A source-grid-load-storage coordinated method

Abstract

This invention relates to the field of smart grid technology and discloses a source-grid-load-storage coordination method. The method includes: fusing real-time collected data on power output fluctuations from distributed generation sources, topology constraint parameters from the grid, time-of-use load demand data from the load side, and state of charge data from the storage side to obtain a multi-dimensional system state set; performing short-, medium-, and long-term multi-scale coupled prediction on the multi-dimensional system state set to obtain a source-load dynamic balance prediction curve; constructing a multi-objective collaborative optimization function based on the power fluctuation range, and generating a target adjustment instruction set based on the multi-objective collaborative optimization function; updating the target operating state based on the adjustment instruction set; correcting the weight parameters of the multi-objective collaborative optimization function based on the deviation between the updated operating state and the predicted state, and performing source-grid-load-storage coordinated regulation of the target power platform based on the corrected multi-objective collaborative optimization function; this invention can improve the efficiency of source-grid-load-storage coordination.

Description

Technical Field

[0001] This invention relates to the field of smart grid technology, and in particular to a method for coordinating power generation, grid, load and storage. Background Technology

[0002] In existing power generation, grid, load, and energy storage coordination technologies, the processing of multi-dimensional data, such as power output fluctuations on the source side, topology constraints on the grid side, load demand on the load side, and state of charge on the energy storage side, is mostly done through single-dimensional independent analysis, lacking an effective data fusion mechanism. This results in a one-sided description of the power system state, making it difficult to accurately reflect the correlation between the various links of power generation, grid, load, and energy storage, thus affecting the accuracy of judging the overall operating status of the system and creating potential risks at the foundational data level for subsequent coordinated regulation and control.

[0003] Meanwhile, existing technologies have significant limitations in prediction and control. Prediction often employs single-timescale models, failing to account for fluctuations across short, medium, and long-term scenarios, resulting in insufficient accuracy in predicting dynamic balance between power sources and loads. Control functions have fixed weight parameters, making dynamic adjustment based on deviations between actual and predicted operating conditions difficult. This leads to poor adaptability of control commands to real-time conditions, ultimately resulting in delayed coordinated response of power sources, grids, loads, and storage, low resource allocation efficiency, and an inability to meet the dual requirements of power system safety, stability, and economic operation. Summary of the Invention

[0004] This invention provides a source-grid-load-storage coordination method, the main purpose of which is to solve the problem of low efficiency in source-grid-load-storage coordination.

[0005] To achieve the above objectives, the present invention provides a source-grid-load-storage coordination method, comprising:

[0006] S1. Data fusion is performed on the source-side distributed power output fluctuation data, grid-side topology constraint parameters, load-side time-of-use load demand data, and storage-side charge status data of the real-time power platform to obtain a multi-dimensional system state set of the power platform;

[0007] S2. Perform short-, medium-, and long-term multi-scale coupled prediction on the multi-dimensional system state set to obtain the source-load dynamic balance prediction curve of the power platform;

[0008] S3. Based on the power fluctuation range of the source-load dynamic balance prediction curve, construct a multi-objective collaborative optimization function for the power platform, and generate a set of adjustment instructions for the target power platform based on the multi-objective collaborative optimization function;

[0009] S4. Update the operating status of the target power platform based on the adjustment instruction set;

[0010] S5. Based on the deviation between the updated operating state and the predicted state, the weight parameters of the multi-objective collaborative optimization function are corrected, and the source-grid-load-storage coordinated regulation of the target power platform is performed based on the corrected multi-objective collaborative optimization function.

[0011] In a preferred embodiment, the data fusion of the source-side distributed power output fluctuation data, grid-side topology constraint parameters, load-side time-of-use load demand data, and storage-side state-of-charge data of the real-time power platform to obtain a multi-dimensional system state set of the power platform includes:

[0012] The power output fluctuation data of the source-side distributed power source is normalized to obtain the fluctuation feature vector of the power platform.

[0013] Convert the network-side topology constraint parameters into a node-branch correlation matrix;

[0014] Align the load-side time-of-use load demand data with the energy storage-side state of charge data based on time series data;

[0015] The fluctuation feature vector, the node-branch correlation matrix, and the aligned load-side time-of-use load demand data and storage-side state of charge data are fused at the feature level to obtain the multi-dimensional system state set of the power platform.

[0016] In a preferred embodiment, the step of performing short-, medium-, and long-term multi-scale coupled prediction on the multi-dimensional system state set to obtain the source-load dynamic balance prediction curve of the power platform includes:

[0017] Based on the time granularity, the multi-dimensional system state set is divided into a short-term prediction subset, a medium-term prediction subset, and a long-term prediction subset;

[0018] Spatial correlation features are extracted from the short-term prediction subset to obtain the minute-level fluctuation components of the short-term prediction subset;

[0019] Empirical mode decomposition is performed on the intermediate prediction subset to obtain the fundamental and harmonic components of the intermediate prediction subset, and the evolution trend of the fundamental and harmonic components is predicted.

[0020] Capture the daily cycle load patterns of the long-term forecast subset to obtain the baseline load curve of the long-term forecast subset;

[0021] The source-load dynamic balance prediction curve of the power platform is obtained by coupling the minute-level fluctuation component, the evolution trend, and the baseline load curve through a sliding time window.

[0022] In a preferred embodiment, capturing the daily cyclical load patterns of the long-term forecast subset to obtain the baseline load curve of the long-term forecast subset includes:

[0023] Extract daily cycle load pattern template sets from the historical database that are of the same category as the long-term forecast subset;

[0024] Templates from the daily cycle load pattern template set whose similarity to the long-term prediction subset exceeds a similarity threshold are selected and weighted fusion is performed to obtain the dynamic baseline pattern of the long-term prediction subset.

[0025] The dynamic baseline model is smoothed and filtered to generate the baseline load curve of the long-term forecast subset.

[0026] In a preferred embodiment, the construction of the multi-objective collaborative optimization function for the power platform includes:

[0027] The power platform is constructed with the following objectives: minimizing power output fluctuations on the source side as the first objective, reducing grid-side line losses as the second objective, achieving a load demand satisfaction rate on the load side as the third objective, and maintaining the storage-side SOC within a safe range as the fourth objective. The multi-objective collaborative optimization function is as follows:

[0028] ;

[0029] In the formula, The adjustment command for the power platform, The weight of the first objective item, The weight of the second objective term, The weight of the third objective term, The weight of the fourth objective term, For the first target item, For the second objective item, For the third target item, For the fourth objective item, The number of time segments after discretizing the power fluctuation range. As a time factor, for The actual output of distributed power sources at all times. for Always refer to the output value. for The effective value of the line current at any given time. The equivalent resistance of the line. for Actual load power at any time for Forecast load values ​​in real time. The rated load capacity of the system, for The state of charge at any given moment.

