Ultra-short-term wind power error correction method and system

Abstract

The invention discloses an ultra-short-term wind power error correction method and system, and relates to the technical field of power dispatching, and the method comprises the steps: collecting the operation data, prediction data and meteorological data of each wind power plant in a region, evaluating the power error of each wind power plant based on the operation data and the prediction data, and extracting the time domain feature of the power error; on the basis of the time domain features and the meteorological data, predicting spatial propagation of errors among the wind power plant clusters, and generating error spatial propagation prediction information; based on the power error and the error space propagation prediction information, generating a cooperative control instruction set oriented to different spectrum characteristic error layers; and performing security verification on the cooperative control instruction set and issuing execution, monitoring the execution effect of the cooperative control instruction set, and evaluating the overall correction performance. The whole-process closed-loop optimization management of the wind power error from space-time prediction to hierarchical correction is realized.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology, and in particular to a method and system for correcting ultra-short-term wind power errors. Background Technology

[0002] As wind power penetration in power systems continues to increase, real-time correction of ultra-short-term wind power forecasting errors has become a core technology for ensuring the safe and economical operation of the power grid. To address the randomness and volatility of wind power output, existing technologies generally employ centralized regulation methods based on automatic generation control frameworks. This involves monitoring the net load of the grid or the overall output deviation of the wind farm, generating unified control commands using optimization algorithms, and scheduling various rapid regulation resources for power compensation. In single wind farms or small-scale grid-connected scenarios, this approach provides fundamental technical support for maintaining system frequency stability and power balance, constituting the current mainstream error correction paradigm.

[0003] Faced with the centralized grid connection of large-scale wind farm clusters and the increasing diversification of flexible adjustment resources in the power system, the limitations of existing correction methods are becoming increasingly apparent. Current strategies fail to deeply analyze and utilize the deterministic spatial correlation and propagation patterns between wind power errors and wind farm clusters under specific meteorological conditions. Furthermore, they fail to finely identify and specifically process the dynamic characteristics of different time scales inherent in the error signals. This results in a lack of forward-looking and collaborative defense capabilities against cross-site error propagation in the spatial context, and difficulty in achieving optimal matching between the spectral characteristics of errors and the dynamic performance of adjustment resources in resource scheduling. Consequently, this restricts further improvement in overall correction efficiency and may increase unnecessary adjustment costs. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an ultra-short-term wind power error correction method that solves the problems of insufficient spatial coordination and low spectrum matching accuracy of existing methods.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an ultra-short-term wind power error correction method, which includes collecting operational data, forecast data and meteorological data of each wind farm in the region, evaluating the power error of each wind farm based on the operational data and the forecast data, and extracting the time-domain features of the power error. Based on the time-domain characteristics and the meteorological data, the spatial propagation of the prediction error among wind farm clusters is predicted, and error spatial propagation prediction information is generated. Based on the power error and the error space propagation prediction information, a cooperative control instruction set is generated for error stratification with different spectral characteristics; The collaborative control instruction set is subjected to security verification and executed, the execution effect of the collaborative control instruction set is monitored, and the overall correction performance is evaluated. Based on the evaluation results of the implementation effect, the parameters of the prediction and control models are adaptively updated.

[0007] As a preferred embodiment of the ultra-short-term wind power error correction method of the present invention, the method includes: collecting operational data, forecast data, and meteorological data of each wind farm in the region; evaluating the power error of each wind farm based on the operational data and the forecast data; and extracting the time-domain features of the power error, comprising the following steps: The system simultaneously acquires real-time active power values, ultra-short-term wind power prediction curves, real-time wind speed and direction data, and numerical weather forecast products for each wind farm in the region through a data acquisition and monitoring system, a wind power prediction system, and a numerical weather forecast system. The power error sequence of each wind farm is evaluated by taking the real-time active power value of each wind farm in the region and the corresponding ultra-short-term wind power prediction curve as input. Using the power error sequence of each wind farm as input, the instantaneous amplitude, rate of change, direction of duration, and time-domain characteristics of the accumulated energy of the power error sequence are extracted.

[0008] As a preferred embodiment of the ultra-short-term wind power error correction method of the present invention, the method includes the following steps: Based on the time-domain characteristics and the meteorological data, the spatial propagation of error among wind farm clusters is predicted, and error spatial propagation prediction information is generated: The time-domain characteristics, real-time wind speed and direction data, and numerical weather prediction products are input into a pre-constructed error contagion network model. The error contagion network model is a directed graph with each wind farm as a node and the historical error propagation probability and intensity between nodes as weights. In the error propagation network model, wind farms whose error amplitude and rate of change in the time domain exceed a preset threshold are marked as error source nodes, and the weights of each directed edge in the error propagation network model are dynamically adjusted based on real-time wind speed and wind direction data. Starting from the error source node, the error contagion network model is driven to simulate the propagation process of error along the directed edges, predict the start time and error intensity contribution value of each downstream wind farm affected, and output error spatial propagation prediction information including downstream wind farm identification, predicted start time of impact and predicted error intensity contribution value.

