Dynamic collaborative optimization method for mobile virtual power plant based on large cross-energy fusion model

Abstract

The invention relates to a dynamic collaborative optimization method for a mobile virtual power plant based on a cross-energy fusion large model, which belongs to the technical field of power system optimization scheduling and comprises the following steps: S1, constructing the cross-energy fusion large model to integrate a dynamic coupling relationship between a traffic network and an energy network; s2, establishing a mobile virtual power plant resource aggregation model; s3, establishing a dynamic collaborative optimization model of the traffic-energy network; s4, adopting a distributed collaborative optimization algorithm to realize collaborative scheduling of the virtual power plant in the traffic-energy network; s5, verifying an optimization result in real time through a digital twin platform, and dynamically updating parameters of the cross-energy fusion large model based on feedback data to form dynamic collaborative optimization; the method has the beneficial effects that the energy exchange fusion large model is constructed, the dynamic collaborative optimization model of the traffic-energy network is established based on the energy exchange fusion large model, collaborative scheduling of the virtual power plant in the traffic-energy network is realized, and the resource collaborative scheduling efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system optimization and dispatching, and in particular relates to a dynamic collaborative optimization method for a mobile virtual power plant based on a large-scale energy fusion model. Background Art

[0002] In recent years, the global energy mix has been rapidly transitioning toward cleaner, low-carbon energy. Renewable energy generation capacity, represented by solar and wind power, has continued to climb. Simultaneously, the electrification of transportation is also accelerating, with electric vehicle ownership exploding. Mobile virtual power plants, as an innovative power system organization model, leverage advanced communication technologies and intelligent control to aggregate the dispersed resources of distributed power sources, controllable loads, and energy storage devices, achieving coordinated and optimized operation. These plants have demonstrated significant potential for enhancing grid flexibility and promoting the integration of renewable energy.

[0003] However, the current practical operation of mobile virtual power plants faces numerous challenges. Regarding the integration of energy and transportation systems, the two have long developed relatively independently, lacking effective information exchange mechanisms. For example, changes in electric vehicle charging demand in the transportation sector cannot be promptly fed back to the energy system, making it difficult for the energy system to rationally allocate resources based on the dynamic changes in the transportation system. This results in inefficient resource coordination and makes it difficult to achieve reliable energy supply and efficient utilization. Summary of the Invention

[0004] The present invention provides a dynamic collaborative optimization method for a mobile virtual power plant based on a large transportation-energy fusion model, which is used to solve the technical problem of low efficiency in resource collaborative scheduling. By constructing a large transportation-energy fusion model to integrate the dynamic coupling relationship between the transportation network and the energy network, a mobile virtual power plant resource aggregation model is established. Based on the large transportation-energy fusion model, a dynamic collaborative optimization model of the transportation-energy network is established through a dynamic optimization objective function of time-space coupling, thereby realizing the collaborative scheduling of virtual power plants in the transportation-energy network and improving the efficiency of resource collaborative scheduling.

[0005] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0006] The dynamic collaborative optimization method of mobile virtual power plant based on the large-scale transportation and energy fusion model includes the following steps:

[0007] Step S1: Construct a large-scale transportation and energy fusion model to integrate the dynamic coupling relationship between the transportation network and the energy network; wherein, the traffic vehicle status and electric energy consumption information are obtained in real time, and the traffic vehicle status and electric energy consumption information are dynamically coupled through multi-source heterogeneous data fusion processing;

[0008] Step S2: Establishing a mobile virtual power plant resource aggregation model; wherein, dimensional resources of traffic vehicle status and electric energy consumption information are aggregated to generate a dimensional resource characteristic matrix;

[0009] Step S3: Based on the large-scale transportation-energy fusion model, a dynamic collaborative optimization model of the transportation-energy network is established through dynamic optimization of the objective function with spatiotemporal coupling;

[0010] Step S4: using a distributed collaborative optimization algorithm to achieve collaborative scheduling of virtual power plants in the transportation-energy network;

[0011] Step S5: Verify the optimization results in real time through the digital twin platform, and dynamically update the parameters of the energy fusion model based on the feedback data to form dynamic collaborative optimization.

