Data communication and mapping method for virtual power plant and multiple scheduling mechanisms

Abstract

The invention relates to a data communication and mapping method of a virtual power plant and multiple scheduling mechanisms, and belongs to the technical field of virtual power plant data interaction. The method comprises the following steps: constructing a communication system based on scheduling hierarchy and demand difference, and introducing a dynamic link priority allocation mechanism and a data desensitization-authority grading binding security mechanism; multi-class data of the virtual power plant are obtained, a standardized data set is generated through multi-source data fusion calibration, potential task requirements are analyzed through a scheduling requirement intention recognition model, and a requirement list is constructed; constructing a mapping rule base by using an adaptive mapping rule generation algorithm based on the demand list, and establishing a cross-mechanism mapping conflict coordination mechanism in combination with task priorities; scheduling demand change and link attenuation risks are predicted through a historical rule and a power grid period, related configuration is pre-updated, and a mapping effect feedback iteration mechanism is established based on data use feedback and a calling efficiency index; and the flexibility of communication between the virtual power plant and the multiple scheduling mechanisms and the accuracy of data mapping are improved.

Description

Technical Field

[0001] This invention belongs to the field of virtual power plant data interaction technology, specifically relating to a data communication and mapping method between a virtual power plant and multiple dispatching agencies. Background Technology

[0002] With the trend of energy structure transformation towards cleaner and distributed energy, virtual power plants, as the core carrier for aggregating decentralized energy resources, need to work collaboratively with multi-level dispatching agencies such as regional power grid dispatching agencies, distribution network dispatching agencies, and microgrid dispatching agencies to achieve flexible energy dispatching and safe and stable grid operation. However, the current data communication and mapping process between virtual power plants and multiple dispatching agencies suffers from multiple technical pain points, which seriously restrict the efficiency of collaboration and the accuracy of dispatching. Specific problems are as follows: Existing technologies mostly establish a single fixed communication link between virtual power plants and various dispatching agencies, without considering the differences in the urgency of tasks among different dispatching agencies. For example, fault response tasks require millisecond-level data transmission, while non-real-time statistical tasks have lower timeliness requirements. However, existing links do not have a priority allocation mechanism, which can easily lead to delays in the transmission of urgent tasks due to insufficient link bandwidth. At the same time, although some solutions are equipped with backup links, the switching response is lagging, and there is no prediction of long-term performance degradation of the link, making it difficult to ensure the continuity and stability of data transmission.

[0003] The data sources of virtual power plants are complex, covering energy data, operating status data and grid connection interaction data collected by multi-source sensing devices. However, existing technologies lack effective multi-source data fusion and calibration mechanisms, and simply splice data together, resulting in deviations in the same type of data due to differences in device accuracy. In addition, the data preprocessing stage is not optimized in accordance with the needs of dispatching agencies, the criteria for removing abnormal data are singular, and the logic for supplementing missing data is crude. The generated dataset is difficult to meet the differentiated data quality requirements of different dispatching agencies.

[0004] There are significant differences in data format standards, accuracy requirements, and update frequencies among different dispatching agencies. Existing mapping schemes mostly use fixed rules, which require repeated manual adjustments to adapt to the needs of different agencies, resulting in low efficiency. More importantly, when multiple dispatching agencies simultaneously call the same data source of the virtual power plant, there is a lack of an effective conflict coordination mechanism, which can easily lead to data source access congestion. This causes some dispatching tasks to be unable to proceed in a timely manner due to data acquisition delays, affecting the overall dispatching efficiency.

[0005] Existing technologies can only passively receive task requirements from dispatching agencies without predicting trends in demand changes. For example, during peak grid load periods, the frequency of demand for load data from dispatching agencies increases significantly, but existing solutions cannot adjust data push strategies in advance. At the same time, the lack of a mapping effect feedback mechanism makes it impossible to optimize mapping rules and communication parameters based on the data usage experience of dispatching agencies and the data call efficiency of virtual power plants. This results in the technical solutions remaining in a "static operation" state for a long time, making it difficult to adapt to the dynamic changes in grid operation scenarios.

[0006] In summary, existing data communication and mapping technologies between virtual power plants and multiple dispatching agencies have significant shortcomings in terms of link adaptation, data processing, rule construction, and demand response. They cannot meet the efficiency and accuracy requirements of multi-agency collaborative dispatching. A technical solution with dynamic adaptation, intelligent mapping, and security and reliability is needed to break through the current collaboration bottleneck. Summary of the Invention

[0007] To address the aforementioned problems in the existing technology, this invention provides a data communication and mapping method between a virtual power plant and multiple dispatching agencies. The objective of this invention can be achieved through the following technical solutions: A data communication and mapping method between a virtual power plant and multiple dispatching agencies includes: S1: Based on the differences in scheduling hierarchy and data interaction requirements of the scheduling agency, an independent communication link including a main link and a backup link is established for the scheduling agency. At the same time, a dynamic link priority allocation mechanism is introduced, and a data desensitization-permission hierarchical binding security mechanism is adopted in the data transmission link. S2: Acquire internal energy data, operating status data, and grid connection interaction data of the virtual power plant; introduce a multi-source data fusion calibration algorithm to generate a standardized dataset and remove abnormal data; based on the scheduling task type of the scheduling agency, deploy a scheduling demand intent recognition model; identify potential task demands through historical interaction data of the scheduling agency, task description keywords, and current grid operating status, and integrate them into the data demand list of the scheduling agency. S3: Based on the data requirement list, an adaptive mapping rule generation algorithm is introduced to generate mapping relationships and integrate them into a callable mapping rule library. The scheduling agency corresponds to an independent subset of mapping rules. Based on the priority of scheduling tasks and the urgency of data requirements, the data source access sequence is allocated, and a cross-agency mapping conflict coordination mechanism is established. S4: By analyzing the historical task change patterns and power grid operation cycle characteristics of the dispatching agencies, predict the trend of dispatching demand changes and the risk of link performance degradation in advance, and update the demand list and mapping rule subset of the corresponding dispatching agencies in advance before demand changes; collect data usage feedback from each dispatching agency, combine the data call efficiency index of the virtual power plant side, regularly optimize the conversion logic, filtering rules and push mechanism in the mapping rule base, and establish a mapping effect feedback and iteration mechanism.

