A low-carbon operation method and system for integrated charging and storage charging piles

Abstract

This invention belongs to the field of power system operation optimization and carbon emission control technology, and relates to a low-carbon operation method and system for integrated charging and energy storage charging piles. The method includes: modeling the uncertainty of grid power supply carbon emission factors and photovoltaic output; establishing system power balance constraints, charging pile energy storage constraints, vehicle-pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints; establishing a rolling optimization model for the low-carbon operation system of charging piles considering electric vehicle charging satisfaction; and solving for the optimal low-carbon operation strategy of the integrated charging and energy storage charging pile. This invention comprehensively considers the probabilistic uncertainty of grid power supply carbon emission factors and photovoltaic output, constructing a low-carbon optimization model integrating multi-source carbon emission modeling and flexible scheduling, effectively improving the carbon emission perception and response capabilities of charging piles during operation; simultaneously, it introduces differentiated modeling of user satisfaction and charging behavior, enhancing the system's operational flexibility while ensuring user experience.

Description

Technical Field

[0001] This invention belongs to the field of power system operation optimization and carbon emission control technology. Specifically, it relates to a low-carbon operation method and system for an integrated charging and storage charging pile. Background Technology

[0002] The electrification of transportation and the green and low-carbon transformation of energy are deeply integrated, and the large-scale integration of electric vehicles has significantly impacted grid dispatching and carbon emission structures. Simultaneously, to accelerate the construction of a new power system dominated by new energy sources, higher requirements are being placed on carbon emission control at the end-user energy consumption side. Charging piles, as a crucial interface connecting electric vehicles and the grid, are evolving from traditional one-way power supply equipment into intelligent terminals that integrate charging and energy storage, and source-load interaction. Against this backdrop, exploring low-carbon operation strategies for integrated charging and energy storage piles not only helps optimize the synergy between electric vehicle charging behavior and grid power supply but also significantly improves the overall carbon reduction efficiency of the transportation and power systems.

[0003] Current optimization methods for charging infrastructure operation primarily focus on economic efficiency or peak shaving and valley filling, neglecting to systematically incorporate carbon emission factors. While a few studies have begun to introduce carbon factors for operational evaluation, most still employ regional annual average emission factors, failing to accurately reflect carbon emission differences during electricity consumption periods and hindering the support for refined, low-carbon-oriented scheduling. Furthermore, the output of renewable energy sources such as photovoltaics exhibits high uncertainty, and grid carbon factors display significant temporal fluctuations and non-Gaussian distribution characteristics, making accurate modeling of energy supply carbon emissions difficult using traditional methods. Additionally, most existing scheduling models assume highly controllable electric vehicle behavior, failing to adequately consider user preferences, satisfaction levels, and the heterogeneity of charging behavior, leading to implementation biases in optimization strategies. With the increasing diversity in the number and types of electric vehicles, a low-carbon operation framework with stronger behavioral recognition capabilities and user response mechanisms is needed to achieve both operational reliability and carbon efficiency guarantees under multi-source uncertainties. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a low-carbon operation method and system for an integrated charging and storage charging pile.

[0005] In a first aspect, the present invention provides a low-carbon operation method for an integrated charging and storage charging pile, comprising:

[0006] Uncertainty modeling of grid-connected power supply carbon emission factor and photovoltaic output is carried out, including establishing a probability distribution model of grid-connected power supply carbon emission factor based on variational autoencoder and establishing a probability distribution model of charging station photovoltaic output based on weather-time joint factor graph;

[0007] With the goal of minimizing carbon emissions within the optimization cycle, a rolling optimization model for the low-carbon operation system of charging piles is established, taking into account system power balance constraints, charging pile energy storage constraints, vehicle-pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints.

[0008] Solve the rolling optimization model of the low-carbon operation system of charging piles that takes into account the charging satisfaction of electric vehicles, and obtain the optimal low-carbon operation strategy of integrated charging and storage charging piles under multi-source uncertainty.

[0009] Secondly, the present invention provides a low-carbon operation system for an integrated charging and storage charging pile, including a model establishment unit, a constraint establishment unit and a model solution unit;

[0010] The model building unit is used to model the uncertainty of grid power supply carbon emission factors and photovoltaic output, including establishing a probability distribution model of grid power supply carbon emission factors based on variational autoencoders and establishing a probability distribution model of charging station photovoltaic output based on weather-time joint factor graphs.