[0009] As a preferred embodiment of the ultra-short-term wind power error correction method of the present invention, the method comprises: generating a collaborative control instruction set oriented towards different spectral characteristic error layers based on the power error and the error spatial propagation prediction information, including the following steps: The power error sequence of each wind farm is superimposed with the derivative error intensity contribution value of the corresponding wind farm in the error space propagation prediction information to form the target error signal to be corrected for each wind farm. The target error signal to be corrected for each wind farm is then subjected to online wavelet transform to decompose it into high-frequency components, mid-frequency components and low-frequency components. Based on the preset matching relationship between the error component frequency band and the controllable and adjustable resource type, a frequency division and coordination control command set is generated to control the energy storage device to offset the high frequency component, control the automatic generation control unit to track the medium frequency component, and control the thermal power unit and the demand side response to smooth the low frequency component.

[0010] As a preferred embodiment of the ultra-short-term wind power error correction method of the present invention, the method includes the following steps: performing security verification on the coordinated control command set and issuing it for execution. The frequency division and coordinated control command set is superimposed on the current power flow of the power grid to perform online power flow assessment. Based on the results of the online power flow assessment, it is verified whether the power of the key transmission lines exceeds the limit, whether the voltage of the key nodes is within the safe range, and whether the unit frequency stability index meets the requirements. After all verifications are passed, the frequency division collaborative control command set will be issued to the energy storage device, automatic generation control unit, thermal power unit and demand-side response resources for execution.

[0011] As a preferred embodiment of the ultra-short-term wind power error correction method of the present invention, the following steps are included: monitoring the execution effect of the cooperative control instruction set and evaluating the overall correction performance: During the execution cycle of the frequency division and coordinated control instruction set, the actual execution power of energy storage devices, automatic generation control units, thermal power units and demand-side response resources is collected; The actual execution power of energy storage devices, automatic generation control units, thermal power units and demand-side response resources is compared with the frequency division and collaborative control command set to evaluate command execution rate, delay time and regulation accuracy. Based on the actual power of the wind farm after command execution and the ultra-short-term wind power prediction curve of the wind power prediction unit, the overall error reduction rate is obtained as an evaluation index of the overall correction performance.

[0012] As a preferred embodiment of the ultra-short-term wind power error correction method of the present invention, the following steps are included: Based on the evaluation results of the execution effect, the parameters of the prediction and control models are adaptively updated. The error propagation prediction information is compared with the actual error propagation observed between wind farm clusters to obtain the error propagation prediction deviation. The expected adjustment effect of the frequency division coordinated control command set is compared with the actual adjustment effect of the executed power to obtain the adjustment effect deviation. The edge weight parameters in the error contagion network model are periodically corrected by using the error propagation prediction bias, and the resource performance parameters in the matching relationship between the error component frequency band and the controllable adjustable resource type are updated online by using the adjustment effect bias.

[0013] Secondly, the present invention provides an ultra-short-term wind power error correction system, comprising: a data acquisition module, which acquires operational data, forecast data and meteorological data of each wind farm in the region, evaluates the power error of each wind farm based on the operational data and the forecast data, and extracts the time-domain features of the power error; The prediction module, based on the time-domain features and the meteorological data, predicts the spatial propagation of errors among wind farm clusters and generates error spatial propagation prediction information. The control module generates a set of cooperative control instructions for error stratification with different spectral characteristics based on the power error and the error space propagation prediction information. The evaluation module performs security verification on the collaborative control instruction set and issues it for execution, monitors the execution effect of the collaborative control instruction set, and evaluates the overall correction performance. The update module adaptively updates the parameters of the prediction and control models based on the evaluation results of the execution effect.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the ultra-short-term wind power error correction method as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the ultra-short-term wind power error correction method as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: by collecting and processing wind farm data to calculate power error and temporal characteristics, the error propagation network model is used to predict the spatial propagation of error among wind farm clusters and generate propagation prediction information. Subsequently, based on the prediction information and the original error, the error to be corrected is decomposed into different frequency band components through spectrum analysis, and a hierarchical collaborative control instruction set is generated according to the preset matching relationship. After completing the security verification of the instruction set, it is issued for execution and the execution effect and overall performance are monitored and evaluated. Finally, the prediction model and control parameters are adaptively updated based on the evaluation results, thereby realizing the closed-loop optimization management of wind power error from spatiotemporal prediction to hierarchical correction. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method for correcting ultra-short-term wind power errors.

