A virtual power plant resource scheduling optimization method based on edge computing

Optimizing the power resource scheduling of virtual power plants through edge computing, solving the problems of virtual power plants control accuracy and real-time, and achieving efficient power resource management and green energy utilization.

CN118659468BActive Publication Date: 2024-11-01SHAANXI SIJI TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411154666.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-11-01
Estimated Expiration
2044-08-22

AI Technical Summary

Technical Problem

The existing virtual power plant control schemes have problems such as poor control accuracy and high uncertainty, especially in large-scale virtual power plant systems, which affects the stability and reliability of the power system, and it is difficult for traditional cloud computing platforms to Meet the real-time scheduling needs of virtual power plants.

Method used

Using the virtual power plant resource scheduling optimization method based on edge computing, obtaining raw data through edge servers for preprocessing and feature extraction, building a demand prediction model, establishing a power resource scheduling optimization model, using the alternating direction multiplier method to solve and optimize the scheduling strategy, and Adjust the optimal scheduling policy through priority assessment, and finally upload the policy to the cloud control center for execution.

Benefits of technology

It improves the real-time and responsiveness of power scheduling in virtual power plants, optimizes the integration and distribution of power resources, reduces energy waste, improves energy utilization efficiency, and promotes the efficient utilization of green energy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118659468B_ABST
    Figure CN118659468B_ABST
Patent Text Reader

Abstract

The present invention discloses a virtual power plant resource scheduling optimization method based on edge computing, which relates to the technical field of virtual power plants, including: an edge server obtains the original data on the user-end device and performs preprocessing; mining key features through feature extraction, building a demand forecasting model, obtaining the power resource demand forecast value through the original data and key features, and establishing a power resource scheduling optimization model in combination with equipment constraints to minimize costs; solving the power resource scheduling optimization model through the alternating direction multiplier method to obtain a scheduling strategy; the edge server prioritizes all scheduling tasks, and adjusts the scheduling strategy according to the priority to obtain the optimal scheduling strategy; the edge server uploads the optimal scheduling strategy to the cloud control center platform for execution, and realizes accurate power resource scheduling. The present invention can efficiently schedule dispersed power resources, thereby optimizing the scheduling of virtual power plants and ensuring the stable operation of the power system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of power generation technology, and in particular to a virtual power plant resource scheduling optimization method based on edge computing. Background Art

[0002] The introduction of Virtual Power Plant (VPP) technology provides a solution for the operation of the power system after the large-scale new energy power generation system is connected to the power grid. As a bridge between the power system and distributed energy, the virtual power plant can ultimately achieve resource integration and peak load shifting of the power system by controlling the flexible resources, electric vehicle loads, controllable loads and energy storage equipment in the power system, thereby ensuring the stable and reliable operation of the power system. However, the current control scheme for virtual power plants still has the defects of poor control accuracy and many uncertain factors in the control process, which makes the current virtual power plant technology have many problems in practical application, which in turn affects the stable and reliable operation of the power system.

[0003] The construction process of a virtual power plant is a process of digitization, intelligence, and Internetization of traditional power grids. Through advanced information and communication technologies and software systems, the aggregation and coordinated optimization of distributed energy sources such as distributed power sources, energy storage systems, controllable loads, and electric vehicles are realized to participate in the power market and power grid operation as a special power plant in a power coordination management system. With the rapid development of information technology and the Internet of Vehicles, the number of scenarios in which electric vehicles and other mobile energy bodies in virtual power plants participate in virtual power plant regulation has increased significantly. Most of these applications generate delay-sensitive and data-intensive tasks, which adds load to traditional cloud computing platforms. In addition, the task characteristics in the network are different. Due to the lack of flexibility, traditional static cloud centers are difficult to flexibly provide services to virtual power plant terminals based on different task characteristics.

