Integrated management method of distributed power prediction system

Abstract

The invention belongs to the technical field of new energy power generation prediction and monitoring, and provides an integrated management method for distributed power prediction systems, which comprises the following steps: acquiring operation monitoring data of each distributed power prediction system; performing data architecture analysis on each distributed power prediction system, and establishing a three-level data model taking station-equipment-parameter as a data architecture; performing cross-system data association and aggregation operation on the operation monitoring data of each distributed power prediction system according to the three-level data model to obtain global integration parameters; and according to the operation monitoring data, the three-level data model and the global integration parameters, establishing a multi-class data statistical graph, and visually displaying the multi-class data statistical graph in the same interface. According to the scheme provided by the invention, the global collaboration, response timeliness and accuracy of the management link of the distributed power prediction system are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of new energy power generation prediction and monitoring, and particularly relates to an integrated management method of a distributed power prediction system. BACKGROUND

[0002] With the rapid development of the new energy industry in the direction of scale and distribution, the number of new energy sites such as wind farms and photovoltaic power stations has increased rapidly, and presents the distribution characteristics of small, scattered, remote and multiple. In order to realize local power generation prediction and basic monitoring, each site generally configures an independent distributed power prediction system. However, the contradiction between the existing distributed management mode and the intensive management and control demand is increasingly prominent, and it is difficult to meet the requirements of efficient operation and maintenance of new energy power stations and safe and stable operation of power grids. The main technical problems are as follows: Firstly, the data management is seriously fragmented. Each distributed power prediction system uses heterogeneous communication protocols and data formats, forming a data island. Management personnel need to log in to multiple independent systems to obtain the predicted power, equipment status, weather data and other information of different sites. Moreover, the data lacks a unified standard and cannot be directly interchanged and analyzed, resulting in low efficiency of cross-site data integration and difficulty in supporting global management decisions.

[0003] Secondly, the system monitoring lacks a global perspective. Under the existing mode, the running state and core parameters of each distributed system need to be monitored through independent interfaces. Management personnel cannot quickly grasp the overall operation situation of multiple sites, and the cross-site collaborative analysis and fault positioning capability is weak.

[0004] Thirdly, the data aggregation capability is missing. Due to the lack of standardized data architecture support, the existing system cannot efficiently complete cross-site and multi-dimensional aggregation calculation, thereby limiting the overall operation response efficiency and prediction accuracy.

[0005] Therefore, the traditional distributed management scheme has the technical problems of poor global collaborative analysis capability, insufficient data aggregation capability, and response efficiency and accuracy that cannot meet the actual demand. SUMMARY

[0006] The present application provides an integrated management method of a distributed power prediction system to solve the defects of traditional distributed management schemes, such as poor global collaborative analysis capability, insufficient data aggregation capability, and response efficiency and accuracy that cannot meet the actual demand.

[0007] The present application provides an integrated management method of a distributed power prediction system, comprising: obtaining running monitoring data of each distributed power prediction system; analyzing the data architecture of each distributed power prediction system, and establishing a three-level data model with a site-equipment-parameter data architecture; According to the three-level data model, cross-system data association and aggregation operation is performed on the operation monitoring data of each distributed power prediction system to obtain global integrated parameters; According to the operation monitoring data, the three-level data model and the global integrated parameters, a plurality of data statistical charts are established, and the plurality of data statistical charts are visually displayed in the same interface.

[0008] According to the three-level data model, cross-system data association and aggregation operation is performed on the operation monitoring data of each distributed power prediction system to obtain global integrated parameters; The original operation data of each distributed power prediction system is received. The original operation data of each distributed power prediction system is respectively subjected to integrity check and precision check to obtain checked data. The format of the checked data is converted into a target format to obtain the operation monitoring data of each distributed power prediction system.

[0009] According to the three-level data model, cross-system data association and aggregation operation is performed on the operation monitoring data of each distributed power prediction system to obtain global integrated parameters; A model basic framework is established according to the data architecture of site as top layer, equipment as middle layer and parameter as bottom layer; The static attribute data and data constraint conditions corresponding to each level in the model basic framework are respectively determined, and the data association rules between the levels are determined; According to the static attribute data, the data constraint conditions and the data association rules, a three-level data model with the data architecture of site-equipment-parameter is established.

[0010] According to the three-level data model, cross-system data association and aggregation operation is performed on the operation monitoring data of each distributed power prediction system to obtain global integrated parameters; The operation monitoring data of each distributed power prediction system is mapped into the three-level data model with the site unique identifier, the equipment unique identifier and the parameter unique identifier as the core association keys to obtain a three-level data model after preliminary association; The data in the three-level data model after preliminary association is subjected to time-space consistency check and logical reasonableness check to obtain a three-level data model after cross-system data association.

