A virtual power plant management method based on multi-source data fusion

Abstract

The application relates to the technical field of power plant management, and discloses a virtual power plant management method based on multi-source data fusion, which is used for solving the problem that response delay occurs between devices when virtual power plant management is performed, and comprises the following steps: obtaining device response behavior characteristic parameters of each distributed energy device according to an initial device operation sequence library; obtaining a response lag index according to the device response behavior characteristic parameters; determining whether a non-synchronous response lag effect occurs for each distributed energy device according to the response lag index; screening all response lag devices; performing delay compensation on all response lag devices; obtaining an actual device operation sequence library; obtaining a fusion input matrix according to the actual device operation sequence library; extracting main characteristic dimensions based on the fusion input matrix; constructing a state characteristic vector sequence; and training a time sequence prediction model, so that the accuracy of virtual power plant management is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of power plant management technology, and more specifically to a virtual power plant management method based on multi-source data fusion. Background Technology

[0002] Virtual power plants, as an important form of aggregation and coordinated management of distributed energy systems, have become a key supporting means in new power systems and integrated energy dispatching systems. They are widely used in various scenarios such as photovoltaic power generation, wind power generation, energy storage regulation, electric vehicle charging, and controllable load management, involving multiple engineering and technical fields such as energy management, power electronics, dispatch optimization, edge computing, and communication protocols.

[0003] In the construction and operation of virtual power plants, to achieve full-process management from distributed resource registration, operational data collection, situation assessment, optimized scheduling to control feedback closed loop, it is necessary to rely on a multi-source data fusion mechanism to build a unified management platform for heterogeneous devices, heterogeneous protocols, and asynchronous communication. This platform collaboratively completes tasks such as multi-device status monitoring, dispatchability assessment, and the generation and execution verification of refined control strategies. Simultaneously, with the deepening development of concepts such as "multi-energy collaboration," "source-load interaction," and "cloud-edge collaboration," virtual power plant platforms are gradually evolving from "static aggregation platforms" to "dynamic prediction and response platforms."

[0004] In recent years, data acquisition protocols, feature standardization methods, and machine learning-based scheduling strategy generation engines for various types of distributed resources have seen significant development. Existing technologies typically predict future system states by establishing a unified data access structure, constructing device operating state vectors, and using time series prediction models, thereby building a multi-device coordinated scheduling instruction set. Scheduling platforms usually rely on a fusion input matrix for unified modeling, combining data from multiple devices at the same point in time as model input to improve the synergy between prediction and control.

[0005] However, the above-mentioned technologies have at least the following technical problems:

[0006] In real-world operating environments, distributed devices exhibit inherent differences in their response times to control commands. For example, energy storage systems typically execute charging and discharging commands rapidly, while load-side responses may exhibit behavioral inertia or control lag. If such causal response time delays between devices are not considered, and all state data are directly aligned to a uniform time step and used for modeling, it can lead to misalignments in the logical relationships between feature data. These implicit lags cannot be detected through conventional data integrity checks. If directly used to construct a fused input matrix, they will disrupt the model's temporal consistency, mislead state assessment and scheduling path generation, and consequently affect prediction accuracy and the effectiveness of control strategies. Summary of the Invention

[0007] In order to overcome the above-mentioned defects of the prior art, the present invention provides a virtual power plant management method based on multi-source data fusion to solve the problems existing in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] A virtual power plant management method based on multi-source data fusion includes the following steps: Step 1: Obtain the set of identification codes for each distributed energy device connected to the virtual power plant, and obtain the equipment information for each distributed energy device, and establish a unified data access structure based on the equipment information; Step 2: Preset a time step, and based on the data access structure, collect historical equipment operating parameters within a set time period according to the time step. The equipment operating parameters include photovoltaic output power, energy storage charge status, and load response power. Construct a historical structured data table based on the historical equipment operating parameters, and establish an initial equipment operation sequence library using the equipment identification code as an index; Step 3: Obtain the equipment response behavior characteristic parameters for each distributed energy device according to the initial equipment operation sequence library. The equipment response behavior characteristic parameters include control command trigger records, state switching records, and equipment operating parameters. Step 4: Response lag index is obtained by evaluating behavioral characteristic parameters. Based on the response lag index, it is determined whether each distributed energy device exhibits asynchronous response lag effects, and all devices with response lag are screened out. Step 5: Delay compensation is performed on all devices with response lag to obtain a library of actual device operation sequences. A fusion input matrix is ​​obtained based on this library. Step 6: Based on the fusion input matrix, the main feature dimensions are extracted to construct a sequence of state feature vectors, and a time series prediction model is trained. Step 7: The state feature vectors of the most recent several time steps are obtained, and the time series prediction model is used to predict the predicted values ​​of the state feature vectors for the next several time steps. A set of dispatch response intervals for the virtual power plant is constructed based on the predicted values ​​of the state feature vectors. Step 8: Based on the set of dispatch response intervals and using the device identifier as an index, a set of dispatch instructions is generated.

