A mobile charging vehicle optimization scheduling method and system

Abstract

The present invention provides an optimization scheduling method and system for mobile charging vehicles. The optimization scheduling method utilizes multi-field and multi-dimensional measurement data to perform multi-dimensional dynamic perception and prediction of traffic, charging load, and power scheduling requirements. At the same time, the optimization scheduling method of the present invention can realize the surplus utilization of mobile charging vehicles by incorporating mobile charging vehicles into the scope of objects of power scheduling, and expand the profit channels of mobile charging vehicles into two dimensions: charging services and power scheduling. The optimization scheduling method includes two stages including two target dimensions: congestion management and control requirements in the power system and electric vehicle charging requirements in the traffic system. Based on the perception and prediction information, multi-objective optimization scheduling and efficient utilization of mobile charging vehicles can be realized.

Description

Technical Field

[0001] The present invention belongs to the field of energy internet, and in particular relates to an optimization scheduling method and system for a mobile charging vehicle. Background Art

[0002] At present, in local power areas such as distribution networks, the demand for electric vehicle charging fluctuates greatly, and mobile charging vehicles face the contradiction between deployment capacity and utilization efficiency; at the same time, the power system is also facing more and more frequent congestion problems, requiring some kind of energy transmission method that can cross traditional power lines. Mobile charging vehicles are a feasible way to mediate the rapid growth of electric vehicles and the insufficient construction of charging piles. Traditional mobile charging vehicle planning and scheduling are oriented to the needs of electric vehicles, and usually only consider mobile charging vehicles to participate in the charging needs of electric vehicles, and the planning and scheduling are single. In addition, the planning and scheduling of traditional mobile charging vehicles lack multi-dimensional accurate data decision support, and it is difficult to cope with the multi-objective operating environment decision needs. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the present invention proposes an optimization scheduling method for a mobile charging vehicle and an optimization scheduling system for a mobile charging vehicle. The technical solution of the present invention is as follows:

[0004] A method for optimizing the scheduling of mobile charging vehicles comprises the following steps:

[0005] Obtain urban planning data and historical multi-source measurement data, create a weighted graph model of the power-traffic dual-flow fusion graph based on the urban planning data, and then pre-process the historical multi-source measurement data;

[0006] Based on the urban planning data and the pre-processed historical multi-source measurement data, a day-ahead prediction model and an intra-day prediction model for charging pile load, OD traffic flow, and power node load are established and trained;

[0007] An optimization scheduling model is constructed with the dispatching goal of maximizing the benefits of mobile charging vehicles in solving power congestion and providing charging services within a day, and then the optimization scheduling model is divided into a two-stage solution model including a day-ahead planning model and an intraday scheduling model; wherein the decision variables corresponding to the day-ahead planning model include the driving route of the mobile charging vehicle; the decision variables corresponding to the intraday scheduling model include the charging and discharging power provided by the mobile charging vehicle to the power grid, the charging and discharging power provided to the electric vehicle, and the driving section of the mobile charging vehicle;

[0008] 4) Acquire real-time multi-source measurement data and perform preprocessing, and alternately iterate and solve the two-stage solution model based on the preprocessed real-time multi-source measurement data and historical multi-source measurement data, and output the solution of the decision variables corresponding to the intraday scheduling model as the optimal scheduling plan for the mobile charging vehicle.

[0009] Preferably, the urban planning data includes the power system grid topology, the transportation network main road topology, and the parking lot node location; the multi-source measurement data includes power load data, traffic travel data, fixed charging pile charging load data, holiday data and meteorological data; among them, the power load data includes two types of data: the power generation and power consumption of each power node; the traffic travel data includes two types of data: the injected traffic flow and the output traffic flow of the traffic node; the fixed charging pile charging load data is the power consumption data of the charging pile; the meteorological data includes four types of data: temperature, humidity, wind speed and light.

[0010] Preferably, the weighted graph model for creating the power-traffic dual-flow fusion graph includes the following specific steps:

[0011] Based on the power system grid topology, traffic network trunk road topology, and parking lot node locations in urban planning data, a weighted graph model G of the power-traffic dual-flow fusion graph is created. Its mathematical model is:

[0012] G = {V, E}

[0013] Among them, V is a node, including the parking lot node as a node in the power-traffic dual-flow fusion diagram, the traffic road node as a road node in the main road topology of the traffic network excluding the traffic road node with vehicles entering and exiting the parking lot, and the power node as a power node in the power system grid topology excluding the power consumption and input connected to the parking lot; E is the connection relationship between nodes, which is a road in the traffic flow and a power line in the power flow.