[0011] According to the three-level data model, cross-system data association and aggregation operation is performed on the operation monitoring data of each distributed power prediction system to obtain global integrated parameters; The operation monitoring data of each distributed power prediction system in the same new energy station is associated with the corresponding station unique identifier in the three-level data model to obtain a station-level association result. The equipment attribution identifier of the same type of equipment in each distributed power prediction system is associated with the equipment unique identifier in the three-level data model to obtain an equipment-level association result. The same type of parameter in the operation monitoring data is associated with the corresponding parameter unique identifier in the three-level data model to obtain a parameter-level association result. According to the station-level association result, the equipment-level association result, and the parameter-level association result, a three-level data model after preliminary association is obtained.

[0012] According to the integrated management method of the distributed power prediction system provided by the application, the operation monitoring data of each distributed power prediction system is aggregated to obtain global integrated parameters, including: The effective parameter data is extracted from the three-level data model after cross-system data association by taking the core association key as a screening condition. The effective parameter data is subjected to attribution consistency verification and data preprocessing to obtain preprocessed data. The preprocessed data is subjected to aggregation operation under a target aggregation dimension to obtain global integrated parameters.

[0013] According to the integrated management method of the distributed power prediction system provided by the application, a plurality of data statistical charts are established according to the operation monitoring data, the three-level data model, and the global integrated parameters, including: According to the three-level data model, a data topology chart for representing the hierarchical relationship among stations, equipment, and parameters is established. According to the operation monitoring data, a data dashboard is established. According to the global integrated data, a data curve chart is established.

[0014] According to the integrated management method of the distributed power prediction system provided by the application, the method further includes: The operation monitoring data is compared with a pre-set data threshold to obtain a comparison result. If it is determined according to the comparison result that there is data anomaly, a multi-party early warning action is triggered.

[0015] According to the integrated management method of the distributed power prediction system provided by the application, the method further includes: After receiving a remote operation instruction initiated by a remote user, the operation authority of the remote user is verified. If the operation permission of the remote user is verified, the data threshold corresponding to the operation monitoring data of at least one distributed power prediction system and / or at least part of the three-level data model is changed in response to the remote operation instruction.

[0016] The integrated management method of the distributed power prediction system provided by the application further comprises: receiving and importing external meteorological data; identifying the site unique identifier in the external meteorological data, and matching the external meteorological data to the power prediction model of the corresponding site according to the site unique identifier; fusing the external meteorological data with the pre-obtained local meteorological data to obtain meteorological fusion data; optimizing the power prediction model according to the meteorological fusion data.

[0017] The integrated management method of the distributed power prediction system provided by the application realizes data standardization and integration by obtaining the operation monitoring data of each distributed power prediction system and establishing a three-level data model, greatly improving data utilization efficiency and management convenience; relying on the three-level data model to carry out cross-system data correlation and aggregation operation can efficiently generate global integrated parameters to provide accurate data support for power grid dispatching decision; based on the operation monitoring data, the three-level data model and the global integrated parameters, multiple types of data statistical charts are established and visualized, a global and intuitive monitoring view is constructed, and management personnel can quickly master the multi-site operation situation and the health status of the prediction system, helping cross-site collaborative analysis and rapid fault positioning, and improving the global collaboration, response timeliness and accuracy of the management link of the distributed power prediction system. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0019] Figure 1 is a flowchart of the integrated management method of the distributed power prediction system provided by the embodiments of the application; Figure 2 is a network architecture schematic diagram of the interaction between the centralized control center and the station A and the station B. DETAILED DESCRIPTION

[0020] In order to make the objects, technical solutions and advantages of the present application clearer, the following will clearly and completely describe the technical solutions in the present application with reference to the drawings in the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall into the protection scope of the present application.

[0021] The details of the integrated management method of the distributed power prediction system provided by the embodiments of the present application will be described below. Figure 1 And Figure 2 The details of the integrated management method of the distributed power prediction system provided by the embodiments of the present application will be described below.

[0022] As Figure 1 shown, the integrated management method of the distributed power prediction system provided by the embodiments of the present application mainly includes the following steps: Step 110: Obtain the operation monitoring data of each distributed power prediction system.

[0023] Specifically, the operation monitoring data mentioned in the present embodiment includes but is not limited to predicted power, meteorological data, equipment state and historical curve.

[0024] Step 120: Analyze the data architecture of each distributed power prediction system, and establish a three-level data model with station-equipment-parameter as the data architecture.

[0025] It can be understood that the establishment of the three-level data model can provide an effective model basis for the storage and subsequent processing of the operation monitoring data of each distributed power prediction system.

[0026] Step 130: According to the three-level data model, perform cross-system data association and aggregation operation on the operation monitoring data of each distributed power prediction system to obtain global integrated parameters.

[0027] In the present embodiment, the three-level data model is combined with cross-system data association and aggregation operation, which can realize global analysis of the data of each distributed power prediction system and improve the global collaborative analysis capability of the data in the management link.

[0028] Step 140: According to the operation monitoring data, the three-level data model and the global integrated parameters, establish multiple types of data statistical charts, and visually display the multiple types of data statistical charts in the same interface.

[0029] In practical application, by visualizing the unified interface of the multiple types of data statistical charts, the key data in the management link can be intuitively presented to the management personnel, and the integrated management effect is further improved.

