Intelligent load scheduling method and system for virtual power plant user side

Abstract

The invention discloses an intelligent load scheduling method and system for a virtual power plant user side. The method comprises the following steps: acquiring original load data of the virtual power plant user side; performing feature extraction according to the original load data to obtain a feature vector; performing load demand prediction according to the feature vector to obtain a predicted load demand; according to the predicted load demand, adopting a load scheduling model based on the game theory to obtain a load scheduling scheme; wherein the load scheduling model comprises a local game layer and a global game layer, and in each game process, the load scheduling model determines the load decision weight of the user side according to the fault-tolerant factor, and adjusts the load decision of the user side based on the fault-tolerant factor. According to the method, the flexibility and accuracy of virtual power plant user side load scheduling can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart grid control, and particularly to an intelligent load scheduling method and system for the user side of a virtual power plant. Background Art

[0002] In the operation of a virtual power plant, load management and control on the user side are one of the key factors for improving the flexibility and stability of the power system. Traditional load scheduling methods generally rely on a centralized control mode, which has problems such as long response time, insufficient accuracy, and network delay when dealing with large-scale distributed loads. With the development of edge computing technology, it can perform data processing and decision-making near the load source, thereby achieving more real-time and accurate load scheduling.

[0003] Most of the existing edge computing methods fail to fully consider the dynamic characteristics and demand fluctuations of power loads, lack an intelligent adaptive adjustment mechanism, and have insufficient flexibility and accuracy in load scheduling. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides an intelligent load scheduling method and system for the user side of a virtual power plant, which can effectively improve the flexibility and accuracy of load scheduling.

[0005] An embodiment of the present invention provides an intelligent load scheduling method for the user side of a virtual power plant, including: Obtaining the original load data of the user side of the virtual power plant; Performing feature extraction on the original load data to obtain a feature vector; Performing load demand prediction based on the feature vector to obtain the predicted load demand; According to the predicted load demand, adopting a load scheduling model based on game theory to obtain a load scheduling plan; Wherein, the load scheduling model includes a local game layer and a global game layer. In each round of the game process, the load scheduling model determines the load decision weight of the user side according to a fault tolerance factor, and adjusts the load decision of the user side based on the fault tolerance factor.

[0006] As an improvement of the above solution, the fault tolerance factor is determined according to the load change amplitude, equipment health status, and external disturbance factors.

[0007] As an improvement of the above solution, the performing feature extraction on the original load data to obtain a feature vector includes: Filtering out noise and removing outliers from the original load data to obtain first load data; Performing data synchronization and normalization processing on the first load data to obtain second load data; The second load data is processed using an adaptive fuzzy rule to obtain third load data; According to the third load data, variational mode decomposition is used for feature extraction to obtain a feature vector; the feature vector includes a long-term trend component and its dynamic change amount, a periodic component and its dynamic change amount, and a random fluctuation component.

[0008] As an improvement of the above solution, the process of using an adaptive fuzzy rule to process the second load data to obtain third load data includes: Based on a preset fuzzy rule, according to the fluctuation amplitude of the second load data, the parameters of the fuzzy rule are adjusted to obtain an adjusted rule width and fuzzy set center; According to the second load data, the adjusted rule width, and the fuzzy set center, the membership degree of each fuzzy rule is calculated; The membership degrees of the fuzzy rules are weighted and averaged to calculate and obtain the third load data.

[0009] As an improvement of the above solution, the process of using variational mode decomposition to extract features from the third load data to obtain a feature vector includes: According to the fluctuation amplitude of the third load data, the decomposition layer number of variational mode decomposition is determined; the greater the fluctuation amplitude, the more the decomposition layer number; According to the decomposition layer number, the third load data is subjected to variational mode decomposition to obtain the long-term trend component, periodic component, and random fluctuation component corresponding to the third load data; Based on the long-term trend component, the periodic component, the random fluctuation component, and the corresponding feature component weights, a feature vector is determined; wherein, the feature component weights are dynamically adjusted according to the fluctuation amplitude of each component.

[0010] As an improvement of the above solution, the process of predicting the load demand according to the feature vector to obtain the predicted load demand includes: According to the feature vector, a long short-term memory network model is used for load demand prediction to obtain the predicted load demand; Wherein, before load demand prediction, a weighted coefficient corresponding to the current feature vector is calculated according to the decay coefficient, fluctuation amplitude, and time difference of the current feature vector, and the parameters of the long short-term memory network model are updated according to the weighted coefficient.

[0011] As an improvement of the above solution, the calculation formula of the feature component weight is as follows: Wherein, is the The weight of each characteristic component is the weighted attenuation coefficient and is the fluctuation amplitude of the characteristic component.

[0012] As an improvement to the above solution, the calculation formula for the load decision weight of the client is: where is the load decision weight of the i-th client and are the utility functions of the i-th client and the j-th client and are the fault tolerance factors of the i-th client and the j-th client, and N is the total number of clients in the system.