[0014] Preferably, the pretreatment includes the following specific steps:

[0015] First, the multi-source measurement data is resampled: the multi-source measurement data is used as the label information of the node, and the original multi-source measurement data is resampled according to the time series of the unified step size to achieve hourly time scale alignment;

[0016] The resampled multi-source measurement data are then normalized: meteorological data, traffic data, power load data, and fixed charging pile charging load data are normalized according to the maximum value of each type of data; holiday data are represented by a binary variable {0, 1}, where 1 indicates that the sample belongs to a holiday and 0 indicates that it belongs to a working day.

[0017] Preferably, the time scale of the day-ahead prediction model for the electric pile load, OD traffic flow, and power node load is one day; the time scale of the intraday prediction model for the electric pile load, OD traffic flow, and power node load is one hour.

[0018] The establishment and training of the day-ahead prediction model for charging pile load, OD traffic flow, and power node load includes the following specific steps:

[0019] Select the preprocessed historical multi-source measurement data, average them within a day according to the length of a day, and obtain the multi-source measurement data with a day as the time scale. The time scale sequence is recorded as d = {1, 2, 3..., N}; N is the total number of days of the historical multi-source measurement data;

[0020] A graph neural network training model is established based on the graph neural network to build a day-ahead prediction model for charging pile load, OD traffic flow, and power node load, where the vertices and edges of the graph neural network training model are the same as V and E of the power-traffic dual-flow fusion graph;

[0021] The multi-source measurement data with a daily time scale is divided into input variables and output variables, and the input variables and output variables are paired with d according to the time scale {dn, d-n+1…, d-1}, where the output variable is the charging load of the charging pile on the dth day, and the rest of the data are all used as input variables; after being divided into a training group and a test group, they are input into the graph neural network training model for training to obtain a day-ahead prediction model for the charging pile load; n is a date bias hyperparameter set by the user, indicating that the day-ahead value data of the charging pile load, OD traffic flow, and power node load on a certain day are predicted based on the measurement data of n days;

[0022] The output variables are replaced by the OD traffic flow on day d and the power node load on day d, respectively, and the day-ahead prediction model of the OD traffic flow and the day-ahead prediction model of the power node load are obtained;

[0023] The establishment and training of the intraday prediction model for charging pile load, OD traffic flow, and power node load includes the following specific steps:

[0024] Select the preprocessed multi-source measurement data and record its time scale as; (t, d) = {(1, 1), (2, 1), (3, 1), ... (M, N)}, where t is the hourly time scale and d is the corresponding daily time scale, and obtain the multi-source measurement data with hourly time scale; M is the maximum value of the daily measurement data;

[0025] Taking multi-source measurement data with a day as the time scale, the time scale of the multi-source measurement data is {dn, d-n+1…, d-1}, inputting it into the day-ahead prediction model of the charging pile load, OD traffic flow, and power node load, and obtaining the day-ahead value prediction information of the charging pile load, OD traffic flow, and power node load on the dth day;

[0026] The multi-source measurement data with hourly time scale is divided into input variables and output variables, and the input variables and output variables are matched according to the time scale t, where the time scale of the input variables is {t-1, t-2, ... tm}; In addition, the dimension of each time scale input variable is increased, which is the day-ahead value prediction information corresponding to t d, and the time scale of the output variable is t, where the output variables are charging pile load, OD traffic flow, and power node load respectively; m is the hourly bias hyperparameter set by the user, indicating that the intraday curve data of the charging pile load, OD traffic flow, and power node load at a certain time is predicted based on the measurement data at time m;

[0027] A graph neural network training model is established based on the graph neural network to establish a daily prediction model for charging pile load, OD traffic flow, and power node load, where the vertices and edges of the graph neural network training model are the same as the nodes V and the connection relationship E between nodes in the power-traffic dual-flow fusion graph;

[0028] The input variables and output variables are input into the graph neural network for training, and the intraday prediction models of charging pile load, OD traffic flow, and power node load are obtained respectively.

[0029] Preferably, the dispatching target of the mobile charging vehicle is

[0030]

[0031] The objectives of the day-ahead planning model are:

[0032]

[0033] The objectives of the intraday scheduling model are:

[0034]

[0035] In the formula, f is the total revenue of the mobile charging vehicle within a day, cx i,t 、cy i,t are the benefits obtained by the i-th mobile charging vehicle from solving power congestion and providing charging services in the t scheduling period, cz i,t is the dispatching cost of the i-th mobile charging vehicle in the t-th dispatching period, I is the set of mobile charging vehicles, T is the total dispatching period, are the charging and discharging power provided by the i-th mobile charging vehicle to the power grid and the electric vehicle in the t-th scheduling period, is the driving section of the i-th mobile charging vehicle in the scheduling period t.