[0010] Preferably, the step of establishing a unified data access structure based on device information is as follows: Based on the device information of each distributed energy device, extract the operating fields supported by each type of device, identify the numerical type and physical unit corresponding to each operating field, and construct a field type mapping table; set standard sampling frequency specifications for each distributed energy device, and uniformly specify the sampling frequency and time granularity of various operating fields; construct a field index structure for each type of distributed energy device, and associate and bind the field structure with the device identification code to form a "device identification code - field template" mapping table; set unified data anomaly handling rules and specify the handling method for missing data; encapsulate the field type mapping table, sampling frequency specifications, field index structure, and data anomaly handling rules into a unified data access structure.

[0011] Preferably, the steps for obtaining the initial device operation sequence library are as follows: Identify the device operation parameters of the devices according to the field type mapping table and sampling frequency specification in the unified data access structure; for each distributed energy device, collect device operation parameters according to a set time step to form a sequence of device operation parameters arranged in chronological order; organize the device operation parameter sequences of each type of device into a historical structured data table; bind the structured data table corresponding to each distributed energy device with an identification code to construct an initial operation sequence library indexed by the device identification code.

[0012] Preferably, the step of obtaining the response lag index is as follows: A detection period is set; within the detection period, control command trigger records of each distributed energy device are acquired; the control trigger frequency fluctuation coefficient is evaluated based on the control command trigger records of each distributed energy device; within the detection period, state switching records of each distributed energy device are acquired; the state switching lag diffusion coefficient is evaluated based on the state switching records; within the detection period, equipment operating parameters of each distributed energy device are acquired; the asynchronous offset coefficient of index change is evaluated based on the equipment operating parameters; the control trigger frequency fluctuation coefficient, state switching lag diffusion coefficient, and asynchronous offset coefficient of index change are normalized; and the response lag index is calculated based on the normalized control trigger frequency fluctuation coefficient, state switching lag diffusion coefficient, and asynchronous offset coefficient of index change. The specific steps are as follows: In the formula, This is expressed as a response lag index. This is expressed as the normalized control trigger frequency fluctuation coefficient. This is expressed as the normalized state transition hysteresis diffusion coefficient. This represents the asynchronous offset coefficient of the index change after normalization. , , These are represented as the weighting coefficients of the control trigger frequency fluctuation coefficient, the state switching hysteresis diffusion coefficient, and the index change asynchronous offset coefficient after normalization.

[0013] Preferably, the step of obtaining the control trigger frequency fluctuation coefficient is as follows: During the detection period, obtain the control command trigger records of each distributed energy device. The trigger records include the device's identification code, command issuance timestamp, and command type information; using a preset time step as a unit, count the number of control commands received by each device within each time step to form a control trigger frequency sequence for each device during the detection period; calculate the average trigger frequency for each device's control trigger frequency sequence during the detection period to obtain the average control trigger frequency for each device in the current period; calculate the standard deviation for each device's control trigger frequency sequence to obtain the standard deviation for each device's control trigger frequency in the current period; divide the standard deviation for each device's control trigger frequency in the current period by the average control trigger frequency to obtain the control trigger frequency fluctuation coefficient.

[0014] Preferably, the steps for obtaining the state switching lag diffusion coefficient are as follows: During the detection period, obtain the state switching records of each distributed energy device. The state switching records include the device's identification code, switching start time, switching completion time, and state information before and after the switching; for each state switching record, calculate the difference between the switching completion time and the switching start time to obtain the lag time of a single state switching; calculate the average of all state switching lag times for each device during the detection period to obtain the average state switching lag time of that device; during the detection period, obtain the maximum and minimum values ​​of the average lag times of all devices respectively, subtract the minimum value from the maximum value of the average lag time between devices, and then divide by the maximum value of the average lag time between devices to obtain the state switching lag diffusion coefficient.

[0015] Preferably, the steps for obtaining the asynchronous offset coefficient of the index change are as follows: During the detection period, extract the equipment operating parameters of each distributed energy device from the initial equipment operation sequence library; for each device's equipment operating parameters, calculate the change in adjacent time steps according to the time step, and record it as the operating change; for each device and each time step, calculate the average difference between the operating change of the device and other devices in the same time step; for each device, construct a difference degree sequence based on the average difference between its operating change and other devices in each time step, and calculate the mean and standard deviation of the difference degree sequence; for each device, calculate the ratio of the standard deviation of the difference degree sequence to the mean of the difference degree sequence to obtain the asynchronous offset coefficient of the index change.

[0016] Preferably, the step of determining whether each distributed energy device exhibits asynchronous response lag effect based on the response lag index and screening out all response lag devices is as follows: compare the response lag index of each distributed energy device with a lag threshold; if the response lag index is greater than or equal to the lag threshold, the device is determined to have an asynchronous response lag effect and is recorded as a response lag device; if the response lag index is less than the lag threshold, the device is determined not to have an asynchronous response lag effect, and all distributed energy devices are traversed to obtain all response lag devices.