[0013] As an improvement to the above solution, adjusting the load decision of the client based on the fault tolerance factor includes: Calculating the load decision of the i-th client in the (t + 1)-th round of the game according to the following formula : where is the load decision of the i-th client in the t-th round of the game is the adjustment step size is the utility function for the load decision gradient is the adjustment parameter is the fault tolerance factor of the i-th client.

[0014] An embodiment of the present invention also provides an intelligent load scheduling system for a virtual power plant client, including: A data acquisition module for acquiring the original load data of the virtual power plant client; A feature extraction module for extracting features from the original load data to obtain a feature vector; A load prediction module for predicting the load demand according to the feature vector to obtain the predicted load demand; A load scheduling module for obtaining a load scheduling plan according to the predicted load demand by using a load scheduling model based on game theory; wherein, the load scheduling model includes a local game layer and a global game layer, and in each round of the game process, the load decision weight of the client is determined according to the fault tolerance factor, and the load decision of the client is adjusted based on the fault tolerance factor.

[0015] Compared with the prior art, the beneficial effects of an intelligent load scheduling method and system for a virtual power plant user side provided by an embodiment of the present invention are as follows: By extracting features from the original load data, predicting the load demand based on the extracted features, and then according to the predicted load demand, obtaining a load scheduling plan by using a load scheduling model based on game theory. In each round of the game process of the load scheduling model, the load decision-making weight of the user side is determined according to the fault tolerance factor, and the load decision of the user side is adjusted based on the fault tolerance factor, realizing the global and local load collaborative scheduling, ensuring the efficient utilization of power resources, guaranteeing the robustness of the system, and improving the accuracy and flexibility of the virtual power plant load scheduling; By adopting an adaptive fuzzy rule adjustment mechanism to optimize the data processing process during the data processing, the adaptability and accuracy of the load data are further improved; By using variational mode decomposition and an adaptive decomposition mechanism to analyze the load data and dynamically adjusting the decomposition layer number according to the data fluctuation amplitude, the modeling accuracy can be improved and the calculation complexity can be reduced; By using an adaptive updated long short-term memory network to predict the load demand, a high-precision load demand prediction is realized, and the accuracy of the virtual power plant load scheduling is further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a schematic flowchart of an intelligent load scheduling method for a virtual power plant user side provided by an embodiment of the present invention; Figure 2 is a schematic structural diagram of an intelligent load scheduling system for a virtual power plant user side provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0018] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of an intelligent load scheduling method for a virtual power plant user side provided by an embodiment of the present invention. The intelligent load scheduling method for the virtual power plant user side includes: S1: Obtain the original load data of the virtual power plant user side; Specifically, by deploying edge computing nodes at the virtual power plant user side, various types of power load data are collected in real time through devices such as smart meters, load monitoring devices, and sensors, including instantaneous power , current , voltage such as data, to form a multi-dimensional load data set , that is, the original load data, and its expression is: wherein, is a time series. The edge computing nodes are deployed close to the user side, which can effectively reduce the data transmission delay and ensure the real-time and integrity of the collected data.

[0019] S2: Extract features from the original load data to obtain a feature vector; As an optional embodiment, the extracting features from the original load data to obtain a feature vector includes: Filter out noise and remove outliers from the original load data to obtain the first load data; Synchronize and standardize the first load data to obtain the second load data; Adopt an adaptive fuzzy rule to process the second load data to obtain the third load data; According to the third load data, use the variational mode decomposition method to extract features to obtain a feature vector; the feature vector includes a long-term trend component and its dynamic change amount, a periodic component and its dynamic change amount, and a random fluctuation component.

[0020] Specifically, after collecting the original load data, in order to eliminate external interference and abnormal data, it is necessary to clean the original load data, including noise filtering and outlier removal; Noise filtering: Use a fuzzy membership function to evaluate the data quality and identify noise data: wherein, is the membership degree of data , is the mean value of data, is the standard deviation. When is less than the set threshold, it is regarded as noise data and is removed.

[0021] Outlier removal: Judge the deviation degree of data through statistical analysis, and remove and replace the data that deviates from the normal range: wherein, is a control parameter that determines the threshold range of data deviation, is the first load data obtained by data cleaning.

[0022] Furthermore, to ensure the consistency of data in time and space, synchronize and standardize the first load data after cleaning; Data synchronization: Align data from different client ends based on timestamps to solve the problem of asynchronous transmission of multi-source data: Among them, is the time offset, and synchronization calibration is achieved by adjusting the timestamps of the data.

[0023] Data standardization: Normalize data with different dimensions to ensure data analysis under a unified standard: Among them, and are the minimum and maximum values of the data respectively, is the second load data after standardization processing.

[0024] Furthermore, to further improve the adaptability and accuracy of data processing, the embodiment of the present invention designs an adaptive fuzzy rule adjustment mechanism to optimize the data processing process, which can dynamically adjust the parameters of the fuzzy rules based on real-time data fluctuations.