[0008] Specifically, the implementation process of the dynamic link priority allocation mechanism includes: dividing the urgency of scheduling tasks into three levels: emergency fault response, regular power scheduling, and non-real-time statistics; and pre-setting corresponding bandwidth allocation ratios and transmission rate thresholds for each level; when the scheduling task type changes or the link load exceeds the preset load threshold, dynamically adjusting the priority of the corresponding link and the bandwidth allocation ratio.

[0009] Specifically, the data desensitization-permission hierarchical binding security mechanism includes: first, classifying the transmitted data according to sensitivity levels, with sensitivity levels divided into core sensitive data, regular sensitive data, and non-sensitive data; adopting differentiated desensitization methods for data of different sensitivity levels; and simultaneously dividing the scheduling agency's permissions into three levels: core permissions, regular permissions, and query permissions.

[0010] Specifically, the process of the multi-source data fusion calibration algorithm includes: classifying the acquired virtual power plant data into three categories: energy data, operating status data, and grid connection interaction data; categorizing the multi-source sensor acquisition results in each category; calculating the weight value of each acquisition result based on the accuracy level of the sensor and the historical data error rate for the same data item; and calculating the fused basic data using a weighted summation formula; and retrieving the historical operating baseline data of the same data item in the virtual power plant and calculating the deviation value from the basic data.

[0011] Specifically, the construction and operation process of the scheduling demand intent recognition model includes: training the intent recognition model using the historical interaction data of the scheduling agency, the power grid operation status data, and the scheduling task description keyword corpus as training data; extracting keywords from the task description text currently submitted by the scheduling agency, and then combining the current power grid operation status data with the historical interaction preferences of the scheduling agency to output the preliminary recognition result of the potential task demand.

[0012] Specifically, the steps for generating the standardized dataset and removing outlier data include: calculating the mean and standard deviation of data items, identifying data that exceeds the range of the mean plus or minus a preset multiple of the standard deviation as outlier data, and removing them directly; and supplementing missing data by using a time-series interpolation algorithm to analyze the historical time-series change patterns of the data item and supplementing the missing values.

[0013] Specifically, the adaptive mapping rule generation algorithm generates conversion logic for data type differences by: obtaining the data type standard preset by the dispatching agency and establishing a correspondence table between the standardized data of the virtual power plant and the data type of the dispatching agency; generating unit conversion logic for differences in numerical units; and generating data dimension splitting logic and data dimension aggregation logic for differences in data dimensions.

[0014] Specifically, the adaptive mapping rule generation algorithm generates processing rules for data accuracy differences by: parsing the accuracy requirements in the data requirement list of the scheduling agency; generating original data extraction rules for errors with an allowable range less than a preset threshold and sampling frequency requirements higher than a preset frequency; and generating data dimensionality reduction processing rules for errors with an allowable range greater than or equal to a preset threshold and sampling frequency requirements lower than a preset frequency. The rules include specific methods for data sampling or data mean aggregation, thereby reducing the amount of data through dimensionality reduction processing.

[0015] Specifically, the implementation process of the cross-agency mapping conflict coordination mechanism includes: real-time monitoring of the scheduling agency's call requests to the same data source; when a mapping conflict occurs, initiating a conflict coordination process; obtaining the scheduling agency's scheduling task priority and data demand urgency; generating a coordination priority ranking result based on the scheduling agency hierarchy; allocating data source access sequence according to the coordination priority ranking result; and feeding back the access sequence allocation result to the scheduling agency.

[0016] Specifically, the process of predicting the trend of changes in scheduling demand in advance includes: performing statistical analysis on the historical task data of the scheduling agency according to the time dimension to extract the periodic change pattern of task types; and then, in combination with the characteristics of the power grid operation cycle, using a time-series prediction mechanism to predict the changes in scheduling task types and data demand parameters within a future preset time period.

[0017] Specifically, the process of predicting the risk of link performance degradation in advance includes: acquiring historical performance data of the communication link, establishing a link performance degradation assessment model, analyzing the changing trend of the historical performance data, and calculating the degradation rate of the link performance parameters; when it is predicted that the performance parameters of a certain link will drop below a preset performance threshold within a preset time period in the future, it is determined that there is a risk of performance degradation.

[0018] Specifically, the implementation steps of the mapping effect feedback iteration mechanism include: designing a feedback data collection form, periodically pushing it to the scheduling agency to collect feedback data; simultaneously, statistical data call efficiency indicators on the virtual power plant side, including data processing time, mapping rule call success rate, and data retransmission rate; and performing comprehensive analysis of the feedback data and efficiency indicators at preset optimization intervals.