[0030] Figure 2The network architecture of the centralized control center of the new energy power station interacting with the station A and the station B is shown, and the core is to realize the remote centralized management of the centralized control center to the station A and the station B. Referring to Figure 2 In the core production area of the centralized control center, power prediction workstations, local databases, backbone network switches, power prediction web servers, power prediction database servers, reverse isolation devices, forward isolation devices, longitudinal encryption devices, and routers and other equipment are deployed to ensure data security and efficient processing. The architecture of the station A and the station B is consistent, and they communicate with the centralized control center through routers and other network equipment. The core components include routers, longitudinal encryption devices, firewalls, station acquisition servers, power prediction servers, AGCs, and electric energy calculation devices.

[0031] In an embodiment, the operation monitoring data of each distributed power prediction system is obtained, specifically including: First, the original operation data of each distributed power prediction system is received.

[0032] In the data acquisition link, this embodiment supports mainstream industrial communication protocols such as OPC, Modbus, MQTT, and HTTP. Through protocol conversion and standardized processing, the original operation data of each distributed power prediction system can be accessed.

[0033] Then, the original operation data of each distributed power prediction system is respectively subjected to integrity check and precision check to obtain checked data.

[0034] In the integrity check link, missing data can be identified and filled in using an interpolation algorithm. Specifically, the data value at the missing position can be calculated according to the change trend and other information of the existing data, so that the data is more complete in the time series and other dimensions.

[0035] In the precision check link, abnormal data that exceeds a reasonable range can be identified and removed. For example, for predicted power, a reasonable range can be set according to the rated power and other information of the equipment. If the data value is much higher or lower than this range, it can be determined as an abnormal value and removed.

[0036] Finally, the format of the checked data is converted into a target format to obtain the operation monitoring data of each distributed power prediction system.

[0037] In actual application, the format of the checked data can be uniformly converted into JSON format. JSON is a lightweight data exchange format, easy to read and write, and convenient for data transmission and interaction between different systems. By converting the data into this format, the consistency and accuracy of the data when used in different scenarios can be ensured, facilitating subsequent cross-system data association and aggregation operation.

[0038] In this embodiment, the operation monitoring data can be classified and stored. Specifically, a distributed database can be used to store real-time data in the operation monitoring data, such as a time series database, and a relational database can be used to store structured data such as device parameters and user configurations, so as to realize efficient read and write of massive data and historical traceability function.

[0039] In an embodiment, data architecture analysis is performed on each distributed power prediction system, and a three-level data model with a data architecture of site-device-parameter is established, which specifically includes: First, a model basic framework is established according to the data architecture of site as top layer, device as middle layer, and parameter as bottom layer.

[0040] In this embodiment, the business requirements of centralized management and control of new energy power stations are used as the guide to first determine the hierarchical positioning and core function boundary of the three-level data architecture, and to build the model basic framework. The top layer site takes the new energy station as the core unit, covers physical stations such as wind farms and photovoltaic power stations, and clearly defines it as the highest dimension of data aggregation, which needs to be associated with all subordinate physical devices and prediction system devices; the middle layer device is the hub connecting the site and the parameter, integrates the physical power generation and monitoring devices in the station and the function devices of the distributed prediction system, and forms the belonging bridge between the site and the parameter; the bottom layer parameter focuses on the smallest unit of data collection, covers core business data such as device status, predicted power, and weather data, and serves as the model data basis. By dividing the three-level function positioning of site planning, device association, and parameter bearing, a model basic framework with clear logic and distinct hierarchy can be initially constructed, providing structural support for subsequent attribute definition and association rule design.

[0041] Then, the static attribute data and data constraint conditions corresponding to each level in the model basic framework are determined, and the data association rules between levels are determined.

[0042] In this step, for the three-level model basic framework that has been built, the static attribute data and data constraint conditions of each level can be sorted out, and the association rules between levels are clarified. The static attribute data to be determined at the site level includes the site unique identifier, descriptive information including the site name, energy type, region to which the site belongs, longitude and latitude, management information, etc., and the data constraint conditions include that the site unique identifier is globally unique, the energy type is only supported for wind power, photovoltaic power, and mixed energy, etc.

[0043] The static attribute data of the device layer includes the device unique identifier, the unique identifier of the belonging site, the device type, the device model, the rated parameter, etc., and the data constraint conditions include that the device unique identifier is globally unique, the device type needs to match the site energy type, etc.

[0044] The static attribute data of the parameter layer includes a parameter unique identifier, a home device unique identifier, a parameter type, a data unit, a parameter name, and the like, and the data constraint condition is that the parameter unique identifier is globally unique, the parameter type needs to be matched with the device function, and the like.

[0045] Meanwhile, in the step of determining the inter-level association rule, the site layer and the device layer establish a 1:N association through the site unique identifier and the home site unique identifier, that is, one site corresponds to multiple devices, the device layer and the parameter layer establish a 1:N association through the device unique identifier and the home device unique identifier, that is, one device corresponds to multiple parameters, and the association field needs to satisfy the constraint condition that the home unique identifier of the child level must exist in the unique identifier list of the parent level, so as to ensure the accuracy of the data home among levels.