[0011] The constraint establishment unit is used to establish system power balance constraints, charging pile energy storage constraints, vehicle-pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints with the goal of minimizing carbon emissions within the optimization cycle. It also establishes a rolling optimization model for the low-carbon operation system of charging piles that takes into account the satisfaction of electric vehicle charging.

[0012] The model solving unit is used to solve the rolling optimization model of the low-carbon operation system of charging piles that takes into account the charging satisfaction of electric vehicles, and obtain the optimal low-carbon operation strategy of charging and storage integrated charging piles under the consideration of multi-source uncertainties.

[0013] Based on the above technical solution, the present invention can be further improved as follows.

[0014] Furthermore, a probability distribution model of grid-connected power supply carbon emission factors based on variational autoencoders is established, including:

[0015] Historical training samples were constructed using the carbon emission factors of grid power supply over several time periods;

[0016] Construct an encoder to perform latent representation on the input data, mapping historical observation data to the parameter space of latent variables;

[0017] A decoder is constructed to transform the given latent variables back into the original data space, generating predicted values ​​of the grid power supply carbon emission factor at each time point.

[0018] Furthermore, let the number of time periods be... , for Historical training sample sets for each time period for The carbon emission factor of grid power supply at any given time, The historical training sample set for each time period is represented as follows: ;

[0019] set up Let N be a latent variable, representing a Gaussian distribution. Represents the mean. Represents variance. This indicates that historical observation data will be used. The parameter space mapped to latent variables. Indicates that given input data Later on latent variables The approximate posterior distribution is represented by a Gaussian distribution: ;

[0020] set up Representing latent variables The prior distribution, Let represent a square matrix with 1s on the diagonal and 0s elsewhere, indicating that each latent variable dimension is independent of the others and has a variance of 1. express Predicted carbon emission factor for electricity supplied by the time network. Indicates latent variables Convert back to the original data space. Indicates that given latent variables Under the condition of the generation probability distribution of emission factors, For parameters The decoder neural network assumes latent variables The prior distribution is a multivariate Gaussian distribution with a mean of 0 and a covariance of the identity matrix. The decoder will take the given latent variable... Transforming back to the original data space, the formula for the latent variables becomes:

[0021] ;

[0022] The predicted carbon emission factor for electricity supply from the TimeNet network is:

[0023] .

[0024] Furthermore, a probability distribution model for photovoltaic power output at charging stations, generated based on a weather-time joint factor graph, is established, including:

[0025] Construct a weather-time feature dataset;

[0026] Design a factor graph structure and establish a joint probability model;

[0027] By learning and training factor graphs using historical sample datasets, the structure of factor graphs and factor node functions are determined, and Gibbs sampling is used to obtain the predicted distribution.

[0028] Furthermore, a weather-time feature dataset is constructed, including:

[0029] Historical data of the area where the charging station is located was collected to construct a multidimensional input dataset in time series form, including power output datasets for several historical periods of photovoltaic power generation and weather feature datasets for several historical periods.

[0030] set up Representing history Temperature at different times Representing history Cloud cover for a given period of time. Representing history Relative humidity at different times Representing history Rainfall in a given period Representing history Solar irradiance for each time period, the number of time periods being: ,history The photovoltaic output data for each time period is as follows ,history The weather feature dataset for each time period is as follows: ,but: ;

[0031] Design the factor graph structure and establish a joint probability model, including:

[0032] History Temperature and historical data for each period Cloud cover for a specific time period, historical data Relative humidity for each time period, historical Rainfall in each period and historical data A two-layer factor graph is constructed using solar irradiance over a period of time as variables. Variable nodes include all input and output variables, and factor nodes are used to represent the local dependencies between variables. Factors include joint factors between weather variables, adjustment factors between time and weather variables, mapping factors between photovoltaic output and weather variables, and smoothing factors between output at adjacent times.

[0033] Using factor graphs to express the overall joint probability distribution, let... This is the partition function, used to normalize the entire distribution. Indicates the first The subset of variables involved in each factor Let the factor node function be a function that reflects local dependencies. Then:

[0034] ;

[0035] set up for Predicted photovoltaic output at any given time for The random variable of photovoltaic output at any given time. For weather-time feature datasets, This is a normalization factor for the weather-time feature dataset. For the first The factor node function for the subset of variables involved in each factor, the predicted value is expressed as:

[0036] .