[0030] In a preferred embodiment, generating the regulation instruction set of the target power platform based on the multi-objective collaborative optimization function includes:

[0031] The power fluctuation range is discretized into multiple optimized time periods;

[0032] Solving the multi-objective collaborative optimization function during the optimization period yields the energy storage charging and discharging commands, distributed power source frequency regulation commands, and load flexible control commands for the target power platform.

[0033] The energy storage charging and discharging commands, the distributed power source frequency regulation commands, and the load flexible control commands are integrated into the regulation command set of the target power platform according to the time sequence.

[0034] In a preferred embodiment, updating the operating status of the target power platform based on the adjustment instruction set includes:

[0035] The energy storage charging and discharging commands are analyzed to obtain the converter optimization parameters. The energy storage converter in the target power platform is controlled according to the converter optimization parameters to adjust the charging and discharging power.

[0036] Adjust the output characteristic curve of the photovoltaic inverter in the target power platform according to the distributed power frequency modulation command;

[0037] The demand response terminal executes the load flexibility control command to adjust the operating time of interruptible loads in the target power platform.

[0038] In a preferred embodiment, the deviation between the updated running state and the predicted state includes:

[0039] Within the market during the evaluation period, the deviation of the target power platform after the update of its operating status is determined based on the measured power of the updated operating status, the predicted power of the predicted status, and the rated power of the target power platform.

[0040] In a preferred embodiment, the step of correcting the weight parameters of the multi-objective collaborative optimization function based on the deviation between the updated running state and the predicted state includes:

[0041] When the deviation is greater than the deviation threshold, the weights of the second target item and the fourth target item remain unchanged, the weight of the first target item is increased, and the weight of the third target item is decreased.

[0042] The multi-objective collaborative optimization function is regenerated based on the new weights, and the dynamic update of the adjustment instruction set is triggered synchronously.

[0043] In a preferred embodiment, the step of performing source-grid-load-storage coordinated regulation of the target power platform based on the modified multi-objective collaborative optimization function includes:

[0044] Using the current moment as a reference, a time window is selected, and dynamic optimization is performed based on the modified multi-objective collaborative optimization function to obtain the optimized control sequence of the target power platform;

[0045] Based on the optimal output sequence of the distributed power source in the optimized control sequence, adjust the MPPT operating point of the photovoltaic inverter and the pitch angle of the wind turbine in the target power platform;

[0046] Based on the energy storage system charging and discharging power sequence in the optimized control sequence, the energy storage converter in the target power platform is adjusted to switch between charging and discharging modes, and the changes in the state of charge are monitored in real time.

[0047] Adjust the power consumption periods of interruptible loads in the target power platform based on the load adjustment sequence in the optimized control sequence.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1. The source-grid-load-storage coordination method provided by this invention significantly improves the accuracy of power system state perception and trend prediction through multi-dimensional data fusion and multi-scale coupled prediction. This method performs feature-level fusion of data from the source side, grid side, load side, and storage side to form a comprehensive multi-dimensional system state set. Simultaneously, it generates source-load dynamic balance prediction curves through short-, medium-, and long-term multi-scale coupled prediction, providing accurate basic data and trend basis for coordinated regulation and control, effectively enhancing the ability to control the power system's operating state.

[0050] 2. This invention achieves high efficiency and adaptability in the coordinated regulation of power generation, grid, load, and storage systems through multi-objective collaborative optimization and dynamic correction mechanisms. The constructed multi-objective collaborative optimization function can comprehensively balance core objectives such as power output on the source side, grid losses on the grid side, load demand on the load side, and storage safety on the storage side, generating targeted adjustment commands. Simultaneously, the optimization function weights are adjusted based on the deviation between the operating state and the predicted state, dynamically updating the regulation strategy to ensure that the regulation commands are always highly matched with the actual operating needs of the system, significantly improving the response speed and regulation effect of the coordinated power generation, grid, load, and storage systems. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating a source-grid-load-storage coordination method provided in an embodiment of the present invention.

[0052] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0053] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0054] This application provides a source-grid-load-storage coordination method. The execution subject of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the source-grid-load-storage coordination method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cluster of cloud servers. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and smart grid platforms.

[0055] Reference Figure 1 The diagram shown is a flowchart illustrating a source-grid-load-storage coordination method according to an embodiment of the present invention. In this embodiment, the source-grid-load-storage coordination method includes:

[0056] S1. Data fusion is performed on the source-side distributed power output fluctuation data, grid-side topology constraint parameters, load-side time-of-use load demand data, and storage-side charge status data of the real-time power platform to obtain a multi-dimensional system state set of the power platform;

[0057] In this embodiment of the invention, the data fusion of the source-side distributed power output fluctuation data, grid-side topology constraint parameters, load-side time-of-use load demand data, and storage-side state-of-charge data of the real-time power platform to obtain the multi-dimensional system state set of the power platform includes:

[0058] The power output fluctuation data of the source-side distributed power source is normalized to obtain the fluctuation feature vector of the power platform.

[0059] Convert the network-side topology constraint parameters into a node-branch correlation matrix;

[0060] Align the load-side time-of-use load demand data with the energy storage-side state of charge data based on time series data;

[0061] The fluctuation feature vector, the node-branch correlation matrix, and the aligned load-side time-of-use load demand data and storage-side state of charge data are fused at the feature level to obtain the multi-dimensional system state set of the power platform.

[0062] Specifically, continuous output data is acquired from the real-time monitoring device of the source-side distributed power source. This data covers the actual output power of equipment such as photovoltaic panels and wind turbines at different times. Due to factors such as changes in light intensity and wind speed, this data will exhibit fluctuating characteristics, i.e., the output fluctuation data of the source-side distributed power source. The maximum and minimum values ​​of the output fluctuation data within this period are statistically analyzed. One output fluctuation data is selected, and the minimum value is subtracted from this data. The result is used as the numerator, and the minimum value is subtracted from the maximum value. The ratio of the numerator to the denominator is calculated. This ratio is the normalized value of the output fluctuation data. All output fluctuation data are processed in the same way. Then, these normalized values ​​are arranged in the order of acquisition time. The resulting vector is the fluctuation characteristic vector of the power platform.

[0063] Furthermore, topological constraint parameters from the network side are collected. These parameters include specific information about all nodes, such as the numbers of generating nodes, load nodes, and interconnection nodes, as well as information about all branches, such as the numbers of transmission lines and transformers. At the same time, the connection status of each branch is clarified, that is, the node numbers connected to both ends of the branch. Based on this information, the size of the matrix is ​​determined. The number of rows in the matrix is ​​the same as the total number of nodes, and the number of columns is the same as the total number of branches. Then, the connection relationship between each branch and each node is checked one by one. When one or both ends of a branch are connected to a node, 1 is filled in at the intersection of the row corresponding to the node and the column corresponding to the branch in the matrix. If the node and the branch have no connection, 0 is filled in at the corresponding position. The matrix constructed in this way is the node-branch association matrix.