[0046] Finally, according to the static attribute data, the data constraint condition, and the data association rule, a three-level data model with a site-device-parameter data architecture is established.

[0047] Based on the determined static attribute data, the data constraint condition, and the inter-level association rule, the logical framework is converted into a three-level data model that can be implemented through database selection and table structure design. First, a relational database is selected to store the static data of the site layer and the device layer, and a time series database is selected to store the dynamic data of the parameter layer. In the relational database, a site table and a device table are created, the fields are defined according to the static attributes, and the inter-level association rule is implemented through foreign key constraints. In the time series database, a parameter time series table is created, the label field and the value field are defined, and the static attribute and the dynamic data storage requirement of the parameter layer are matched. Then, the static attribute data of each level is imported, and finally, the three-level data model is formed, which provides a structured data basis for cross-system data association and aggregation operation.

[0048] In an embodiment, according to the three-level data model, the operation monitoring data of each distributed power prediction system is cross-system associated, specifically including: First, the operation monitoring data of each distributed power prediction system is mapped into the three-level data model by taking the site unique identifier, the device unique identifier, and the parameter unique identifier as the core association key, to obtain a three-level data model after preliminary association.

[0049] In a specific implementation, the operation monitoring data of each distributed power prediction system is mapped into the three-level data model by taking the site unique identifier, the device unique identifier, and the parameter unique identifier as the core association key, to obtain a three-level data model after preliminary association, specifically including: On the one hand, the operation monitoring data of each distributed power prediction system in the same new energy station is associated with the corresponding site unique identifier in the three-level data model, to obtain a site-level association result.

[0050] In this step, the unique site identifiers of all new energy power plants can be extracted from the site layer of the three-level data model first, clarifying the basic information such as the site name and energy type corresponding to each identifier. Then, the operation monitoring data uploaded by each distributed power prediction system within the same new energy power plant are preprocessed to filter out the implicit site affiliation information, such as the site code in the data packet header and the site identifier carried in the communication protocol. Subsequently, a mapping relationship between the site affiliation information and the unique site identifiers in the model is established. The operation monitoring data of all distributed power prediction systems within the site are bound to the corresponding unique site identifiers through a matching algorithm, ensuring that each piece of monitoring data is clearly attributed to a specific site in the three-level model. Finally, a site-level association result containing the unique site identifiers and the set of operation monitoring data is generated, thereby completing the initial classification of data at the site dimension.

[0051] On the other hand, the device affiliation identifiers of similar devices in each distributed power prediction system are associated with the unique device identifiers in the three-level data model to obtain device-level association results.

[0052] In this step, we can first sort out the equipment attribution identifiers of similar devices in each distributed power prediction system, and at the same time extract the unique equipment identifiers of all devices and their corresponding unique attribution site identifiers, equipment types and other attribute data from the equipment layer of the three-level data model; then, we can classify the equipment attribution identifiers of the distributed power prediction system according to equipment type, and establish matching rules between equipment attribution identifiers and unique equipment identifiers for each type of equipment; through rule verification, we can associate the attribution identifiers of similar devices in each distributed power prediction system with the unique equipment identifiers in the model one by one, forming the equipment-level association results between the unique equipment identifiers and the monitoring data of similar devices, so as to achieve accurate positioning of data at the equipment dimension.

[0053] On the other hand, parameters of the same type in the operational monitoring data are associated with the corresponding unique identifiers of the parameters in the three-level data model to obtain parameter-level association results.

[0054] In this step, the operation monitoring data of each distributed power prediction system can be classified into parameters first, and parameters of the same type can be identified and their names and data units can be standardized. Then, the unique identifiers of all parameters and their corresponding unique identifiers of the affiliated equipment and parameter types can be extracted from the parameter layer of the three-level data model. Subsequently, for each type of parameter, a correspondence between the monitoring data parameter type and the model parameter unique identifier is established. Through data format verification, the operation monitoring data of the same type is associated with the corresponding parameter unique identifier, generating parameter-level association results between the parameter unique identifier and the monitoring data of the same type of parameter, thus completing the standardized matching of data in the parameter dimension.

[0055] Finally, based on the site-level association results, device-level association results, and parameter-level association results, a three-level data model after preliminary association is obtained.

[0056] This step requires integrating the site-level, device-level, and parameter-level association results. Based on the original data architecture of the three-level data model, the unique site identifier in the site-level association results and the operation monitoring dataset are used as the top-level data support of the model to ensure that all monitoring data is first attributed to the corresponding site. Next, the unique device identifier in the device-level association results and the monitoring data of similar devices are embedded in the middle layer of the model. Through the attribution relationship between the unique device identifier and the unique site identifier, the associated data of the device dimension is bound to the corresponding site, forming a two-level association structure of site-device. Then, the unique parameter identifier in the parameter-level association results and the monitoring data of similar parameters are filled into the bottom layer of the model. Based on the attribution relationship between the unique parameter identifier and the unique device identifier, the associated data of the parameter dimension is bound to the corresponding device, constructing a complete association chain of site-device-parameter, generating a three-level data model after preliminary association, and realizing the deep integration of distributed system operation monitoring data with each level of the model.