[0017] Preferably, the step of performing delay compensation on all delayed response devices to obtain the actual device operation sequence library is as follows: for each delayed response device, the ratio of the response lag index to the lag threshold is calculated to obtain the compensation factor; an initial compensation duration is set, and the compensation factor is multiplied by the initial compensation duration to obtain the actual compensation time; based on the actual compensation time, the operating parameters of the delayed response devices in the initial device operation sequence library are time-stamped and compensated; the operating parameters of the delayed response devices after delay compensation are reorganized and indexed according to the device identification code to generate the actual device operation sequence library.

[0018] The technical effects and advantages of this invention are as follows:

[0019] Based on the initial equipment operation sequence library, the equipment response behavior characteristic parameters of each distributed energy device are obtained. The response lag index is evaluated based on the equipment response behavior characteristic parameters. The response lag index is used to determine whether each distributed energy device has asynchronous response lag effect. All response lag devices are screened out, and delay compensation is performed on all response lag devices to obtain the actual equipment operation sequence library. The fusion input matrix is ​​obtained based on the actual equipment operation sequence library. Based on the fusion input matrix, the main feature dimensions are extracted, a state feature vector sequence is constructed, and a time series prediction model is trained to effectively improve the accuracy of virtual power plant management. Attached Figure Description

[0020] Figure 1 A flowchart illustrating a virtual power plant management method based on multi-source data fusion, provided for embodiments of this application. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The virtual power plant management method based on multi-source data fusion involved in the present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] This invention provides a virtual power plant management method based on multi-source data fusion, such as... Figure 1 As shown, it includes the following steps:

[0023] Step 1: Obtain the set of identification codes for each distributed energy device connected to the virtual power plant, obtain the device information for each distributed energy device, and establish a unified data access structure based on the device information;

[0024] Building a unified data access structure enables standardized field management, time synchronization, and structured template management of operational data from various distributed energy devices connected to the virtual power plant. This resolves inconsistencies in data format, time accuracy, sampling frequency, and field naming among multi-source devices. The unified data access structure facilitates standardized collection, stable storage, and efficient retrieval of device operational data, providing structurally consistent and dimensionally unified foundational data support for subsequent time series construction, state feature extraction, and scheduling model training. This effectively enhances the system's fusion processing capabilities and the accuracy of scheduling modeling.

[0025] In this embodiment, it should be specifically explained that the steps for establishing a unified data access structure based on device information are as follows:

[0026] Based on the equipment information of each distributed energy device, the operating fields supported by each type of device are extracted. The operating fields include voltage value, current value, active power, reactive power and operating status field. The numerical type and physical unit corresponding to each operating field are identified, and a field type mapping table is constructed.

[0027] Standard sampling frequency specifications are set for each distributed energy device, and the sampling frequency and time granularity of various operating fields are uniformly specified to ensure the consistency and alignment of data from multiple devices in the time dimension.

[0028] For each type of distributed energy device, a field index structure is constructed, and the field structure is associated and bound with the device's identification code to form a "device identification code - field template" mapping table, ensuring that the data collected subsequently can be classified, indexed and retrieved according to the device number;

[0029] Establish unified data anomaly handling rules, including illegal value removal, null value imputation, and outlier identification, and specify the handling method for missing data, such as forward imputation or linear interpolation, to improve the stability and integrity of data access.

[0030] The field type mapping table, sampling frequency specification, field index structure, and data anomaly handling rules are encapsulated into a unified data access structure and stored in categories based on device type for subsequent historical operating parameter collection and standardized processing.

[0031] Step 2: Preset time step. Based on the data access structure, collect historical equipment operating parameters within the set time period according to the time step. The equipment operating parameters include photovoltaic output power, energy storage charge status and load response power. Construct a historical structured data table based on the historical equipment operating parameters, and establish an initial equipment operating sequence library with the equipment identification code as the index.

[0032] It should be noted that the time step is the smallest time unit for data acquisition. The time step and the set time period can be adjusted according to the actual situation. For example, the time step can be 0.5S or 1S, and the set time period can be 1H or 2H.

[0033] In this embodiment, it should be specifically explained that the steps for obtaining the initial device runtime sequence library are as follows:

[0034] Based on the field type mapping table and sampling frequency specification in the unified data access structure, the equipment operating parameters of the device are identified. The equipment operating parameters include photovoltaic output power, energy storage state of charge and load response power, to ensure the consistency of the collected fields.

[0035] For each connected distributed energy device, the device operating parameters are collected at a set time step to form a sequence of device operating parameters arranged in chronological order, ensuring that the operating parameters of each device have a correspondence between timestamps and values;

[0036] The equipment operating parameter sequences of each type of equipment are organized into a historical structured data table, which is divided into columns by field and sorted by time to form a standard time series structure. Each row in the data table corresponds to a time step, and each column corresponds to an operating parameter.

[0037] By binding the structured data table corresponding to each distributed energy device with its identification code, an initial operation sequence library indexed by the device identification code is constructed, enabling unified management and rapid retrieval of the time-series structure of the operating parameters of all distributed energy devices.