[0019] The beneficial effects of this invention are as follows: At the communication security level, a triple mechanism of "primary and backup links + dynamic priority allocation + data anonymization and permission binding" enhances the security and stability of data transmission. The automatic switching design of primary and backup links avoids communication paralysis caused by the interruption of a traditional single link. Dynamic priority allocation reserves high-bandwidth resources for emergency fault response tasks, ensuring that critical scheduling instructions are transmitted without delay. The data anonymization and permission-level binding mechanism performs differentiated anonymization based on data sensitivity level and precise authorization based on organizational permissions. This prevents the leakage of core data (such as grid-connected power strategy data) and avoids the abuse of permissions. Compared with traditional general encryption methods, the risk of data leakage is reduced and the transmission interruption rate is lowered.

[0020] At the data adaptation level, relying on multi-source data fusion calibration algorithms and adaptive mapping rules, the problem of "incompatible formats and mismatched precision" in cross-agency data is solved. Multi-source data fusion eliminates data bias from multiple sensors through weighted calculation and historical baseline calibration, generating highly reliable standardized datasets and avoiding the accumulation of errors from traditional single data sources. Adaptive mapping rules can automatically match the data types and precision requirements of dispatching agencies. For example, it can generate low-precision aggregated data for distribution network dispatching and extract high-precision raw data for regional dispatching, without the need for repeated manual format adjustments, thus improving data adaptation efficiency and increasing the data reuse rate across agencies.

[0021] In terms of dispatch efficiency, the response cycle of dispatch instructions is shortened by leveraging dispatch demand intent identification and cross-agency conflict coordination. The demand intent identification model predicts potential demand through historical data and grid status, such as pre-linking meteorological data for load forecasting tasks to avoid secondary requests from dispatching agencies. The conflict coordination mechanism allocates data source access sequence according to task priority, solving the bottleneck problem of multiple agencies competing for the same data source. The instruction issuance time for emergency fault response tasks is shortened, dispatch decision efficiency is improved, and virtual power plants can quickly respond to grid peak shaving and fault recovery needs.

[0022] At the system optimization level, predictive updates and feedback iteration mechanisms enable long-term adaptive evolution of the system. Predictive updates adjust the demand list and link parameters in advance based on historical patterns, avoiding the lag of traditional "passive correction after problems occur." Feedback iteration combines dispatcher evaluations and virtual power plant efficiency indicators to continuously optimize the mapping logic. For example, it adjusts the data push frequency based on feedback, reduces the amount of invalid data transmission, reduces system maintenance costs, and continuously improves long-term operational stability and adaptability, providing sustainable technical support for the normalized collaboration between virtual power plants and multiple dispatchers. Attached Figure Description

[0023] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0024] Figure 1This is a flowchart illustrating the overall process of data communication and mapping between a virtual power plant and multiple dispatching agencies according to the present invention. Figure 2 This is a diagram illustrating the system architecture and data flow in this invention. Detailed Implementation

[0025] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0026] Please see Figures 1-2 A data communication and mapping method between a virtual power plant and multiple dispatching agencies, comprising: S1: Based on the differences in scheduling hierarchy and data interaction requirements of the scheduling agency, an independent communication link including a main link and a backup link is established for the scheduling agency. At the same time, a dynamic link priority allocation mechanism is introduced, and a data desensitization-permission hierarchical binding security mechanism is adopted in the data transmission link. S2: Acquire internal energy data, operating status data, and grid connection interaction data of the virtual power plant; introduce a multi-source data fusion calibration algorithm to generate a standardized dataset; based on the scheduling task type of the scheduling agency, deploy a scheduling demand intent recognition model; identify potential task demands through historical interaction data of the scheduling agency, task description keywords, and current grid operating status, and integrate them into the data demand list of the scheduling agency. S3: Based on the data requirement list, an adaptive mapping rule generation algorithm is introduced to generate mapping relationships and integrate them into a callable mapping rule library. The scheduling agency corresponds to an independent subset of mapping rules. Based on the priority of scheduling tasks and the urgency of data requirements, the data source access sequence is allocated, and a cross-agency mapping conflict coordination mechanism is established. S4: By analyzing the historical task change patterns and power grid operation cycle characteristics of the dispatching agencies, predict the trend of dispatching demand changes and the risk of link performance degradation in advance, and update the demand list and mapping rule subset of the corresponding dispatching agencies in advance before demand changes; collect data usage feedback from each dispatching agency, combine the data call efficiency index of the virtual power plant side, regularly optimize the conversion logic, filtering rules and push mechanism in the mapping rule base, and establish a mapping effect feedback and iteration mechanism.

[0027] This embodiment uses a virtual power plant in a certain region as the application subject. This virtual power plant needs to achieve data communication and mapping with the regional power grid dispatching agency (denoted as Agency A), the distribution network dispatching agency (denoted as Agency B), and the microgrid dispatching agency (denoted as Agency C). The specific implementation process is as follows: Specifically, the implementation process of the dynamic link priority allocation mechanism is as follows: First, the urgency of scheduling tasks is divided into three levels: emergency fault response, regular power scheduling, and non-real-time statistics. For each level, a corresponding bandwidth resource allocation ratio and transmission rate threshold are preset. Among them, the links corresponding to emergency fault response tasks are allocated the highest proportion of bandwidth, and the transmission rate must meet the real-time interaction requirements. Links for regular power scheduling tasks are allocated a medium proportion of bandwidth, and links for non-real-time statistics tasks are allocated a basic proportion of bandwidth. When the scheduling task type changes or the link load exceeds the preset load threshold, the priority and bandwidth allocation ratio of the corresponding link are dynamically adjusted. After adjustment, the transmission performance is verified by the link status monitoring module to see if it meets the standard. If it meets the standard, the adjusted configuration is maintained; if it does not meet the standard, the allocation scheme is re-optimized.