[0037] Furthermore, the energy storage constraints of charging piles include charging and discharging power constraints and energy conservation constraints; vehicle-pile matching constraints include: each vehicle can only be allocated one charging pile at the same time, each pile can only serve one vehicle, and the vehicle can only be charged after it arrives; electric vehicle charging behavior constraints include classifying electric vehicles into immediate charging type and waiting charging type; the charging behavior of immediate charging type is characterized by the shortest desired charging time, charging immediately when there is a charging pile available, and queuing when there is no available one, and charging at the rated power, which is not adjustable, and stopping charging and leaving after reaching the desired amount of power; the charging target is reached within the desired time period, and the charging power is adjustable during the charging period.

[0038] Furthermore, a rolling optimization model for the low-carbon operation system of charging piles, taking into account the charging satisfaction of electric vehicles, is established, including:

[0039] set up Indicates time period Carbon emission factors of grid power supply Indicates the power supply capacity of the charging station network. These are the weighting coefficients. Let represent the overall satisfaction of electric vehicle charging users during the optimization period. Then, the objective function of the optimization model is expressed as:

[0040] .

[0041] Furthermore, system power balance constraints, charging pile energy storage constraints, vehicle-charging pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints are established, including:

[0042] set up for Predicted photovoltaic output at any given time for Time of the first The energy storage discharge power of each charging pile This is a collection of all charging stations. express Time of the first The charging power of each charging station for electric vehicles. express Time of the first The energy storage charging power of each charging pile is then expressed as follows:

[0043] ;

[0044] set up express Time of the first The energy storage capacity of each charging station Indicates the first The maximum energy storage capacity of each charging station Indicates the first Energy storage charging efficiency of each charging pile Indicates the first The energy storage discharge efficiency of a charging pile is then expressed as the energy storage constraint of the charging pile:

[0045] ;

[0046] set up Indicates the first The car is Charging power at any time For the collection of all charging vehicles, Indicates electric vehicle With charging piles exist The matching relation variable at time t is a Boolean variable, when Shike Electric Vehicle At the charging station When charging, ,otherwise ; Indicates electric vehicle If the arrival time is given, then the vehicle-pile matching constraint is expressed as:

[0047] ;

[0048] set up: This refers to a collection of electric vehicles that can be charged immediately. Let it be a Boolean variable representing an electric vehicle. The queuing status A value of 1 indicates that the user is in a queue. A value of 0 indicates that no queue has been formed; This indicates the maximum waiting time. After the maximum waiting time is exceeded, charging will be abandoned and the vehicle will leave. Let it be a Boolean variable representing an electric vehicle. The service status, A value of 1 indicates that the charging service has been accepted. A value of 0 indicates that charging has been abandoned and the vehicle has departed; It is an infinitely large constant. Indicates the rated charging power; Indicates electric vehicle The expected charging capacity; Indicates electric vehicle The initial charge; Indicates electric vehicle Maximum battery capacity; Indicates electric vehicle The departure time is then constrained by the electric vehicle charging behavior as follows:

[0049]

[0050] set up The satisfaction reward coefficient, This represents the waiting time penalty coefficient. Let represent the penalty coefficient for users not receiving charging services. Then, the user satisfaction constraint is expressed as:

[0051] .

[0052] Optionally, when solving the rolling optimization model of the low-carbon operation system of charging piles that considers the satisfaction of electric vehicle charging, based on the existing probability distribution model of grid power supply carbon emission factors and the probability distribution model of photovoltaic power output of charging stations, multiple random samplings are performed for each rolling optimization cycle to generate several typical scenarios; each scenario is transformed into a deterministic scheduling problem, and a commercial optimization solver is used to solve it one by one to obtain the optimal scheduling scheme of charging piles and energy storage systems under each scenario; based on the rolling optimization strategy obtained from the system rolling optimization model, the input data is updated in chronological order to iteratively advance the overall optimization process.

[0053] The beneficial effects of this invention are as follows: This invention comprehensively considers the probabilistic uncertainty of the carbon emission factor of power grid supply and photovoltaic power output, and constructs a low-carbon optimization model that integrates multi-source carbon emission modeling and flexible scheduling, which effectively improves the carbon emission perception and response capabilities of charging piles during operation; at the same time, it introduces differentiated modeling of user satisfaction and charging behavior, which enhances the system's operational flexibility while ensuring user experience. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of a low-carbon operation method for an integrated charging and storage charging pile provided in Embodiment 1 of the present invention.