[0038] Step 3: Obtain the device response behavior characteristic parameters of each distributed energy device according to the initial device operation sequence library. The device response behavior characteristic parameters include control command trigger records, state switching records and device operation parameters. Calculate the response lag index based on the device response behavior characteristic parameters. Determine whether each distributed energy device has an asynchronous response lag effect based on the response lag index and screen out all devices with response lag.

[0039] In this embodiment, it should be specifically explained that the step of obtaining the response lag index is as follows:

[0040] Set a detection period. During the detection period, acquire the control command trigger records of each distributed energy device. Evaluate the control trigger frequency fluctuation coefficient based on the control command trigger records of each distributed energy device. It should be noted that the detection period can be changed according to the actual situation. For example, the detection period can be one month or two months.

[0041] During the detection period, the state switching records of each distributed energy device are acquired, and the state switching hysteresis diffusion coefficient is evaluated based on the state switching records.

[0042] During the detection period, the equipment operating parameters of each distributed energy device are acquired, and the asynchronous offset coefficient of the index change is obtained based on the equipment operating parameters.

[0043] The control trigger frequency fluctuation coefficient, state transition hysteresis diffusion coefficient, and index change asynchronous offset coefficient are normalized. The response hysteresis index is then calculated based on these normalized coefficients. The specific steps for obtaining the index are as follows:

[0044] ;

[0045] In the formula, This is expressed as a response lag index. This represents the normalized control trigger frequency fluctuation coefficient. The greater the fluctuation in the frequency of control commands received by each distributed energy device per unit time, i.e., the more unstable the control rhythm or the more obvious the difference in response intervals between devices, the higher the risk of asynchronous response of the entire system, thus leading to an increase in the response lag index. This is represented by the normalized state transition lag diffusion coefficient. When distributed energy devices respond to the same control command, the completion time of their state transitions varies significantly. That is, some devices respond quickly while others show significant delays. This indicates that the response lag phenomenon between devices is diffuse and inconsistent, leading to an increase in the overall system response lag index. This is represented as the asynchronous offset coefficient of the index change after normalization. When multiple distributed energy devices exhibit significant differences in the magnitude of changes in their operating parameters within the same time step, it indicates inconsistent responses and poor coordination to control commands or external disturbances. This reflects an asynchronous offset in the system's internal operating behavior, leading to an increase in the response lag index. , , These are the weighting coefficients for the normalized control trigger frequency fluctuation coefficient, the normalized state transition hysteresis diffusion coefficient, and the normalized asynchronous offset coefficient for index changes. , , Obtained through the Analytic Hierarchy Process (AHP), and , , , The values ​​can be 0.4, 0.2, or 0.4. The Analytic Hierarchy Process (AHP) is a structured, multi-criteria decision-making method used to quantitatively compare the relative importance of multiple evaluation factors. Its basic principle is: first, the complex problem is decomposed into multiple levels, such as the goal layer, criterion layer, and indicator layer; then, by constructing a pairwise comparison judgment matrix, experts are guided to score the importance between any two indicators (e.g., using a 1–9 scale), forming a judgment matrix; next, eigenvector calculation and consistency checks are performed on this matrix to finally obtain the weight value of each evaluation factor relative to the overall goal.

[0046] In this embodiment, it should be specifically explained that the step of obtaining the control trigger frequency fluctuation coefficient is as follows:

[0047] During the detection period, the control command trigger records of each distributed energy device are obtained. The trigger records include the device identification code, command issuance timestamp and command type information, which are used to calculate the number of control triggers for each device in the future.

[0048] Using a preset time step as the unit, the number of control commands received by each device within each time step is counted to form a control trigger frequency sequence of each device within the detection time period.

[0049] The average trigger frequency is calculated for the control trigger frequency sequence of each device within the detection time period. The average control trigger frequency of each device in the current time period is obtained. The average value is used to measure the overall level of the device control rhythm.

[0050] The standard deviation of the control trigger frequency sequence for each device is calculated to obtain the standard deviation of the control trigger frequency of each device in the current time period. This standard deviation is used to measure the fluctuation range of the device control trigger frequency. The larger the standard deviation, the more unstable the trigger frequency.

[0051] Divide the standard deviation of the control trigger frequency of each device in the current time period by the average control trigger frequency to obtain the control trigger frequency fluctuation coefficient. The larger the coefficient value, the more unstable the rhythm of the control command and the greater the possibility of asynchronous response between devices.

[0052] In this embodiment, it should be specifically explained that the steps for obtaining the state switching hysteresis diffusion coefficient are as follows:

[0053] During the detection period, the status switching records of each distributed energy device are acquired. The status switching records include the device's identification code, switching start time, switching completion time, and status information before and after the switching, which are used to analyze the time characteristics of the switching process.

[0054] For each state transition record, the difference between the transition completion time and the transition start time is calculated to obtain the lag time of a single state transition. The lag time is used to characterize the degree of delay in the device's response to control commands.