[0004] The patent document with the publication number CN117350425A and the name of a virtual power plant optimization scheduling method comprehensively considers the correlation and uncertainty of wind and solar output with scenario technology, establishes an optimization scheduling model with the expected value of the net profit of the virtual power plant operation as the optimization goal, and uses the improved artificial bee colony algorithm to solve the optimization scheduling model to obtain the optimal operation plan of the virtual power plant. This method considers the impact of wind and solar output correlation and the operating profit risk brought to the virtual power plant system by the uncertainty of wind and solar resources, constructs a virtual power plant optimization scheduling model, and uses the improved artificial bee colony algorithm to solve it. It solves the problem that the uncertainty of the output and load demand of renewable energy sources such as wind and solar resources leads to a certain risk in their benefits, but does not specifically mention how to use the real-time data processing capabilities of edge computing for optimization scheduling. In addition, the improved artificial bee colony algorithm may be high in complexity and computing resource requirements, especially in large-scale virtual power plant systems. This complexity may increase computing delays and affect the efficiency and accuracy of real-time scheduling.

[0005] The patent document with the publication number CN116227128A and the name of the patent document is an edge computing method for virtual power plant services. The method deploys a physical model of typical distributed resources on the edge side; deploys an economic model of typical application scenarios on the edge side; obtains data of various distributed resources; extracts features of user electricity consumption data and obtains numerical models of key equipment; aggregates and analyzes all distributed resources of users to obtain external characteristics of virtual power plants; calculates the economic efficiency of participating in various power market services in real time and synchronizes them to the virtual power plant management and control platform; receives the dispatching requirements of the upper virtual power plant management and control platform, and independently chooses whether to accept the dispatching and how to participate in the dispatching through economic analysis, and achieves the balance between user-side benefit maximization and power safety through multi-objective optimization methods. It can enable users to independently choose to participate in demand response, ancillary services, spot markets or wall-to-wall electricity sales, and independently balance the relationship between peak-valley arbitrage, capacity electricity prices and virtual power plants. However, such complex model deployment and data processing tasks may make the coordination and management of edge computing resources difficult, especially when dealing with high loads or sudden demands, which places high demands on the performance of edge computing devices and may restrict the expansion capabilities of the system. Summary of the invention

[0006] In order to solve the above technical problems, a technical solution adopted by the present invention is to provide a virtual power plant resource scheduling optimization method based on edge computing, the method comprising:

[0007] S10 edge server obtains the original data on the user terminal device, and pre-processes the original data;

[0008] S20. Mining key features from the preprocessed data through feature extraction, constructing a demand forecasting model, and obtaining a power resource demand forecast value through the original data and key features;

[0009] S30. Establishing a power resource scheduling optimization model based on the power resource demand forecast value and equipment constraints to minimize the operating costs of the virtual power plant;

[0010] S40. Solving the power resource scheduling optimization model by alternating direction multiplier method to obtain a scheduling strategy;

[0011] S50 edge server for all scheduling tasks to prioritize, by the priority to adjust the scheduling strategy to obtain the optimal scheduling strategy;

[0012] S60. The edge server uploads the optimal scheduling strategy to the cloud control center platform for execution to achieve accurate power resource scheduling;

[0013] The raw data includes user activity data and real-time data of power resources;

[0014] The preprocessing includes cleaning and repairing outliers or missing values ​​in the data, and standardizing different types of data to make them have the same scale;

[0015] The equipment constraints include power generation equipment output limitation, energy storage system charging and discharging power, energy storage capacity limitation, and power demand balance constraint.

[0016] Furthermore, the key features include load features, power generation features, and energy storage system features, including:

[0017] The load characteristics include peak load, average load, load fluctuation and load duration;

[0018] The power generation characteristics include power generation, power generation efficiency, generator operation status and renewable energy power generation ratio;

[0019] The energy storage system characteristics include energy storage capacity, charge and discharge times, charge and discharge rates, and the health of the energy storage system.