[0019] Figure 2 This is a schematic diagram of an ultra-short-term wind power error correction system. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Reference Figures 1-2 This is one embodiment of the present invention, which provides an ultra-short-term wind power error correction method, including the following steps: S1. Collect operational data, forecast data, and meteorological data of each wind farm in the region. Based on the operational data and forecast data, evaluate the power error of each wind farm and extract the time-domain characteristics of the power error.

[0024] S1.1. Real-time active power values, ultra-short-term wind power prediction curves, real-time wind speed and direction data, and numerical weather forecast products of each wind farm in the region are simultaneously acquired through the data acquisition and monitoring system, wind power prediction system, and numerical weather forecast system.

[0025] Furthermore, data acquisition operations are performed synchronously through the data acquisition and monitoring system, the wind power prediction system, and the numerical weather prediction system. The data acquisition and monitoring system collects and records the real-time active power value at the grid connection point of each grid-connected wind farm in the region at a high-frequency sampling period, such as once per second or every few seconds. The wind power prediction system outputs a series of continuous power values ​​with a time range of several hours in the future and a time resolution of minutes. This series is the ultra-short-term wind power prediction curve. At the same time, the numerical weather prediction system provides high spatiotemporal resolution gridded meteorological data covering the entire region, i.e., numerical weather prediction products, and synchronously accesses the real-time wind speed and wind direction data measured by the wind measuring devices deployed in each wind farm. These data together form the basis for subsequent analysis.

[0026] S1.2. The power error sequence of each wind farm is evaluated by taking the real-time active power value of each wind farm in the region and the corresponding ultra-short-term wind power prediction curve as input.

[0027] Furthermore, the real-time active power values ​​of each wind farm in the region are compared point by point with the predicted values ​​on the ultra-short-term wind power prediction curve corresponding to the same wind farm at the same time point. For each aligned time point, the difference is calculated, that is, the real-time active power value is subtracted from the predicted value of the corresponding ultra-short-term wind power prediction curve, thus obtaining a set of difference sequences arranged in time sequence. This sequence is defined as the power error sequence of the wind farm. Positive values ​​in the power error sequence indicate that the actual power generation at that time is higher than the predicted value, and negative values ​​indicate that the actual power generation is lower than the predicted value.

[0028] S1.3. Using the power error sequence of each wind farm as input, extract the time-domain characteristics of the instantaneous amplitude, rate of change, direction of duration, and cumulative energy of the power error sequence.

[0029] Furthermore, for the power error sequence of each wind farm, calculations are performed within a continuous time window to extract its characteristic quantities. Specifically, this includes calculating the arithmetic mean of the absolute values ​​of the power error sequence within the selected time window, which serves as the instantaneous amplitude characteristic representing the strength of fluctuations; the standard deviation of the differences between consecutive time points in the power error sequence, used to quantify the severity of power changes, i.e., the rate of change characteristic; statistically analyzing the length and frequency of consecutively positive or negative periods in the power error sequence to describe the direction of error persistence; and integrating the absolute value of the power error sequence within the time window to obtain an approximate value of the area enclosed by the time axis, which serves as the cumulative energy characteristic reflecting the total accumulated error over a period of time.

[0030] S2. Based on time-domain characteristics and meteorological data, predict the spatial propagation of errors among wind farm clusters and generate error spatial propagation prediction information.

[0031] S2.1 Input the time-domain characteristics, real-time wind speed and direction data, and numerical weather forecast products into the pre-constructed error propagation network model. The error propagation network model is a directed graph with each wind farm as a node and the historical error propagation probability and intensity between nodes as weights.

[0032] Furthermore, the instantaneous amplitude, rate of change, direction of persistence, and time-domain characteristics of accumulated energy of the power error sequence, along with real-time wind speed and direction data and numerical weather prediction products, are imported as input data into a pre-constructed error contagion network model. The structure of this error contagion network model is defined as a directed graph, where each independent vertex corresponds to a specific wind farm. The directed edges from the vertex representing one wind farm to the vertex representing another wind farm are used to represent the path through which the error may propagate from the former wind farm to the latter. Each directed edge is assigned one or more weight values, which are calculated based on long-term historical operating data through statistical analysis of factors such as the lead-lag relationship, spatial orientation correlation, and terrain shading effect between the power error sequences of two wind farms. These weight values ​​are used to quantify the probability and intensity of historical error propagation between nodes.

[0033] S2.2 In the error contagion network model, wind farms whose error amplitude and rate of change in the time domain exceed a preset threshold are marked as error source nodes, and the weights of each directed edge in the error contagion network model are dynamically adjusted based on real-time wind speed and wind direction data.