[0025] As one of the optional embodiments, using the adaptive fuzzy rules to process the second load data to obtain the third load data includes: Based on the preset fuzzy rules, adjust the parameters of the fuzzy rules according to the fluctuation amplitude of the second load data to obtain the adjusted rule width and the center of the fuzzy set; Calculate the membership degree of each fuzzy rule according to the second load data, the adjusted rule width and the center of the fuzzy set; Perform weighted averaging on the membership degrees of the fuzzy rules to calculate and obtain the third load data.

[0026] Specifically, the embodiment of the present invention designs a rule base containing multiple rules, covering key factors such as load fluctuations, demand prediction errors, and power supply changes: Rule 1: When the load fluctuates greatly, increase the reserve power. Specifically, when the load volatility > 10%, charge the energy storage device or start the standby generator to ensure system stability.

[0027] Rule 2: When the load is stable and low, optimize the load distribution and reduce the standby demand. Specifically, when the load volatility < 5%, optimize the load distribution, reduce the activation of standby power supplies, and improve system efficiency.

[0028] Rule 3: When the load rises rapidly, schedule distributed energy in advance. Specifically, when the load growth rate > 5% / minute, schedule distributed energy or energy storage systems to relieve the load pressure.

[0029] Rule 4: Adjust the load distribution when the grid frequency fluctuates. When the grid frequency is abnormal (such as the frequency < 49Hz or > 51Hz), adjust the load distribution, activate the standby generator or adjust the load response to ensure the stability of the power grid.

[0030] Based on the above fuzzy rules, perform fuzzy inference on the second load data and calculate the membership degree of each fuzzy rule. The membership degree reflects the matching degree between the input data (i.e., the second load data) and the fuzzy set.

[0031] Specifically, set the fuzzy set of the second load data as a combination of multiple rules, and each rule has a corresponding membership function for calculating the matching degree of the input data to each rule.

[0032] Calculate the membership degree for the i-th rule: Where, is the center of the fuzzy set of the -th rule, representing the central value of the load range described by this rule; is the width of the fuzzy rule. By calculating the membership degree of each rule, the matching degree of the input data under each rule can be determined, and then the action intensity of each rule can be obtained.

[0033] Furthermore, in order to cope with different load environments, the embodiments of the present invention adopt an adaptive adjustment mechanism to update the parameters of the fuzzy rules in real time. Whenever new second load data is obtained, adjust the rule parameters according to the fluctuation of the second load data, especially the rule width and the center of the fuzzy set , to ensure that the fuzzy rules can accurately reflect the current load state.

[0034] Specifically, update the rule parameters according to the following formula: Where, and are the change amounts of the center of the fuzzy set and the width of the fuzzy rule, and are the adjustment coefficients, is the fluctuation amplitude of the second load data: The embodiments of the present invention automatically optimize the fuzzy rules according to the load change trend, enhancing the adaptability of the system to abnormal fluctuations and dynamic data.

[0035] Further, defuzzify the data processed by the fuzzy rules, and output the data processing result through the weighted average method, that is, the third load data : wherein is the membership degree of the i-th rule, is the weight of the fuzzy rule.

[0036] After data acquisition, cleaning, synchronous standardization, and adaptive fuzzy rule processing, the edge computing node outputs high-quality data results: the third load data , and transmits it to the virtual power plant control center to provide accurate data support for subsequent load analysis and scheduling.

[0037] Further, in view of the nonlinear, time-varying, and periodic characteristics of the load at the virtual power plant user side, the embodiment of the present invention adopts variational mode decomposition (VMD) and an adaptive decomposition mechanism to decompose the original load data into different components, so as to accurately extract the multi-level characteristics of the user-side load data and improve the accuracy of data modeling.

[0038] As one optional embodiment, feature extraction is performed on the third load data by using the variational mode decomposition method to obtain a feature vector, including: Determine the decomposition layer number of the variational mode decomposition according to the fluctuation amplitude of the third load data; the greater the fluctuation amplitude, the more the decomposition layer number; Perform variational mode decomposition on the third load data according to the decomposition layer number to obtain the long-term trend component, periodic component, and random fluctuation component corresponding to the third load data; Determine the feature vector based on the long-term trend component, the periodic component, the random fluctuation component, and the corresponding feature component weights; wherein, the feature component weights are dynamically adjusted according to the fluctuation amplitude of each component.

[0039] Specifically, decompose the third load data into a long-term trend component , periodic component and random fluctuation component : wherein, the long-term trend component reflects the overall growth or decline trend of the load; the periodic component captures the daily, weekly, and seasonal change rules of the load; the random fluctuation component represents short-term fluctuations and noises caused by external disturbances.