[0036] Preferably, step 4) comprises the following specific steps:

[0037] Start prediction: set the current scheduling time as t=1, record the current scheduling time and the current scheduling day as (t, d), preprocess the historical multi-source measurement data n days before the current scheduling day d, and obtain the preprocessed data;

[0038] The pre-processed data is averaged within a day according to the length of a day; the data within n days before the current scheduling day d is taken and input into the charging pile load, OD traffic flow, and power node load day-ahead value prediction model to obtain the charging pile load, OD traffic flow, and power node load day-ahead value prediction information for the current scheduling day d;

[0039] Obtain preprocessed data with hourly time scale; take data within m time before the current scheduling time t, input it into the intraday prediction model of charging pile load, OD traffic flow, and power node load, and obtain the intraday curve prediction information of charging pile load, OD traffic flow, and power node load on the current scheduling day d;

[0040] Alternating iterative solution:

[0041] The initial revenue of the mobile charging vehicle in solving power congestion and providing charging services is set, and the day-ahead planning model in the two-stage solution model is solved based on the day-ahead value prediction information to obtain the driving route of the mobile charging vehicle. Today’s planning scheme;

[0042] The day-ahead planning scheme is input into the intraday scheduling model in the two-stage solution model to solve the day-ahead planning scheme, and the optimal value of the decision variable of the corresponding intraday planning scheme is obtained. and At the same time, the corresponding mobile charging vehicle can solve the power congestion and provide charging services.

[0043] If the revised revenue is equal to the set initial revenue, then the charging and discharging power provided by each mobile charging vehicle to the grid in each dispatch period is output. Charging and discharging power for electric vehicles and driving route And as the charging and discharging plan of each mobile charging vehicle to the power grid, the charging and discharging plan for providing electric vehicle charging services and the driving route, the intraday dispatch plan of the mobile charging vehicle at time t is obtained, replacing the original day-ahead planning plan; otherwise, the revised income is used as the initial income, and the revised income is recalculated according to the initial income;

[0044] The current time reaches time t+1: if t=max(T), the current scheduling day ends, and the intraday scheduling plan of the mobile charging vehicle on the current scheduling day is obtained.

[0045] Preferably, the urban planning data, historical multi-source measurement data and real-time multi-source measurement data are all retrieved through the energy integrated service station.

[0046] The data acquisition module is used to acquire urban planning data and historical multi-source measurement data; wherein the urban planning data includes the power system grid topology, the traffic network trunk road topology, and the parking lot node location; the multi-source measurement data includes power load data, traffic travel data, fixed charging pile charging load data, holiday data, and meteorological data; wherein the power load data includes two types of data: the power generation and power consumption of each power node; the traffic travel data includes two types of data: the injected traffic flow and the output traffic flow of the traffic node; the fixed charging pile charging load data is the power consumption data of the charging pile; the meteorological data includes four types of data: temperature, humidity, wind speed, and light;

[0047] The first model building module is used to create a weighted graph model of the power-traffic dual-flow fusion graph based on urban planning data, and then pre-process the historical multi-source measurement data;

[0048] A second model building module is used to build and train a day-ahead prediction model and an intraday prediction model for charging pile load, OD traffic flow, and power node load based on the urban planning data and pre-processed historical multi-source measurement data;

[0049] The third model building module is used to build an optimization scheduling model with the dispatching goal of maximizing the benefits of mobile charging vehicles solving power congestion and providing charging services within a day, and then divide the optimization scheduling model into a two-stage solution model including a day-ahead planning model and an intraday scheduling model;

[0050] The solution module is used to obtain and preprocess the real-time multi-source measurement data, and based on the preprocessed real-time multi-source measurement data and the historical multi-source measurement data, alternately and iteratively solve the two-stage solution model, and output the intraday scheduling model as the optimized scheduling plan for the mobile charging vehicle.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] The present invention proposes an optimization scheduling method and system for mobile charging vehicles. The optimization scheduling method utilizes multi-field and multi-dimensional measurement data to perform multi-dimensional dynamic perception and prediction of traffic, charging load, and power scheduling requirements. At the same time, the optimization scheduling method of the present invention can realize the surplus utilization of mobile charging vehicles by incorporating mobile charging vehicles into the scope of objects of power scheduling, and expand the profit path of mobile charging vehicles into two dimensions of charging services and power scheduling. The optimization scheduling method includes two stages including two target dimensions of congestion management and control requirements in the power system and electric vehicle charging requirements in the traffic system. Based on the perception and prediction information, multi-objective optimization scheduling and efficient utilization of mobile charging vehicles can be realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A flow chart of the scheduling optimization method of the present invention;