[0057] The second step is to perform spatiotemporal consistency verification and logical rationality verification on the data within the three-level data model after initial association, so as to obtain the three-level data model after cross-system data association.

[0058] In the spatiotemporal consistency verification stage, parameter monitoring data of all devices under the same site can be extracted from the three-level data model after initial association. First, time consistency is verified. Based on the preset unified time zone in the three-level data model, the sampling timestamp deviation of all data is checked to see if it is within the set time threshold, such as within 1 minute, to avoid time misalignment caused by communication delay. Data exceeding the deviation range is marked as time anomaly. Then, spatial consistency is verified. Specifically, based on the association relationship between the unique identifier of the affiliated site of the middle-level device and the unique identifier of the top-level site in the three-level data model, the device unique identifier and the unique identifier of the affiliated site corresponding to each parameter data are checked to see if they are consistent with the unique identifier of the site to which the data is bound. Data with contradictory affiliations are marked as spatial anomaly.

[0059] After completing the spatiotemporal consistency verification, further logical rationality verification can be performed. On the one hand, based on the rated parameters of the equipment layer and the parameter constraint range of the parameter layer in the three-level data model, parameter data exceeding the constraint range is filtered out and marked as numerical anomalies. On the other hand, combined with the business logic of new energy power generation, such as the actual power of photovoltaic inverters should be close to 0 when the irradiance is 0, and the predicted power should be 0 when the wind turbine wind speed is lower than the cut-in wind speed, the logical coherence between related data is verified, and data that violates the business logic is marked as logical anomalies. Finally, all marked anomaly data is temporarily stored in the anomaly data pool, and a verification report is generated synchronously. The verification report includes the site unique identifier, equipment unique identifier, parameter type, and anomaly reason of the anomaly data, while compliant data is retained in the three-level data model, forming a three-level data model after cross-system data association, providing an accurate and consistent data foundation for subsequent aggregation operations and visualization.

[0060] In one embodiment, the operational monitoring data of each distributed power prediction system are aggregated to obtain global integration parameters, specifically including: First, using the core association key as the filtering condition, effective parameter data is extracted from the three-level data model after cross-system data association.

[0061] Using site unique identifiers, device unique identifiers, and parameter unique identifiers as core association keys, and combining user needs or business scenarios, the target filtering scope is determined, and filtering is initiated from the three-level data model after cross-system data association. First, at the site level, the target site unique identifiers are matched to lock the range of sites from which data needs to be extracted; then, based on the matched site unique identifiers, the device layer is associated to filter out the device unique identifiers of the target type under these sites; finally, the parameter layer is associated through the device unique identifiers to extract the parameter data corresponding to the parameter unique identifiers of the specified type under these devices, while only retaining parameter data with valid data quality identifiers and sampling timestamps within the target time period, and removing invalid, missing, or expired data to obtain the initially filtered valid parameter data.

[0062] Then, the valid parameter data undergoes consistency verification and data preprocessing to obtain preprocessed data.

[0063] In the attribution consistency verification stage, based on the hierarchical association rules preset in the three-level data model, it can be checked whether the unique identifier of the attribution device corresponding to the unique identifier of the parameter in each piece of extracted valid parameter data matches the unique identifier of the attribution site of the device. At the same time, the logical matching between the device type and the parameter type is verified, and attribution abnormal data is moved to the abnormal pool while compliant data is retained.

[0064] In the data preprocessing stage, short-term missing values ​​in compliant data, such as missing data in a 1-minute sample, can be filled by interpolation; outliers that exceed the rated range of the equipment or the reasonable range of business operations can be removed or corrected by combining historical data trends and equipment parameter thresholds; finally, the data format is unified to obtain preprocessed data with a unified format and complete data.

[0065] Finally, the preprocessed data is aggregated under the target aggregation dimension to obtain the global integration parameters.

[0066] In this embodiment, it is necessary to first clarify the target aggregation dimension and the corresponding global integration parameter requirements, and formulate the aggregation calculation rules for each indicator. For example, the total predicted power is calculated by summing all preprocessed data within the same time period. The average error rate can be calculated by first calculating the error rate of a single device within a single time period, and then calculating the arithmetic mean of the error rates of all devices within the same region within the same time period. Subsequently, the preprocessed data is grouped according to the target aggregation dimension, and aggregation operations are performed within each group according to the calculation rules. For example, when calculating the total predicted power of a certain region for a certain hour, the preprocessed data of the predicted power of all wind turbines in the region within that hour can be summarized and summed to obtain the total predicted power for that time period. When calculating the average error rate, the error rate of each wind turbine in the region for that hour is calculated first, and then the extreme values ​​with error rates exceeding 30% are excluded before averaging.

[0067] After the initial calculations are completed, the calculation results under each aggregation dimension are organized into structured global integration parameters. The global integration parameters include information such as aggregation dimension identifier, indicator name, value, calculation period, and amount of data involved in the calculation. These parameters are stored in a distributed database and synchronized to the application layer, thereby providing data support for subsequent visualization and management decisions.