[0020] Furthermore, the S20 includes:

[0021] S21. Build demand forecasting model through LSTM neural network;

[0022] The demand forecasting model includes an input layer, an LSTM unit layer, and an output layer;

[0023] The LSTM unit layer includes a candidate unit state, a unit state, a hidden state, a forget gate, an input gate, and an output gate;

[0024] S22. Taking the load characteristics, power generation characteristics, and energy storage system characteristics as the input feature vector at time t;

[0025] The input feature vector at time t is expressed as:

[0026] X t =[ X load , t , X gen , t , X storage , t ] ;

[0027] in, is the input feature vector at time t, is the load characteristic, For the power generation characteristics, It is the characteristics of energy storage system;

[0028] S23. Calculate the activation values ​​of the forget gate, input gate, and output gate at time t;

[0029] The calculation formula of the activation value of the forget gate, input gate, and output gate at time t is:

[0030] F t = δ ( W F [ H t -1 , X t ]+ b F ) ;

[0031] I t = δ ( W I [ H t -1 , X t ]+ b I ) ;

[0032] O t = δ ( W O [ H t -1 , X t ]+ b O ) ;

[0033] in, is the activation value of the forget gate at time t, is the activation value of the input gate at time t, is the activation value of the output gate at time t, is the Sigmoid activation function, , , are the weight matrices of the forget gate, input gate, and output gate respectively. , , are the bias vectors of the forget gate, input gate, and output gate respectively, is the hidden state at time t-1;

[0034] S24. Calculate the candidate unit state and unit state at time t;

[0035] The candidate unit state at time t , Unit Status The calculation formula is:

[0036] C t ̃ = tanh ( W C [ H t -1 , X t ]+ b C ) ;

[0037] ;

[0038] in, is the unit state at time t-1, is the candidate unit state at time t-1, is the weight matrix of the candidate unit state, is the bias vector of the candidate cell state, is the activation function;

[0039] S25. Output the hidden state at time t;

[0040] The hidden state at time t The calculation formula is:

[0041] ;

[0042] S26. Predict the power demand forecast value at time t;

[0043] The power demand forecast value at time t The calculation formula is:

[0044] ;

[0045] in, is the output layer weight matrix, is the output layer bias vector;

[0046] S27. Calculating the mean square error between the power demand forecast value and the actual power demand, and optimizing the demand forecasting model by minimizing the mean square error;

[0047] The calculation formula of the mean square error is:

[0048] ;

[0049] in, is the actual power demand at time t, is the predicted value of electricity demand at time t, and K is the number of training samples.

[0050] Furthermore, the step S30 includes:

[0051] S31. Establish an optimization model for power resource scheduling to minimize the operating costs of virtual power plants;

[0052] S32. Setting output limit of power generation equipment;

[0053] The output limit of the power generation equipment is expressed as:

[0054] ;

[0055] in, is the minimum output of the ith power generation equipment, is the output of the i-th generating equipment at the t-th moment, is the maximum output of the i-th power generation equipment;

[0056] S33. Setting the energy storage system charging and discharging power and energy storage capacity limits;

[0057] The energy storage system charging and discharging power and energy storage capacity limit are expressed as:

[0058] ;

[0059] ;

[0060] ;

[0061] in, is the maximum charging power of the energy storage system, is the maximum discharge power of the energy storage system, is the charging power of the energy storage system at time t, is the discharge power of the energy storage system at time t, is the energy storage capacity of the energy storage system at time t, is the maximum limit of energy storage capacity, The minimum limit for energy storage capacity;

[0062] S34. Setting power demand balancing constraints;

[0063] The power demand balance constraint is expressed as:

[0064] ;

[0065] in, is the predicted value of electricity demand at time t, is the output of the i-th power generation equipment at the t-th moment, and N is the total number of power generation equipment.