[0055] Figure 2This is a schematic diagram of a low-carbon operation system for an integrated charging and storage charging pile provided in Embodiment 1 of the present invention. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0057] Example 1

[0058] As an example, see the attached document. Figure 1 As shown, to solve the above-mentioned technical problems, this embodiment provides a low-carbon operation method for an integrated charging and storage charging pile, including:

[0059] Uncertainty modeling of grid-connected power supply carbon emission factor and photovoltaic output is carried out, including establishing a probability distribution model of grid-connected power supply carbon emission factor based on variational autoencoder and establishing a probability distribution model of charging station photovoltaic output based on weather-time joint factor graph;

[0060] With the goal of minimizing carbon emissions within the optimization cycle, a rolling optimization model for the low-carbon operation system of charging piles is established, taking into account system power balance constraints, charging pile energy storage constraints, vehicle-pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints.

[0061] Solve the rolling optimization model of the low-carbon operation system of charging piles that takes into account the charging satisfaction of electric vehicles, and obtain the optimal low-carbon operation strategy of integrated charging and storage charging piles under multi-source uncertainty.

[0062] This invention first addresses the uncertainty of source-side energy supply by establishing a probabilistic modeling framework for two key inputs: First, it constructs a probability distribution model based on a variational autoencoder to address the non-Gaussianity and temporal correlation of grid-connected carbon emission factors; second, it constructs a probabilistic generation model for photovoltaic output based on a weather-time joint factor graph to address the strong dependence of photovoltaic output on weather information, thereby achieving uncertainty modeling of source-side energy supply carbon emissions. Then, considering the charging preferences of electric vehicle users, it categorizes users into immediate charging and waiting-to-charge types, establishing charging behavior constraint mechanisms for each, and further introducing queuing and early departure mechanisms to enhance the model. The invention enhances the responsiveness to dynamic changes in user behavior. Furthermore, aiming to minimize carbon emissions within the optimization cycle, a rolling optimization model is established, incorporating power balance constraints, charging / discharging constraints, energy conservation constraints, vehicle-pile matching constraints, and user satisfaction constraints to ensure the model's operational rationality and user experience. Finally, by solving the rolling optimization model, the optimal low-carbon operation strategy for charging piles under multiple uncertainties is obtained, achieving synergistic optimization among photovoltaic power consumption, low-carbon power supply from the grid, and reasonable charging of electric vehicles. This effectively reduces carbon emissions during the charging process and improves user participation, system flexibility, and low-carbon operation. This invention helps improve the carbon perception and scheduling response capabilities of charging infrastructure during operation, achieving synergistic optimization of electric vehicle user satisfaction and low-carbon operation goals, and providing effective technical support for promoting carbon reduction in end-user energy consumption under the deep integration of transportation and power systems.

[0063] Optionally, a probability distribution model of grid-connected carbon emission factors based on a variational autoencoder is established, including:

[0064] Historical training samples were constructed using the carbon emission factors of grid power supply over several time periods;

[0065] Construct an encoder to perform latent representation on the input data, mapping historical observation data to the parameter space of latent variables;

[0066] A decoder is constructed to transform the given latent variables back into the original data space, generating predicted values ​​of the grid power supply carbon emission factor at each time point.

[0067] The grid-connected carbon emission factor and photovoltaic output are key factors influencing the operation strategy of integrated charging and energy storage piles. Therefore, uncertainty modeling of the grid-connected carbon emission factor and photovoltaic output is first performed. Considering that the grid-connected carbon emission factor of integrated charging and energy storage piles has significant time-series correlation, periodicity, and non-Gaussian properties, a probability distribution model of the grid-connected carbon emission factor based on a variational autoencoder is established.

[0068] Optionally, the number of time periods is set to 1. , for Historical training sample sets for each time period for The carbon emission factor of grid power supply at any given time, The historical training sample set for each time period is represented as follows: ;

[0069] set up Let N be a latent variable, representing a Gaussian distribution. Represents the mean. Represents variance. This indicates that historical observation data will be used. The parameter space mapped to latent variables. Indicates that given input data Later on latent variables The approximate posterior distribution is represented by a Gaussian distribution: ;

[0070] set up Representing latent variables The prior distribution, Let represent a square matrix with 1s on the diagonal and 0s elsewhere, indicating that each latent variable dimension is independent of the others and has a variance of 1. express Predicted carbon emission factor for electricity supplied by the time network. Indicates latent variables Convert back to the original data space. Indicates that given latent variables Under the condition of the generation probability distribution of emission factors, For parameters The decoder neural network assumes latent variables The prior distribution is a multivariate Gaussian distribution with a mean of 0 and a covariance of the identity matrix. The decoder will take the given latent variable... Transforming back to the original data space, the formula for the latent variables becomes:

[0071] ;

[0072] The predicted carbon emission factor for electricity supply from the TimeNet network is:

[0073] .