[0064] Further, examine the timestamps of the load-side time-of-use load demand data to determine its recording period, such as recording every 15 minutes. Simultaneously, examine the timestamps of the storage-side state of charge data to confirm if its recording period matches that of the load side. If they do not match, split the data with the longer recording period or merge the data with the shorter recording period to make their recording periods the same. Arrange the load-side time-of-use load demand data in chronological order from earliest to latest to form a load-side data sequence. Arrange the storage-side state of charge data in the same chronological order to form a storage-side data sequence. Then, based on the timestamps, find the positions in the two sequences where the times completely correspond. Match the time-of-use load demand data at that position in the load-side data sequence with the corresponding state of charge data in the storage-side data sequence. If there is only data from one side at a certain time point, supplement the missing data by analyzing the data trends of adjacent time points to ensure that each time point has corresponding load-side and storage-side data, thus completing the alignment.

[0065] Furthermore, the fluctuation feature vector, the node-branch correlation matrix, and the aligned load-side time-of-use load demand data and storage-side state of charge data are imported into the same data integration system. This system uses time as the core correlation clue. First, it establishes a correspondence between the value of each time point in the fluctuation feature vector and the values ​​of the same time point in the aligned load-side time-of-use load demand data and storage-side state of charge data, so that the fluctuation characteristics of the source side and the time-series data of the load side and storage side are mutually correlated in the time dimension. Then, the node-branch correlation matrix is ​​incorporated into the system. The nodes in the matrix correspond to the actual nodes on the grid side, while the equipment on the load side and storage side is usually connected to specific nodes. The node-branch correlation matrix is ​​associated with the load-side and storage-side data through the node number. At the same time, the distributed power sources on the source side are also connected to the nodes on the grid side. The fluctuation feature vector is associated with the node-branch correlation matrix through the node number. Finally, a data set containing source-side fluctuations, grid-side topology, load-side load, and storage-side state, with each part being mutually correlated, is formed. This set is the multi-dimensional system state set of the power platform.

[0066] In summary, by normalizing source-side data, matrix-based transformation of grid-side parameters, and time alignment of load and storage data, standardized processing of different types and formats of data has been achieved. This eliminates format differences and dimensional barriers between data, enabling the originally scattered single-side data to form an organic whole that can comprehensively reflect the overall operating status of the power platform and provide complete and consistent basic data support for subsequent forecasting and regulation.

[0067] In summary, the feature-level fusion approach preserves the core characteristics of data from each side, such as the fluctuation characteristics of the source side, the topological correlation of the grid side, and the temporal patterns of the load-storage side, ensuring that the fused data is both holistic and retains key details. This enables subsequent analysis and prediction of system status to be based on more comprehensive and accurate information, effectively improving the ability to perceive the complex operating states of the power platform and laying a high-quality data foundation for subsequent dynamic balance prediction of source and load and coordinated optimization and control.

[0068] S2. Perform short-, medium-, and long-term multi-scale coupled prediction on the multi-dimensional system state set to obtain the source-load dynamic balance prediction curve of the power platform;

[0069] In this embodiment of the invention, the step of performing short-, medium-, and long-term multi-scale coupled prediction on the multi-dimensional system state set to obtain the source-load dynamic balance prediction curve of the power platform includes:

[0070] Based on the time granularity, the multi-dimensional system state set is divided into a short-term prediction subset, a medium-term prediction subset, and a long-term prediction subset;

[0071] Spatial correlation features are extracted from the short-term prediction subset to obtain the minute-level fluctuation components of the short-term prediction subset;

[0072] Empirical mode decomposition is performed on the intermediate prediction subset to obtain the fundamental and harmonic components of the intermediate prediction subset, and the evolution trend of the fundamental and harmonic components is predicted.

[0073] Capture the daily cycle load patterns of the long-term forecast subset to obtain the baseline load curve of the long-term forecast subset;

[0074] The source-load dynamic balance prediction curve of the power platform is obtained by coupling the minute-level fluctuation component, the evolution trend, and the baseline load curve through a sliding time window.

[0075] The process of capturing the daily cyclical load patterns of the long-term forecast subset to obtain the baseline load curve of the long-term forecast subset includes:

[0076] Extract daily cycle load pattern template sets from the historical database that are of the same category as the long-term forecast subset;

[0077] Templates from the daily cycle load pattern template set whose similarity to the long-term prediction subset exceeds a similarity threshold are selected and weighted fusion is performed to obtain the dynamic baseline pattern of the long-term prediction subset.

[0078] The dynamic baseline model is smoothed and filtered to generate the baseline load curve of the long-term forecast subset.

[0079] Specifically, time granularity standards are set, where the time range corresponding to short-term forecasts is 0-60 minutes from the future, the time range corresponding to medium-term forecasts is 1-24 hours from the future, and the time range corresponding to long-term forecasts is 1-7 days from the future. All data belonging to the future 0-60 minutes are extracted from the multi-dimensional system state set. The set of these data is the short-term forecast subset. All data belonging to the future 1-24 hours are extracted. The set of these data is the medium-term forecast subset. All data belonging to the future 1-7 days are extracted. The set of these data is the long-term forecast subset.

[0080] Furthermore, all data included in the short-term prediction subset are selected. These data cover various aspects such as source-side fluctuations, grid-side topology, load-side load, and storage-side status. The spatial correlations of these data are analyzed. For example, the correlation between the source-side output fluctuation of a node and the load-side load change of adjacent nodes, and the correlation between the storage-side state of charge and the branch connection status of the node. By identifying these spatially interdependent data relationships, features that can reflect rapid changes in a short period of time are extracted. These features are the minute-level fluctuation components of the short-term prediction subset.

[0081] Furthermore, the intermediate-term prediction subset is obtained, and empirical mode decomposition is performed on all data within this subset. Specifically, the data of the intermediate-term prediction subset is decomposed into multiple components of different frequencies according to its own fluctuation pattern. Among them, the components with lower frequencies and stable changes are fundamental components, and the components with higher frequencies and periodic small fluctuations are harmonic components. After obtaining the fundamental and harmonic components of the intermediate-term prediction subset, the direction and amplitude of the fundamental component's changes over a period of time are observed, such as continuous rise or fall and the magnitude of the rise or fall, to determine its future trend. Similarly, the fluctuation period and amplitude of the harmonic components over a period of time are observed, such as the occurrence of fluctuations at regular intervals and the difference between the maximum and minimum values ​​of the fluctuations, to determine its future fluctuation situation. This is the prediction of the evolution trend of the fundamental and harmonic components.

[0082] Furthermore, examine all load-side data included in the long-term forecast subset, statistically analyze the load values ​​for each hour within a 24-hour period, average the load values ​​for the same hour to obtain the average load for each hour of the day, and arrange these average loads sequentially from 0:00 to 23:00 to form a curve that reflects the load variation pattern within a day. This curve is the baseline load curve of the long-term forecast subset.