[0054] Figure 2 4 is a diagram of the data format of multi-source measurement data fusion in an embodiment. Specific implementation plan

[0055] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0056] Embodiment 1:

[0057] This example provides an optimization scheduling method for mobile charging vehicles. The method uses multi-domain and multi-dimensional measurement data provided by the energy integrated service station to perform multi-dimensional dynamic perception and prediction of traffic, charging load, and power scheduling needs. At the same time, the scheduling method includes two stages including the congestion management and control needs in the power system and the electric vehicle charging needs in the traffic system. Based on the perception and prediction information, the multi-objective optimization scheduling of mobile charging vehicles can be achieved, such as Figure 1 and Figure 2 As shown, the optimization scheduling method includes the following steps:

[0058] 1) Obtain urban planning data and historical multi-source measurement data, create a weighted graph model of the power-traffic dual-flow fusion graph based on the urban planning data, and then pre-process the historical multi-source measurement data; among them, urban planning data and historical multi-source measurement data are retrieved through the energy comprehensive service station. Urban planning data includes power system grid topology, traffic network trunk road topology, and parking lot node location; multi-source measurement data includes power load data, traffic travel data, fixed charging pile charging load data, holiday data, and meteorological data; power load data includes two types of data: power generation and power consumption of each power node; traffic travel data includes two types of data: injected traffic flow and output traffic flow of traffic nodes; fixed charging pile charging load data is the power consumption data of charging piles; meteorological data includes four types of data: temperature, humidity, wind speed, and light.

[0059] The weighted graph model for creating the power-traffic dual-flow fusion graph in this step includes the following specific steps:

[0060] Based on the power system grid topology, traffic network trunk road topology, and parking lot node locations in urban planning data, a weighted graph model G of the power-traffic dual-flow fusion graph is created. Its mathematical model is:

[0061] G = {V, E}

[0062] Among them, V is a node, including the parking lot node as a node in the power-traffic dual-flow fusion diagram, the traffic road node as a road node in the main road topology of the traffic network excluding the traffic road node with vehicles entering and exiting the parking lot, and the power node as a power node in the power system grid topology excluding the power consumption and input connected to the parking lot; E is the connection relationship between nodes, which is a road in the traffic flow and a power line in the power flow.

[0063] 2) Based on urban planning data and pre-processed historical multi-source measurement data, establish and train the day-ahead prediction model and intraday scheduling model of charging pile load, OD traffic flow, and power node load; the time scale of the day-ahead prediction model of charging pile load, OD traffic flow, and power node load is one day; establishing and training the day-ahead prediction model of charging pile load, OD traffic flow, and power node load includes the following specific steps:

[0064] Step 2-1, select the pre-processed historical multi-source measurement data, average them within a day according to the length of a day, and obtain multi-source measurement data with a day as the time scale, and record its time scale sequence as d = {1, 2, 3..., N}; N is the total number of days of the historical multi-source measurement data;

[0065] Step 2-2, a graph neural network training model of the day-ahead prediction model of charging pile load, OD traffic flow, and power node load is established based on the graph neural network, wherein the vertices and edges of the graph neural network training model are the same as V and E of the power-traffic dual-flow fusion graph;

[0066] Step 2-3, divide the multi-source measurement data with a daily time scale into input variables and output variables, and pair the input variables and output variables according to the time scale {dn, d-n+1…, d-1} with d. In this example, n can be taken as a month length of 30, where the output variable is the charging load of the charging pile on the dth day, and the rest of the data are all used as input variables; after being divided into a training group and a test group, they are input into the graph neural network training model for training to obtain a day-ahead prediction model for the charging pile load; n is a date bias hyperparameter set by the user, indicating that the day-ahead value data of the charging pile load, OD traffic flow, and power node load on a certain day are predicted based on the measurement data of n days;

[0067] In step 2-4, the output variables are replaced by the OD traffic flow and the power node load within d days, respectively. According to the method in step 2-3, the day-ahead prediction model of OD traffic flow and the day-ahead prediction model of power node load are obtained. Among them, the time scale of the intraday prediction model of charging pile load, OD traffic flow, and power node load is one hour.