[0034] Furthermore, in the error propagation network model with loaded input data, for each node corresponding to a wind farm, the temporal characteristics of its input are checked. Specifically, the instantaneous amplitude of the power error sequence is compared with a preset amplitude threshold, and the rate of change of the power error sequence is also compared with a preset rate threshold. For wind farm nodes whose instantaneous amplitude exceeds the amplitude threshold and whose rate of change also exceeds the rate threshold, their status is marked as error source nodes in the error propagation network model. Then, based on the current dominant direction and intensity of atmospheric flow reflected by real-time wind speed and direction data, the weights of all directed edges connecting each wind farm node in the error propagation network model are dynamically adjusted. For example, when the real-time wind direction indicates that the wind is blowing from the first wind farm to the second wind farm, the weight of the directed edge from the node representing the first wind farm to the node representing the second wind farm is increased; otherwise, the weight is decreased or maintained. This adjustment enables the network topology to reflect the impact of meteorological conditions on the error propagation path in real time.

[0035] S2.3 Starting from the error source node, drive the error contagion network model to simulate the propagation process of error along the directed edge, predict the start time and error intensity contribution value of each downstream wind farm affected, and output error spatial propagation prediction information including downstream wind farm identification, predicted start time of impact and predicted error intensity contribution value.

[0036] Furthermore, using all error source nodes as the set of starting points for simulated propagation, in the error contagion network model with dynamically adjusted weights, the dynamic process of error propagating from the error source nodes along the directed edges to adjacent downstream nodes is simulated according to the direction of the directed edges and the adjusted weights. During this simulation, the predicted time delay of error propagation to each downstream wind farm node is calculated by combining the weight values ​​of the directed edges and the electrical distance between nodes, which is the predicted start time of the impact. At the same time, the cumulative attenuation is calculated based on the weights of each directed edge on the propagation path to obtain the predicted intensity ratio when the error propagates to the downstream node, which is the error intensity contribution value. Finally, a structured list is compiled and output as error spatial propagation prediction information. Each record in this list clearly contains a unique identifier of a downstream wind farm, the predicted start time of the wind farm being affected by the error, and the predicted error intensity contribution value received from the upstream error source.

[0037] S3. Based on power error and error space propagation prediction information, generate a set of cooperative control instructions for error stratification with different spectral characteristics.

[0038] S3.1. The power error sequence of each wind farm is superimposed with the derivative error intensity contribution value of the corresponding wind farm in the error space propagation prediction information to form the target error signal to be corrected for each wind farm. The target error signal to be corrected for each wind farm is subjected to online wavelet transform and decomposed into high-frequency components, mid-frequency components and low-frequency components.

[0039] Furthermore, for each wind farm within the region, the power error sequence of that wind farm is first read. This sequence is a time series containing historical and current error values. Simultaneously, error spatial propagation prediction information is retrieved, and all records marking this wind farm as a downstream affected object are extracted from this information. The predicted error intensity contribution values ​​in these records are converted into a time-aligned future error contribution sequence based on their corresponding predicted impact start time. Then, the power error sequence of the wind farm and all relevant future error contribution sequences are algebraically superimposed on the same time axis to form a new comprehensive time series that integrates its own existing errors and predicted derivative errors. This comprehensive sequence is defined as the target error signal to be corrected for the wind farm. Next, an online wavelet transform algorithm is applied to this target error signal to be corrected. By selecting appropriate wavelet basis functions and decomposition scales, the signal is analyzed in the time and frequency domain using multi-resolution analysis. The fast fluctuation components in the signal are separated into high-frequency components, the medium-speed fluctuation components into medium-frequency components, and the slow-changing trend components into low-frequency components, thus completing the spectral decomposition of the target error signal to be corrected.

[0040] S3.2 Based on the preset matching relationship between the error component frequency band and the controllable and adjustable resource type, generate a frequency division and coordination control command set to control the energy storage device to offset the high frequency component, control the automatic generation control unit to track the medium frequency component, and control the thermal power unit and the demand side response to smooth the low frequency component.

[0041] Furthermore, based on a predefined and stored matching table, which clarifies which type of controllable and adjustable resource should handle the error components of different frequency bands, specifically, the matching relationship stipulates that the high-frequency component is matched with the adjustment capability of the energy storage device, the medium-frequency component is matched with the adjustment capability of the automatic generation control unit, and the low-frequency component is matched with the adjustment capability of the thermal power unit and demand-side response resources. Based on this matching relationship, for each wind farm, corresponding power control commands are generated for the high-frequency, medium-frequency, and low-frequency components obtained after decomposition. The commands generated for the energy storage device aim to make its output power match the high-frequency component. Small equal and opposite directions are used to achieve cancellation. The commands generated to control the automatic generation control units are intended to adjust their output to track the changes in the medium frequency component. The commands generated to control the thermal power units and demand-side response resources are intended to adjust their planned output or load to smooth the trend of the low frequency component. Finally, the power control commands generated for all wind farms, all frequency band components, and all corresponding controllable and adjustable resources in the region are summarized and arranged according to time sequence and resource object to form a unified, time-synchronized command set that includes specific control objects, control time points, and control values. This command set constitutes the final frequency division cooperative control command set.