[0040] Based on traditional time series decomposition, the present invention designs an adaptive decomposition mechanism to dynamically adjust the decomposition parameters, which can automatically adapt to the changing characteristics of different load data. In particular, it dynamically adjusts the number of decomposition layers according to the fluctuation amplitude of the load data, effectively improving the modeling accuracy and reducing the computational complexity. First, the fluctuation amplitude of the load data is monitored in real time, and the volatility of the data is evaluated based on statistical features (such as standard deviation, mean, fluctuation amplitude, etc.). The evaluation formula for the fluctuation amplitude is as follows: where, is the fluctuation amplitude of the third load data at time t; is the value of the i-th third load data at time t; is the window size for calculating the local fluctuation. According to the above formula, the fluctuation amplitude of the third load data can be obtained. When the fluctuation amplitude A(t) is greater than the preset threshold, it indicates that the current load data fluctuates greatly, and the number of decomposition layers is automatically increased; otherwise, the number of decomposition layers is decreased.

[0041] In the case of large fluctuation amplitude of the load data, the number of layers of VMD decomposition is increased to refine the decomposition process. Higher layers can better capture the fluctuations and trends in different frequency bands, but at the same time, it will also increase the computational complexity. To avoid overly complex calculations, the embodiments of the present invention dynamically adjust the most suitable number of decomposition layers according to the change of the fluctuation amplitude . The specific adjustment formula is as follows: where, is the minimum number of decomposition layers, preferably set to 2 layers; is the adjustment coefficient. When the data fluctuates greatly, the number of decomposition layers is increased to improve the modeling accuracy; when the fluctuation is small, the number of decomposition layers is decreased to reduce the computational complexity.

[0042] Furthermore, based on the time series decomposition, the features of each component are extracted to construct a feature vector containing multi-dimensional information: First, the decomposed components are integrated with the historical change amount to construct a feature vector : where, is the long-term trend component, is the periodic component is the random fluctuation component, is the dynamic change amount of the long-term trend component, is the dynamic change amount of the periodic component, reflecting the change trend of the load.

[0043] Furthermore, the feature vector is determined by a long-term trend component, a periodic component, a random fluctuation component, and corresponding feature component weights, where the feature component weights are dynamically adjusted according to the fluctuation amplitude of each component. Specifically, by calculating the importance scores of each feature component, the weights of the feature components are dynamically adjusted to ensure that the model can strengthen the influence of key features when the load changes violently, thereby improving the prediction response speed. Specifically, the calculation formula for the feature component weights is as follows: Where, is the weight of the th feature component, is the weighted decay coefficient, is the fluctuation amplitude of the feature component.

[0044] S3: Perform load demand prediction based on the feature vector to obtain the predicted load demand; As one optional embodiment, the performing load demand prediction based on the feature vector to obtain the predicted load demand includes: Perform load demand prediction using a long short-term memory network model based on the feature vector to obtain the predicted load demand; Where, before performing load demand prediction, the weighted coefficient corresponding to the current feature vector is calculated based on the decay coefficient, fluctuation amplitude, and time difference of the current feature vector, and the parameters of the long short-term memory network model are updated according to the weighted coefficient.

[0045] Specifically, to achieve high-precision load prediction, the embodiment of the present invention uses a long short-term memory (LSTM) network to perform load demand prediction, and realizes high-precision load demand prediction through real-time data update and adaptive optimization of model parameters.

[0046] The LSTM network is used to model the time series characteristics of the load, and its input is the feature vector of the load , and the output is the predicted load demand. Information is transmitted inside the LSTM through memory cells, and the state of the memory cells determines the output at the current moment. The update formula for the output is as follows: Where, is the hidden state, representing the feature information at the current moment; is the input feature vector; is the weight matrix, is the bias, and f is the activation function. Each time step of the LSTM processes the input feature vector, and captures the long-term and short-term dependence characteristics of the load by transmitting the hidden state and cell state at the current moment in the network.

[0047] Furthermore, during the load demand forecasting process, the long short-term memory network model adaptively updates the model through a preset weighted sliding window. Specifically, assume there is a window containing n input data points, and each data point in the window will have a corresponding weighted coefficient , and this weighted coefficient determines the degree of influence of the data point on model update.

[0048] Specifically, the weighted coefficient is calculated according to the decay coefficient , the fluctuation amplitude , and the time difference , and the formula is: where represents the time difference between the current moment and the moment of the rd data point ; is the decay coefficient used to control the intensity of time decay and can be set to 1; is the fluctuation amplitude of the rd data point, indicating the degree of change of the load data. The decay coefficient controls the decay effect of the load data over time, that is, the closer the data point is to the current moment, the greater its weight should be. The fluctuation amplitude of the data will affect its weight, and the data points with larger fluctuations have a greater impact on model update. The time difference between the data point and the current moment will also affect the weight. As time goes by, the influence of data points farther from the current moment on the model should gradually decrease.