[0028] Specifically, the data anonymization-permission hierarchical binding security mechanism includes: firstly, classifying transmitted data according to sensitivity levels, which are divided into core sensitive data (including control parameters of key equipment in virtual power plants and user privacy energy consumption data), regular sensitive data (including operating parameters of non-core equipment and regional energy consumption statistics), and non-sensitive data (including publicly available energy policy data and regular dispatch notification data); adopting differentiated anonymization methods for data of different sensitivity levels, using character replacement combined with range fuzzification for core sensitive data, partial field hiding for regular sensitive data, and no anonymization required for non-sensitive data; simultaneously, dividing dispatching agency permissions into three levels: core permissions, regular permissions, and query permissions. Core permissions correspond to access to core sensitive data (after anonymization) and all regular and non-sensitive data, regular permissions correspond to access to regular sensitive data (after anonymization) and non-sensitive data, and query permissions correspond to access only non-sensitive data, thus achieving precise binding between the degree of data anonymization and dispatching agency permissions.

[0029] I. Construction and Implementation of Dynamic Priority Communication System Independent communication links and dynamic priority allocation Establish primary links (denoted as L_A1, L_B1, L_C1) and backup links (denoted as L_A2, L_B2, L_C2) for organizations A, B, and C respectively, and deploy link status monitoring modules to collect the transmission rate (denoted as v) and packet loss rate (denoted as r) of each link in real time.

[0030] A dynamic link priority allocation mechanism is introduced: the urgency of scheduling tasks is divided into emergency fault response (level 1), regular power scheduling (level 2), and non-real-time statistics (level 3). The bandwidth allocation ratio of each level is preset to k1:k2:k3 (satisfying k1+k2+k3=1), and the transmission rate thresholds are v1, v2, and v3 (v1>v2>v3).

[0031] When organization A initiates an emergency fault response task (level 1), it allocates a bandwidth ratio k1 to its main link L_A1. If the load rate of L_A1 (denoted as s) is detected to exceed the preset threshold s0 (s>s0), the bandwidth ratio is dynamically adjusted to k1'=k1+Δk (Δk is the bandwidth increment), while ensuring that the transmission rate v_A1≥v1. When the task switches to regular power scheduling (level 2), the bandwidth ratio is restored to k2, and the transmission rate is maintained at v_A1≥v2.

[0032] Data anonymization - secure transmission with hierarchical access control First, classify the transmitted data according to sensitivity level: core sensitive data (such as the upper limit of energy storage capacity, denoted as D1), regular sensitive data (such as photovoltaic output data, denoted as D2), and non-sensitive data (such as equipment running time data, denoted as D3). For D1, field replacement desensitization is used (the replacement rule is D1'=D1×a, where a is the desensitization coefficient), for D2, range fuzzy desensitization is used (the fuzzy rule is D2'∈[D2-b,D2+b], where b is the fuzzy amplitude), and for D3, no desensitization is performed.

[0033] The permissions for the three organizations are divided into core permissions (Organization A, denoted as P1), regular permissions (Organization B, denoted as P2), and query permissions (Organization C, denoted as P3): Organization A can access D1', D2', and D3; Organization B can access D2' and D3; Organization C can only access D3. After verifying permissions through multi-factor authentication (including organization identification code and dynamic key), data is encrypted and transmitted using an encryption algorithm (key length c). During transmission, integrity is confirmed through CRC check (checksum d). If the check fails, retransmission is triggered, with a maximum of n retransmissions. If n is exceeded, the system switches to a backup link.

[0034] Specifically, the multi-source data fusion calibration algorithm operates as follows: First, the collected virtual power plant data is categorized into three types: energy data, operating status data, and grid connection interaction data. The data collected by multi-source sensing devices in each category is then classified. Second, for the same data item, the weight of each collected result is calculated based on the accuracy level of the sensing device and its historical error rate. Devices with higher accuracy levels and lower historical error rates have larger weights. The fused basic data is calculated using a weighted summation formula. Third, the historical operating baseline data for that data item in the virtual power plant is retrieved, and the deviation between the basic data and the baseline data is calculated. If the deviation is within a preset reasonable range, the basic data is used as the calibrated data. If the deviation exceeds a reasonable range, the basic data is corrected based on the trend of the baseline data. The deviation is then recalculated until it meets the requirements, ultimately forming a standardized dataset.

[0035] Specifically, the construction and operation process of the dispatch demand intent recognition model is as follows: First, using the historical interaction data of the dispatching agency over the past three years (including task type, data request content, and feedback), the power grid operation status data over the past five years (including load fluctuation data, fault occurrence data, and seasonal operation characteristic data), and a corpus of dispatching task description keywords as training data, a deep learning algorithm is used to train the intent recognition model. When the model is running, keywords are first extracted from the task description text currently submitted by the dispatching agency. Then, combined with the current power grid operation status data (such as real-time load rate and whether it is in the fault recovery period) and the dispatching agency's historical interaction preferences (such as whether similar historical tasks need to be associated with additional meteorological data), a preliminary identification result of the potential task demand is output. The preliminary identification result is fed back to the dispatching agency through the interaction interface. If the dispatching agency confirms that there is no error, the potential demand is included in the data demand list. If the dispatching agency proposes a correction, the model recognition logic is adjusted according to the correction, the potential demand is regenerated and confirmed, ensuring that the identification result meets the actual needs of the dispatching agency.