[0068] Among them, establishing and training the intraday prediction model of charging pile load, OD traffic flow, and power node load includes the following specific steps:

[0069] Step 3-1, select the preprocessed multi-source measurement data, record its time scale as; (t, d) = {(1, 1), (2, 1), (3, 1), ... (M, N)}, M is a day time 24, where t is the hourly time scale, d is the corresponding daily time scale, and obtain the multi-source measurement data with hourly time scale; M is the maximum time of the daily measurement data;

[0070] Step 3-2, taking the multi-source measurement data with a daily time scale of {dn, d-n+1…, d-1} after processing in step 2-1, inputting it into the day-ahead prediction model of the charging pile load, OD traffic flow, and power node load obtained in step 2, and obtaining the day-ahead value prediction information of the charging pile load, OD traffic flow, and power node load on the dth day;

[0071] Step 3-3, divide the multi-source measurement data with hourly time scale into input variables and output variables, and correspond the input variables and output variables according to the time scale t, where the time scale of the input variables is {t-1, t-2, ... tm}; set m = 24; in addition, the dimension of each time scale input variable is increased, which is the day-ahead value prediction information corresponding to t d, and the time scale of the output variable is t, where the output variables are respectively charging pile load, OD traffic flow, and power node load; m is the hourly bias hyperparameter set by the user, indicating that the intraday curve data of the charging pile load, OD traffic flow, and power node load at a certain time is predicted based on the measurement data at time m;

[0072] Step 3-4, a graph neural network training model of the intraday prediction model of charging pile load, OD traffic flow, and power node load is established based on the graph neural network, wherein the vertices and edges of the graph neural network training model are the same as the nodes V and the connection relationship E between nodes of the power-traffic dual flow fusion graph;

[0073] In step 3-5, the input variables and output variables are input into the graph neural network for training, and the intraday prediction models of charging pile load, OD traffic flow, and power node load are obtained respectively.

[0074] 3) An optimization scheduling model is constructed with the dispatching goal of maximizing the benefits of mobile charging vehicles in solving power congestion and providing charging services within a day. Then, based on the Bender decoupling method, the optimization scheduling model is divided into a two-stage solution model (two-stage solution model) including a day-ahead planning model and an intraday scheduling model;

[0075] The dispatching target of the mobile charging vehicle is

[0076]

[0077] The objectives of the day-ahead planning model are:

[0078]

[0079] The objectives of the intraday scheduling model are:

[0080]

[0081] In the formula, f is the total revenue of the mobile charging vehicle within a day, cx i,t 、cy i,t are the benefits obtained by the i-th mobile charging vehicle from solving power congestion and providing charging services in the t scheduling period, cz i,t is the dispatching cost of the i-th mobile charging vehicle in the t-th dispatching period, I is the set of mobile charging vehicles, T is the total dispatching period, are the charging and discharging power provided by the i-th mobile charging vehicle to the power grid and the electric vehicle in the t-th scheduling period, is the driving section of the i-th mobile charging vehicle in the scheduling period t;

[0082] 4) Obtain real-time multi-source measurement data, and alternately iterate to solve the two-stage solution model based on the real-time multi-source measurement data and historical multi-source measurement data, and output the intraday scheduling plan for mobile charging vehicles, where the real-time multi-source measurement data is retrieved through the energy comprehensive service station. This step includes the following specific steps:

[0083] Step 4-1, start prediction: set the current scheduling time as t=1, record the current scheduling time and the current scheduling day as (t, d), preprocess the historical multi-source measurement data n days before the current scheduling day d, and obtain the preprocessed data;

[0084] The preprocessed data is processed according to the method in step 2-1, and the intraday average is calculated according to the length of one day; the data within n days before the current scheduling day d is taken, and input into the charging pile load, OD traffic flow, and power node load day-ahead value prediction model to obtain the charging pile load, OD traffic flow, and power node load day-ahead value prediction information for the current scheduling day d;

[0085] Step 4-2, the preprocessed data is processed according to the method in step 3-1 to obtain the preprocessed data with hours as the time scale; the data within m time before the current scheduling time t is taken and input into the intraday prediction model of the charging pile load, OD traffic flow, and power node load to obtain the intraday curve prediction information of the charging pile load, OD traffic flow, and power node load on the current scheduling day d;

[0086] Step 4-3, alternate iterative solution:

[0087] Step 4-3-1, set the initial revenue of the mobile charging vehicle to solve power congestion and provide charging services, solve the day-ahead planning model in the two-stage solution model in step 3 based on the day-ahead value prediction information in step 4-1, and obtain the mobile charging vehicle driving route l i tra , that is, the current planning scheme.

[0088] Step 4-3-2, input the day-ahead planning scheme into the intraday scheduling model in the two-stage solution model in step 3 to solve it, and obtain the optimal value of the decision variable of the corresponding intraday planning scheme for the day-ahead planning scheme and At the same time, the corresponding mobile charging vehicle can solve the power congestion and provide charging services.