[0083] Furthermore, a fixed-length sliding time window is set, which can cover a certain time range. The minute-level fluctuation components, the evolution trend, and the baseline load curve are placed in chronological order. The sliding time window moves sequentially through various time points starting from the initial time. At each time point, the sliding time window simultaneously contains relevant data on the minute-level fluctuation components, evolution trend, and baseline load curve corresponding to that time point. By integrating these data, the three are matched and influenced by each other at each time point, ultimately forming a continuous curve. This curve is the power platform's source-load dynamic balance prediction curve.

[0084] Specifically, the category of the long-term forecast subset is determined by the characteristics of the daily cycle load data it contains, such as the weekday or weekend load data category. All daily cycle load data of the same category are searched from the historical database. These data have been pre-organized into daily cycle load pattern templates, each template corresponding to the load change pattern of a day. By bringing these templates together, a set of daily cycle load pattern templates that are the same as the category of the long-term forecast subset is obtained.

[0085] Furthermore, a fixed similarity threshold is defined, such as 80%. The daily cycle load pattern of the long-term forecast subset is compared with each template in the daily cycle load pattern template set. The comparison starts with the difference in load values ​​for each hour of the day, and the similarity between the two in terms of overall trend and specific values ​​is calculated. Templates with similarity exceeding the similarity threshold are selected, and the number of these selected templates is counted. A weight is assigned to each template according to its similarity with the long-term forecast subset. The higher the similarity, the greater the weight. The load value of each template is multiplied by the corresponding weight, and all products are added together. The result is the dynamic baseline pattern of the long-term forecast subset.

[0086] Furthermore, the dynamic baseline pattern is obtained, which is a curve containing the load values ​​for each hour of the day. This curve is then smoothed and filtered. Specifically, the load values ​​of adjacent hours on the curve are checked. If the difference between the load values ​​of two adjacent hours is too large, such as exceeding a certain reasonable range, the load value of the next hour is adjusted to make it closer to the average level of the load value of the previous hour and the load values ​​of other related hours in the surrounding area, reducing the abrupt changes in the curve. After such processing, the resulting curve is the baseline load curve of the long-term forecast subset.

[0087] In summary, by segmenting the multi-dimensional system state set into different prediction subsets according to time granularity, and extracting short-term minute-level fluctuation components, medium-term fundamental and harmonic component evolution trends, and long-term baseline load curves, the operating characteristics of the power system at different time scales can be accurately captured. This segmented processing method avoids the omission of key fluctuation characteristics in single-time-dimensional predictions, enabling the prediction results to simultaneously cover short-term instantaneous fluctuations, medium-term trend changes, and long-term periodic patterns, significantly improving the comprehensiveness and precision of source-load dynamic balance prediction.

[0088] In summary, by coupling and integrating prediction results at various scales based on a sliding time window, the organic fusion of characteristics across different time dimensions is achieved. The resulting source-load dynamic balance prediction curve better reflects the continuity and correlation of actual power system operation. This prediction curve provides an accurate basis for the construction of subsequent multi-objective collaborative optimization functions, ensuring that the generation of regulation command sets is more forward-looking and targeted. This lays the foundation for the high efficiency of source-grid-load-storage coordinated regulation from the prediction stage.

[0089] S3. Based on the power fluctuation range of the source-load dynamic balance prediction curve, construct a multi-objective collaborative optimization function for the power platform, and generate a set of adjustment instructions for the target power platform based on the multi-objective collaborative optimization function;

[0090] In this embodiment of the invention, the construction of the multi-objective collaborative optimization function of the power platform includes:

[0091] The power platform is constructed with the following objectives: minimizing power output fluctuations on the source side as the first objective, reducing grid-side line losses as the second objective, achieving a load demand satisfaction rate on the load side as the third objective, and maintaining the storage-side SOC within a safe range as the fourth objective. The multi-objective collaborative optimization function is as follows:

[0092] ;

[0093] In the formula, The adjustment command for the power platform, The weight of the first objective item, The weight of the second objective term, The weight of the third objective term, The weight of the fourth objective term, For the first target item, For the second objective item, For the third target item, For the fourth objective item, The number of time segments after discretizing the power fluctuation range. As a time factor, for The actual output of distributed power sources at all times. for Always refer to the output value. for The effective value of the line current at any given time. The equivalent resistance of the line. for Actual load power at any time for Forecast load values ​​in real time. The rated load capacity of the system, for The state of charge at any given moment.

[0094] The method for generating the regulation instruction set for the target power platform based on the multi-objective collaborative optimization function includes:

[0095] The power fluctuation range is discretized into multiple optimized time periods;

[0096] Solving the multi-objective collaborative optimization function during the optimization period yields the energy storage charging and discharging commands, distributed power source frequency regulation commands, and load flexible control commands for the target power platform.

[0097] The energy storage charging and discharging commands, the distributed power source frequency regulation commands, and the load flexible control commands are integrated into the regulation command set of the target power platform according to the time sequence.

[0098] Specifically, the first objective is to minimize the power output fluctuation on the source side. Power output data of distributed power sources on the source side are collected during the prediction period. The actual fluctuation of these data is calculated. By adjusting the power output arrangement, the amplitude of the power output data change over time is minimized as much as possible. This serves as the first part of the multi-objective collaborative optimization function.

[0099] Furthermore, reducing network-side line loss is determined as the second objective. The transmission power data of each network-side line is obtained, and the current loss status is determined based on the transmission status of the line. By optimizing the power allocation on the line, the amount of power loss during transmission is reduced. This optimization direction is taken as the second part of the multi-objective collaborative optimization function.

[0100] Furthermore, with the load demand satisfaction rate as the third objective, the total load demand and the actual load being satisfied during the predicted time period are statistically analyzed to calculate the load demand satisfaction rate. By rationally allocating power resources, the proportion of the actual load being satisfied to the total load demand is increased, which is used as the third part of the multi-objective collaborative optimization function.

[0101] Furthermore, maintaining the SOC of the storage side within a safe range is set as the fourth objective. The safe range of the SOC of the storage side is defined, the current SOC value of the storage side is monitored in real time, and the SOC value is always kept within the set safe range by controlling the charging and discharging operation of the storage side. This control requirement is set as the fourth part of the multi-objective collaborative optimization function.

[0102] Furthermore, the four objectives mentioned above are integrated into the same functional framework in the order of first to fourth, so that the function can simultaneously reflect the optimization direction of these four objectives. The objectives are interconnected and coordinated with each other in the function, and together they constitute the multi-objective collaborative optimization function of the power platform.

[0103] Specifically, the daily load data within the long-term forecast subset is carefully analyzed to determine its category from multiple key dimensions. For example, observing the dates the data falls on within a week, if the data is concentrated from Monday to Friday, it likely belongs to the weekday load data category; if it is concentrated on Saturday and Sunday, it belongs to the weekend load data category. Simultaneously, considering special holiday information, if the data covers load conditions during statutory holidays such as National Day and Spring Festival, it is classified as holiday load data. After determining the category, a search is conducted in a historical database, which has pre-organized daily load data according to categories. All daily load data consistent with the long-term forecast subset category are selected. This data has been processed into daily load pattern templates, each template corresponding to the load changes over 24 hours. All selected templates are integrated to obtain a set of daily load pattern templates identical to the long-term forecast subset category.