[0036] Specifically, the generation of the standardized dataset also includes detailed operations for outlier removal and missing data supplementation: Outlier removal involves calculating the mean and standard deviation of data items, and identifying data that exceeds the mean plus or minus a preset multiple of the standard deviation as outlier data, which is then directly removed; Missing data supplementation uses a time-series interpolation algorithm, first analyzing the historical time-series variation pattern of the data item. If the data shows a linear variation trend, linear interpolation is used to supplement missing values; if the data shows a non-linear variation trend, polynomial interpolation is used to supplement missing values; After supplementation, the standardized dataset is subjected to integrity verification. If the verification passes (i.e., the proportion of missing data is lower than a preset proportion), the subsequent steps are performed; if the verification fails, the data collection and preprocessing process is re-executed.

[0037] II. Analysis of the Intent for Multi-Source Data Fusion and Scheduling Requirements Multi-source data fusion calibration generates standardized datasets Collect virtual power plant data: energy data (photovoltaic output data D_pv1, D_pv2, energy storage charging and discharging data D_st1, D_st2), operating status data (inverter temperature D_t1, D_t2, dispatch command execution data D_ex1, D_ex2), and grid connection interaction data (grid-connected power D_p1, D_p2, voltage data D_u1, D_u2).

[0038] Adopt a multi-source data fusion and calibration algorithm: After classifying by data type, for the multi-source acquisition results (D_pv1, D_pv2) of the same data item (such as photovoltaic output), calculate the weights according to the accuracy levels of the sensing devices (denoted as q1, q2) and the historical data error rates (denoted as e1, e2): w1=(q1×(1 - e1)) / (q1×(1 - e1)+q2×(1 - e2)), w2 = 1 - w1, and the fused basic data D_pvr = w1×D_pv1+w2×D_pv2.

[0039] Retrieve the historical operation baseline data D_pvj of this data item, calculate the deviation value ΔD_pv = |D_pvr - D_pvj|. If ΔD_pv ≤ ΔD0 (ΔD0 is the deviation threshold), then retain D_pvr; otherwise, recalibrate the weights. At the same time, eliminate abnormal data (such as D_u1 > D_umax or D_u1 < D_umin, where D_umax and D_umin are the reasonable voltage ranges), and use the time series interpolation algorithm (interpolation step size is t) to supplement missing data, finally forming the standardized data set D_b.

[0040] Dispatch demand intention recognition and list construction Deploy a dispatch demand intention recognition model: Use the historical interaction data of three institutions (sample size is m), the power grid operation status data (dimension is p), and the dispatch task description keyword corpus (number of words is n) as training data to train the model parameters.

[0041] When Institution B submits the task description "Obtain the distribution network load regulation data this week", extract the keywords (denoted as K1, K2), combine with the current power grid operation status data (such as the distribution network load rate D_load) and the historical interaction preferences of Institution B (such as the need to supplement load prediction related data every Wednesday), and the model outputs the potential demand: "Need to synchronously provide distribution network load prediction data (for the next t days) and associated energy storage regulation advice data". Integrate this demand with the basic data requirements of Institution B (such as conventional power data, equipment operation data) to form the exclusive data demand list L_B of Institution B. Similarly, generate the list L_A of Institution A and the list L_C of Institution C.

[0042] Specifically, the adaptive mapping rule generation algorithm generates conversion logic for data type differences as follows: First, it obtains the preset data type standards (including numerical unit standards and data dimension definition standards) of each dispatching agency and establishes a correspondence table between the standardized data of the virtual power plant and the data types of the dispatching agency. For numerical unit differences, it automatically generates unit conversion logic based on the unit conversion formula in the correspondence table. The conversion logic includes a dynamic update mechanism for the unit conversion coefficient. When the dispatching agency updates the unit standard, the conversion coefficient is automatically updated synchronously. For data dimension differences, if the dispatching agency requires data dimensions higher than the standardized data dimensions of the virtual power plant (i.e., more refined data is required), it generates data dimension splitting logic to decompose the standardized data into dimension data that meets the requirements according to preset splitting rules (such as splitting by time granularity or equipment grouping granularity). If the dispatching agency requires data dimensions lower than the standardized data dimensions of the virtual power plant (i.e., more aggregated data is required), it generates data dimension aggregation logic to merge the standardized data into dimension data that meets the requirements according to preset aggregation rules (such as aggregation by day / week or aggregation by region).

[0043] Specifically, the adaptive mapping rule generation algorithm generates processing rules for data precision differences as follows: First, it parses the precision requirements (including the allowable range of data error and the data sampling frequency requirements) in the data requirement list of each scheduling agency; for high precision requirements (i.e., the allowable range of error is less than a preset threshold and the sampling frequency requirement is higher than a preset frequency), it generates original data extraction rules, which clearly select original data in standardized data whose timestamps match the scheduling requirements and whose data integrity meets the standards, and directly pushes them to the scheduling agency; for low precision requirements (i.e., the allowable range of error is greater than or equal to a preset threshold and the sampling frequency requirement is lower than a preset frequency), it generates data dimensionality reduction processing rules, which include specific methods for data sampling (extracting data according to a preset sampling interval) or data mean aggregation (calculating the mean according to a preset time period), reducing the amount of data through dimensionality reduction processing, while ensuring that the error of the processed data is within the allowable range of the scheduling agency.