[0068] In one embodiment, multiple types of data statistical charts are established based on operational monitoring data, a three-level data model, and global integration parameters, specifically including: On the one hand, based on the three-level data model, a data topology diagram is established to represent the hierarchical relationship between sites, devices and parameters.

[0069] This step centers on the hierarchical relationship of the three-level data model. Using a visualization development tool, the unique identifiers of each site at the site level are first mapped to top-level nodes in the topology diagram, with different colors distinguishing energy types (e.g., blue for wind farms, yellow for photovoltaic power plants). Next, the unique identifiers of each device at the equipment level are used as mid-level nodes, establishing connections between them and the corresponding top-level site nodes based on their respective site identifiers, and using different icons to distinguish equipment types (e.g., wind turbines for wind turbines, current for inverters). Finally, the unique identifiers of each parameter at the parameter level are used as bottom-level nodes, associating them with the corresponding mid-level device nodes based on their respective device identifiers, and labeling the core parameter information. Interactive node functions are also implemented; for example, clicking a site node expands its subordinate devices and parameters, while clicking a parameter node displays real-time data. Ultimately, a data topology diagram that intuitively presents the hierarchical relationships between sites, devices, and parameters is generated, facilitating rapid data attribution.

[0070] On the other hand, data dashboards are established based on operational monitoring data.

[0071] In this step, key operational monitoring data for each site and equipment can be extracted from the three-level data model after cross-system association. Based on the monitoring needs of management personnel, core indicator modules for the dashboard are determined, such as site operation overview, equipment health status, and real-time parameter monitoring. In the dashboard development tool, visualization components are designed for each indicator module. For example, digital cards can be used to display core values ​​such as the overall equipment online rate and current total predicted power; color-coded indicator lights can indicate equipment status (e.g., green for online, red for fault); and progress bars can display the comparison between parameter values ​​and thresholds, such as the prediction error rate as a percentage of the threshold. Simultaneously, a real-time data refresh mechanism is set up to ensure that dashboard indicators are consistent with actual operating status, forming a data dashboard that intuitively reflects the real-time operation of the site.

[0072] On the other hand, data curves are created based on globally integrated data.

[0073] In this step, indicators requiring trend analysis can be selected from the globally integrated parameters obtained through aggregation calculations, such as the daily total predicted power for a certain region and the weekly average error rate of a single station, to determine the time and comparison dimensions of the curve charts. In the charting tool, with the time axis as the X-axis and the global integrated parameter values ​​as the Y-axis, differentiated curve styles are set for data of different comparison dimensions; data annotation functions and trend warning lines are added, and multi-curve overlay comparison is supported. Finally, a data curve chart that clearly reflects the trend and comparison relationship of the global integrated parameters over time is generated, providing support for operational pattern analysis and prediction accuracy assessment.

[0074] In one embodiment, the integrated management method for the above-described distributed power prediction system may further include: First, the operational monitoring data is compared with the pre-set data thresholds to obtain the comparison results.

[0075] In practical applications, operational monitoring data can be extracted in batches by level from the three-level data model after cross-system association, while pre-set corresponding data thresholds can be retrieved from the model parameter layer or system configuration module. A one-to-one correspondence between parameters and thresholds is then established. Through automated comparison algorithms, such as numerical comparison and cumulative time comparison, each piece of operational monitoring data is verified against its corresponding threshold in real time. The normal or abnormal comparison results for each piece of data are recorded, and the anomaly type and specific deviation value are labeled, forming a structured comparison result.

[0076] Then, if the comparison results indicate that there is data anomaly, multiple warning actions will be triggered.

[0077] If the comparison result of a certain data item or group is abnormal in this step, and a second verification confirms that there is a real data anomaly, the early warning process will be automatically triggered.

[0078] In the early warning process, standardized early warning information can be generated first, including the three-level model identifier of the anomaly association, the anomaly type, the time of the anomaly occurrence, the specific deviation value, and the snapshot of the associated data; then, multiple early warning actions can be initiated simultaneously.

[0079] Specifically, the system interface can pop up tiered pop-ups, trigger voice broadcasts in the duty room, and push early warning messages to designated terminals of management personnel, with the messages including an anomaly location link. In addition, the early warning information and processing status are synchronously written to the system log and the anomaly recording module of the three-level data model, supporting subsequent traceability and statistical analysis, ensuring that anomaly information reaches relevant personnel in a timely manner, and saving time for rapid fault diagnosis and handling.

[0080] In one embodiment, the integrated management method for the above-described distributed power prediction system may further include: First, upon receiving a remote operation command initiated by a remote user, the operation permissions of the remote user are verified.

[0081] After receiving a remote operation command from an administrator, maintenance personnel, or other remote user requesting to modify the prediction error alarm threshold of a wind farm or update the rated power parameters of equipment, the system first extracts the user identification information and operation content identifier carried in the remote operation command. Then, it retrieves the preset role and permission list for that remote user. For example, the administrator role has permission to modify parameters at all sites, while ordinary operators only have permission to view the status of equipment at a specified site. The system then verifies the matching of permissions against the operation content identifier.