[0066] Furthermore, the optimization objective function of the power resource scheduling optimization model is expressed as:

[0067] ;

[0068] in, is the virtual power plant operating cost, is the power generation cost of the ith power generation equipment at the tth moment, is the charging and discharging cost of the energy storage system at time t, is the output of the i-th generating equipment at the t-th moment, is the energy storage capacity of the energy storage system at time t, T is the scheduling time range, N is the total number of power generation equipment, is the charging power of the energy storage system at time t, is the discharge power of the energy storage system at the tth moment.

[0069] Further, the S40 includes:

[0070] S41. Transform the power resource dispatch optimization model into an equivalent form with Lagrange multipliers, introduce a Lagrange multiplier for each constraint condition, and construct the Lagrange function;

[0071] The constraints include power generation equipment output limitation, energy storage system charging and discharging power, energy storage capacity limitation and power demand balance constraint;

[0072] S42. Initializing the Lagrangian multiplier, minimizing the Lagrangian function by optimizing the original variables;

[0073] The original variables include the energy storage capacity of the energy storage system at time t , the charging power of the energy storage system at time t , the discharge power of the energy storage system at time t ;

[0074] S43. updating the Lagrange multiplier according to the original variables and constraints;

[0075] The calculation formula for updating the Lagrange multiplier is:

[0076] ;

[0077] in, is the updated Lagrange multiplier in the k+1th iteration, A is the constraint matrix, is the Lagrange multiplier in the kth iteration, is the original variable updated in the k+1th iteration, is the penalty factor, is a constant vector;

[0078] S44. Determine whether the convergence condition is met, the convergence condition being that the objective function change is less than a set threshold or the maximum number of iterations is reached. If the convergence condition is met, stop the iteration; if not, return to S42;

[0079] S45. Decode the result and obtain the optimized scheduling strategy based on the optimal original variables and Lagrange multipliers.

[0080] Furthermore, the S50 includes:

[0081] S51 edge server creates a priority set, all tasks are stored in the priority set;

[0082] S52 edge server calculates the priority of all tasks;

[0083] S53 sorts all tasks in the priority set in descending order according to the priority level;

[0084] S54. Each time a power resource scheduling decision is made, the task ranked first in the priority set is scheduled first.

[0085] Furthermore, the calculation formula of the priority is:

[0086] ;

[0087] in, , , Represent the weight coefficients of importance, urgency and impact range respectively, is the importance of the i-th task, is the urgency of the ith task, is the influence range of the i-th task.

[0088] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0089] 1. The virtual power plant resource scheduling optimization method based on edge computing provided by the present invention can greatly meet the real-time and high responsiveness requirements of virtual power plant power scheduling by assisting virtual power plant power resource scheduling through edge computing. The application of edge computing makes the data processing and decision-making process closer to the data source, thereby reducing delays and improving the response speed of the scheduling system.

[0090] 2. The virtual power plant resource scheduling optimization method based on edge computing provided by the present invention significantly improves the performance of virtual power plant power resource scheduling by introducing edge computing. By utilizing the distributed processing capability of edge computing, a large amount of power data can be more efficiently integrated and analyzed, the scheduling algorithm can be optimized, and the efficient utilization and stable supply of power resources can be ensured.

[0091] 3. The virtual power plant resource scheduling optimization method based on edge computing provided by the present invention can optimize the discrete power resource scheduling of the virtual power plant and reasonably allocate power resources. Through precise resource scheduling strategies, the allocation of power resources can be flexibly adjusted under different time and load conditions, thereby maximizing the economic benefits of the virtual power plant.

[0092] 4. The virtual power plant resource scheduling optimization method based on edge computing provided by the present invention not only improves energy utilization efficiency but also effectively reduces energy waste by optimizing power resource scheduling. Through intelligent scheduling and resource allocation, it can ensure that while meeting electricity demand, power loss and unnecessary energy consumption can be minimized, further promoting the efficient use of green energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 It is a flow chart of a virtual power plant resource scheduling optimization method based on edge computing provided by the present invention.

[0094] Figure 2 It is a flowchart of constructing a demand forecasting model for a virtual power plant resource scheduling optimization method based on edge computing provided by the present invention.