[0074] Optionally, a probability distribution model of photovoltaic power output at charging stations is established based on a weather-time joint factor map, including:

[0075] Construct a weather-time feature dataset;

[0076] Design a factor graph structure and establish a joint probability model;

[0077] By learning and training factor graphs using historical sample datasets, the structure of factor graphs and factor node functions are determined, and Gibbs sampling is used to obtain the predicted distribution.

[0078] Photovoltaic power output is highly dependent on weather information and has a clear structured dependency relationship. Therefore, a probability distribution model of photovoltaic power output of charging stations can be established based on weather-time joint factor graphs.

[0079] Optionally, construct a weather-time feature dataset, including:

[0080] Historical data of the area where the charging station is located was collected to construct a multidimensional input dataset in time series form, including power output datasets for several historical periods of photovoltaic power generation and weather feature datasets for several historical periods.

[0081] set up Representing history Temperature at different times Representing history Cloud cover for a given period of time. Representing history Relative humidity at different times Representing history Rainfall in a given period Representing history Solar irradiance for each time period, the number of time periods being: ,history The photovoltaic output data for each time period is as follows ,history The weather feature dataset for each time period is as follows: ,but: ;

[0082] Design the factor graph structure and establish a joint probability model, including:

[0083] History Temperature and historical data for each period Cloud cover for a specific time period, historical data Relative humidity for each time period, historical Rainfall in each period and historical data A two-layer factor graph is constructed using solar irradiance over a period of time as variables. Variable nodes include all input and output variables, and factor nodes are used to represent the local dependencies between variables. Factors include joint factors between weather variables, adjustment factors between time and weather variables, mapping factors between photovoltaic output and weather variables, and smoothing factors between output at adjacent times.

[0084] Using factor graphs to express the overall joint probability distribution, let... This is the partition function, used to normalize the entire distribution. Indicates the first The subset of variables involved in each factor Let the factor node function be a function that reflects local dependencies. Then:

[0085] ;

[0086] set up for Predicted photovoltaic output at any given time for The random variable of photovoltaic output at any given time. For weather-time feature datasets, This is a normalization factor for the weather-time feature dataset. For the first The factor node function for the subset of variables involved in each factor, the predicted value is expressed as:

[0087] .

[0088] Optionally, the energy storage constraints of charging piles include charging and discharging power constraints and energy conservation constraints; vehicle-pile matching constraints include: each vehicle can only be assigned one charging pile at the same time, each pile can only serve one vehicle, and the vehicle can only be charged after it arrives; electric vehicle charging behavior constraints include classifying electric vehicles into immediate charging type and waiting charging type; the charging behavior of immediate charging type is characterized by the shortest desired charging time, charging immediately when there is a charging pile available, and queuing when there is no available one, and charging at the rated power, which is not adjustable, and stopping charging and leaving after reaching the desired amount of power; the charging target is reached within the desired time period, and the charging power is adjustable during the charging period.

[0089] Optionally, a rolling optimization model for the low-carbon operation system of charging piles, taking into account the charging satisfaction of electric vehicles, can be established, including:

[0090] set up Indicates time period Carbon emission factors of grid power supply Indicates the power supply capacity of the charging station network. These are the weighting coefficients. Let represent the overall satisfaction of electric vehicle charging users during the optimization period. Then, the objective function of the optimization model is expressed as:

[0091] .