[0104] Furthermore, a fixed similarity threshold is defined, taking 80% as an example. A detailed comparison is then made between the daily load patterns of the long-term forecast subset and each template in the daily load pattern template set. During the comparison, the 24 hours of a day are divided into hours, and the load values ​​of the two are compared hourly. First, the overall trend is analyzed, such as whether the load values ​​gradually increase, decrease, or increase first and then decrease from dawn to dusk, to determine if the trends are consistent. Then, specific values ​​are compared, calculating the difference in load values ​​for each hour and comparing it with the difference in load values ​​of all templates in the template set. Combining the trend and numerical differences, the degree of similarity between the two is determined. Templates with a similarity exceeding 80% are selected, and the number of these selected templates is counted.

[0105] Furthermore, weights are assigned based on the similarity between the template and the long-term prediction subset. The higher the similarity, the greater the weight. For example, a template with a similarity of 90% is assigned a weight of 0.6, and one with 85% is assigned a weight of 0.4. Next, the hourly load value corresponding to each template for 24 hours is multiplied by its corresponding weight, and these products are all added together. The final result is the dynamic baseline pattern of the long-term prediction subset.

[0106] Furthermore, a dynamic baseline pattern is obtained, which is presented as a curve containing hourly load values ​​throughout the day. This curve is then smoothed and filtered, and starting from the first hour of the curve, the load values ​​of adjacent hours are examined.

[0107] Furthermore, a reasonable range is set; for example, a difference of more than 10% between adjacent hourly load values ​​is considered too large.

[0108] Furthermore, when this occurs, the load value for the next hour is adjusted. For example, if the load value for the second hour is 20% higher than that for the first hour, exceeding a reasonable range, the load values ​​for the first hour, the third hour, and other relevant hours (such as the 0th and 4th hours) are referenced. The average of these values ​​is calculated, and the load value for the second hour is then adjusted to be close to this average to reduce abrupt changes in the curve. This process is repeated for all adjacent hours on the curve, and after complete processing, the resulting new curve is the baseline load curve for the long-term forecast subset.

[0109] Specifically, in the multi-objective collaborative optimization function, , , , The settings need to be combined with the actual operating scenario of the power platform. For example, during peak electricity consumption periods, if it is necessary to prioritize ensuring the stability of the load-side power supply, the settings can be adjusted accordingly. Set to 0.4. , , Set them to 0.2, 0.2, and 0.2 respectively; if it is during power grid maintenance and line losses need to be reduced, you can... The value was adjusted to 0.4, and the other three values ​​were each set to 0.2. These values ​​were entered into the control module and fixed before the system started.

[0110] Furthermore, Determining the duration requires first clarifying the total duration of the power fluctuation range, such as 24 hours, and then setting a fixed time interval, such as 1 hour. Dividing the total duration by the interval duration yields 24. It is 24. Data is collected in real time by sensors on the source-side photovoltaic panels, wind turbines, and other equipment, and recorded once at set intervals (e.g., every hour). The data collected in the first hour, The corresponding data collected in the second hour, until... . Referring to the daily power generation plan formulated by the power dispatch center, this plan is based on historical output data under the same weather and load levels during the same period. For example, if the ideal output at 9:00 AM on a similar workday in history was 500kW, then... Set to 500kW.

[0111] Furthermore, The current transformers on the grid side lines monitor the data in real time, recording the effective value once per hour. If the line current is monitored to be 100A, then =100A.

[0112] Furthermore, The determination requires first obtaining the resistivity of the circuit material, such as the resistivity of copper being 0.0175 Ω・mm² / m. Then, the circuit length is measured to be 1000m and the cross-sectional area is 25mm². The resistivity is multiplied by the length and then divided by the cross-sectional area to calculate the value. Once this value is determined, it is stored in the system as a fixed parameter.

[0113] Furthermore, The power meter installed on the load side of the main distribution box is read once per hour. For example, if the power meter shows an actual power consumption of 300kW at 2 PM, then... =300kW.

[0114] Furthermore, This is derived from a previously generated load-side load forecast curve, which is based on historical load data for the same period and current user electricity consumption trends. For example, if the predicted load at 2 PM is 290kW, then... =290kW.

[0115] Furthermore, According to the power system design documents, if the maximum load capacity of the system is 1000kW, then... Fixed at 1000kW. The battery management system on the storage side performs an hourly check, calculating the current voltage and current of the battery and its capacity. If the detected state of charge is 0.6, then =0.6.

[0116] Furthermore, The corresponding formula integrates the four objective items into a single indicator through weight allocation. For example, when =0.3、 =0.2、 =0.3、 When =0.2, if =100、 =50、 =80、 =60, =0.3×100+0.2×50+0.3×80+0.2×60=74. This value is directly used to generate adjustment instructions. The smaller the value, the better the overall optimization effect.

[0117] Furthermore, In the corresponding formula, the squared difference of each time segment can amplify the fluctuation amplitude, for example... hour =480kW =500kW, the square of the difference is 400; The square of the time difference is 100, and the sum is 500. The smaller the sum, the smaller the deviation between the output and the reference value in each time period, and the more stable the source side operation.

[0118] Furthermore, In the corresponding formula, the product of the square of the current and the resistance is the line loss during that time period, such as... hour =10A、 The loss during this period is 200. The total loss is obtained by adding up the losses of all periods. The smaller the total, the less power is wasted on the grid side.

[0119] Furthermore, In the corresponding formula, the absolute value of the ratio can prevent positive and negative deviations from canceling each other out, for example... hour =320kW , =1000kW, the absolute value of the ratio is 0.02. After adding the absolute values ​​of the ratios for all time periods, the smaller the value, the lower the proportion of the deviation between the actual load and the predicted load to the rated capacity, and the more matched the supply and demand on the load side.

[0120] Furthermore, In the corresponding formula, 0.5 is the center value of the safety interval (e.g., 0.2-0.8), and the squared difference reflects the degree of deviation. When the value is 0.6, the squared difference is 0.01. When the sum is 0.3, the squared difference is 0.04. The smaller the sum, the closer the state of charge on the storage side is to the center value, thus avoiding overcharging or over-discharging.

[0121] Furthermore, when As the output gradually decreases from 500 to 100, it means that the deviation between the actual output and the reference output in each period is narrowing. For example, the original maximum deviation of 20kW gradually decreases to 5kW, and the output curves of photovoltaic and wind turbines on the source side are flatter, reducing the impact on the power grid.

[0122] Furthermore, when When the value drops from 1000 to 300, it indicates that the effective value of the line current has decreased from 20A to about 10A, and the hourly power loss has decreased from 800Wh to 200Wh, resulting in improved grid-side transmission efficiency and reduced power supply costs.