[0095] Figure 3It is a flow chart of establishing an electric power resource scheduling optimization model for a virtual power plant resource scheduling optimization method based on edge computing provided by the present invention.

[0096] Figure 4 It is a scheduling strategy solution flow chart of a virtual power plant resource scheduling optimization method based on edge computing provided by the present invention.

[0097] Figure 5 It is a priority assessment flow chart of a virtual power plant resource scheduling optimization method based on edge computing provided by the present invention. DETAILED DESCRIPTION

[0098] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.

[0099] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present invention, rather than all of the embodiments.

[0100] Figure 1 : is a flow chart of a virtual power plant resource scheduling optimization method based on edge computing provided by an embodiment of the present invention, the method comprising:

[0101] S10 edge server obtains the original data on the user terminal device, and pre-processes the original data;

[0102] The raw data includes user activity data and real-time data of power resources;

[0103] The preprocessing includes cleaning and repairing outliers or missing values ​​in the data, and standardizing different types of data to make them have the same scale.

[0104] S20. Mining key features from the preprocessed data through feature extraction, constructing a demand forecasting model, and obtaining a power resource demand forecast value through the original data and key features;

[0105] The key characteristics include load characteristics, power generation characteristics, and energy storage system characteristics;

[0106] The load characteristics include peak load, average load, load fluctuation and load duration;

[0107] The power generation characteristics include power generation, power generation efficiency, generator operation status and renewable energy power generation ratio;

[0108] The energy storage system characteristics: including energy storage capacity, charge and discharge times, charge and discharge rates, and health of the energy storage system;

[0109] Further, refer to Figure 2 , the S20 comprises:

[0110] S21. Build demand forecasting model through LSTM neural network;

[0111] The demand forecasting model includes an input layer, an LSTM unit layer, and an output layer;

[0112] The LSTM unit layer includes a candidate unit state, a unit state, a hidden state, a forget gate, an input gate, and an output gate;

[0113] S22. Taking the load characteristics, power generation characteristics, and energy storage system characteristics as the input feature vector at time t;

[0114] The input feature vector at time t is expressed as:

[0115] X t =[ X load , t , X gen , t , X storage , t ] ;

[0116] in, is the input feature vector at time t, is the load characteristic, For the power generation characteristics, It is the characteristics of energy storage system;

[0117] S23. Calculate the activation values ​​of the forget gate, input gate, and output gate at time t;

[0118] The calculation formula of the activation value of the forget gate, input gate, and output gate at time t is:

[0119] F t = δ ( W F [ H t -1 , X t ]+ b F ) ;

[0120] I t = δ ( W I [ H t -1 , X t ]+ b I ) ;

[0121] O t = δ ( W O [ H t -1 , X t ]+ b O ) ;

[0122] in, is the activation value of the forget gate at time t, is the activation value of the input gate at time t, is the activation value of the output gate at time t, is the Sigmoid activation function, , , are the weight matrices of the forget gate, input gate, and output gate respectively. , , are the bias vectors of the forget gate, input gate, and output gate respectively, is the hidden state at time t-1;

[0123] S24. Calculate the candidate unit state and unit state at time t;

[0124] The candidate unit state at time t , Unit Status The calculation formula is:

[0125] C t ̃ = tanh ( W C [ H t -1 , X t ]+ b C ) ;

[0126] ;

[0127] in, is the unit state at time t-1, is the candidate unit state at time t-1, is the weight matrix of the candidate unit state, is the bias vector of the candidate cell state, is the activation function;

[0128] S25. Output the hidden state at time t;

[0129] The hidden state at time t The calculation formula is:

[0130] ;

[0131] S26. Predict the power demand forecast value at time t;

[0132] The power demand forecast value at time t The calculation formula is:

[0133] ;

[0134] in, is the output layer weight matrix, is the output layer bias vector;

[0135] S27. Calculating the mean square error between the power demand forecast value and the actual power demand, and optimizing the demand forecasting model by minimizing the mean square error;

[0136] The calculation formula of the mean square error is:

[0137] ;

[0138] in, is the actual power demand at time t, is the predicted value of electricity demand at time t, and K is the number of training samples.