[0089] Step 4-3-3: If the corrected benefit in step 4-3-2 is equal to the initial benefit set in step 4-3-1, then output the charging and discharging power provided by each mobile charging vehicle to the grid in each dispatching period. Charging and discharging power for electric vehicles and driving route And as the charging and discharging plan of each mobile charging vehicle to the power grid, the charging and discharging plan for providing electric vehicle charging services and the driving route, the intraday scheduling plan of the mobile charging vehicle at time t is obtained to replace the original day-ahead planning plan; otherwise, return to step 4-3-1 and replace the set income in step 4-3-1 with the revised income in step 4-3-2;

[0090] Step 4-4, the current time reaches time t+1: if t=max(T), the current scheduling day ends, and the daily scheduling plan of the mobile charging vehicle of the current scheduling day is obtained; otherwise, t=t+1, return to step 4-2. Example 2:

[0091] An optimization dispatching system for a mobile charging vehicle, comprising:

[0092] The data acquisition module is used to acquire urban planning data and historical multi-source measurement data; wherein the urban planning data includes the power system grid topology, the traffic network trunk road topology, and the parking lot node location; the multi-source measurement data includes power load data, traffic travel data, fixed charging pile charging load data, holiday data, and meteorological data; wherein the power load data includes two types of data: the power generation and power consumption of each power node; the traffic travel data includes two types of data: the injected traffic flow and the output traffic flow of the traffic node; the fixed charging pile charging load data is the power consumption data of the charging pile; the meteorological data includes four types of data: temperature, humidity, wind speed, and light;

[0093] The first model building module is used to create a weighted graph model of the power-traffic dual-flow fusion graph based on urban planning data, and then pre-process the historical multi-source measurement data;

[0094] A second model building module is used to build and train a day-ahead prediction model and an intraday prediction model for charging pile load, OD traffic flow, and power node load based on the urban planning data and pre-processed historical multi-source measurement data;

[0095] The third model building module is used to build an optimization scheduling model with the dispatching goal of maximizing the benefits of mobile charging vehicles solving power congestion and providing charging services within a day, and then divide the optimization scheduling model into a two-stage solution model including a day-ahead planning model and an intraday scheduling model;

[0096] The solution module is used to obtain real-time multi-source measurement data and perform preprocessing, and based on the preprocessed real-time multi-source measurement data and historical multi-source measurement data, alternately and iteratively solve the two-stage solution model, and output the solution of the decision variables corresponding to the intraday scheduling model of the intraday scheduling model as the optimized scheduling plan for the mobile charging vehicle.

[0097] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0098] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0099] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0100] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