[0012] Optionally, in step S1, the traffic vehicle status and electric energy consumption information include: obtaining the vehicle's traffic status through the vehicle-mounted GPS module positioning technology, and displaying the vehicle's traffic status on the map; the vehicle's charging status is charging, charging completed, and charging interrupted; the vehicle's real-time energy consumption is the vehicle's power consumption per kilometer or per hour in the current driving state.

[0013] Optionally, multi-source heterogeneous data fusion processing can be performed using principal component analysis as follows:

[0014] With indivual Dimensional data points , centralize the data: , , calculate the covariance matrix ;

[0015] Pair covariance matrix Perform eigenvalue decomposition, ,in, is a matrix of eigenvectors, is a diagonal matrix of eigenvalues.

[0016] Optionally, in step S2, dimensional resources are determined: dimensional resources related to traffic vehicle status and electric energy consumption are clarified, data are collected, data are pre-processed, and a dimensional resource characteristic matrix is generated.

[0017] Optionally, the dimension resource property matrix is:

[0018] For two-dimensional resources and , the correlation coefficient between them The calculation formula is:

[0019] ;in, and They are and The mean of

[0020] If there is Dimensional resources , then the dimension resource characteristic matrix Elements for and The correlation coefficient of ,in, Indicates the The first dimension sample observations, Indicates the The first dimension sample observations.

[0021] Optionally, in step S3, the dynamic optimization objective function of the spatiotemporal coupling is:

[0022] ;

[0023] in, represents the time step, is the total time period, Is the transportation network in time state variables; Is the energy network in time The state variables, Is the transportation network in time The cost function depends on the state of the transportation network. ; Is the energy network in time The cost function depends on the state of the energy network ; It is a measure of the time A function of the degree of coupling, 、 and is the weight coefficient, which is used to balance the importance of transportation network cost, energy network cost and the coupling relationship between the two in the objective function.

[0024] Optionally, in step S4, the distributed collaborative optimization algorithm is as follows:

[0025] One with A system of distributed nodes, and a local objective function and local constraints ;

[0026] The distributed collaborative optimization problem can be formulated as:

[0027] ;

[0028] in, is the global objective function, which is obtained by adding the local objective functions of each node; is a node The local constraints of Represents global coupling constraints, which are used to describe the relationships between nodes.

[0029] Optionally, in step S5, during the entire dynamic collaborative optimization process, the operation status of the energy fusion and the digital twin platform are continuously monitored.

[0030] Beneficial effects of the present invention:

[0031] The present invention establishes a mobile virtual power plant resource aggregation model by constructing a large-scale transportation-energy fusion model to integrate the dynamic coupling relationship between the transportation network and the energy network. Based on the large-scale transportation-energy fusion model, a dynamic collaborative optimization model of the transportation-energy network is established through the dynamic optimization objective function of time-space coupling, which realizes the collaborative scheduling of virtual power plants in the transportation-energy network and improves the efficiency of resource collaborative scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0033] Figure 1 Schematic diagram of the workflow of the present invention. DETAILED DESCRIPTION

[0034] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0035] Example 1

[0036] Reference Figure 1 This embodiment provides a dynamic collaborative optimization method for a mobile virtual power plant based on a large-scale energy fusion model, comprising the following steps:

[0037] Step S1: Construct a large-scale transportation and energy fusion model to integrate the dynamic coupling relationship between the transportation network and the energy network; wherein, the traffic vehicle status and electric energy consumption information are obtained in real time, and the traffic vehicle status and electric energy consumption information are dynamically coupled through multi-source heterogeneous data fusion processing;

[0038] Step S2: Establishing a mobile virtual power plant resource aggregation model; wherein, dimensional resources of traffic vehicle status and electric energy consumption information are aggregated to generate a dimensional resource characteristic matrix;

[0039] Step S3: Based on the large-scale transportation-energy fusion model, a dynamic collaborative optimization model of the transportation-energy network is established through dynamic optimization of the objective function with spatiotemporal coupling;

[0040] Step S4: using a distributed collaborative optimization algorithm to achieve collaborative scheduling of virtual power plants in the transportation-energy network;

[0041] Step S5: Verify the optimization results in real time through the digital twin platform, and dynamically update the parameters of the energy fusion model based on the feedback data to form dynamic collaborative optimization.