[0139] S30. Establishing a power resource scheduling optimization model based on the power resource demand forecast value and equipment constraints to minimize the operating costs of the virtual power plant;

[0140] The equipment constraints include power generation equipment output limitation, energy storage system charging and discharging power, energy storage capacity limitation, and power demand balance constraint;

[0141] Further, refer to Figure 3 , the S30 includes:

[0142] S31. Establish an optimization model for power resource scheduling to minimize the operating costs of virtual power plants;

[0143] S32. Setting output limit of power generation equipment;

[0144] The output limit of the power generation equipment is expressed as:

[0145] ;

[0146] in, is the minimum output of the ith power generation equipment, is the output of the i-th generating equipment at the t-th moment, is the maximum output of the i-th power generation equipment;

[0147] S33. Setting the energy storage system charging and discharging power and energy storage capacity limits;

[0148] The energy storage system charging and discharging power and energy storage capacity limit are expressed as:

[0149] ;

[0150] ;

[0151] ;

[0152] in, is the maximum charging power of the energy storage system, is the maximum discharge power of the energy storage system, is the charging power of the energy storage system at time t, is the discharge power of the energy storage system at time t, is the energy storage capacity of the energy storage system at time t, is the maximum limit of energy storage capacity, The minimum limit for energy storage capacity;

[0153] S34. Setting power demand balancing constraints;

[0154] The power demand balance constraint is expressed as:

[0155] ;

[0156] in, is the predicted value of electricity demand at time t, is the output of the i-th power generation equipment at the t-th moment, and N is the total number of power generation equipment.

[0157] Furthermore, the optimization objective function of the power resource scheduling optimization model is expressed as:

[0158] ;

[0159] in, is the virtual power plant operating cost, is the power generation cost of the ith power generation equipment at the tth moment, is the charging and discharging cost of the energy storage system at time t, is the output of the i-th generating equipment at the t-th moment, is the energy storage capacity of the energy storage system at time t, T is the scheduling time range, N is the total number of power generation equipment, is the charging power of the energy storage system at time t, is the discharge power of the energy storage system at the tth moment.

[0160] S40. Solving the power resource scheduling optimization model by alternating direction multiplier method to obtain a scheduling strategy;

[0161] Further, refer to Figure 4 , the S40 comprises:

[0162] S41. Transform the power resource dispatch optimization model into an equivalent form with Lagrange multipliers, introduce a Lagrange multiplier for each constraint condition, and construct the Lagrange function;

[0163] The constraints include power generation equipment output limitation, energy storage system charging and discharging power, energy storage capacity limitation and power demand balance constraint;

[0164] S42. Initializing the Lagrangian multiplier, minimizing the Lagrangian function by optimizing the original variables;

[0165] The original variables include the energy storage capacity of the energy storage system at time t , the charging power of the energy storage system at time t , the discharge power of the energy storage system at time t ;

[0166] S43. updating the Lagrange multiplier according to the original variables and constraints;

[0167] The calculation formula for updating the Lagrange multiplier is:

[0168] ;

[0169] in, is the updated Lagrange multiplier in the k+1th iteration, A is the constraint matrix, is the Lagrange multiplier in the kth iteration, is the original variable updated in the k+1th iteration, is the penalty factor, is a constant vector;

[0170] S44. Determine whether the convergence condition is met, the convergence condition being that the objective function change is less than a set threshold or the maximum number of iterations is reached. If the convergence condition is met, stop the iteration; if not, return to S42;

[0171] S45. Decode the result and obtain the optimized scheduling strategy based on the optimal original variables and Lagrange multipliers.