Claims

1. A method for optimizing the scheduling of mobile charging vehicles, characterized in that: The steps include: Obtain urban planning data and historical multi-source measurement data; A weighted graph model of a power-traffic dual-flow fusion graph is created based on urban planning data, and then historical multi-source measurement data is preprocessed; the weighted graph model of a power-traffic dual-flow fusion graph is created, including the following specific steps: Based on the power system grid topology, traffic network trunk road topology, and parking lot node locations in urban planning data, a weighted graph model G of the power-traffic dual-flow fusion graph is created. Its mathematical model is: G = {V, E} Among them, V is a node, including the parking lot node as a node in the power-traffic dual flow fusion diagram, the traffic road node with vehicles entering and exiting the main road topology of the traffic network as a road node, and the power node with power consumption and input connected to the parking lot in the power system grid topology as a power node; E is the connection relationship between nodes, which is a road in the traffic flow and a power line in the power flow; Based on the urban planning data and the pre-processed historical multi-source measurement data, a day-ahead prediction model and an intra-day prediction model for charging pile load, OD traffic flow, and power node load are established and trained; The establishment and training of the day-ahead prediction model for charging pile load, OD traffic flow, and power node load includes the following specific steps: Select the preprocessed historical multi-source measurement data, average them within a day according to the length of a day, and obtain the multi-source measurement data with a day as the time scale. The time scale sequence is recorded as d = {1, 2, 3..., N}; N is the total number of days of the historical multi-source measurement data; A graph neural network training model is established based on the graph neural network to build a day-ahead prediction model for charging pile load, OD traffic flow, and power node load, where the vertices and edges of the graph neural network training model are the same as V and E of the power-traffic dual-flow fusion graph; The multi-source measurement data with a daily time scale is divided into input variables and output variables, and the input variables and output variables are paired with d according to the time scale {dn, d-n+1…, d-1}, where the output variable is the charging load of the charging pile on the dth day, and the rest of the data are all used as input variables; after being divided into a training group and a test group, they are input into the graph neural network training model for training to obtain a day-ahead prediction model for the charging pile load; n is a date bias hyperparameter set by the user, indicating that the day-ahead value data of the charging pile load, OD traffic flow, and power node load on a certain day are predicted based on the measurement data of n days; The output variables are replaced by the OD traffic flow on day d and the power node load on day d, respectively, and the day-ahead prediction model of the OD traffic flow and the day-ahead prediction model of the power node load are obtained; The establishment and training of the intraday prediction model for charging pile load, OD traffic flow, and power node load includes the following specific steps: Select the preprocessed multi-source measurement data and record its time scale as; (t, d) = {(1, 1), (2, 1), (3, 1), ... (M, N)}, where t is the hourly time scale and d is the corresponding daily time scale, and obtain the multi-source measurement data with hourly time scale; M is the maximum value of the daily measurement data; Taking multi-source measurement data with a day as the time scale, the time scale of the multi-source measurement data is {dn, d-n+1…, d-1}, inputting it into the day-ahead prediction model of the charging pile load, OD traffic flow, and power node load, and obtaining the day-ahead value prediction information of the charging pile load, OD traffic flow, and power node load on the dth day; The multi-source measurement data with hourly time scale is divided into input variables and output variables, and the input variables and output variables are matched according to the time scale t, where the time scale of the input variables is {t-1, t-2, ... tm}; In addition, the dimension of each time scale input variable is increased, which is the day-ahead value prediction information corresponding to t d, and the time scale of the output variable is t, where the output variables are charging pile load, OD traffic flow, and power node load respectively; m is the hourly bias hyperparameter set by the user, indicating that the intraday curve data of the charging pile load, OD traffic flow, and power node load at a certain time is predicted based on the measurement data at time m; A graph neural network training model is established based on the graph neural network to establish a daily prediction model for charging pile load, OD traffic flow, and power node load, where the vertices and edges of the graph neural network training model are the same as the nodes V and the connection relationship E between nodes in the power-traffic dual-flow fusion graph; Input variables and output variables are input into the graph neural network for training, and the intraday prediction models of charging pile load, OD traffic flow, and power node load are obtained respectively; The dispatching target of the mobile charging vehicle is The objectives of the day-ahead planning model are: The objectives of the intraday scheduling model are: In the formula, f is the total revenue of the mobile charging vehicle within a day, cx i,t 、cy i,t are the benefits obtained by the i-th mobile charging vehicle from solving power congestion and providing charging services in the t scheduling period, cz i,t is the dispatching cost of the i-th mobile charging vehicle in the t-th dispatching period, I is the set of mobile charging vehicles, T is the total dispatching period, are the charging and discharging power provided by the i-th mobile charging vehicle to the power grid and the electric vehicle in the t-th scheduling period, is the driving section of the i-th mobile charging vehicle in the scheduling period t; An optimization scheduling model is constructed with the dispatching goal of maximizing the benefits of mobile charging vehicles in solving power congestion and providing charging services within a day, and then the optimization scheduling model is divided into a two-stage solution model including a day-ahead planning model and an intraday scheduling model; wherein the decision variables corresponding to the day-ahead planning model include the driving route of the mobile charging vehicle; the decision variables corresponding to the intraday scheduling model include the charging and discharging power provided by the mobile charging vehicle to the power grid, the charging and discharging power provided to the electric vehicle, and the driving section of the mobile charging vehicle; Real-time multi-source measurement data is acquired and preprocessed. Based on the preprocessed real-time multi-source measurement data and historical multi-source measurement data, the two-stage solution model is solved alternately and iteratively, and the solution of the decision variables corresponding to the intraday scheduling model is output as the optimized scheduling plan for the mobile charging vehicle.

2. The optimization scheduling method of mobile charging vehicles according to claim 1 is characterized in that: The urban planning data includes the power system grid topology, the traffic network main road topology, and the parking lot node location; the multi-source measurement data includes power load data, traffic travel data, fixed charging pile charging load data, holiday data, and meteorological data; among them, the power load data includes two types of data: the power generation and power consumption of each power node; the traffic travel data includes two types of data: the injected traffic flow and the output traffic flow of the traffic node; the fixed charging pile charging load data is the power consumption data of the charging pile; the meteorological data includes four types of data: temperature, humidity, wind speed, and light.

3. The optimization scheduling method of mobile charging vehicles according to claim 2 is characterized in that: The pretreatment comprises the following specific steps First, the multi-source measurement data is resampled: the multi-source measurement data is used as the label information of the node, and the original multi-source measurement data is resampled according to the time series of the unified step size to achieve hourly time scale alignment; The resampled multi-source measurement data are then normalized: meteorological data, traffic data, power load data, and fixed charging pile charging load data are normalized according to the maximum value of each type of data; holiday data are represented by a binary variable {0, 1}, where 1 indicates that the sample belongs to a holiday and 0 indicates that it belongs to a working day.

