simultaneously meet the above three conditions. To avoid misjudgment, judgment criteria such as a voltage drop rate threshold of 0.2V / min and a current release direction duration threshold of 15 minutes can be set to ensure that the selected nodes have data sufficiency and behavior consistency in intervention determination, which is convenient for the precise positioning of subsequent control strategies and the input of scheduling strategies, and generate an intervention node trend set.

[0032] The specific steps for obtaining the energy storage release freeze instruction set are as follows: S501: Based on the intervention node trend set, extract the change amount value of the release rate of the node within the time period, and call the release change limit interval in the release curve adjustment setting. Compare the node change amount value with the upper and lower limits of the interval, and analyze the adjustment range of the release curve by judging whether it exceeds the upper and lower boundaries, and generate a release change determination trend value; Deeply analyze the release behavior of the intervention node within a specific period, extract the value of the change in its release rate, read the current release rate of adjacent time slices from the node operation data, calculate the rate difference between the current time slice and the previous time slice, and form a release rate change sequence. If the release rate of the node is 15 A / min at 10:00 and 18 A / min at 10:05, the change in the release rate is +3 A / min, indicating an increase in the release rate. Retrieve the preset release curve adjustment setting parameters, which are defined by the engineer based on the device's carrying capacity and operational safety, including a release rate change limit range set as [-5 A / min, +5 A / min]. Compare the rate change value of each node with the upper and lower boundaries of this limit range to determine whether it exceeds the boundary. If the release rate change exceeds the upper limit of +5 A / min, it is regarded as a too-fast release; if it is lower than the lower limit of -5 A / min, it is regarded as too low a release rate or reverse recovery. Accordingly, generate a release change determination trend value for each intervention node, marked in symbolic form, such as "+" indicating exceeding the upper limit, "-" indicating being lower than the lower limit, and "0" indicating that the change amount is within the range. The change determination trend values of the intervention nodes form structured output data for the next amplitude limiting process operation to generate the release change determination trend value.

[0033] S502: Perform amplitude limiting processing on the real-time node release current rate according to the release change determination trend value, extract the release amplitude parameter corresponding to the rate value, and perform interval clipping operation with the boundary values of the release change limit range to update the rate value of the release current and generate the node release limit rate; Perform amplitude limiting processing on the current real-time current release rate. This processing process uses the release rate as the basic data and ensures operational safety by limiting its change amplitude. Identify the difference between the actual release rate of each node in the current time slice and the rate in the previous time slice, extract this difference as the "release amplitude parameter", compare this parameter with the release change limit range, and perform interval clipping operation to ensure that the change rate does not exceed the set boundary. Set that if the release rate of a certain node increases from 20 A / min to 28 A / min, the release amplitude is +8 A / min, exceeding the set upper limit of +5 A / min, then update it to 25 A / min, forcing the rate to be clipped within the allowable range. If the release rate decreases from 14 A / min to 10 A / min, with a change of -4 A / min, no adjustment is made within the allowable range. The amplitude-limited current release rate will be updated and written into the operation parameters of the current node to control the release behavior of the node during the current period, ensuring that the stability is not affected by rate mutations. The entire processing flow runs independently for each node and retains the original rate and adjusted rate records for subsequent scheduling or evaluation analysis to form the node release limit rate.

[0034] S503: The calling node releases the limit rate, matches the energy storage units in the energy storage path that form a connected relationship with the real-time intervention node, compares the scheduling direction value of the energy storage unit with the direction parameters in the constraint release rate set, locks the scheduling direction and writes the direction value into the freeze instruction field to generate an energy storage release freeze instruction set; Perform an association process with the energy storage path graph, match the energy storage units in the path that form a direct or indirect connected relationship with the intervention node, call the path structure diagram, identify the energy storage units in the same path as the intervention node, and extract their current scheduling direction values, i.e., the charge and discharge directions, mark them as "in" or "out", compare this direction value with the direction parameters in the node release limit rate to determine if they are consistent. Set the limit rate direction to release (out). If the scheduling direction value is also "out", they are consistent. If the directions are inconsistent, it indicates that there is a deviation in the scheduling direction, and the scheduling instruction needs to be corrected or frozen. For the consistent nodes, write their scheduling directions into the "freeze instruction field" to ensure that the energy storage unit maintains its existing direction unchanged during the current scheduling cycle, avoiding the instability caused by frequent switching. The freeze instruction field will correspond one by one with the nodes in the energy storage path, including parameter information such as node numbers, freeze directions, upper and lower rate limits, etc., providing a constraint basis for the subsequent execution layer to issue scheduling commands, ensuring the continuity and consistency of the scheduling behavior during the disturbance response period, and forming an energy storage release freeze instruction set.