[0055] The average state switching lag time of each device is calculated by averaging all state switching lag times within the detection period. This average state switching lag time is used to measure the overall response level of the device during that period.

[0056] During the detection period, the maximum and minimum values ​​of the average lag time of all devices are obtained to reflect the distribution differences in response delay among devices.

[0057] The state transition lag diffusion coefficient is obtained by subtracting the minimum value from the maximum value of the average lag time between devices, and then dividing by the maximum value of the average lag time between devices. The larger the state transition lag diffusion coefficient value, the more significant the difference in response lag between devices, and the stronger the asynchrony.

[0058] In this embodiment, it should be specifically explained that the steps for obtaining the asynchronous offset coefficient of the indicator change are as follows:

[0059] During the detection period, the equipment operation parameters of each distributed energy device are extracted from the initial equipment operation sequence library. The equipment operation parameters include photovoltaic output power, energy storage state of charge, and load response power.

[0060] For the operating parameters of each device, the change in the operating parameters between adjacent time steps is calculated according to the time step and recorded as the operating change. The operating change is the operating parameter value of the current time step minus the operating parameter value of the previous time step, which is used to reflect the dynamic fluctuation of the operating parameters.

[0061] For each device and each time step, calculate the average difference in operational changes between the device and other devices at the same time step. The specific steps are as follows:

[0062] ;

[0063] In the formula, It represents the average difference in the amount of change in operation between device i and other devices at time step t, where N is the total number of devices. This is expressed as the change in the operation of device i at time step t. This is expressed as the change in the operation of device j at time step t;

[0064] For each device, a difference series is constructed based on the average difference between its operation and that of other devices at each time step, and the mean and standard deviation of the difference series are calculated.

[0065] For each device, the ratio of the standard deviation of the difference series to the mean of the difference series is calculated to obtain the asynchronous offset coefficient of the index change. The larger the asynchronous offset coefficient of the index change, the more asynchronous the index changes of the device and the group are, and the higher the risk of asynchronous response.

[0066] At the same time step, the operational changes of all devices are statistically analyzed, and the standard deviation is calculated to measure the degree of difference in the changes of indicators among devices at that time step. The larger the standard deviation, the stronger the asynchrony.

[0067] At the same time step, the average value is calculated after taking the absolute value of the changes in the indicators of all devices. The average value is used to characterize the overall level of changes in the device indicators at that time step.

[0068] Divide the standard deviation of the change in the index at each time step by the corresponding average value to obtain the asynchronous offset coefficient of the index change at that time step. The larger the coefficient value, the more asynchronous the changes in the operating index between devices are, and the higher the possibility of asynchronous response.

[0069] In this embodiment, it should be specifically explained that the step of determining whether each distributed energy device exhibits asynchronous response lag effect based on the response lag index and screening out all response lag devices is as follows:

[0070] The response lag index of each distributed energy device is compared with a lag threshold. If the response lag index is greater than or equal to the lag threshold, the device is determined to have an asynchronous response lag effect and is recorded as a lag device. If the response lag index is less than the lag threshold, the device is determined not to have an asynchronous response lag effect. This process is repeated for all distributed energy devices to obtain all lag devices. The lag threshold is obtained using an adaptive thresholding method, an algorithm that dynamically determines the judgment threshold based on historical data statistical characteristics. This method is suitable for automatically adjusting the threshold standard under different scenarios to improve judgment accuracy. Its basic principle is: within a set data observation window, the distribution characteristic parameters of the target variable (such as the response lag index), such as mean, standard deviation, maximum value, quantile, etc., are extracted, and then combined with the trend or degree of fluctuation to construct a dynamically updated threshold function. Unlike the fixed threshold method, the adaptive thresholding method can adjust the lag judgment threshold in real time according to the historical fluctuation range of the system's operating status, thereby improving the judgment sensitivity and robustness under complex operating conditions.

[0071] Step 4: Perform delay compensation on all devices with delayed response to obtain the actual device operation sequence library. Time-label the execution time of the device operation parameters in the actual device operation sequence library, remove abnormal data segments, and use the moving average calculation method to convert the output values ​​of each device at each time step within the set time period into an equal-dimensional structure to generate a fusion input matrix.

[0072] The moving average is a commonly used time series smoothing technique to eliminate short-term fluctuations and enhance the stability of data trends. Its basic principle is as follows: a fixed-length sliding window is set in the continuous time series. The value at each time point, together with the data from several time points before and after it, forms a local dataset. The arithmetic mean of this dataset is calculated and used as the smoothed value for that time point. The window slides across the entire time series, forming a continuous sequence of means.

[0073] It should be noted that the method of using the moving average to calculate the output values ​​of each device at each time step within a set time period to convert them into an equal-dimensional structure and generate a fusion input matrix is ​​an existing technology, and this embodiment will not describe its specific steps in detail.