[0042] S4. Perform security verification on the collaborative control instruction set and issue it for execution.

[0043] S4.1. Overlay the frequency division cooperative control command set onto the current power flow of the power grid in its base state to perform online power flow assessment. Based on the results of the online power flow assessment, verify whether the power of key transmission lines exceeds the limit, whether the voltage of key nodes is within the safe range, and whether the unit frequency stability index meets the requirements.

[0044] Furthermore, the current ground-state power flow data of the power grid is obtained from the power grid energy management system or state estimation function. This ground-state power flow data describes the power distribution of each line in the power grid and the voltage amplitude and phase angle of each node under the current generator output and load levels. All power regulation commands included in the frequency division coordinated control command set, including power commands for controlling energy storage devices, power commands for controlling automatic generation control units, power commands for controlling thermal power units, and power commands for controlling demand-side response resources, are converted into corresponding node injection power changes according to the future execution time points specified in the commands. These injected power changes are then superimposed on the current ground-state power flow of the power grid to form a reflection of the execution. A predictive power flow model of the expected power grid state after the instruction is generated. Based on this predictive power flow model, online power flow calculation and evaluation are performed. After obtaining the evaluation results, three key safety checks are performed. First, it is checked whether the active power of the key transmission lines under this predicted power flow exceeds the line's own thermal stability limit or static safety limit. Second, it is checked whether the voltage amplitude of key nodes, such as the bus of the hub substation, is within the safe operating range formed by the preset upper and lower limits under this predicted power flow. Finally, it is checked whether the frequency stability indicators such as the system frequency change rate and the frequency minimum point calculated by the contingency analysis or frequency response model under the predicted power injection change meet the relevant requirements of the power grid safety and stability guidelines.

[0045] S4.2 After all verifications pass, the frequency division collaborative control command set is sent to the energy storage device, automatic generation control unit, thermal power unit and demand-side response resources for execution.

[0046] Furthermore, after verifying that all three indicators of critical transmission line power, critical node voltage, and system frequency stability have passed (i.e., all within limits, within safe ranges, and meeting requirements), the frequency division coordinated control command set generated in step S3.2 is deemed safe to execute. Subsequently, this safe-to-execute frequency division coordinated control command set is transmitted to its corresponding execution units through the command distribution channel of the power grid dispatch automation system. Specifically, control commands for energy storage devices are distributed to the corresponding energy storage power station control units, control commands for automatic generation control units are distributed to the automatic generation control units of the corresponding power plants, control commands for thermal power units are distributed to the control units of the corresponding thermal power plants, and aggregated control commands for demand-side response resources are distributed to the demand-side response aggregation platform. Each unit and platform then receives and begins executing the power regulation tasks specified in the commands.

[0047] S5. Monitor the execution effect of the collaborative control instruction set and evaluate the overall correction performance.

[0048] S5.1 During the execution cycle of the frequency division and coordinated control instruction set, collect the actual execution power of energy storage devices, automatic generation control units, thermal power units and demand-side response resources.

[0049] Furthermore, the execution period is defined as the time from the moment the frequency division coordinated control instruction set is issued and executed until the time when all the adjustment actions specified in the frequency division coordinated control instruction set are completed. Within this execution period, through the power metering devices deployed at the grid connection points of energy storage devices, automatic generation control units, and thermal power units, as well as the monitoring terminals connected to the demand-side response resource aggregation points, the actual output or absorbed power of energy storage devices, the actual output of automatic generation control units, the actual output of thermal power units, and the actual aggregated power changes of demand-side response resources are continuously collected and recorded at fixed time intervals. These continuously collected data sequences are collectively referred to as the actual executed power of energy storage devices, automatic generation control units, thermal power units, and demand-side response resources. These actual executed powers are the original basis for evaluating the execution effect of the instructions.

[0050] S5.2 Compare the actual execution power of energy storage devices, automatic generation control units, thermal power units and demand-side response resources with the frequency division and collaborative control command set to evaluate command execution rate, delay time and regulation accuracy.