[0123] Furthermore, when When the value drops from 1.2 to 0.3, it means that the total deviation between the actual load and the predicted load decreases from 120% of the rated capacity to 30%. For example, when the predicted load was originally 300kW, the actual deviation was 400kW, which gradually decreased to 320kW, indicating that the proportion of load-side users' electricity demand being met has increased.

[0124] Furthermore, when When the value drops from 0.5 to 0.1, it means that the SOC on the storage side gradually stabilizes between 0.4 and 0.6 from deviations such as 0.8 and 0.2, resulting in fewer charge and discharge cycles and extended battery life.

[0125] Furthermore, when When the value drops from 200 to 50, it indicates that under the weight allocation, the four objectives of source-side stability, grid-side energy saving, load-side matching, and storage-side safety are improved simultaneously. The regulation command can more accurately coordinate the operation of each side, and the overall efficiency of the power platform is improved.

[0126] In summary, the multi-objective collaborative optimization function takes minimizing power output fluctuations on the source side, reducing line losses on the grid side, achieving load demand satisfaction on the load side, and maintaining the state of charge (SOC) on the storage side within a safe range as its core objectives. This achieves a comprehensive consideration of key operational indicators across the power platform's power generation, grid, load, and storage components. This objective setting directly addresses the core requirements for the safe, stable, and economical operation of the power system, ensuring the comprehensiveness of the optimization direction and avoiding the local optima problem that may result from single-objective optimization.

[0127] In summary, the regulation instruction set generated based on the optimization function, by discretizing the power fluctuation range into optimized time periods and integrating instructions for energy storage charging and discharging, distributed power generation frequency regulation, and load flexible control, achieves precise scheduling of various regulation resources on the power platform. The instruction set can specifically respond to power fluctuations in the predicted curve, enabling synergistic effects in the regulation actions of the power source, grid, load, and storage components. This effectively enhances the power platform's rapid response capability to power fluctuations and provides an executable operational basis for system dynamic balancing.

[0128] S4. Update the operating status of the target power platform based on the adjustment instruction set;

[0129] In this embodiment of the invention, updating the operating status of the target power platform based on the adjustment instruction set includes:

[0130] The energy storage charging and discharging commands are analyzed to obtain the converter optimization parameters. The energy storage converter in the target power platform is controlled according to the converter optimization parameters to adjust the charging and discharging power.

[0131] Adjust the output characteristic curve of the photovoltaic inverter in the target power platform according to the distributed power frequency modulation command;

[0132] The demand response terminal executes the load flexibility control command to adjust the operating time of interruptible loads in the target power platform.

[0133] Specifically, the energy storage charging and discharging command is obtained. This command contains the charging and discharging status (charging or discharging) of the energy storage device in each optimized time period and the corresponding power level. When parsing the command, the charging and discharging status of each optimized time period is first determined. If it is a charging state, the charging power of that time period is extracted; if it is a discharging state, the discharging power of that time period is extracted. These charging and discharging statuses and power levels are the converter optimization parameters. These parameters are input into the control module of the energy storage converter in the target power platform. The control module controls the converter to input the grid power into the energy storage device according to the power parameters corresponding to the charging state, thereby achieving charging; according to the power parameters corresponding to the discharging state, the converter controls the converter to output the energy storage device's power to the grid, thereby achieving discharging, thus adjusting the charging and discharging power of the energy storage device.

[0134] Furthermore, the distributed power source frequency regulation command is obtained. This command contains the output adjustment target that the distributed power source (such as photovoltaic power source) needs to achieve in each optimization period, that is, the output range that needs to be increased or decreased. According to the command, the output control unit of the photovoltaic inverter in the target power platform is located. This unit is responsible for adjusting the output voltage and current of the inverter, thereby changing the output power. When the command requires an increase in output, the control unit adjusts the inverter to increase the effective value of the output voltage and current, so that the output power increases; when the command requires a decrease in output, the control unit adjusts the inverter to decrease the effective value of the output voltage and current, so that the output power decreases. Through such adjustment, the output characteristic curve of the photovoltaic inverter is changed to meet the requirements of the frequency regulation command.

[0135] Furthermore, the load flexibility control command is obtained, which specifies the operation schedule of interruptible loads (such as non-critical equipment in industrial production, some lighting equipment in commercial venues, etc.) in each optimized period, including the periods when operation needs to be stopped and the periods when operation needs to be resumed. The command is sent to the demand response terminal. After receiving the command, the terminal identifies the control switch corresponding to the interruptible load. During the period when the command requires operation to be stopped, the control switch is opened to cut off the power supply to the interruptible load and stop its operation. During the period when the command requires operation to be resumed, the control switch is closed to connect the power supply to the interruptible load and resume its operation, thereby adjusting the operation period of the interruptible load in the target power platform.

[0136] In summary, by analyzing various instructions in the control instruction set and translating them into specific operations, precise adjustments to the operating status of the power platform are achieved. For example, the energy storage converter is controlled to adjust its charging and discharging power based on energy storage charging and discharging instructions; the output characteristics of the photovoltaic inverter are adjusted according to the distributed power source frequency regulation instructions; and the operating periods of interruptible loads are adjusted through load flexible control instructions. This enables the power source, grid, load, and storage components to act synchronously according to the goal of coordinated optimization, ensuring a complete closed loop from the "generation" to the "implementation" of control instructions, and effectively improving the accuracy and timeliness of instruction execution.

[0137] In summary, this instruction set-based state update method can quickly respond to power fluctuation demands reflected in the dynamic balance prediction curve of power sources and loads. Synchronous updates of the operating status of each component can promptly mitigate power fluctuations, optimize resource allocation, and avoid system imbalances caused by lags in the adjustment of a single component. This provides direct support for the power platform to maintain a stable and efficient operating state, further ensuring the practical effectiveness of coordinated regulation of power sources, grids, loads, and storage.

[0138] S5. Based on the deviation between the updated operating state and the predicted state, the weight parameters of the multi-objective collaborative optimization function are corrected, and the source-grid-load-storage coordinated regulation of the target power platform is performed based on the corrected multi-objective collaborative optimization function.

[0139] In this embodiment of the invention, the deviation between the updated running state and the predicted state includes:

[0140] Within the market during the evaluation period, the deviation of the target power platform after the update of its operating status is determined based on the measured power of the updated operating status, the predicted power of the predicted status, and the rated power of the target power platform.

[0141] The step of correcting the weight parameters of the multi-objective collaborative optimization function based on the deviation between the updated running state and the predicted state includes:

[0142] When the deviation is greater than the deviation threshold, the weights of the second target item and the fourth target item remain unchanged, the weight of the first target item is increased, and the weight of the third target item is decreased.

[0143] The multi-objective collaborative optimization function is regenerated based on the new weights, and the dynamic update of the adjustment instruction set is triggered synchronously.