[0074] In this embodiment, it is necessary to specifically explain the steps for performing delay compensation on all devices with delayed responses to obtain the actual device operation sequence library:

[0075] For each device with a delayed response, the ratio of the delayed response index to the delayed response threshold is calculated to obtain a compensation factor, which is used to characterize the severity of the delayed response of the current distributed energy device.

[0076] Set the initial compensation duration, multiply the compensation factor with the initial compensation duration to calculate the actual compensation time, which is used to dynamically adjust the response timing of distributed energy devices.

[0077] Based on the actual compensation time, the operating parameters of the lagging devices in the initial equipment operation sequence library are time-labeled and compensated to correct the positional deviation of the operating parameters of each lagging device in the time series.

[0078] The delayed response parameters of the equipment are reorganized after delay compensation and indexed according to the equipment identification code to generate an actual equipment operation sequence library.

[0079] By calculating the ratio of the response lag exponent to the lag threshold to obtain the compensation factor, and dynamically adjusting the initial compensation time based on the compensation factor, the delay compensation magnitude can be adaptively determined according to the response lag degree of different devices and different operating scenarios. This allows for precise correction of timing deviations in the initial device operation sequence library. It improves the consistency and alignment accuracy of multi-device data in the actual device operation sequence library across the time dimension, providing a more stable and reliable data foundation for subsequent state feature extraction, predictive modeling, and scheduling optimization.

[0080] Step 5: Based on the fused input matrix, the principal component analysis method is used to extract the main feature dimensions, construct a state feature vector sequence, and use the state feature vector sequence as input to train the time series prediction model;

[0081] Principal Component Analysis (PCA) is a commonly used dimensionality reduction and feature extraction technique. It aims to project original multidimensional data onto a set of uncorrelated new variables (principal components) through linear transformation, preserving as much of the original data's variance information as possible. Its basic principle is to perform covariance analysis on the fused input matrix and extract the eigenvectors along the direction of maximum variance as principal components, thereby reducing dimensionality while preserving the main features of the data structure to the greatest extent possible. PCA can effectively remove redundant features, reduce inter-dimensional correlation, and improve the discriminative power of state feature vector sequences and model training efficiency.

[0082] In this embodiment, it should be specifically explained that the construction steps of the time series prediction model are as follows:

[0083] Based on the state feature vector sequence, historical state feature vectors within a set time period are selected as training input samples, and state feature vectors corresponding to several future time steps are selected as training output samples to construct the training dataset for the time series prediction model.

[0084] Based on the dimension and time step of the state feature vector sequence, the structural parameters of the time series prediction model are initialized. The structural parameters include the number of input layer nodes, the number of hidden layer units, the number of output layer nodes, and the time step window length. The model type is determined to be a neural network structure with time series prediction capabilities, such as a recurrent neural network or a long short-term memory network.

[0085] The training dataset is input into the time series prediction model, and the model weight parameters are iteratively updated using forward computation and backpropagation algorithms. The training process optimizes the model weights that achieve the best fit on the current training dataset by minimizing the error function between the predicted output and the true state feature vector.

[0086] Using a validation dataset that was not used in training, the accuracy of the trained time series prediction model is validated. If the prediction error is less than the set convergence threshold, the time series prediction model and its final weight parameters are saved to obtain the time series prediction model.

[0087] Forward computation and backpropagation is a commonly used parameter optimization method in neural network model training. It consists of two main processes: First, through forward computation, the input data is passed to each layer of neurons in sequence and the output results are calculated; then, by calculating the error between the predicted output and the true value, the error is propagated forward from the output layer to the input layer using the chain rule, and the weights and bias parameters of each layer are updated in sequence to minimize the error function.

[0088] Step 6: Obtain the state feature vectors of the most recent time steps, call the time series prediction model to predict the predicted values ​​of the state feature vectors of the future time steps, and construct the set of scheduling response intervals of the virtual power plant based on the predicted values ​​of the state feature vectors.

[0089] In this embodiment, it should be specifically explained that the steps for obtaining the set of scheduling response intervals of the virtual power plant are as follows:

[0090] Based on the predicted values ​​of the state feature vectors, key operating parameters of each distributed energy device are extracted in each prediction time step. The key operating parameters include the predicted output power of the device, the predicted charge state of the energy storage system, the predicted load value of the adjustable load device, and the upper limit of the rated output power and the minimum operating threshold of various devices, which serve as the basic input parameters required for scheduling capability assessment.

[0091] Based on the key operating parameters in each prediction time step, the maximum adjustable power within the current time step is calculated. Specifically, the adjustable power of the energy storage system is the remaining charging capacity divided by the allowable charging rate per unit time; the adjustable power of the adjustable load is the predicted load value minus the minimum load threshold of the equipment; and the adjustable power of the distributed generation equipment is the difference between the rated power of the equipment and the predicted output power. The adjustable power of the above three types of equipment is summed to obtain the maximum adjustable power for that time step.