[0049] When new data arrives, the LSTM model is updated according to the calculated weighted coefficient . The update process is as follows: On the one hand, by calculating the weighted coefficient of each data point, the learning rate of the model is adjusted. A larger weighted coefficient corresponds to a larger learning rate, which means that the new load data has a greater impact on the model and the update will be faster. On the other hand, for the data at the current moment, the system dynamically updates the parameters of the LSTM according to the weighted coefficient , thereby improving the prediction accuracy. By adjusting the weights and biases of the model, the model can better fit the newly arrived data points. The update formula is expressed as: where and denotes the model parameters at the current moment, and is the parameter update amount obtained through backpropagation calculation.

[0050] S4: According to the predicted load demand, a load scheduling model based on game theory is adopted to obtain a load scheduling plan; wherein, the load scheduling model includes a local game layer and a global game layer. In each round of the game process of the load scheduling model, the load decision-making weight of the user side is determined according to the fault tolerance factor, and the load decision of the user side is adjusted based on the fault tolerance factor.

[0051] Specifically, the embodiment of the present invention combines game theory with a fault tolerance mechanism, and realizes the dynamic allocation and optimization of the load through multi-level games and real-time fault tolerance adjustment. During the load scheduling process, the load demand prediction value obtained through the LSTM network will be used as the input for the load scheduling of each user side, thereby realizing the reasonable distribution of the load.

[0052] The load scheduling model adopts a strategy combining local game and global game. The local game layer optimizes its load scheduling by considering the load demand and device status of a single user side, and the global game layer coordinates the load decisions of different user sides to ensure the overall stability of the system.

[0053] Each user side formulates a scheduling strategy in the local game layer according to its own load demand, device status, and predicted load change trend (i.e., ). The goal of the local game is to maximize its utility function by optimizing the load scheduling, thereby reducing load fluctuations, avoiding overload, and improving the operating stability of the device. The utility function of the local game is as follows: wherein, is the current load of the th user side; is the load demand predicted by the LSTM; is the load change amount, indicating the change of the load; and are adjustment coefficients, which respectively control the stability of the target load and the smoothness of the load fluctuation; is the utility function of the th user side. Through this utility function, the user side will optimize its load scheduling strategy, aiming to reduce device damage and load imbalance, and at the same time ensure that the load fluctuation is within a reasonable range to avoid overload or power waste.

[0054] The global game layer needs to coordinate the load decisions of all client terminals to ensure system stability and security. The goal of the global game is to reasonably allocate the loads of all client terminals in the virtual power plant to avoid the impact of load fluctuations of any single client terminal on global stability. The specific formula is as follows: where, is the total number of client terminals in the system; and are adjustment coefficients used to control the balance between the load target and the load change; is the load decision value of the i-th client terminal at time t; is the target load value of the i-th client terminal at time t; is the load change amount of the i-th client terminal at time t, . The key to the global game is to coordinate the load demands of different client terminals to ensure load balance between different levels of the system, thereby reducing the impact of power fluctuations on equipment and avoiding system instability or equipment damage.

[0055] The local game and the global game interact with each other through a feedback mechanism. The load decisions generated by the local game are fed back into the global game to optimize the load scheduling of the entire system, ensuring global stability while meeting the demands of each client terminal. Conversely, the scheduling results of the global game are also fed back to each client terminal to adjust its load decision. Through multiple game processes, the system gradually optimizes the load scheduling and improves the rationality and stability of the overall load distribution.

[0056] Furthermore, during the actual load scheduling process, due to factors such as equipment failures, power fluctuations, or information delays, the system may encounter uncertainty problems. To address these emergencies, the present invention adopts a dynamic fault-tolerant game mechanism, which can flexibly adjust the scheduling strategy in the face of faults or load fluctuations to maintain the stability of the system.

[0057] As one of the optional embodiments, the fault-tolerant factor is determined according to the load change amplitude, equipment health status, and external disturbance factors.

[0058] Specifically, during the load scheduling process, by introducing the fault-tolerant factor , the game strategy is dynamically adjusted. The fault-tolerant factor is determined according to factors such as the amplitude of load change, equipment health status, and external disturbances. The formula for the fault-tolerant factor is: where, represents the equipment health factor, which reflects the impact of the equipment health status on load scheduling. If the equipment is in a faulty state, then has a smaller value, indicating that the load-bearing capacity of the equipment is reduced; Denotes the load fluctuation factor, which reflects the impact of the amplitude of load changes on the scheduling strategy. The greater the load fluctuation, the greater the fault tolerance factor, and the system will tend to adjust the load distribution strategy; Denotes the external environment factor, considering the impact of external disturbances on load scheduling. If the external conditions are unstable, the fault tolerance factor increases to improve the robustness of the system; 、 、 Are adjustment coefficients used to control the contribution of each factor to the fault tolerance factor. In the embodiments of the present invention, through the dynamic adjustment of the fault tolerance factor, it is ensured that in the face of sudden load fluctuations, equipment failures or external interferences, the system can flexibly adjust the load distribution strategy, thereby ensuring the robustness of the scheduling.