[0144] The method of performing source-grid-load-storage coordinated regulation of the target power platform based on the modified multi-objective collaborative optimization function includes:

[0145] Using the current moment as a reference, a time window is selected, and dynamic optimization is performed based on the modified multi-objective collaborative optimization function to obtain the optimized control sequence of the target power platform;

[0146] Based on the optimal output sequence of the distributed power source in the optimized control sequence, adjust the MPPT operating point of the photovoltaic inverter and the pitch angle of the wind turbine in the target power platform;

[0147] Based on the energy storage system charging and discharging power sequence in the optimized control sequence, the energy storage converter in the target power platform is adjusted to switch between charging and discharging modes, and the changes in the state of charge are monitored in real time.

[0148] The power consumption periods of interruptible loads in the target power platform are adjusted based on the load adjustment sequence in the optimized control sequence.

[0149] Specifically, the evaluation period is determined to be of a specific duration, such as 24 hours. The market within this period refers to the operation of the target power platform and related power dispatch scenarios within these 24 hours. During this period, the power data updated after the operation status is collected in real time through the power monitoring device of the target power platform. These data are generated in actual operation, that is, the measured power of the updated operation status.

[0150] Furthermore, the predicted power data corresponding to the assessment period is extracted from the previously generated source-load dynamic balance prediction curve. These data are the pre-estimates of the target power platform power within the assessment period, i.e., the predicted power of the predicted state.

[0151] Furthermore, the rated power is obtained from the design documents of the target power platform. This power is the upper limit of the power that the platform can operate stably, which is determined during the design phase and serves as a benchmark value for measuring the platform's operating status.

[0152] Furthermore, the measured power of the updated operating state is compared with the predicted power of the predicted state, the difference between the two is calculated, and the ratio of the difference to the rated power is observed. If the ratio of the difference to the rated power is small, it indicates that the deviation is small; if the ratio is large, it indicates that the deviation is large. Through such comparison and proportional analysis, the result obtained is the deviation of the target power platform after the operating state is updated.

[0153] Specifically, a pre-set deviation threshold is determined based on the stable operation requirements of the target power platform, for example, 5%. The previously obtained deviation is compared with this deviation threshold. When the deviation value is greater than the deviation threshold value, the weights of the second and fourth target items remain unchanged. A certain value is added to the current weight of the first target item, for example, α is increased from 0.2 to 0.3. At the same time, the weight of the third target item is reduced by the corresponding value, for example, γ is reduced from 0.3 to 0.2, ensuring that the total weight of the four target items after adjustment is consistent with that before adjustment.

[0154] Furthermore, the adjusted weights of the first objective term, the unchanged weights of the second and fourth objective terms, and the reduced weights of the third objective term are substituted into the construction framework of the multi-objective collaborative optimization function to replace the original weights and form a new function expression. This is the multi-objective collaborative optimization function regenerated based on the new weights. At the same time, an update signal is sent to the system that generates the adjustment instruction set. After receiving the signal, the system immediately recalculates the energy storage charging and discharging instructions, distributed power frequency regulation instructions, and load flexible control instructions according to the new multi-objective collaborative optimization function, and replaces the original adjustment instruction set, thus completing the dynamic update of the adjustment instruction set.

[0155] Specifically, starting from the current moment, the duration of a time window is set, for example, 1 hour. This time window includes all time points within the current moment and the next 1 hour. The modified multi-objective collaborative optimization function is applied to the power platform operation data within this time window. This data includes source-side output, grid-side line status, load-side load, and storage side. Through calculation, a control scheme that achieves the optimal synergy between the four objectives of source-side output fluctuation, grid-side line loss, load-side demand satisfaction rate, and storage side safety is found. The control parameters in these control schemes are arranged in chronological order, and the resulting sequence is the optimized control sequence of the target power platform.

[0156] Furthermore, the optimal output sequence of distributed power sources is extracted from the optimized control sequence. This sequence contains the optimal output values ​​that photovoltaic and wind turbines should achieve at each moment within the future time window. For photovoltaic inverters, their MPPT operating point is adjusted according to the optimal output value of photovoltaics at the corresponding moment in the sequence. Specifically, this is done by changing the voltage and current output ratio of the inverter to make the actual output of the photovoltaic panels consistent with the optimal output. For wind turbines, the pitch angle is adjusted according to the optimal output value of wind turbines at the corresponding moment in the sequence. That is, the angle between the wind turbine blades and the wind direction is changed. When it is necessary to increase the output, the angle is decreased, and when it is necessary to decrease the output, the angle is increased to ensure that the actual output of the wind turbine matches the optimal output sequence.

[0157] Furthermore, the charging and discharging power sequence of the energy storage system is extracted from the optimized control sequence. This sequence clearly defines whether the energy storage system should charge or discharge at each moment within the future time window, as well as the corresponding power level. When the sequence indicates that charging is required at a certain moment, the energy storage converter is controlled to switch to charging mode, inputting grid power into the energy storage battery. When discharging is required, the system switches to discharging mode, outputting energy from the energy storage battery to the grid. Simultaneously, the state of charge data is collected in real time through the energy storage-side battery management system, recording its changes during the charging and discharging process to ensure consistency with the expected changes in the sequence.

[0158] Furthermore, a load adjustment sequence is extracted from the optimized control sequence. This sequence marks the specific time periods within a future time window during which interruptible loads need to stop or resume power consumption. This time period information is sent to the demand response terminal corresponding to the interruptible load. After receiving the information, the terminal cuts off the power supply circuit of the corresponding load during the time period when power consumption needs to be stopped, and connects the power supply circuit during the time period when power consumption needs to be resumed. Through this operation, the power consumption periods of the interruptible load are perfectly matched with the load adjustment sequence.

[0159] In summary, a dynamic correlation mechanism between prediction and actual operation was established through deviation assessment. When the deviation exceeds a threshold, the weight parameters of the multi-objective collaborative optimization function are adjusted accordingly (keeping the weights of the grid-side and storage-side objectives unchanged, increasing the weight of the source-side output fluctuation optimization, and decreasing the weight of the load-side demand fulfillment rate), enabling the optimization function to adapt to the actual operating state of the system in real time. This weight adjustment mechanism avoids the problem of insufficient adaptability of fixed parameters to dynamic operating conditions, making the optimization objectives more aligned with the core needs of the current system, and ensuring that the direction of regulation always remains consistent with actual operating requirements.

[0160] In summary, based on the coordinated control strategy generated by the modified optimization function, an optimized control sequence is obtained through dynamic optimization and applied to distributed power sources, energy storage systems, and loads, achieving closed-loop iteration of the control strategy. This dynamic adjustment mode can quickly compensate for the deviation between prediction and reality, improve the coordinated response accuracy of the power source, grid, load, and storage systems, effectively enhance the resilience of the power platform to fluctuations in operating conditions, and ensure that the system can maintain efficient and stable operation under complex conditions.

[0161] In the several embodiments provided by this invention, it should be understood that the disclosed method can be implemented in other ways.