[0082] On one hand, it verifies whether the user role has the permissions for the corresponding operation type. On the other hand, it verifies whether the user's permission scope covers the site, device, and parameters associated with the command. At the same time, it also verifies the legality of the operation command, such as whether there are problems such as incorrect command format or non-existent target identifier. If the permissions match and the command is legal, a permission verification result is generated; otherwise, a prompt message indicating insufficient permissions or invalid command is returned and the operation log is recorded.

[0083] Then, if the remote user's operation permission verification passes, in response to the remote operation command, the data thresholds and / or at least some data in the three-level data model corresponding to the operation monitoring data of at least one distributed power prediction system are modified.

[0084] In this embodiment, if the remote user's operation permission verification passes, the remote operation command is first encrypted. The SSL / TLS protocol can be used to ensure the security of command transmission. At the same time, the command content is digitally signed to prevent tampering. Then, data changes are performed according to the command operation type.

[0085] For instructions to modify data thresholds, the system can locate the threshold configuration field of the corresponding parameter in the three-level data model parameter layer, or the local threshold storage module of the distributed power prediction system. The system updates the relevant configuration according to the new threshold value in the instruction and automatically records traceability information such as the value before and after the threshold modification, the modification time, and the operator.

[0086] For instructions that modify data in the Level 3 data model, the system accesses the relational database storing the static data of the model, locates the target data field at the corresponding site or device layer, and performs a data update operation. After the data change is completed, the update results are synchronously fed back to the remote user interface, such as displaying that the threshold has been successfully modified and the model data has been updated. At the same time, the changed data is synchronized to the associated distributed power prediction system, and the complete operation process, including the instruction content, the data before and after the change, and the execution results, is written to an immutable operation log for subsequent auditing and traceability.

[0087] In one embodiment, the integrated management method for the above-described distributed power prediction system may further include: First, receive and import external meteorological data.

[0088] In practical applications, external meteorological data (such as meteorological departments or third-party meteorological service platforms) can be received through a pre-defined standardized data interface. During import, the core fields of the external meteorological data are first extracted through the data parsing module, such as station identifier, collection time, wind speed, wind direction, irradiance, temperature, and precipitation. Then, the data undergoes preliminary format validation, such as checking whether the number of fields matches and whether the data type conforms to the definition. If there are format errors, an exception message is returned and the import is rejected. After the validation passes, the external meteorological data is temporarily stored in a temporary data buffer to prepare for subsequent station matching and data fusion.

[0089] Then, the unique identifier of the station in the external meteorological data is identified, and the external meteorological data is matched to the power prediction model of the corresponding station based on the unique identifier of the station.

[0090] In this embodiment, the unique identifier of a station can be extracted from the external meteorological data in the temporary buffer. Then, the station-level data of the three-level data model is called, and a matching algorithm is used to associate the unique identifier of the station in the external meteorological data with the unique identifier of the station stored in the three-level data model to determine the new energy power station corresponding to the external meteorological data. Next, according to the preset binding relationship between the station and the power prediction model, the external meteorological data of the matched station is pushed to the data input port of the power prediction model of that station, completing the accurate matching between the external meteorological data and the power prediction model of the corresponding station, ensuring that the data is only used for the prediction calculation of the corresponding station.

[0091] Subsequently, external meteorological data is fused with pre-acquired local meteorological data to obtain fused meteorological data.

[0092] In this embodiment, local meteorological data of the corresponding station can be extracted from the parameter layer of the three-level data model or the database of the local monitoring equipment of the station. Then, the external meteorological data and the local meteorological data are spatiotemporally aligned. Then, the same type of meteorological parameters of the same station and the same period are fused and calculated by the data fusion algorithm. Specifically, the weighted average method can be used to calculate and generate meteorological fusion data that takes into account both external predictability and local real-time performance. Finally, the meteorological fusion data is checked for rationality, outliers are removed, and the data is stored in the data source library of the station power prediction model.

[0093] Finally, the power prediction model was optimized based on meteorological fusion data.

[0094] In practical applications, meteorological fusion data can be used as the core input and substituted into the power prediction model of the corresponding station. First, the parameter learning module built into the power prediction model is used to compare the prediction results before and after the meteorological fusion data input. If the deviation is reduced, the meteorological fusion data is deemed effective. Then, the internal calculation parameters of the model are adjusted to make the model more in line with the power generation pattern under the current meteorological conditions.

[0095] like Figure 2 As shown, for the main station's external network, after obtaining data from the core production area through a forward isolation device, it connects to the external meteorological system, which includes a meteorological center, firewall, and meteorological data server, to introduce external meteorological data and provide a basis for optimizing the power prediction model.

[0096] In summary, compared with the prior art, the integrated management method for distributed power prediction systems provided by this invention has at least the following beneficial effects: First, it breaks through the protocol barriers of decentralized systems, achieves data standardization and centralized storage, eliminates information silos, and improves data utilization.

[0097] Secondly, a unified interface enables centralized display of the status and parameters of multiple systems, supports cross-site collaborative analysis, and reduces several management complexities.