[0051] Furthermore, the actual power data of energy storage devices, automatic generation control units, thermal power units, and demand-side response resources are compared with the pre-set expected power values ​​for each resource at each time point in the frequency division coordinated control command set. Through comparison, the command execution rate is calculated, which refers to the total percentage of energy completion of the trajectory between the actual power and the expected power of the command within the execution cycle. The delay time is calculated, which refers to the time difference from the start time of regulation specified in the command to the moment when the actual power of the energy storage device, automatic generation control unit, thermal power unit, or demand-side response resource first reaches a specific proportion of the expected regulation amount, such as 90 percent. The regulation accuracy is evaluated, which refers to the root mean square error or maximum deviation value of the fluctuation of the actual power around the expected power of the command during the steady-state regulation phase. The command execution rate, delay time, and regulation accuracy together constitute a quantitative evaluation of the execution effect of the frequency division coordinated control command set.

[0052] S5.3. Based on the actual power of the wind farm after the command is executed and the ultra-short-term wind power prediction curve of the wind power prediction unit, the overall error reduction rate is obtained as an evaluation index of the overall correction performance.

[0053] Furthermore, within an evaluation time window after the completion of the frequency-division coordinated control command set, actual power data of all wind farms in the region are collected, and ultra-short-term wind power prediction curves published by the wind power prediction unit within the same time period are obtained. The actual power data of each wind farm within the evaluation time window is subtracted point by point from the corresponding ultra-short-term wind power prediction curve of the wind farm within the same time period to calculate the wind power error sequence after the execution of the command. This wind power error sequence after execution is compared with the original wind power error sequence within the same evaluation time window before the execution of the frequency-division coordinated control command set. The overall error reduction rate is obtained by calculating the degree of improvement of the original error sequence and the error sequence after execution on the evaluation index. This evaluation index is usually the root mean square value or mean absolute error of the error. The overall error reduction rate is the ratio of the original error evaluation index value minus the error evaluation index value after execution to the original error evaluation index value. This overall error reduction rate is used as the final evaluation index to measure the overall correction performance of wind power error after the implementation of the frequency-division coordinated control command set.

[0054] S6. Based on the evaluation results of the execution effect, adaptively update the parameters of the prediction and control models.

[0055] S6.1 Compare the error space propagation prediction information with the actual error propagation observed between wind farm clusters to obtain the error propagation prediction deviation. Compare the expected adjustment effect of the frequency division coordinated control command set with the actual adjustment effect of the executed power to obtain the adjustment effect deviation.

[0056] Furthermore, from the error spatial propagation prediction information, prediction data regarding downstream wind farm identification, predicted start time of impact, and predicted error intensity contribution are extracted. Simultaneously, after the monitoring period ends, based on the actually collected wind farm power data, the actual power error between wind farm clusters is analyzed in terms of temporal occurrence and spatial transfer patterns. The true path of error propagation between wind farm clusters, the true start time of impact, and the true error intensity transmission relationship are then inverted and recorded. This record represents the actual error propagation situation observed between wind farm clusters. The downstream wind farm identification, predicted start time of impact, and predicted error intensity contribution for each prediction item in the error spatial propagation prediction information are compared one by one with the true start time of impact and the true error intensity contribution for the same wind farm in the actual error propagation situation observed between wind farm clusters. The time difference between the predicted start time and the actual start time is recorded. The absolute value of the error difference, as well as the relative error or absolute difference between the predicted error intensity contribution value and the actual error intensity contribution value, together constitute the error propagation prediction bias. On the other hand, the expected power adjustment amount specified in the frequency division coordinated control instruction set for energy storage devices, automatic generation control units, thermal power units, and demand-side response resources at each time point is defined as the expected adjustment effect of the frequency division coordinated control instruction set. At the same time, the difference between the actual executed power of energy storage devices, automatic generation control units, thermal power units, and demand-side response resources at the actual corresponding time point and the initial power state of these resources before the execution of the instruction is defined as the adjustment effect of the actual executed power. The expected adjustment effect of the frequency division coordinated control instruction set and the adjustment effect of the actual executed power are compared on the same time coordinate, and the root mean square error or average absolute deviation of the two on the adjustment amount time series is calculated. This deviation is the adjustment effect deviation.

[0057] S6.2. Periodically correct the edge weight parameters in the error contagion network model by using the error propagation prediction bias, and update the resource performance parameters in the matching relationship between the error component frequency band and the controllable adjustable resource type online by using the adjustment effect bias.