[0172] S50 edge server for all scheduling tasks to prioritize, by the priority to adjust the scheduling strategy to obtain the optimal scheduling strategy;

[0173] Further, refer to Figure 5 , the S50 comprises:

[0174] S51 edge server creates a priority set, all tasks are stored in the priority set;

[0175] S52 edge server calculates the priority of all tasks;

[0176] S53 sorts all tasks in the priority set in descending order according to the priority level;

[0177] S54. Each time a power resource scheduling decision is made, the task ranked first in the priority set is scheduled first.

[0178] Furthermore, the calculation formula of the priority is:

[0179] ;

[0180] in, , , Represent the weight coefficients of importance, urgency and impact range respectively, is the importance of the i-th task, is the urgency of the ith task, is the influence range of the i-th task.

[0181] S60. The edge server uploads the optimal scheduling strategy to the cloud control center platform for execution to achieve accurate power resource scheduling.

[0182] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0183] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A virtual power plant resource scheduling optimization method based on edge computing, characterized in that: include: S10 edge server obtains the original data on the user terminal device, and pre-processes the original data; S20. Mining key features from the preprocessed data through feature extraction, constructing a demand forecasting model, and obtaining a power resource demand forecast value through the original data and key features; S30. Establishing a power resource scheduling optimization model based on the power resource demand forecast value and equipment constraints to minimize the operating costs of the virtual power plant; S40. Solving the power resource scheduling optimization model by alternating direction multiplier method to obtain a scheduling strategy; S50 edge server for all scheduling tasks to prioritize, by the priority to adjust the scheduling strategy to obtain the optimal scheduling strategy; S60. The edge server uploads the optimal scheduling strategy to the cloud control center platform for execution to achieve accurate power resource scheduling; The raw data includes user activity data and real-time data of power resources; The preprocessing includes cleaning and repairing outliers or missing values ​​in the data, and standardizing different types of data to make them have the same scale; The equipment constraints include power generation equipment output limitation, energy storage system charging and discharging power, energy storage capacity limitation, and power demand balance constraint; The key characteristics include load characteristics, power generation characteristics, and energy storage system characteristics; The load characteristics include peak load, average load, load fluctuation and load duration; The power generation characteristics include power generation, power generation efficiency, generator operation status and renewable energy power generation ratio; The energy storage system characteristics: including energy storage capacity, charge and discharge times, charge and discharge rates, and health of the energy storage system; The S20 includes: S21. Build demand forecasting model through LSTM neural network; The demand forecasting model includes an input layer, an LSTM unit layer, and an output layer; The LSTM unit layer includes a candidate unit state, a unit state, a hidden state, a forget gate, an input gate, and an output gate; S22. Taking the load characteristics, power generation characteristics, and energy storage system characteristics as the input feature vector at time t; The input feature vector at time t is expressed as: ; in, is the input feature vector at time t, is the load characteristic, For the power generation characteristics, It is the characteristics of energy storage system; S23. Calculate the activation values ​​of the forget gate, input gate, and output gate at time t; The calculation formula of the activation value of the forget gate, input gate, and output gate at time t is: ; ; ; in, is the activation value of the forget gate at time t, is the activation value of the input gate at time t, is the activation value of the output gate at time t, is the Sigmoid activation function, , , are the weight matrices of the forget gate, input gate, and output gate respectively. , , are the bias vectors of the forget gate, input gate, and output gate respectively, is the hidden state at time t-1; S24. Calculate the candidate unit state and unit state at time t; The candidate unit state at time t , Unit Status The calculation formula is: ; ; in, is the unit state at time t-1, is the candidate unit state at time t-1, is the weight matrix of the candidate unit state, is the bias vector of the candidate cell state, is the activation function; S25. Output the hidden state at time t; The hidden state at time t The calculation formula is: ; S26. Predict the power demand forecast value at time t; The power demand forecast value at time t The calculation formula is: ; in, is the output layer weight matrix, is the output layer bias vector; S27. Calculating the mean square error between the power demand forecast value and the actual power demand, and optimizing the demand forecasting model by minimizing the mean square error; The calculation formula of the mean square error is: ; in, is the actual power demand at time t, is the predicted value of electricity demand at time t, and K is the number of training samples.