[0098] Third, it integrates multiple alarm methods such as messages, voice, and pop-ups to ensure timely response to alarm information and improve fault handling efficiency.

[0099] Fourth, it enables centralized operation of remote parameter configuration and program start / stop, reducing on-site operation and maintenance costs and shortening response time.

[0100] Fifth, it supports the standardized import of external meteorological data, improves the input accuracy of power prediction models, and optimizes prediction results.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An integrated management method for a distributed power prediction system, characterized in that, include: Obtain operational monitoring data from each distributed power prediction system; Data architecture analysis was performed on each distributed power prediction system, and a three-level data model with site-equipment-parameter as the data architecture was established; Based on the aforementioned three-level data model, cross-system data association and aggregation operations are performed on the operation monitoring data of each distributed power prediction system to obtain global integrated parameters; Based on the operational monitoring data, the three-level data model, and the global integration parameters, multiple types of data statistical charts are established and visualized on the same interface.

2. The integrated management method for the distributed power prediction system according to claim 1, characterized in that, Obtain operational monitoring data for each distributed power prediction system, including: Receive raw operating data from each distributed power prediction system; The integrity and accuracy of the original operating data of each distributed power prediction system are verified to obtain the verified data. The format of the verified data is converted to the target format to obtain the operation monitoring data of each distributed power prediction system.

3. The integrated management method for a distributed power prediction system according to claim 1, characterized in that, The data architecture of each distributed power prediction system is analyzed, and a three-level data model with site-equipment-parameter as the data architecture is established, including: The basic framework of the model is established based on a data architecture with the site as the top layer, the device as the middle layer, and the parameter as the bottom layer; Determine the static attribute data and data constraints corresponding to each level in the basic framework of the model, and determine the data association rules between each level; Based on the static attribute data, data constraints, and data association rules, a three-level data model with a site-device-parameter data architecture is established.

4. The integrated management method for the distributed power prediction system according to claim 1, characterized in that, Based on the aforementioned three-level data model, cross-system data correlation is performed on the operational monitoring data of each distributed power prediction system, including: Using the unique site identifier, unique device identifier, and unique parameter identifier as the core association keys, the operation monitoring data of each distributed power prediction system are mapped to the three-level data model to obtain the three-level data model after preliminary association. The data within the three-level data model after initial association are subjected to spatiotemporal consistency verification and logical rationality verification to obtain the three-level data model after cross-system data association.

5. The integrated management method for a distributed power prediction system according to claim 4, characterized in that, Using site unique identifiers, device unique identifiers, and parameter unique identifiers as core association keys, the operation monitoring data of each distributed power prediction system are mapped to the three-level data model, resulting in a preliminary associated three-level data model, including: The operation monitoring data of each distributed power prediction system within the same new energy power station are associated with the unique site identifier in the three-level data model to obtain the site-level association result. The device affiliation identifier of similar devices in each distributed power prediction system is associated with the unique device identifier in the three-level data model to obtain the device-level association result; The parameters of the same type in the operation monitoring data are associated with the corresponding unique identifiers of the parameters in the three-level data model to obtain parameter-level association results; Based on the site-level association results, the device-level association results, and the parameter-level association results, a three-level data model after preliminary association is obtained.

6. The integrated management method for a distributed power prediction system according to claim 4, characterized in that, The operational monitoring data of each distributed power prediction system are aggregated and calculated to obtain global integration parameters, including: Using the core association key as the filtering condition, effective parameter data is extracted from the three-level data model after cross-system data association; The valid parameter data is subjected to affiliation consistency verification and data preprocessing to obtain preprocessed data; The preprocessed data is aggregated under the target aggregation dimension to obtain global integration parameters.

7. The integrated management method for a distributed power prediction system according to claim 1, characterized in that, Based on the operational monitoring data, the three-level data model, and the global integration parameters, multiple types of data statistical charts are established, including: Based on the aforementioned three-level data model, a data topology diagram is established to represent the hierarchical relationship between sites, devices, and parameters; Based on the aforementioned operational monitoring data, a data dashboard is established; Based on the aforementioned globally integrated data, a data curve graph is created.

8. The integrated management method for a distributed power prediction system according to claim 1, characterized in that, The method further includes: The operational monitoring data is compared with a pre-set data threshold to obtain the comparison result; If the comparison results indicate that there is data anomaly, then multiple warning actions will be triggered.

9. The integrated management method for a distributed power prediction system according to claim 1, characterized in that, The method further includes: Upon receiving a remote operation command initiated by a remote user, the operation permissions of the remote user are verified. If the remote user's operation permission verification passes, then in response to the remote operation command, the data threshold corresponding to the operation monitoring data of at least one distributed power prediction system and / or at least some data in the three-level data model are modified.

10. The integrated management method for a distributed power prediction system according to claim 1, characterized in that, The method further includes: Receive and import external meteorological data; Identify the unique station identifier in the external meteorological data, and match the external meteorological data to the power prediction model of the corresponding station based on the unique station identifier; The external meteorological data is fused with the pre-acquired local meteorological data to obtain fused meteorological data; The power prediction model is optimized based on the meteorological fusion data.

Images