[0092] Based on the equipment operating parameters extracted at the same time step, the maximum adjustable power of the energy storage system, adjustable load equipment and distributed generation equipment is calculated respectively. Among them, the adjustable power of the energy storage system is the maximum discharge capacity allowed by the current charge state, the adjustable power of the adjustable load is the difference between the rated load limit of the equipment and the current predicted load, and the adjustable power of the generation equipment is the predicted output power minus the minimum operating power value. The adjustable power of each type of equipment is summed to obtain the maximum adjustable power of the time step.

[0093] Using the maximum adjustable power and maximum adjustable power at each predicted time step as the lower and upper limits of power regulation, the dispatch response interval of the virtual power plant at the current time step is constructed. The dispatch response intervals corresponding to all predicted time steps are arranged in chronological order to form a set of dispatch response intervals, which is used to describe the dispatch capability boundary of the virtual power plant in each future time period and serves as the constraint input for subsequent dispatch path optimization.

[0094] Step 7: Based on the predicted value of the state feature vector of the future time step and the scheduling response interval, a target scheduling path is generated using a dynamic programming method to obtain a set of scheduling target values. Then, a scheduling instruction set is generated using the device identifier as an index. The scheduling instruction set includes the target power value, execution time point, device number, and priority identifier.

[0095] Dynamic programming is an algorithmic approach for solving multi-stage optimal decision problems. It decomposes the original problem into interconnected subproblems, solves the optimal solutions to these subproblems step-by-step, and reuses the results to construct the optimal solution path for the overall problem. Its core principles are "optimal substructure" and "state transition," meaning that the optimal decision at each stage depends on the state and optimal result of the previous stage. When applied to virtual power plant scheduling, dynamic programming can comprehensively consider the predicted values ​​of state feature vectors and the scheduling response intervals over multiple future time steps. Under the conditions of power constraints and equipment capacity constraints, it seeks the path with the lowest overall scheduling cost or the highest benefit, thereby generating a set of target scheduling values ​​for continuous time periods and achieving multi-time-period coordinated optimization.

[0096] It should be noted that the method of generating the target scheduling path using dynamic programming based on the predicted value of the state feature vector of the future time step and the scheduling response interval is an existing technology, and this embodiment will not describe its specific steps in detail.

[0097] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A virtual power plant management method based on multi-source data fusion, characterized in that, The method comprises the following steps: Step 1: obtaining a set of identification codes of various distributed energy equipment accessed by a virtual power plant, and obtaining device information of each distributed energy equipment, and establishing a unified data access structure according to the device information; Step 2: presetting a time step, collecting historical device operation parameters in a set time period according to the time step based on the data access structure, the device operation parameters including photovoltaic output power, energy storage charge state and load response power, constructing a historical structured data table according to the historical device operation parameters, and establishing an initial device operation sequence library with the identification code of the device as an index; Step 3: obtaining device response behavior characteristic parameters of each distributed energy equipment according to the initial device operation sequence library, the device response behavior characteristic parameters including control instruction trigger records, state switching records and device operation parameters, evaluating a response lag index according to the device response behavior characteristic parameters, determining whether a non-synchronous response lag effect occurs for each distributed energy equipment according to the response lag index, and screening all response lag devices; Step 4: compensating for the delay of all response lag devices to obtain an actual device operation sequence library, and obtaining a fusion input matrix according to the actual device operation sequence library; Step 5: based on the fusion input matrix, extracting main feature dimensions, constructing a state feature vector sequence, and training a time series prediction model; Step 6: obtaining state feature vectors of recent time steps, calling the time series prediction model to predict state feature vector prediction values of future time steps, and constructing a scheduling response interval set of the virtual power plant according to the state feature vector prediction values; Step 7: generating a scheduling instruction set according to the scheduling response interval set and taking the identification code of the device as an index. 2.The virtual power plant management method based on multi-source data fusion of claim 1, wherein: The step of establishing a unified data access structure according to device information comprises the following steps: According to the device information of each distributed energy equipment, extracting the running fields supported by each type of equipment, identifying the numerical type and physical unit corresponding to each running field, and constructing a field type mapping table; Setting a standard sampling frequency specification for each distributed energy equipment, and uniformly specifying the sampling frequency and time granularity of each type of running field; Constructing a field index structure for each type of distributed energy equipment, and associating and binding the field index structure with the identification code of the equipment; Setting a unified data exception processing rule, and specifying the processing method for missing data; Encapsulating the field type mapping table, the sampling frequency specification, the field index structure and the data exception processing rule into a unified data access structure. 3.The virtual power plant management method based on multi-source data fusion of claim 2, wherein, The step of obtaining the initial device operation sequence library comprises the following steps: According to the field type mapping table and the sampling frequency specification in the unified data access structure, identifying the device operation parameters of the equipment; Collecting the device operation parameters of each distributed energy equipment according to the set time step to form a device operation parameter sequence arranged in chronological order; Organizing the device operation parameter sequence of each type of equipment into a historical structured data table; Binding the structured data table corresponding to each distributed energy equipment with the identification code to construct an initial operation sequence library with the identification code of the equipment as an index. 4.The virtual power plant management method based on multi-source data fusion of claim 1, wherein, The step of obtaining the response lag index comprises the following steps: A detection time period is set, and control instruction trigger records of each distributed energy equipment are acquired in the detection time period; a control trigger frequency fluctuation coefficient is evaluated according to the control instruction trigger records of each distributed energy equipment; In the detection time period, state switching records of each distributed energy equipment are acquired, and a state switching lag diffusion coefficient is evaluated according to the state switching records; In the detection time period, equipment operation parameters of each distributed energy equipment are acquired, and an index change asynchronous offset coefficient is evaluated according to the equipment operation parameters; The control trigger frequency fluctuation coefficient, the state switching lag diffusion coefficient and the index change asynchronous offset coefficient are normalized, and a response lag index is calculated according to the normalized control trigger frequency fluctuation coefficient, the state switching lag diffusion coefficient and the index change asynchronous offset coefficient; the specific acquisition steps are as follows: ； In the formula, is expressed as a response lag index, is expressed as a normalized control trigger frequency fluctuation coefficient, is expressed as a normalized state switching lag diffusion coefficient, is expressed as a normalized index change asynchronous offset coefficient, , , is expressed as a weight coefficient of the normalized control trigger frequency fluctuation coefficient, a weight coefficient of the normalized state switching lag diffusion coefficient, and a weight coefficient of the normalized index change asynchronous offset coefficient.