2. A virtual power plant resource scheduling optimization method based on edge computing as claimed in claim 1, characterized in that: The S30 includes: S31. Establish an optimization model for power resource scheduling to minimize the operating costs of virtual power plants; S32. Setting output limit of power generation equipment; The output limit of the power generation equipment is expressed as: ; in, is the minimum output of the ith power generation equipment, is the output of the i-th generating equipment at the t-th moment, is the maximum output of the i-th power generation equipment; S33. Setting the energy storage system charging and discharging power and energy storage capacity limits; The energy storage system charging and discharging power and energy storage capacity limit are expressed as: ; ; ; in, is the maximum charging power of the energy storage system, is the maximum discharge power of the energy storage system, is the charging power of the energy storage system at time t, is the discharge power of the energy storage system at time t, is the energy storage capacity of the energy storage system at time t, is the maximum limit of energy storage capacity, The minimum limit for energy storage capacity; S34. Setting power demand balancing constraints; The power demand balance constraint is expressed as: ; in, is the predicted value of electricity demand at time t, is the output of the i-th power generation equipment at the t-th moment, and N is the total number of power generation equipment.

3. A virtual power plant resource scheduling optimization method based on edge computing as claimed in claim 2, characterized in that: The optimization objective function of the power resource scheduling optimization model is expressed as: ; in, is the virtual power plant operating cost, is the power generation cost of the ith power generation equipment at the tth moment, is the charging and discharging cost of the energy storage system at time t, is the output of the i-th generating equipment at the t-th moment, is the energy storage capacity of the energy storage system at time t, T is the scheduling time range, N is the total number of power generation equipment, is the charging power of the energy storage system at time t, is the discharge power of the energy storage system at the tth moment.

4. A virtual power plant resource scheduling optimization method based on edge computing as claimed in claim 1, characterized in that: The S40 comprises: S41. Transform the power resource dispatch optimization model into an equivalent form with Lagrange multipliers, introduce a Lagrange multiplier for each constraint condition, and construct the Lagrange function; The constraints include power generation equipment output limitation, energy storage system charging and discharging power, energy storage capacity limitation and power demand balance constraint; S42. Initializing the Lagrangian multiplier, minimizing the Lagrangian function by optimizing the original variables; The original variables include the energy storage capacity of the energy storage system at time t , the charging power of the energy storage system at time t , the discharge power of the energy storage system at time t ; S43. updating the Lagrange multiplier according to the original variables and constraints; The calculation formula for updating the Lagrange multiplier is: ; in, is the updated Lagrange multiplier in the k+1th iteration, A is the constraint matrix, is the Lagrange multiplier in the kth iteration, is the original variable updated in the k+1th iteration, is the penalty factor, is a constant vector; S44. Determine whether the convergence condition is met, the convergence condition being that the objective function change is less than a set threshold or the maximum number of iterations is reached. If the convergence condition is met, the iteration is stopped, if not, return to S42; S45. Decode the result and obtain the optimized scheduling strategy based on the optimal original variables and Lagrange multipliers.

5. The method for optimizing resource scheduling of a virtual power plant based on edge computing according to claim 1, characterized in that: The S50 comprises: S51 edge server creates a priority set, all tasks are stored in the priority set; S52 edge server calculates the priority of all tasks; S53 sorts all tasks in the priority set in descending order according to the priority level; S54. Each time a power resource scheduling decision is made, the task ranked first in the priority set is scheduled first.

6. A virtual power plant resource scheduling optimization method based on edge computing as claimed in claim 5, characterized in that: The priority calculation formula is: ; in, , , Represent the weight coefficients of importance, urgency and impact range respectively, is the importance of the i-th task, is the urgency of the ith task, is the influence range of the i-th task.