[0042] Example 2

[0043] Based on Example 1, specifically, in step S1, the traffic vehicle status and electric energy consumption information include: obtaining the vehicle's traffic status through the vehicle-mounted GPS module positioning technology, and displaying the vehicle's traffic status on the map; the vehicle's charging status is charging, charging completed, and charging interrupted, so as to control the vehicle's charging progress; the vehicle's real-time energy consumption is the vehicle's power consumption per kilometer or per hour in the current driving state.

[0044] The following principal component analysis (PCA) is used for multi-source heterogeneous data fusion processing:

[0045] With indivual Dimensional data points , centralize the data, that is, , Indicates the Sample data, is the sample size, The mean is obtained by summing all sample data and dividing the sum by the number of samples. , for a set of data The specific approach of centralization is to convert each original data Subtract the sample mean , get new data , , calculate the covariance matrix ,in, yes The transpose of Is a matrix multiplication operation, for a dimensional sample vector , You will get one The matrix of .

[0046] Covariance matrix Perform eigenvalue decomposition, ,in, is a matrix of eigenvectors, is a diagonal matrix of eigenvalues, Represents a matrix composed of eigenvectors The transpose of .

[0047] Before selection The eigenvector corresponding to the largest eigenvalue , forming the projection matrix , then the original data Projection onto principal component space for: ,in, It is the eigenvector matrix obtained by principal component analysis (PCA). The eigenvectors are extracted from the eigendecomposition of the covariance matrix of the original data and are arranged in descending order according to the corresponding eigenvalues. Each column is a principal component direction, which means that the data has a large variance in the principal component direction and the degree of change in the principal component direction is large. yes The transposed matrix of .

[0048] For the data to be centralized, suppose there is samples, each sample has features, forming a dataset ,in, , , standardize the data and change the mean of each feature to 0 and the variance to 1. The standardization formula is: ,in, is the standardized data value. After calculation by this formula, the original data is converted to a specific scale. is the original data value, Indicates the samples, Indicates the variables, It is The mean of a variable reflects the average level of all sample values of this variable. The calculation of , gets the deviation between the original data value and the mean, It is The standard deviation of a variable.

[0049] The covariance matrix is calculated as:

[0050] Data after data processing Recorded as , calculate its covariance matrix , the elements of the covariance matrix The calculation formula is:

[0051] ;

[0052] in, , ,so ,

[0053] is the first in the covariance matrix Rank The elements of the column. It measures the random variable and The degree of linear correlation between them.

[0054] For the In the sample The value of a variable can be a certain dimension feature in multidimensional data; For the The mean of a variable is the value of the The average value of the variable values is calculated as , For the In the sample The deviation of a variable value from its mean reflects the The degree of deviation from the average level on a dimensional characteristic.

[0055] It is the product of the deviations of two variables. When the changing trends of the two variables are consistent (that is, when one variable is greater than the mean, the other is also greater than the mean, or when one is less than the mean, the other is also less than the mean), the product is positive; when the changing trends are opposite, the product is negative.

[0056] It is the sum of the deviation products of all samples, reflecting the overall co-variation of the two variables in all samples.

[0057] Normalize the summation result.

[0058] Covariance matrix Perform eigenvalue decomposition, that is ,in, is an orthogonal matrix composed of eigenvectors, is a diagonal matrix consisting of eigenvalues, and .

[0059] The principal components are determined according to the size of the eigenvalues, usually the first The eigenvector corresponding to the largest eigenvalue As the principal component, The choice can be determined based on the cumulative contribution rate, which is calculated as follows: Generally, when the cumulative contribution rate reaches 80%-90%, it can be considered that the selected The principal components can better represent the information of the original data.