[0092] Optionally, establish system power balance constraints, charging pile energy storage constraints, vehicle-charging pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints, including:

[0093] set up for Predicted photovoltaic output at any given time for Time of the first The energy storage discharge power of each charging pile This is a collection of all charging stations. express Time of the first The charging power of each charging station for electric vehicles. express Time of the first The energy storage charging power of each charging pile is then expressed as follows:

[0094] ;

[0095] set up express Time of the first The energy storage capacity of each charging station Indicates the first The maximum energy storage capacity of each charging station Indicates the first Energy storage charging efficiency of each charging pile Indicates the first The energy storage discharge efficiency of a charging pile is then expressed as the energy storage constraint of the charging pile:

[0096] ;

[0097] set up Indicates the first The car is Charging power at any time For the collection of all charging vehicles, Indicates electric vehicle With charging piles exist The matching relation variable at time t is a Boolean variable, when Shike Electric Vehicle At the charging station When charging, ,otherwise ; Indicates electric vehicle If the arrival time is given, then the vehicle-pile matching constraint is expressed as:

[0098] ;

[0099] set up: This refers to a collection of electric vehicles that can be charged immediately. Let it be a Boolean variable representing an electric vehicle. The queuing status A value of 1 indicates that the user is in a queue. A value of 0 indicates that no queue has been formed; This indicates the maximum waiting time. After the maximum waiting time is exceeded, charging will be abandoned and the vehicle will leave. Let it be a Boolean variable representing an electric vehicle. The service status, A value of 1 indicates that the charging service has been accepted. A value of 0 indicates that charging has been abandoned and the vehicle has departed; It is an infinitely large constant. Indicates the rated charging power; Indicates electric vehicle The expected charging capacity; Indicates electric vehicle The initial charge; Indicates electric vehicle Maximum battery capacity; Indicates electric vehicle If the departure time is given, then the electric vehicle charging behavior constraint is expressed as:

[0100]

[0101] set up The satisfaction reward coefficient, This represents the waiting time penalty coefficient. Let represent the penalty coefficient for users not receiving charging services. Then, the user satisfaction constraint is expressed as:

[0102] .

[0103] Optionally, when solving the rolling optimization model of the low-carbon operation system of charging piles that considers the satisfaction of electric vehicle charging, based on the existing probability distribution model of grid power supply carbon emission factors and the probability distribution model of photovoltaic power output of charging stations, multiple random samplings are performed for each rolling optimization cycle to generate several typical scenarios; each scenario is transformed into a deterministic scheduling problem, and a commercial optimization solver is used to solve it one by one to obtain the optimal scheduling scheme of charging piles and energy storage systems under each scenario; based on the rolling optimization strategy obtained from the system rolling optimization model, the input data is updated in chronological order to iteratively advance the overall optimization process.

[0104] This invention proposes a low-carbon operation method for integrated charging and energy storage charging piles. First, addressing source-side uncertainties, this invention constructs a non-Gaussian probability modeling method for grid-connected carbon emission factors, introducing a variational autoencoder to characterize their temporal correlation and periodic fluctuations. Simultaneously, considering the strong dependence of photovoltaic output on weather information, a photovoltaic output probability generation model based on a weather-time joint factor graph is constructed to achieve structured modeling of renewable output. Second, addressing load-side flexible response behavior, a behavioral modeling framework is established that includes both immediate-charging and waiting-to-charge electric vehicles, considering queuing and early departure mechanisms to precisely constrain the electric vehicle charging scheduling process. Furthermore, an electric vehicle charging satisfaction function is introduced to quantify the user's responsiveness to waiting time, constructing user satisfaction constraints. Finally, a rolling optimization model with the goal of minimizing carbon emissions is proposed, systematically integrating photovoltaic output prediction, dynamic carbon factor estimation, integrated charging and energy storage scheduling, and vehicle-pile matching mechanisms to construct a fully carbon-aware charging pile operation system, achieving low-carbon and friendly collaboration between electric vehicles and the power grid.

[0105] This invention comprehensively considers the probabilistic uncertainties of carbon emission factors from power grid supply and photovoltaic output, constructing a low-carbon optimization model that integrates multi-source carbon emission modeling and flexible scheduling. This effectively improves the carbon emission perception and response capabilities of charging piles during operation. Simultaneously, it introduces differentiated modeling of user satisfaction and charging behavior, enhancing the system's operational flexibility while ensuring a positive user experience. This method possesses good scalability and practical value, providing strong support for carbon reduction in end-user energy consumption in transportation-power integration scenarios.

[0106] Example 2

[0107] Based on the same principle as the method shown in Embodiment 1 of the present invention, as illustrated in the appendix. Figure 2 As shown, an embodiment of the present invention also provides a system, including a low-carbon operation system for an integrated charging and storage charging pile, comprising a model establishment unit, a constraint establishment unit, and a model solving unit;

[0108] The model building unit is used to model the uncertainty of grid power supply carbon emission factors and photovoltaic output, including establishing a probability distribution model of grid power supply carbon emission factors based on variational autoencoders and establishing a probability distribution model of charging station photovoltaic output based on weather-time joint factor graphs.