[0044] Specifically, the specific implementation process of the cross-institutional mapping conflict coordination mechanism is as follows: First, deploy a conflict detection module to monitor in real time the call requests of multiple scheduling institutions for the same data source. When it is detected that two or more scheduling institutions request to call the same data source within the same time period, it is determined as a mapping conflict; start the conflict coordination process. First, obtain the scheduling task priorities and the urgency of data requirements of each requesting institution, and determine the coordination priority ranking in combination with the levels of the scheduling institutions (for example, the priority of the regional power grid scheduling institution is higher than that of the distribution network scheduling institution); According to the ranking results, allocate the data source access time sequence for each scheduling institution. The high-priority scheduling institution preferentially obtains the data access permission, and the access duration is set according to the data processing requirements. The low-priority scheduling institution obtains the permission in turn after the high-priority institution finishes accessing; At the same time, feedback the access time sequence allocation result to each scheduling institution, inform it of the specific time window for data acquisition, and avoid subsequent conflicts. If a scheduling institution cannot obtain data within the allocated time window, restart the coordination process to adjust the time sequence.

[0045] III. Adaptive Mapping Rule Construction and Conflict Coordination Adaptive Mapping Rule Generation and Rule Library Construction Based on L_A, L_B, and L_C, introduce an adaptive mapping rule generation algorithm: Data type difference processing: Institution A requires "grid-connected power data (unit: MW)", and the grid-connected power unit in the standardized data of the virtual power plant is kW. Generate the conversion logic: D_p_A = D_p_standard × 10^ (-3) (The conversion coefficient is 10^ (-3) ); Institution B requires "distribution network partition load data", and the standardized data of the virtual power plant is the overall load data. Generate the dimension splitting logic: D_load_Bi = D_load_standard × f_i (f_i is the proportion of the partition load, Σf_i = 1).

[0046] Data accuracy difference processing: Analyze "the allowable range of fault response data error ≤ ΔE1" in L_A (ΔE1 < ΔE0, ΔE0 is the accuracy threshold), and generate the original data extraction rule: directly call the fault-related original data in D_standard; Analyze "the allowable range of statistical data error ≥ ΔE2" in L_C (ΔE2 ≥ ΔE0), and generate the data dimensionality reduction rule: perform random sampling with a sampling ratio g (0 < g < 1), and the data volume after dimensionality reduction D_C volume = D_standard volume × g.

[0047] Integrate the above mapping relationships into a mapping rule library R, where Institution A corresponds to the subset R_A, Institution B corresponds to R_B, and Institution C corresponds to R_C.

[0048] Cross-institutional Mapping Conflict Coordination Real-time monitoring of data source call requests from three institutions: When institution A (task priority p_A) and institution B (task priority p_B) simultaneously call "energy storage charging and discharging raw data" (denoted as D_st), the conflict coordination process is initiated.

[0049] Obtain the task priorities (p_A>p_B) and data urgency of the two organizations (the urgency coefficient of organization A is e_A, that of organization B is e_B, e_A>e_B), and combine them with the organizational hierarchy (organization A is at a higher level than organization B) to generate a coordination priority ranking: organization A>organization B.

[0050] Assign data source access sequence: Institution A accesses D_st first (access duration is t_A), and after t_A ends, Institution B accesses it (access duration is t_B), and t_A+t_B≤T (T is the total duration that the data source can occupy at one time). The sequence allocation result (t_A, t_B) is fed back to the two institutions.

[0051] Specifically, the process of predicting the trend of scheduling demand changes in advance is as follows: First, the historical task data of the scheduling agency is statistically analyzed according to the time dimension (daily, weekly, monthly, seasonal) to extract the periodic change pattern of task types (such as the increase in the frequency of summer load forecasting tasks and the frequent adjustment of power scheduling tasks before and after holidays); then, combined with the characteristics of the power grid operation cycle (such as the power grid maintenance cycle and the seasonal fluctuation cycle of new energy output), a time series prediction model is used to predict the changes in scheduling task types and data demand parameters within a preset time period in the future; when the prediction results show that the demand probability of a certain type of task exceeds the preset probability threshold, or the change amplitude of a certain data demand parameter exceeds the preset amplitude threshold, it is determined that there is a trend of scheduling demand change, and the pre-update process of the data demand list and mapping rule subset is initiated in advance. The pre-update content needs to be adjusted according to the predicted task type and parameter changes. After adjustment, it is temporarily stored in the backup database. When the scheduling agency formally proposes a demand change, the pre-update content is directly called and verified. After verification, the formal update is completed.

[0052] Specifically, the process of predicting the risk of link performance degradation in advance is as follows: First, historical performance data (including transmission rate, packet loss rate, and latency) of each communication link is collected through the link status monitoring module to establish a link performance degradation assessment model; when the model is running, the changing trend of historical performance data is analyzed, and the degradation rate of link performance parameters (transmission rate, packet loss rate, and latency) is calculated; when it is predicted that the performance parameters of a certain link will drop below a preset performance threshold within a preset time period in the future, it is determined that there is a risk of performance degradation; for links with risks, their transmission protocol parameters are adjusted in advance (such as optimizing the TCP window size and adjusting the data frame length), or link expansion operations are initiated (such as increasing bandwidth resources and deploying relay nodes); after adjustment or expansion, the performance changes are continuously monitored through the link status monitoring module. If the performance recovers to above the threshold, the current configuration is maintained; if the performance does not meet the standard, the adjustment plan is further optimized until the risk is eliminated.