4. The method for optimizing the scheduling of mobile charging vehicles according to claim 3 is characterized in that: The time scale of the day-ahead prediction model for the electric pile load, OD traffic flow, and power node load is one day; The time scale of the daily prediction model for the electric pile load, OD traffic flow, and power node load is one hour.

5. The method for optimizing the scheduling of mobile charging vehicles according to claim 4, characterized in that: Step 4) includes the following specific steps: Start prediction: set the current scheduling time as t=1, record the current scheduling time and the current scheduling day as (t, d), preprocess the historical multi-source measurement data n days before the current scheduling day d, and obtain the preprocessed data; The pre-processed data is averaged within a day according to the length of a day; the data within n days before the current scheduling day d is taken and input into the charging pile load, OD traffic flow, and power node load day-ahead value prediction model to obtain the charging pile load, OD traffic flow, and power node load day-ahead value prediction information for the current scheduling day d; Obtain preprocessed data with hourly time scale; take data within m time before the current scheduling time t, input it into the intraday prediction model of charging pile load, OD traffic flow, and power node load, and obtain the intraday curve prediction information of charging pile load, OD traffic flow, and power node load on the current scheduling day d; Alternating iterative solution: The initial revenue of the mobile charging vehicle in solving power congestion and providing charging services is set, and the day-ahead planning model in the two-stage solution model is solved based on the day-ahead value prediction information to obtain the driving route of the mobile charging vehicle. Today’s planning scheme; The day-ahead planning scheme is input into the intraday scheduling model in the two-stage solution model for solution, and the optimal value of the decision variable of the corresponding intraday planning scheme is obtained. and At the same time, the corresponding mobile charging vehicle can solve the power congestion and provide charging services. If the revised revenue is equal to the set initial revenue, then the charging and discharging power provided by each mobile charging vehicle to the grid in each dispatch period is output. Charging and discharging power for electric vehicles and driving route And as the charging and discharging plan of each mobile charging vehicle to the power grid, the charging and discharging plan for providing electric vehicle charging services and the driving route, the intraday dispatch plan of the mobile charging vehicle at time t is obtained, replacing the original day-ahead planning plan; otherwise, the revised income is used as the initial income, and the revised income is recalculated according to the initial income; The current time reaches time t+1: if t=max(T), the current scheduling day ends, and the intraday scheduling plan of the mobile charging vehicle on the current scheduling day is obtained.

6. The method for optimizing the scheduling of mobile charging vehicles according to claim 1, characterized in that: The urban planning data, historical multi-source measurement data and real-time multi-source measurement data are all retrieved through the energy comprehensive service station.

7. An optimization dispatching system for mobile charging vehicles, characterized in that: include The data acquisition module is used to acquire urban planning data and historical multi-source measurement data; wherein the urban planning data includes the power system grid topology, the traffic network trunk road topology, and the parking lot node location; the multi-source measurement data includes power load data, traffic travel data, fixed charging pile charging load data, holiday data, and meteorological data; wherein the power load data includes two types of data: the power generation and power consumption of each power node; the traffic travel data includes two types of data: the injected traffic flow and the output traffic flow of the traffic node; the fixed charging pile charging load data is the power consumption data of the charging pile; the meteorological data includes four types of data: temperature, humidity, wind speed, and light; The first model building module is used to create a weighted graph model of the power-traffic dual-flow fusion graph based on urban planning data, and then pre-process the historical multi-source measurement data; A second model building module is used to build and train a day-ahead prediction model and an intraday prediction model for charging pile load, OD traffic flow, and power node load based on the urban planning data and pre-processed historical multi-source measurement data; The third model building module is used to construct an optimization scheduling model with the maximum benefit obtained by the mobile charging vehicle solving power congestion and providing charging services within a day as the scheduling goal, and then divide the optimization scheduling model into a two-stage solution model including a day-ahead planning model and an intraday scheduling model; wherein the decision variables corresponding to the day-ahead planning model include the driving route of the mobile charging vehicle, and the decision variables corresponding to the intraday scheduling model include the charging and discharging power provided by the power grid, the charging and discharging power provided for the electric vehicle, and the driving section of the mobile charging vehicle; The solution module is used to obtain real-time multi-source measurement data and perform preprocessing. Based on the preprocessed real-time multi-source measurement data and historical multi-source measurement data, the two-stage solution model is alternately iterated to output the solution of the decision variables corresponding to the intraday scheduling model as the optimized scheduling plan for the mobile charging vehicle.

Images