[0059] Furthermore, the fault tolerance factor will directly affect the weight of the load decision-making of the user side during the game process. When the system is in an unstable state, the user side with a larger fault tolerance factor will assume more load distribution responsibilities to ensure the stability of the overall load scheduling of the system. User side load decision-making weight Is dynamically adjusted by the fault tolerance factor, and the calculation formula for the load decision-making weight of the user side is: Wherein, Is the load decision-making weight of the i-th user side, And Are the utility functions of the i-th user side and the j-th user side, And Are the fault tolerance factors of the i-th user side and the j-th user side, and N is the total number of user sides in the system. The user side with a larger fault tolerance factor will be allocated more load to ensure that the user side can cope with sudden equipment failures or load fluctuations and avoid over-concentration or uneven distribution of the system load.

[0060] Furthermore, during each round of the game process, the system dynamically adjusts the load distribution of each user side according to the value of the fault tolerance factor. For each user side, the adjustment of its load decision depends not only on the result of the current game strategy but also on the fault tolerance factor . The specific adjustment formula is: Wherein, Is the load decision of the i-th user side in the (t + 1)-th round of the game, Is the load decision of the i-th user side in the t-th round of the game, Is the adjustment step size, Is the utility function For the load decision Gradient, Is the adjustment parameter, is the fault tolerance factor for the i-th client. Through this formula, the system can adjust the load distribution strategy according to the change of the fault tolerance factor, so that the system can maintain good stability and robustness in the face of equipment failures, load fluctuations or external disturbances.

[0061] Further, after obtaining the scheduling plan using the load scheduling model, the client devices are controlled according to the scheduling plan to adjust the load distribution, such as adjusting the output power of the power generation equipment or changing the charge and discharge strategy of the energy storage equipment. The client devices execute according to the instructions, implement load control, and feedback the operating status to the virtual power plant control center. Furthermore, the control center adjusts the load prediction for the future period according to the deviation between the feedback data and the prediction result, and its control strategy is fine-tuned according to the actual situation and environmental changes to optimize the load distribution of the entire system.

[0062] The embodiments of the present invention aim to improve the load management and scheduling capabilities of the virtual power plant, ensure precise scheduling and efficient control in dynamic power demand and supply fluctuations, optimize the operating efficiency and system stability of the virtual power plant through advanced edge computing technologies and intelligent scheduling algorithms, and meet the requirements for flexible scheduling and precise control in modern power systems; By deploying edge computing nodes at the client side, load data is collected in real time, and combined with noise filtering and data synchronization technologies to ensure the accuracy and timeliness of the data. To improve the reliability and processing efficiency of the data, fuzzy rules are used to intervene in the loads that meet the adjustment conditions. The rule base covers multiple aspects such as load regulation, load prediction, and load optimization. By formulating precise adjustment thresholds and optimization strategies, the system can efficiently respond to load fluctuations, effectively avoiding the problems of data delay and error that may exist in traditional centralized computing, providing a reliable basis for subsequent load analysis and scheduling; Through the adaptive analysis and prediction of historical load data and real-time collected load change information, the changing trend of load demand is accurately captured. By dynamically adjusting the prediction results, it can flexibly respond to the load demands in different time periods and automatically adjust the control strategy to cope with large load fluctuations. This adaptive control strategy not only significantly improves the accuracy of load scheduling, but also enhances the regulation ability of the virtual power plant in a complex power environment and the stability of the system, ensuring that the power supply and demand can be balanced during peak periods or when the demand fluctuates greatly; Through game theory and fault tolerance mechanisms, the load coordination and scheduling process at the user side of the virtual power plant is further optimized. The load scheduling model realizes global and local load collaborative scheduling by establishing a real-time information feedback mechanism between the edge computing nodes and the virtual power plant control center. The game theory model is used to optimize the scheduling strategies of each load unit to maximize the interests of all parties; while the fault tolerance mechanism ensures the robustness of the system. Even if some nodes fail, the system can still automatically adjust and maintain stable operation. In complex power environments such as power supply fluctuations and sudden demand changes, the system can automatically adjust the load distribution strategy to ensure the efficient utilization of power resources while maintaining the stable operation of the system.

[0063] Through this intelligent distributed control mechanism, the virtual power plant can achieve precise load scheduling and efficient control in a dynamic and complex power environment, thus significantly improving the comprehensive operation efficiency, system stability and reliability of the virtual power plant, providing technical support for the intelligent scheduling and load management of the virtual power plant, and laying a foundation for the flexible scheduling and optimal operation of future power systems.

[0064] Correspondingly, the present invention also provides an intelligent load scheduling system for the user side of the virtual power plant, which can implement all the processes of the intelligent load scheduling method for the user side of the virtual power plant in the above embodiments.