[0053] Specifically, the implementation process of the mapping effect feedback iteration mechanism is as follows: First, a feedback data collection form is designed, which includes data accuracy satisfaction score (1-5 points), transmission timeliness score (1-5 points), data integrity feedback (meets / does not meet), and optimization suggestion fields. The form is periodically pushed to various dispatching agencies to collect feedback data. At the same time, statistical data call efficiency indicators are collected on the virtual power plant side, including data processing time (the time from receiving the request to generating the mapping data), mapping rule call success rate (the proportion of the number of times the rule was successfully called to generate data to the total number of calls), and data retransmission rate (the proportion of the number of times retransmission is required due to data problems to the total number of transmissions). (Proportion of times); at each preset optimization cycle, a comprehensive analysis of feedback data and efficiency indicators is conducted. If the data accuracy satisfaction score is lower than the preset score, or the transmission timeliness score is lower than the preset score, or the data processing time exceeds the preset time threshold, or the mapping rule call success rate is lower than the preset success rate threshold, then the mapping rule base optimization process is initiated. During optimization, the conversion coefficients in the conversion logic are adjusted, the judgment conditions of the filtering rules are corrected, and the frequency of the push mechanism is optimized. After optimization, a pilot verification is conducted in a small-scale scheduling agency. After successful verification, the optimized rule base is fully applied, and the optimization log records the adjusted content and effects.

[0054] IV. Predictive Updates and Iterative Optimization of Mapping Effects Scheduling requirements and predictive updates of link performance Prediction of the Trend of Scheduling Requirement Changes: Statistically analyze the historical task data of the three institutions according to the time dimension (with a period of T0), extract the change rules of task types (such as institution A adds the requirement of "monthly grid connection plan data" at the beginning of each month), combine the characteristics of the power grid operation cycle (such as the peak power consumption cycle is T1), and use a time series prediction model (with a prediction window of t_pre) to predict that within the next t_pre: institution A will add the requirement of "peak-hour fault response data", and institution B will increase the "update frequency of distribution network load data". Pre-update L_A to L_A' (add the peak fault data item), and R_A to R_A' (add the frequency adaptation rule); pre-update L_B to L_B' (adjust the update frequency parameter), and R_B to R_B' (optimize the data push cycle).

[0055] Prediction of the Risk of Link Performance Degradation: Obtain the historical performance data of each link (transmission rates v_hist1, v_hist2,..., v_histk), establish a link performance degradation evaluation model, and calculate the degradation rate v_dec = (v_hist1 - v_histk) / k (k is the number of historical data points). If it is predicted that the transmission rate of L_B1 will drop below v2 within the next t_r (v_LB1_pre = v_histk - v_dec × t_r < v2), it is determined that there is a degradation risk, and the transmission protocol parameters of L_B1 are adjusted in advance (such as adjusting the frame length to l' = l × h, where h is the frame length coefficient), or the link bandwidth is expanded to b' = b + Δb.

[0056] Iterative Optimization of Mapping Effect Feedback Design a feedback data collection form and push it to the three institutions regularly (with a period of t_feed) to collect feedback data (satisfaction score s_A of institution A, accuracy score s_B of institution B, timeliness score s_C of institution C). At the same time, count the efficiency indicators on the virtual power plant side: data processing time t_proc, mapping rule call success rate r_suc, and data retransmission rate r_re.

[0057] Every optimization period t_opt, calculate the comprehensive optimization index S = α1×(s_A + s_B + s_C) / 3 + α2×(1 - t_proc / t_proc0) + α3×r_suc + α4×(1 - r_re) (α1, α2, α3, α4 are weight coefficients, satisfying Σαi = 1; t_proc0 is the time consumption threshold). If S < S0 (S0 is the qualified threshold), then optimize the mapping rule library: adjust the unit conversion coefficient in R_A, the data splitting ratio in R_B, and the sampling frequency in R_C, and at the same time optimize the data push mechanism (such as shortening the data push interval of institution C to t_push' = t_push × k_push, k_push < 1). After the iteration, re-monitor the value of S until S ≥ S0.

[0058] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A data communication and mapping method between a virtual power plant and multiple dispatching agencies, characterized in that, include: S1: Based on the differences in scheduling hierarchy and data interaction requirements of the scheduling agency, an independent communication link including a main link and a backup link is established for the scheduling agency. At the same time, a dynamic link priority allocation mechanism is introduced, and a data desensitization-permission hierarchical binding security mechanism is adopted in the data transmission link. S2: Acquire internal energy data, operating status data, and grid connection interaction data of the virtual power plant; introduce a multi-source data fusion calibration algorithm to generate a standardized dataset and remove abnormal data; based on the scheduling task type of the scheduling agency, deploy a scheduling demand intent recognition model; identify potential task demands through historical interaction data of the scheduling agency, task description keywords, and current grid operating status, and integrate them into the data demand list of the scheduling agency. S3: Based on the data requirement list, an adaptive mapping rule generation algorithm is introduced to generate mapping relationships and integrate them into a callable mapping rule library. The scheduling agency corresponds to an independent subset of mapping rules. Based on the priority of scheduling tasks and the urgency of data requirements, data source access sequence is allocated, and a cross-organizational mapping conflict coordination mechanism is established. S4: By analyzing the historical task change patterns and power grid operation cycle characteristics of the dispatching agency, the trend of dispatching demand changes and the risk of link performance degradation can be predicted in advance, and the demand list and mapping rule subset of the corresponding dispatching agency can be updated in advance before the demand changes. Collect data usage feedback from various dispatching agencies, combine it with data call efficiency indicators from the virtual power plant side, regularly optimize the conversion logic, filtering rules and push mechanism in the mapping rule base, and establish a mapping effect feedback and iteration mechanism.