5. The virtual power plant management method based on multi-source data fusion according to claim 4, characterized in that: The acquisition steps of the control trigger frequency fluctuation coefficient are as follows: In the detection time period, control instruction trigger records of each distributed energy equipment are acquired, and the trigger records include the identification code of the equipment, the instruction issue timestamp and the instruction type information; In units of a preset time step, the number of control instructions received by each equipment in each time step is counted to form a control trigger frequency sequence of each equipment in the detection time period; The average trigger frequency of the control trigger frequency sequence of each equipment in the detection time period is calculated to obtain the average control trigger frequency of each equipment in the current period; The standard deviation of the control trigger frequency sequence of each equipment is calculated to obtain the control trigger frequency standard deviation of each equipment in the current period; The control trigger frequency standard deviation of each equipment in the current period is divided by the average control trigger frequency to obtain the control trigger frequency fluctuation coefficient. 6.The virtual power plant management method based on multi-source data fusion of claim 4, wherein: The acquisition steps of the state switching lag diffusion coefficient are as follows: In the detection time period, state switching records of each distributed energy equipment are acquired, and the state switching records include the identification code of the equipment, the switching start time, the switching completion time and the state information before and after the switching; For each state switching record, the difference between the switching completion time and the switching start time is calculated to obtain the lag time of single state switching; The average value of all state switching lag times of each equipment in the detection time period is calculated to obtain the average state switching lag time of the equipment; In the detection time period, the maximum value and the minimum value of the average lag time of all equipment are acquired, the maximum value of the average lag time between equipment is subtracted from the minimum value, and then divided by the maximum value of the average lag time between equipment to obtain the state switching lag diffusion coefficient.

7. The virtual power plant management method based on multi-source data fusion according to claim 4, characterized in that: The acquisition steps of the index change asynchronous offset coefficient are as follows: In the detection time period, the equipment operation parameters of each distributed energy equipment are extracted from the initial equipment operation sequence library; For the equipment operation parameters of each equipment, the change amount of adjacent time steps is calculated according to the time step, which is recorded as the operation change amount; For each equipment and each time step, the average difference of the operation change amount of the equipment and other equipment in the same time step is calculated; For each equipment, a difference degree sequence is constructed according to the average difference of the operation change amount of the equipment and other equipment in each time step, and the mean value of the difference degree sequence and the standard deviation of the difference degree sequence are calculated. For each device, the standard deviation of the difference degree sequence is divided by the mean of the difference degree sequence to obtain an index change asynchronous offset coefficient. 8.The virtual power plant management method based on multi-source data fusion of claim 1, wherein: The step of determining whether each distributed energy source device has a non-synchronous response lag effect according to the response lag index is: The response lag index of each distributed energy source device is compared with the lag threshold value. If the response lag index is greater than or equal to the lag threshold value, it is determined that the device has a non-synchronous response lag effect, and the device is recorded as a response lag device. If the response lag index is less than the lag threshold value, it is determined that the device does not have a non-synchronous response lag effect. All distributed energy source devices are traversed to obtain all response lag devices. 9.The virtual power plant management method based on multi-source data fusion of claim 8, wherein: The step of performing delay compensation on all response lag devices to obtain an actual device operation sequence library is: For each response lag device, the response lag index is divided by the lag threshold value to obtain a compensation factor. An initial compensation duration is set, and the compensation factor is multiplied by the initial compensation duration to obtain an actual compensation time. According to the actual compensation time, the operation parameters of the response lag device in the initial device operation sequence library are time label shifted and compensated. The operation parameters of the response lag device after delay compensation are reorganized and indexed according to the device identification code to generate an actual device operation sequence library.

Images