[0162] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0163] The embodiments of this application can acquire and process relevant data based on smart grid technology. Smart grid refers to the theory, methods, and technologies that utilize digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A source-grid-load-storage coordinated method, characterized in that, The method includes: S1. Data fusion is performed on the source-side distributed power output fluctuation data, grid-side topology constraint parameters, load-side time-of-use load demand data, and storage-side charge status data of the real-time power platform to obtain a multi-dimensional system state set of the power platform; S2. Perform short-, medium-, and long-term multi-scale coupled prediction on the multi-dimensional system state set to obtain the source-load dynamic balance prediction curve of the power platform; S3. Based on the power fluctuation range of the source-load dynamic balance prediction curve, construct a multi-objective collaborative optimization function for the power platform, and generate a set of adjustment instructions for the target power platform based on the multi-objective collaborative optimization function; S4. Update the operating status of the target power platform based on the adjustment instruction set; S5. Based on the deviation between the updated operating state and the predicted state, the weight parameters of the multi-objective collaborative optimization function are corrected, and the source-grid-load-storage coordinated regulation of the target power platform is performed based on the corrected multi-objective collaborative optimization function.

2. The source-grid-load-storage coordination method as described in claim 1, characterized in that, The process involves fusing data from the real-time power platform, including distributed generation output fluctuation data on the source side, grid-side topology constraint parameters, load-side time-of-use load demand data, and energy storage-side state of charge data, to obtain a multi-dimensional system state set for the power platform, comprising: The power output fluctuation data of the source-side distributed power source is normalized to obtain the fluctuation feature vector of the power platform. Convert the network-side topology constraint parameters into a node-branch correlation matrix; Align the load-side time-of-use load demand data with the energy storage-side state of charge data based on time series data; The fluctuation feature vector, the node-branch correlation matrix, and the aligned load-side time-of-use load demand data and storage-side state of charge data are fused at the feature level to obtain the multi-dimensional system state set of the power platform.

3. The source-grid-load-storage coordination method as described in claim 1, characterized in that, The process of performing short-, medium-, and long-term multi-scale coupled prediction on the multi-dimensional system state set to obtain the source-load dynamic balance prediction curve of the power platform includes: Based on the time granularity, the multi-dimensional system state set is divided into a short-term prediction subset, a medium-term prediction subset, and a long-term prediction subset; Spatial correlation features are extracted from the short-term prediction subset to obtain the minute-level fluctuation components of the short-term prediction subset; Empirical mode decomposition is performed on the intermediate prediction subset to obtain the fundamental and harmonic components of the intermediate prediction subset, and the evolution trend of the fundamental and harmonic components is predicted. Capture the daily cycle load patterns of the long-term forecast subset to obtain the baseline load curve of the long-term forecast subset; The source-load dynamic balance prediction curve of the power platform is obtained by coupling the minute-level fluctuation component, the evolution trend, and the baseline load curve through a sliding time window.

4. The source-grid-load-storage coordination method as described in claim 3, characterized in that, The process of capturing the daily cyclical load patterns of the long-term forecast subset to obtain the baseline load curve of the long-term forecast subset includes: Extract daily cycle load pattern template sets from the historical database that are of the same category as the long-term forecast subset; Templates from the daily cycle load pattern template set whose similarity to the long-term prediction subset exceeds a similarity threshold are selected and weighted fusion is performed to obtain the dynamic baseline pattern of the long-term prediction subset. The dynamic baseline model is smoothed and filtered to generate the baseline load curve of the long-term forecast subset.

5. The source-grid-load-storage coordination method as described in claim 1, characterized in that, The multi-objective collaborative optimization function for constructing the power platform includes: The power platform is constructed with the following objectives: minimizing power output fluctuations on the source side as the first objective, reducing grid-side line losses as the second objective, achieving a load demand satisfaction rate on the load side as the third objective, and maintaining the storage-side SOC within a safe range as the fourth objective. The multi-objective collaborative optimization function is as follows: ； In the formula, The adjustment command for the power platform, The weight of the first objective item, The weight of the second objective term, The weight of the third objective term, The weight of the fourth objective term, For the first target item, For the second objective item, For the third target item, For the fourth objective item, The number of time segments after discretizing the power fluctuation range. As a time factor, for The actual output of distributed power sources at all times. for Always refer to the output value. for The effective value of the line current at any given time. The equivalent resistance of the line. for Actual load power at any time for Forecast load values ​​in real time. The rated load capacity of the system, for The state of charge at any given moment.

6. The source-grid-load-storage coordination method as described in claim 1, characterized in that, The method for generating the regulation instruction set for the target power platform based on the multi-objective collaborative optimization function includes: The power fluctuation range is discretized into multiple optimized time periods; Solving the multi-objective collaborative optimization function during the optimization period yields the energy storage charging and discharging commands, distributed power source frequency regulation commands, and load flexible control commands for the target power platform. The energy storage charging and discharging commands, the distributed power source frequency regulation commands, and the load flexible control commands are integrated into the regulation command set of the target power platform according to the time sequence.

7. The source-grid-load-storage coordination method as described in claim 6, characterized in that, The step of updating the operating status of the target power platform based on the adjustment instruction set includes: The energy storage charging and discharging commands are analyzed to obtain the converter optimization parameters. The energy storage converter in the target power platform is controlled according to the converter optimization parameters to adjust the charging and discharging power. Adjust the output characteristic curve of the photovoltaic inverter in the target power platform according to the distributed power frequency modulation command; The demand response terminal executes the load flexibility control command to adjust the operating time of interruptible loads in the target power platform.

8. The source-grid-load-storage coordination method as described in claim 1, characterized in that, The deviation between the updated running state and the predicted state includes: Within the market during the evaluation period, the deviation of the target power platform after the update of its operating status is determined based on the measured power of the updated operating status, the predicted power of the predicted status, and the rated power of the target power platform.

9. The source-grid-load-storage coordination method as described in claim 5, characterized in that, The step of correcting the weight parameters of the multi-objective collaborative optimization function based on the deviation between the updated running state and the predicted state includes: When the deviation is greater than the deviation threshold, the weights of the second target item and the fourth target item remain unchanged, the weight of the first target item is increased, and the weight of the third target item is decreased. The multi-objective collaborative optimization function is regenerated based on the new weights, and the dynamic update of the adjustment instruction set is triggered synchronously.

10. The source-grid-load-storage coordination method as described in claim 1, characterized in that, The method of performing source-grid-load-storage coordinated regulation of the target power platform based on the modified multi-objective collaborative optimization function includes: Using the current moment as a reference, a time window is selected, and dynamic optimization is performed based on the modified multi-objective collaborative optimization function to obtain the optimized control sequence of the target power platform; Based on the optimal output sequence of the distributed power source in the optimized control sequence, adjust the MPPT operating point of the photovoltaic inverter and the pitch angle of the wind turbine in the target power platform; Based on the energy storage system charging and discharging power sequence in the optimized control sequence, the energy storage converter in the target power platform is adjusted to switch between charging and discharging modes, and the changes in the state of charge are monitored in real time. The power consumption periods of interruptible loads in the target power platform are adjusted based on the load adjustment sequence in the optimized control sequence.

Images