[0060] The data Projected onto the selected principal component, the principal component score matrix is obtained For the sample , Projection onto principal component space for:

[0061] ,in, ;

[0062] Through the above steps, the original multi-source heterogeneous data can be fused into the principal component space to achieve data dimensionality reduction and feature extraction, remove noise and redundant information in the data, and retain the main features and information of the data, namely the vehicle's traffic status, vehicle charging status and vehicle's real-time energy consumption.

[0063] Example 3

[0064] Based on Example 1-Example 2, in step S2, the dimension resources are determined:

[0065] 1. Identify the dimensional resources related to traffic vehicle status and electric energy consumption; that is, vehicle traffic status dimensions: speed, acceleration, mileage, driving time, vehicle position, driving mode (such as: economy mode, sports mode); vehicle charging status: charging, charging completed, and charging interrupted; electric energy consumption dimensions: battery level, charging status, power consumption rate (power consumption per unit mileage or unit time), remaining range, and charging time.

[0066] 2. Data collection: Data collected from various sources (e.g., vehicle sensors, charging station records), ensuring data accuracy and completeness;

[0067] 3. Data preprocessing: Perform preprocessing operations such as cleaning, conversion, and normalization on the collected data to facilitate subsequent analysis;

[0068] 4. Generate a dimension resource feature matrix, take each dimension resource as the row and column of the matrix, and calculate the correlation between them.

[0069] The dimension resource characteristic matrix based on correlation analysis is:

[0070] For two-dimensional resources and , the correlation coefficient between them The calculation formula is:

[0071] ;in, and They are and The mean of , .

[0072] If there is Dimensional resources , then the dimension resource characteristic matrix Elements for and The correlation coefficient of and They are variables and No. observations, by calculating each observation point Value and The difference in means, multiplied by Value and The difference in means, and then summing it, reflects and The degree of co-variation between two variables. Greater than the mean and Also greater than the mean ,or Less than the mean and Also less than the mean hour, is positive; when and When one is greater than the mean and the other is less than the mean, is negative, that is ,in, Indicates the The first dimension (resource characteristics) sample observations, Indicates the The first dimension (resource characteristics) sample observations.

[0073] It is The mean of all sample observations under dimensions, that is, , It is The mean of all sample observations under dimensions, that is, , the mean reflects the average level of sample observations under this dimension.

[0074] There are three dimensional resources, the speed of the vehicle , the battery capacity of the vehicle after charging is completed , the vehicle's power consumption rate Taking the correlation coefficient between calculation speed and battery power as an example, , , ;

[0075] ;

[0076] ;

[0077] but, Similarly, the correlation coefficients between speed and mileage, battery power and mileage can be calculated, thus obtaining the dimensional resource characteristic matrix .

[0078] matrix Elements in Indicates the Dimensional resources and The correlation coefficient between the resources in each dimension. For example, Indicates the degree of correlation between the first dimension resource and the second dimension resource.

[0079] when hour, , because any dimension resource is completely related to itself.

[0080] Positive correlation: If , explain the Dimensional resources and The three dimensions of resources are positively correlated, meaning that when the value of one dimension increases, the value of another dimension also tends to increase. For example, a vehicle's mileage and energy consumption are generally positively correlated: the longer the mileage, the greater the energy consumption, and the correlation coefficient between them is positive.

[0081] Negative correlation: If , indicating a negative correlation between two dimensional resources. This means that as the value of one dimensional resource increases, the value of the other dimensional resource tends to decrease. For example, in some cases, a vehicle's driving speed and remaining range may be negatively correlated. The faster the speed, the faster the remaining range may decrease, resulting in a negative correlation coefficient.

[0082] Not relevant: When When Dimensional resources and There is no linear correlation between the resources in each dimension.

[0083] Dimensional Resource Characteristics Matrix It can fully and intuitively understand the relationship between multiple dimensional resources. By observing the elements in the matrix, it can quickly determine whether there is a strong correlation between dimensional resources and a weak correlation between dimensional resources.