[0035] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for optimizing the dispatching of energy storage in a distribution network based on dynamic load prediction, characterized in that, It includes the following steps: S1: Collect the power data of consecutive time slices of the energy storage access nodes in the distribution network, detect the power change direction of adjacent time slices, extract the direction reversal points as mutation candidate nodes, calibrate the starting node after comparing and judging the mutation recognition reference direction, locate the end node of the reversal, and generate a load disturbance section sequence; S2: Based on the load disturbance section sequence, identify the time window, extract the state of charge records of the energy storage access nodes in the distribution network, compare the state of charge change directions of adjacent cycles in turn, judge whether the trend direction shows a continuous decline and eliminate the continuously declining units, and generate a list of energy storage units; S3: Call the list of energy storage units, map the state of charge trend sequence and the load disturbance section sequence on the time axis, according to the connectivity relationship of the energy storage nodes in the distribution network, retrieve the unit combinations with path connection ability and consistent trend directions, and construct the scheduling links in turn according to the connection order between nodes to generate an energy storage path map; S4: Based on the energy storage path map, calculate the main vector composed of the fluctuation direction and the fluctuation rate, combine the current release direction of the energy storage unit and the state of charge change direction to construct an auxiliary vector, and generate an intervention node trend set.

2. The optimized scheduling method for the energy storage of a distribution network based on dynamic load prediction according to claim 1, wherein The load disturbance section sequence includes the disturbance starting point position, the disturbance end point number, and the number of disturbance direction changes. The list of energy storage units includes the energy storage node number, the state of charge trend direction, and the average state of charge within the cycle. The energy storage path map includes the node connection order, the trend consistency identifier, and the path scheduling priority sequence. The intervention node trend set includes the main vector of voltage change, the auxiliary vector of current release, and the node trend aggregation label.

3. The optimized scheduling method for energy storage in a distribution network based on dynamic load prediction according to claim 1, characterized in that, The specific steps for obtaining the load disturbance section sequence are as follows: S101: Collect the power data of consecutive time slices of the energy storage access nodes in the distribution network, detect the power values between adjacent two time slices, and construct a power change direction sequence according to the difference signs of the front and back power values. Screen the points where the change direction changes from positive to negative or from negative to positive as the direction reversal points and extract them as mutation candidate nodes to generate a mutation candidate node sequence; S102: Based on the node positions in the mutation candidate node sequence, call the power values of the two time slices before and after the node, and calculate the forward change amount and the backward change amount respectively. Compare the two with the node change direction, and judge the node with the change amplitude direction consistent with the original change direction as the starting node for identifying the reference direction to obtain a set of direction reference positioning nodes; S103: According to the set of direction reference positioning nodes, detect whether the power change directions of adjacent nodes show continuous reversals. If the number of continuous reversals exceeds the set number of reversals, locate it as the disturbance end node, mark the time slices between the starting node and the disturbance end node, and obtain the load disturbance section sequence.

4. The optimal scheduling method for distribution network energy storage based on dynamic load prediction according to claim 3, characterized in that The specific steps for obtaining the list of energy storage units are as follows: S201: Based on the interval marked by the load disturbance section sequence, extract the state of charge records of the energy storage access nodes in the distribution network within the corresponding time period, arrange the state of charge values of the nodes in chronological order, and obtain the state of charge trend data of the energy storage nodes; S202: Call the state-of-charge values of adjacent cycles in the state-of-charge trend data of the energy storage nodes, compare each group of values successively, record the change direction markers and connect them to form a trend change sequence, identify the continuous change sections of the state-of-charge values, and obtain the time slice set of the rising trend sections; S203: According to the time positions corresponding to the time slice set of the rising trend sections, filter the energy storage access node numbers with an upward state-of-charge trend, sort out the compliant nodes and perform deduplication and summarization to generate an energy storage unit list.