[0065] Please refer to Figure 2 , Figure 2 which is a schematic structural diagram of an intelligent load scheduling system for the user side of the virtual power plant provided by an embodiment of the present invention. The intelligent load scheduling system for the user side of the virtual power plant includes: A data acquisition module 201, configured to acquire the original load data of the user side of the virtual power plant; A feature extraction module 202, configured to perform feature extraction according to the original load data to obtain a feature vector; A load prediction module 203, configured to perform load demand prediction according to the feature vector to obtain the predicted load demand; A load scheduling module 204, configured to adopt a load scheduling model based on game theory according to the predicted load demand to obtain a load scheduling plan; Wherein, the load scheduling model includes a local game layer and a global game layer. In each round of the game process, the load scheduling model determines the load decision weight of the user side according to the fault tolerance factor and adjusts the load decision of the user side based on the fault tolerance factor.

[0066] Preferably, the fault tolerance factor is determined according to the load change amplitude, equipment health status and external disturbance factors.

[0067] Preferably, the feature extraction module 202 is specifically configured to: Filter the noise and remove outliers from the original load data to obtain the first load data; Synchronize and standardize the first load data to obtain the second load data; Process the second load data using an adaptive fuzzy rule to obtain the third load data; Based on the third load data, use the variational mode decomposition method for feature extraction to obtain a feature vector; the feature vector includes a long-term trend component and its dynamic change amount, a periodic component and its dynamic change amount, and a random fluctuation component.

[0068] Preferably, the process of using an adaptive fuzzy rule to process the second load data to obtain the third load data includes: Based on a preset fuzzy rule, adjust the parameters of the fuzzy rule according to the fluctuation amplitude of the second load data to obtain an adjusted rule width and a fuzzy set center; Calculate the membership degree of each fuzzy rule according to the second load data, the adjusted rule width, and the fuzzy set center; Perform weighted averaging on the membership degrees of the fuzzy rules to calculate and obtain the third load data.

[0069] Preferably, the process of using the variational mode decomposition method for feature extraction based on the third load data to obtain a feature vector includes: Determine the decomposition layer number of the variational mode decomposition according to the fluctuation amplitude of the third load data; the greater the fluctuation amplitude, the more the decomposition layer number; Perform variational mode decomposition on the third load data according to the decomposition layer number to obtain the long-term trend component, the periodic component, and the random fluctuation component corresponding to the third load data; Determine a feature vector based on the long-term trend component, the periodic component, the random fluctuation component, and the corresponding feature component weights; wherein, the feature component weights are dynamically adjusted according to the fluctuation amplitudes of each component.

[0070] Preferably, the load prediction module 203 is specifically configured to: Perform load demand prediction using a long short-term memory network model based on the feature vector to obtain the predicted load demand; Wherein, before performing load demand prediction, calculate the weighted coefficient corresponding to the current feature vector according to the decay coefficient, the fluctuation amplitude, and the time difference of the current feature vector, and update the parameters of the long short-term memory network model according to the weighted coefficient.

[0071] Preferably, the calculation formula for the feature component weights is as follows: Among them, is the weight of the th feature component, is the weighted attenuation coefficient, is the fluctuation amplitude of the feature component.

[0072] Preferably, the calculation formula for the load decision weight of the client is: Among them, is the load decision weight of the and are the utility functions of the and th client and the

[0073] th client, and and are the fault tolerance factors of the th client and the th client. N is the total number of clients in the system. is the adjustment step size, is the utility function for the load decision is the adjustment parameter,

[0074] is the fault tolerance factor of the In specific implementation, the working principle, control process, and achieved technical effects of the intelligent load scheduling system for the client of the virtual power plant provided by the embodiments of the present invention are the same as those of the intelligent load scheduling method for the client of the virtual power plant in the above embodiments, and will not be elaborated here.

[0075] ​An intelligent load scheduling method and system for the user side of a virtual power plant are provided in an embodiment of the present invention. The beneficial effects are as follows: By extracting features from the original load data, predicting the load demand based on the extracted features, and then obtaining a load scheduling plan using a load scheduling model based on game theory. In each round of the game process of the load scheduling model, the load decision weight of the user side is determined according to the fault tolerance factor, and the load decision of the user side is adjusted based on the fault tolerance factor, realizing the global and local load collaborative scheduling, ensuring the efficient utilization of power resources, guaranteeing the robustness of the system, and improving the accuracy and flexibility of the virtual power plant load scheduling; By adopting an adaptive fuzzy rule adjustment mechanism to optimize the data processing process during data processing, the adaptability and accuracy of the load data are further improved; By using variational mode decomposition and an adaptive decomposition mechanism to analyze the load data and dynamically adjusting the decomposition layer number according to the data fluctuation amplitude, the modeling accuracy can be improved and the calculation complexity can be reduced; By using an adaptive updated long short-term memory network to predict the load demand, high-precision load demand prediction is realized, and the accuracy of the virtual power plant load scheduling is further improved.