2. The method according to claim 1, characterized in that, In S1, the specific implementation process of the dynamic link priority allocation mechanism is as follows: the urgency of scheduling tasks is divided into three levels: emergency fault response, regular power scheduling and non-real-time statistics, and a corresponding bandwidth allocation ratio and transmission rate threshold are preset for each level. When the scheduling task type changes or the link load exceeds the preset load threshold, the priority of the corresponding link and the bandwidth allocation ratio are dynamically adjusted.

3. The method according to claim 1, characterized in that, In S1, the data desensitization-permission hierarchical binding security mechanism includes: first, classifying the transmitted data according to the sensitivity level, which is divided into core sensitive data, regular sensitive data, and non-sensitive data; adopting differentiated desensitization methods for data of different sensitivity levels; and dividing the scheduling agency permissions into three levels: core permissions, regular permissions, and query permissions.

4. The method according to claim 1, characterized in that, In S2, the specific process of the multi-source data fusion calibration algorithm includes: classifying the acquired virtual power plant data into three categories according to type: energy data, operating status data, and grid connection interaction data; classifying the multi-source sensor acquisition results in each category; calculating the weight value of each acquisition result based on the accuracy level of the sensor device and the historical data error rate for the multi-source acquisition results of the same data item; and calculating the fused basic data through a weighted summation formula; and retrieving the historical operating baseline data of the same data item of the virtual power plant and calculating the deviation value from the basic data.

5. The method according to claim 1, characterized in that, In S2, the construction and operation process of the scheduling demand intent recognition model includes: training the intent recognition model using the historical interaction data of the scheduling agency, the power grid operation status data, and the scheduling task description keyword corpus as training data; extracting keywords from the task description text currently submitted by the scheduling agency, and then combining the current power grid operation status data with the historical interaction preferences of the scheduling agency to output the preliminary recognition result of the potential task demand.

6. The method according to claim 1, characterized in that, In S2, the specific steps for generating the abnormal data removal of the standardized dataset include: by calculating the mean and standard deviation of the data items, data that exceed the range of the mean plus or minus a preset multiple of the standard deviation are identified as abnormal data and directly removed; for missing data supplementation, a time-series interpolation algorithm is used to analyze the historical time-series change pattern of the data item and supplement the missing values.

7. The method according to claim 1, characterized in that, In S3, the adaptive mapping rule generation algorithm generates conversion logic for data type differences, including: obtaining the data type standard preset by the dispatching agency and establishing a correspondence table between the standardized data of the virtual power plant and the data type of the dispatching agency; generating unit conversion logic for numerical unit differences; and generating data dimension splitting logic and data dimension aggregation logic for data dimension differences.

8. The method according to claim 1, characterized in that, In S3, the adaptive mapping rule generation algorithm generates processing rules for data accuracy differences, including: parsing the accuracy requirements in the data requirement list of the scheduling agency; generating original data extraction rules for errors with an allowable range less than a preset threshold and sampling frequency requirements higher than a preset frequency; and generating data dimensionality reduction processing rules for errors with an allowable range greater than or equal to a preset threshold and sampling frequency requirements lower than a preset frequency. The rules include specific methods for data sampling or data mean aggregation, thereby reducing the amount of data through dimensionality reduction processing.

9. The method according to claim 1, characterized in that, In S3, the specific implementation steps of the cross-agency mapping conflict coordination mechanism include: real-time monitoring of the scheduling agency's call requests to the same data source; when a mapping conflict occurs, initiating a conflict coordination process; obtaining the scheduling agency's scheduling task priority and data demand urgency; generating a coordination priority ranking result based on the scheduling agency hierarchy; allocating data source access sequence according to the coordination priority ranking result; and feeding back the access sequence allocation result to the scheduling agency.

10. The method according to claim 1, characterized in that, In S4, the process of predicting the trend of scheduling demand changes in advance includes: performing statistical analysis on the historical task data of the scheduling agency according to the time dimension to extract the periodic change pattern of task types; and then, in combination with the characteristics of the power grid operation cycle, using a time-series prediction mechanism to predict the changes in scheduling task types and data demand parameters within a future preset time period.

11. The method according to claim 1, characterized in that, In S4, the process of predicting the risk of link performance degradation in advance includes: acquiring historical performance data of the communication link, establishing a link performance degradation assessment model, analyzing the changing trend of the historical performance data, and calculating the degradation rate of the link performance parameters; when it is predicted that the performance parameters of a certain link will drop below a preset performance threshold within a preset time period in the future, it is determined that there is a risk of performance degradation.

12. The method according to claim 1, characterized in that, In S4, the implementation steps of the mapping effect feedback iteration mechanism include: designing a feedback data collection form, periodically pushing it to the scheduling agency to collect feedback data; simultaneously, statistical data call efficiency indicators on the virtual power plant side, including data processing time, mapping rule call success rate, and data retransmission rate; and performing comprehensive analysis of the feedback data and efficiency indicators at preset optimization intervals.

Images