5. The optimal scheduling method for distribution network energy storage based on dynamic load prediction according to claim 4, characterized in that The steps for obtaining the energy storage path map are specifically as follows: S301: Call the energy storage unit list, correspond the corresponding charge trend sequence with the load disturbance section sequence in terms of time axis position, calculate the state value of the peak interval of the energy storage unit, filter the energy storage units whose state of charge shows an upward trend and the state value of the state of charge is within the peak interval during the disturbance time slice, and obtain the peak interval trend energy storage node set; S302: Based on the state-of-charge trend sequence of the nodes in the peak interval trend energy storage node set, count the number of time slices of the continuous upward trend of the nodes, and number and mark the nodes in descending order according to the trend duration to obtain the trend sorted energy storage node sequence; S303: According to each pair of nodes in the trend sorted energy storage node sequence, retrieve the node pairs with a path connection relationship in the distribution network structure, judge whether the trend directions of the nodes on the connection path are consistent, and if so, arrange them in sequence according to the connection order in the network path to generate an energy storage path map.

6. The optimized scheduling method for the energy storage of a distribution network based on dynamic load prediction according to claim 5, characterized in that, The calculation of the state value of the peak interval of the energy storage unit adopts the formula: ； Among them, represents the state value of the energy storage unit in the peak interval, represents the state of charge value of the i-th time slice, represents the state of charge value of the previous time slice, represents the time interval, represents the load disturbance value of the i-th time slice, is the minimum value, is the total number of time slices.

7. The optimal scheduling method for distribution network energy storage based on dynamic load prediction according to claim 5, wherein The steps for obtaining the intervention node trend set are specifically as follows: S401: Based on the timing information corresponding to the node paths in the energy storage path map, combine the real-time voltage values of the nodes during the time period, calculate the ratio of the voltage value difference to the time interval in continuous time slices, and combine it with the voltage change direction of the adjacent time period to construct a main vector, and obtain the node main vector set; S402: According to the current release direction and state-of-charge change direction data of the nodes in the same time period in the node main vector set, construct an auxiliary vector, filter the nodes that simultaneously meet the conditions that the voltage rate is in a growing state, the voltage value is continuously in a descending section, and the current release direction is consistent, and generate an intervention node trend set.

8. The optimized dispatching method for energy storage in a distribution network based on dynamic load prediction according to claim 7, wherein, The calculation of the ratio of the voltage value difference to the time interval in continuous time slices adopts the formula: ； Among them, represents the node and the node The ratio of the voltage difference to the time interval within the time period, and respectively represent the nodes and the node At the time point and The voltage values of, represents the node and the node The time interval between, represents the The voltage change direction within the th time period, represents the total number of voltage change directions.

9. The optimal dispatching method for energy storage in a distribution network based on dynamic load prediction according to claim 1, wherein, The method further includes step S5: S5: Call the intervention node trend set, adjust the release change limit according to the release curve regulation standard, perform amplitude limiting processing on the current release rate, and at the same time perform a scheduling behavior direction locking operation on the energy storage units that form a connected relationship with the nodes in the energy storage path to generate an energy storage release freeze instruction set; The energy storage release freeze instruction set includes the release rate limit value, the path direction locking number, and the freeze response instruction code.

10. The optimized scheduling method for a distribution network energy storage based on dynamic load prediction according to claim 9, characterized in that, The steps for obtaining the energy storage release freeze instruction set are specifically as follows: S501: Based on the intervention node trend set, extract the change amount value of the release rate of the node within the time period, and call the release change limit range in the release curve adjustment setting. Compare the node change amount value with the upper and lower limits of the range, and analyze the adjustment range of the release curve by determining whether it exceeds the upper and lower boundaries, and generate a release change determination trend value; S502: According to the release change determination trend value, perform a limiting process on the real-time node release current rate, extract the release amplitude parameter corresponding to the rate value, and perform an interval clipping operation with the boundary value of the release change limit range to update the rate value of the release current and generate a node release limit rate; S503: Call the node release limit rate, match the energy storage unit in the energy storage path that forms a connection relationship with the real-time intervention node, compare the scheduling direction value of the energy storage unit with the direction parameter in the constraint release rate set, lock the scheduling direction and write the direction value into the freeze instruction field to generate an energy storage release freeze instruction set.

Images