[0084] Example 4

[0085] Based on Example 1 to Example 3, in step S3, the dynamic optimization objective function of time-space coupling is:

[0086] ;

[0087] in, represents the time step, is the total time period, The transportation network in time state variables; Is the energy network in time The state variables, The transportation network in time The cost function depends on the state of the transportation network. , including operating costs (such as vehicle fuel consumption) and congestion costs (related to the degree of traffic congestion, such as time delays caused by congestion); Is the energy network in time The cost function depends on the state of the energy network , including the cost of energy production (e.g., generation, generation method, and amount of power generated), losses in energy transmission and distribution, and the cost of energy storage (investment in storage equipment, maintenance, and energy losses); It is a measure of the time The coupling degree function reflects the coupling relationship between the transportation system and the energy system over time. The costs or benefits brought about by the transportation system's energy consumption will affect the demand for the energy system, thus generating certain costs; the supply capacity and price of the energy system will also affect the operation of the transportation system. This part of the cost function is usually a relatively complex function and needs to be determined according to the specific coupling mechanism; 、 and is the weight coefficient, which is used to balance the importance of transportation network cost, energy network cost and the coupling relationship between them in the objective function;

[0088] Throughout the time period By optimizing the state variables of the transportation network and energy network and , minimizing the sum of transportation network cost, energy network cost and the coupling cost between the two.

[0089] The spatiotemporal coupled dynamic optimization objective function helps to achieve a rational allocation of transportation and energy resources in time and space. By comprehensively considering the needs of the transportation system (such as travel demand in different time periods and regions) and the supply capacity of the energy system (such as the spatiotemporal distribution of energy production and the limitations of energy storage), the optimal resource allocation plan can be determined to avoid waste or excessive concentration of resources.

[0090] Example 5

[0091] Based on Examples 1 to 4, in step S4, the distributed collaborative optimization algorithm is as follows:

[0092] One with A system of distributed nodes, and a local objective function and local constraints ;

[0093] The distributed collaborative optimization problem can be formulated as:

[0094] ;

[0095] Among them, each It's about variables function, is the global objective function, which is obtained by adding the local objective functions of each node; is a node The local constraints of Represents global coupling constraints, which are used to describe the relationships between nodes.

[0096] For each variable Both Inequality constraint function , the values of these constraint functions must be less than or equal to 0.

[0097] Indicates that there is Equality constraint function , these functions act on all variables , and the function value is equal to 0.

[0098] In practical applications, the alternating direction multiplier method (ADMM) is used: the basic iterative formula is as follows:

[0099] For the above optimization problem, auxiliary variables are introduced , transforming the original problem into an equivalent augmented Lagrangian function form:

[0100] ;

[0101] in, is the Lagrange multiplier, is the penalty parameter.

[0102] The iterative steps of ADMM are as follows:

[0103] renew : ;

[0104] renew : ;

[0105] renew : ;

[0106] Through continuous iteration, it eventually converges to the optimal solution to the original problem.

[0107] Variables for the optimization problem Under the premise of satisfying the constraints, solve the objective function achieve The minimum values in .

[0108] Constraints define the range of variables and their relationships, limiting the solution to a specific feasible region. For example, in resource allocation, there is a limit on the total amount of resources. Constraints can limit the amount of resources allocated to each component to no more than the total amount, ensuring that the optimization is in line with the actual situation.

[0109] Example 6

[0110] Based on Examples 1 to 5, in step S5, the operation status of the energy fusion and the digital twin platform are continuously monitored during the entire dynamic collaborative optimization process.

[0111] During the entire monitoring and optimization process, all relevant data and operation information are recorded in detail, including collected data, analysis results, early warning information, optimization strategies and implementation effects, forming a complete monitoring log and optimization record.

[0112] Regularly evaluate the operation of the energy-transportation fusion system and digital twin platform, comparing various indicators before and after optimization to assess the effectiveness of optimization measures. Based on the evaluation results, summarize the lessons learned to provide reference and improvement basis for subsequent dynamic collaborative optimization, and continuously improve the monitoring and optimization process.