[0076] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. An intelligent load scheduling method for the user side of a virtual power plant, characterized in that, Including: Obtain the original load data of the virtual power plant user side; Extract features from the original load data to obtain a feature vector; Perform load demand prediction based on the feature vector to obtain the predicted load demand; According to the predicted load demand, adopt a load scheduling model based on game theory to obtain a load scheduling plan; Wherein, the load scheduling model includes a local game layer and a global game layer. In each round of the game process, the load decision weight of the user side is determined according to the fault tolerance factor, and the load decision of the user side is adjusted based on the fault tolerance factor.

2. The intelligent load scheduling method for the user side of the virtual power plant according to claim 1, wherein The fault tolerance factor is determined according to the load change amplitude, equipment health status, and external disturbance factors.

3. The intelligent load scheduling method for the user side of the virtual power plant according to claim 2, wherein, The extracting features from the original load data to obtain a feature vector includes: Perform noise filtering and outlier removal on the original load data to obtain the first load data; Perform data synchronization and normalization processing on the first load data to obtain the second load data; Perform data processing on the second load data using an adaptive fuzzy rule to obtain the third load data; According to the third load data, adopt the variational mode decomposition method for feature extraction to obtain a feature vector; the feature vector includes a long-term trend component and its dynamic change amount, a periodic component and its dynamic change amount, and a random fluctuation component.

4. The intelligent load scheduling method for the user side of the virtual power plant according to claim 3, characterized in that, The performing data processing on the second load data using an adaptive fuzzy rule to obtain the third load data includes: Based on a preset fuzzy rule, adjust the parameters of the fuzzy rule according to the fluctuation amplitude of the second load data to obtain the adjusted rule width and fuzzy set center; Calculate the membership degree of each fuzzy rule according to the second load data, the adjusted rule width, and the fuzzy set center; Perform weighted averaging on the membership degrees of the fuzzy rules to calculate and obtain the third load data.

5. The intelligent load scheduling method for the user side of a virtual power plant according to claim 4, characterized in that The adopting the variational mode decomposition method for feature extraction according to the third load data to obtain a feature vector includes: Determine the decomposition layer number of the variational mode decomposition according to the fluctuation amplitude of the third load data; the greater the fluctuation amplitude, the more the decomposition layer number; Perform variational mode decomposition on the third load data according to the decomposition layer number to obtain the long-term trend component, periodic component, and random fluctuation component corresponding to the third load data; Based on the long-term trend component, the periodic component, the random fluctuation component, and the corresponding feature component weights, determine a feature vector; wherein, the feature component weights are dynamically adjusted according to the fluctuation amplitude of each component.

6. The intelligent load scheduling method for the user side of the virtual power plant according to claim 5, characterized in that, The performing load demand prediction according to the feature vector to obtain the predicted load demand includes: According to the feature vector, adopt a long short-term memory network model for load demand prediction to obtain the predicted load demand; Wherein, before performing load demand prediction, calculate the weighted coefficient corresponding to the current feature vector according to the decay coefficient, fluctuation amplitude, and time difference of the current feature vector, and update the parameters of the long short-term memory network model according to the weighted coefficient.

7. The intelligent load scheduling method for the user side of a virtual power plant according to claim 5, characterized in that, The calculation formula of the feature component weight is as follows: Among them, is the weight of the th feature component, is the weighted attenuation coefficient, is the fluctuation amplitude of the feature component.

8. The intelligent load scheduling method for the user side of the virtual power plant according to claim 2, characterized in that The calculation formula of the load decision weight of the user side is: Among them, is the load decision weight of the i-th client, and are the utility functions of the i-th client and the j-th client, and are the fault tolerance factors of the i-th client and the j-th client, and N is the total number of clients in the system.

9. The intelligent load scheduling method for the virtual power plant user side according to claim 8, characterized in that, Adjusting the load decision of the user side based on the fault tolerance factor includes: The load decision of the i-th client in the (t + 1)-th round of the game is calculated according to the following formula :[[]]END]] wherein, the load decision of the i-th client in the t-th round of the game is the adjustment step size, is the utility function for the load decision gradient of is the adjustment parameter, is the fault tolerance factor of the i-th client.

10. An intelligent load scheduling system for the user side of a virtual power plant, characterized in that, Including: A data acquisition module for acquiring the original load data of the user side of the virtual power plant; A feature extraction module for extracting features from the original load data to obtain a feature vector; A load prediction module for predicting the load demand based on the feature vector to obtain the predicted load demand; A load scheduling module for obtaining a load scheduling plan by using a load scheduling model based on game theory according to the predicted load demand; Wherein, the load scheduling model includes a local game layer and a global game layer. In each round of the game process of the load scheduling model, the load decision weight of the user side is determined according to the fault tolerance factor, and the load decision of the user side is adjusted based on the fault tolerance factor.

Images