[0058] Furthermore, the edge weight parameters in the error contagion network model are periodically corrected by utilizing the prediction bias of error propagation. Specifically, for each directed edge in the error contagion network model, if there is an error propagation relationship between the upstream and downstream wind farms it connects in actual observation, the weight parameters of this directed edge are adjusted in reverse according to the degree of deviation between the prediction error intensity contribution value reflected in the current error propagation prediction bias and the true value, based on a preset learning rate, such as a decimal between zero and one. If the predicted value is higher than the true value, the weight is reduced proportionally; if the predicted value is lower than the true value, the weight is increased proportionally. If the deviation between the predicted start time and the true start time is large, it may also trigger the adjustment of relevant parameters reflecting the propagation time characteristics. This correction process is not performed in real time, but rather after accumulating a certain number of deviation samples, such as every 24 hours or after completing 100 predictions, a batch update is performed. Simultaneously, by utilizing the adjustment effect deviation, the resource performance parameters implicit or stored in the matching relationship between the error component frequency band and the controllable adjustable resource type are updated online. For example, for energy storage devices matching high-frequency components, their performance parameters may include response delay time and upper limit of adjustment rate. Based on the difference between the actual response delay time and the expected command time, and the difference between the actual adjustment rate and the command-required rate revealed by the current adjustment effect deviation, the nominal values ​​of the response delay time and upper limit of adjustment rate of the energy storage device in the matching relationship are dynamically fine-tuned. For automatic generation control units matching medium-frequency components, their performance parameters may include adjustment dead zone and tracking error coefficient. These parameter values ​​are updated based on the deviation between the actual adjustment effect and the expected effect. These parameter updates based on adjustment effect deviation are performed in near real-time to ensure that resource performance parameters that better fit the actual situation can be used when generating the frequency division coordinated control command set next time.

[0059] This embodiment also provides an ultra-short-term wind power error correction system, including: a data acquisition module, which collects the operation data, forecast data and meteorological data of each wind farm in the area, evaluates the power error of each wind farm based on the operation data and forecast data, and extracts the time-domain characteristics of the power error; The prediction module, based on time-domain characteristics and meteorological data, predicts the spatial propagation of errors among wind farm clusters and generates error spatial propagation prediction information. The control module generates a set of cooperative control instructions for error stratification with different spectral characteristics, based on power error and error space propagation prediction information. The evaluation module performs security verification on the collaborative control instruction set and issues it for execution, monitors the execution effect of the collaborative control instruction set, and evaluates the overall corrective performance. The update module adaptively updates the parameters of the prediction and control models based on the evaluation results of the execution effect.

[0060] This embodiment also provides a computer device applicable to the ultra-short-term wind power error correction method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the ultra-short-term wind power error correction method proposed in the above embodiment.

[0061] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0062] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the ultra-short-term wind power error correction method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0063] In summary, this invention collects and processes wind farm data to calculate power error and temporal characteristics. Then, it uses an error contagion network model to predict the spatial propagation of errors among wind farm clusters and generates propagation prediction information. Subsequently, based on the prediction information and the original error, the error to be corrected is decomposed into different frequency band components through spectrum analysis, and a hierarchical collaborative control instruction set is generated according to a preset matching relationship. After completing the security verification of the instruction set, it is issued for execution and the execution effect and overall performance are monitored and evaluated. Finally, the prediction model and control parameters are adaptively updated based on the evaluation results, thereby realizing the closed-loop optimization management of wind power error from spatiotemporal prediction to hierarchical correction.

[0064] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for correcting ultra-short-term wind power error, characterized in that: This includes collecting operational data, forecast data, and meteorological data from each wind farm within the region; evaluating the power error of each wind farm based on the operational data and forecast data; and extracting the time-domain features of the power error. Based on the time-domain characteristics and the meteorological data, the spatial propagation of the prediction error among wind farm clusters is predicted, and error spatial propagation prediction information is generated. Based on the power error and the error space propagation prediction information, a cooperative control instruction set is generated for error stratification with different spectral characteristics; The collaborative control instruction set is subjected to security verification and executed, the execution effect of the collaborative control instruction set is monitored, and the overall correction performance is evaluated. Based on the evaluation results of the implementation effect, the parameters of the prediction and control models are adaptively updated.

2. The ultra-short-term wind power error correction method as described in claim 1, characterized in that: The process involves collecting operational data, forecast data, and meteorological data from various wind farms within a region, evaluating the power error of each wind farm based on the operational data and forecast data, and extracting the time-domain characteristics of the power error. This includes the following steps: The system simultaneously acquires real-time active power values, ultra-short-term wind power prediction curves, real-time wind speed and direction data, and numerical weather forecast products for each wind farm in the region through a data acquisition and monitoring system, a wind power prediction system, and a numerical weather forecast system. The power error sequence of each wind farm is evaluated by taking the real-time active power value of each wind farm in the region and the corresponding ultra-short-term wind power prediction curve as input. Using the power error sequence of each wind farm as input, the instantaneous amplitude, rate of change, direction of duration, and time-domain characteristics of the accumulated energy of the power error sequence are extracted.