[0113] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope of the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A dynamic collaborative optimization method for mobile virtual power plants based on a large-scale transportation and energy fusion model, characterized by: The steps include: Step S1: Construct a large-scale transportation and energy fusion model to integrate the dynamic coupling relationship between the transportation network and the energy network; wherein, the traffic vehicle status and electric energy consumption information are obtained in real time, and the traffic vehicle status and electric energy consumption information are dynamically coupled through multi-source heterogeneous data fusion processing; Step S2: Establishing a mobile virtual power plant resource aggregation model; wherein, dimensional resources of traffic vehicle status and electric energy consumption information are aggregated to generate a dimensional resource characteristic matrix; Step S3: Based on the large-scale transportation-energy fusion model, a dynamic collaborative optimization model of the transportation-energy network is established through dynamic optimization of the objective function with spatiotemporal coupling; Step S4: using a distributed collaborative optimization algorithm to achieve collaborative scheduling of virtual power plants in the transportation-energy network; Step S5: Verify the optimization results in real time through the digital twin platform, and dynamically update the parameters of the energy fusion model based on the feedback data to form dynamic collaborative optimization.

2. The method for dynamic collaborative optimization of mobile virtual power plants based on a large-scale transportation and energy fusion model according to claim 1 is characterized in that: In step S1, the traffic vehicle status and electric energy consumption information include: obtaining the vehicle's traffic status through the vehicle-mounted GPS module positioning technology and displaying the vehicle's traffic status on the map; the vehicle's charging status is charging, charging completed, and charging interrupted; the vehicle's real-time energy consumption is the vehicle's power consumption per kilometer or per hour in the current driving state.

3. The method for dynamic collaborative optimization of mobile virtual power plants based on a large-scale transportation and energy fusion model according to claim 2 is characterized in that: The multi-source heterogeneous data fusion process adopts the following principal component analysis: With indivual Dimensional data points , centralize the data: , , calculate the covariance matrix ; Pair covariance matrix Perform eigenvalue decomposition, ,in, is a matrix of eigenvectors, is a diagonal matrix of eigenvalues.

4. The method for dynamic collaborative optimization of mobile virtual power plants based on a large-scale transportation and energy fusion model according to claim 1 is characterized in that: In the step S2, the dimensional resources are determined: the dimensional resources related to the traffic vehicle status and the electric energy consumption are clarified, data is collected, data is pre-processed, and a dimensional resource characteristic matrix is generated.

5. The method for dynamic collaborative optimization of mobile virtual power plants based on a large-scale transportation and energy fusion model according to claim 4 is characterized in that: The dimension resource characteristic matrix is: For two-dimensional resources and , the correlation coefficient between them The calculation formula is: ;in, and They are and The mean of If there is Dimensional resources , then the dimension resource characteristic matrix Elements for and The correlation coefficient of ,in, Indicates the The first dimension sample observations, Indicates the The first dimension sample observations.

6. The method for dynamic collaborative optimization of mobile virtual power plants based on a large-scale transportation and energy fusion model according to claim 1 is characterized in that: In step S3, the dynamic optimization objective function of the spatiotemporal coupling is: ； in, represents the time step, is the total time period, The transportation network in time state variables; Is the energy network in time The state variables, Is the transportation network in time The cost function depends on the state of the transportation network. ; Is the energy network in time The cost function depends on the state of the energy network ; It is a measure of the time A function of the degree of coupling, 、 and is the weight coefficient, which is used to balance the importance of transportation network cost, energy network cost and the coupling relationship between the two in the objective function.

7. The method for dynamic collaborative optimization of mobile virtual power plants based on a large-scale transportation and energy fusion model according to claim 1 is characterized in that: In step S4, the distributed collaborative optimization algorithm is as follows: One with A system of distributed nodes, and a local objective function and local constraints ; The distributed collaborative optimization problem can be formulated as: ； in, is the global objective function, which is obtained by adding the local objective functions of each node; is a node The local constraints of Represents global coupling constraints, which are used to describe the relationships between nodes.

8. The method for dynamic collaborative optimization of mobile virtual power plants based on a large-scale transportation and energy fusion model according to claim 1 is characterized in that: In step S5, during the entire dynamic collaborative optimization process, the operation status of the energy fusion and the digital twin platform are continuously monitored.

Images