3. The ultra-short-term wind power error correction method as described in claim 2, characterized in that: Based on the aforementioned time-domain characteristics and meteorological data, the spatial propagation of prediction errors among wind farm clusters is predicted, generating spatial propagation prediction information for errors, including the following steps: The time-domain characteristics, real-time wind speed and direction data, and numerical weather prediction products are input into a pre-constructed error contagion network model. The error contagion network model is a directed graph with each wind farm as a node and the historical error propagation probability and intensity between nodes as weights. In the error propagation network model, wind farms whose error amplitude and rate of change in the time domain exceed a preset threshold are marked as error source nodes, and the weights of each directed edge in the error propagation network model are dynamically adjusted based on real-time wind speed and wind direction data. Starting from the error source node, the error contagion network model is driven to simulate the propagation process of error along the directed edges, predict the start time and error intensity contribution value of each downstream wind farm affected, and output error spatial propagation prediction information including downstream wind farm identification, predicted start time of impact and predicted error intensity contribution value.

4. The ultra-short-term wind power error correction method as described in claim 3, characterized in that: Based on the power error and the error spatial propagation prediction information, a cooperative control instruction set is generated for error stratification with different spectral characteristics, including the following steps: The power error sequence of each wind farm is superimposed with the derivative error intensity contribution value of the corresponding wind farm in the error space propagation prediction information to form the target error signal to be corrected for each wind farm. The target error signal to be corrected for each wind farm is then subjected to online wavelet transform to decompose it into high-frequency components, mid-frequency components and low-frequency components. Based on the preset matching relationship between the error component frequency band and the controllable and adjustable resource type, a frequency division and coordination control command set is generated to control the energy storage device to offset the high frequency component, control the automatic generation control unit to track the medium frequency component, and control the thermal power unit and the demand side response to smooth the low frequency component.

5. The ultra-short-term wind power error correction method as described in claim 4, characterized in that: The process of performing security verification on the collaborative control instruction set and issuing it for execution includes the following steps: The frequency division and coordinated control command set is superimposed on the current power flow of the power grid to perform online power flow assessment. Based on the results of the online power flow assessment, it is verified whether the power of the key transmission lines exceeds the limit, whether the voltage of the key nodes is within the safe range, and whether the unit frequency stability index meets the requirements. After all verifications are passed, the frequency division collaborative control command set will be issued to the energy storage device, automatic generation control unit, thermal power unit and demand-side response resources for execution.

6. The ultra-short-term wind power error correction method as described in claim 5, characterized in that: Monitoring the execution effect of the collaborative control instruction set and evaluating the overall corrective performance includes the following steps: During the execution cycle of the frequency division and coordinated control instruction set, the actual execution power of energy storage devices, automatic generation control units, thermal power units and demand-side response resources is collected; The actual execution power of energy storage devices, automatic generation control units, thermal power units and demand-side response resources is compared with the frequency division and collaborative control command set to evaluate command execution rate, delay time and regulation accuracy. Based on the actual power of the wind farm after command execution and the ultra-short-term wind power prediction curve of the wind power prediction unit, the overall error reduction rate is obtained as an evaluation index of the overall correction performance.

7. The ultra-short-term wind power error correction method as described in claim 6, characterized in that: Based on the evaluation results of the implementation effect, the parameters of the prediction and control models are adaptively updated, including the following steps: The error propagation prediction information is compared with the actual error propagation observed between wind farm clusters to obtain the error propagation prediction deviation. The expected adjustment effect of the frequency division coordinated control command set is compared with the actual adjustment effect of the executed power to obtain the adjustment effect deviation. The edge weight parameters in the error contagion network model are periodically corrected by using the error propagation prediction bias, and the resource performance parameters in the matching relationship between the error component frequency band and the controllable adjustable resource type are updated online by using the adjustment effect bias.

8. A short-term wind power error correction system, based on the short-term wind power error correction method according to any one of claims 1 to 7, characterized in that: It includes a data acquisition module that collects operational data, forecast data, and meteorological data of each wind farm in the area, evaluates the power error of each wind farm based on the operational data and forecast data, and extracts the time-domain features of the power error. The prediction module, based on the time-domain features and the meteorological data, predicts the spatial propagation of errors among wind farm clusters and generates error spatial propagation prediction information. The control module generates a set of cooperative control instructions for error stratification with different spectral characteristics based on the power error and the error space propagation prediction information. The evaluation module performs security verification on the collaborative control instruction set and issues it for execution, monitors the execution effect of the collaborative control instruction set, and evaluates the overall correction performance. The update module adaptively updates the parameters of the prediction and control models based on the evaluation results of the execution effect.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the ultra-short-term wind power error correction method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the ultra-short-term wind power error correction method according to any one of claims 1 to 7.

Images