[0109] The constraint establishment unit is used to establish system power balance constraints, charging pile energy storage constraints, vehicle-pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints with the goal of minimizing carbon emissions within the optimization cycle. It also establishes a rolling optimization model for the low-carbon operation system of charging piles that takes into account the satisfaction of electric vehicle charging.

[0110] The model solving unit is used to solve the rolling optimization model of the low-carbon operation system of charging piles that takes into account the charging satisfaction of electric vehicles, and obtain the optimal low-carbon operation strategy of charging and storage integrated charging piles under the consideration of multi-source uncertainties.

[0111] Optionally, a probability distribution model of grid-connected carbon emission factors based on a variational autoencoder is established, including:

[0112] Historical training samples were constructed using the carbon emission factors of grid power supply over several time periods;

[0113] Construct an encoder to perform latent representation on the input data, mapping historical observation data to the parameter space of latent variables;

[0114] A decoder is constructed to transform the given latent variables back into the original data space, generating predicted values ​​of the grid power supply carbon emission factor at each time point.

[0115] Optionally, the number of time periods is set to 1. , for Historical training sample sets for each time period for The carbon emission factor of grid power supply at any given time, The historical training sample set for each time period is represented as follows: ;

[0116] set up Let N be a latent variable, representing a Gaussian distribution. Represents the mean. Represents variance. This indicates that historical observation data will be used. The parameter space mapped to latent variables. Indicates that given input data Later on latent variables The approximate posterior distribution is represented by a Gaussian distribution: ;

[0117] set up Representing latent variables The prior distribution, Let represent a square matrix with 1s on the diagonal and 0s elsewhere, indicating that each latent variable dimension is independent of the others and has a variance of 1. express Predicted carbon emission factor for electricity supplied by the time network. Indicates latent variables Convert back to the original data space. Indicates that given latent variables Under the condition of the generation probability distribution of emission factors, For parameters The decoder neural network assumes latent variables The prior distribution is a multivariate Gaussian distribution with a mean of 0 and a covariance of the identity matrix. The decoder will take the given latent variable... Transforming back to the original data space, the formula for the latent variables becomes:

[0118] ;

[0119] The predicted carbon emission factor for electricity supply from the TimeNet network is:

[0120] .

[0121] Optionally, a probability distribution model of photovoltaic power output at charging stations is established based on a weather-time joint factor map, including:

[0122] Construct a weather-time feature dataset;

[0123] Design a factor graph structure and establish a joint probability model;

[0124] By learning and training factor graphs using historical sample datasets, the structure of factor graphs and factor node functions are determined, and Gibbs sampling is used to obtain the predicted distribution.

[0125] Optionally, construct a weather-time feature dataset, including:

[0126] Historical data of the area where the charging station is located was collected to construct a multidimensional input dataset in time series form, including power output datasets for several historical periods of photovoltaic power generation and weather feature datasets for several historical periods.

[0127] set up Representing history Temperature at different times Representing history Cloud cover for a given period of time. Representing history Relative humidity at different times Representing history Rainfall in a given period Representing history Solar irradiance for each time period, the number of time periods being: ,history The photovoltaic output data for each time period is as follows ,history The weather feature dataset for each time period is as follows: ,but: ;

[0128] Design the factor graph structure and establish a joint probability model, including:

[0129] History Temperature and historical data for each period Cloud cover for a specific time period, historical data Relative humidity for each time period, historical Rainfall in each period and historical data A two-layer factor graph is constructed using solar irradiance over a period of time as variables. Variable nodes include all input and output variables, and factor nodes are used to represent the local dependencies between variables. Factors include joint factors between weather variables, adjustment factors between time and weather variables, mapping factors between photovoltaic output and weather variables, and smoothing factors between output at adjacent times.

[0130] Using factor graphs to express the overall joint probability distribution, let... This is the partition function, used to normalize the entire distribution. Indicates the first The subset of variables involved in each factor Let the factor node function be a function that reflects local dependencies. Then:

[0131] ;

[0132] set up for Predicted photovoltaic output at any given time for The random variable of photovoltaic output at any given time. For weather-time feature datasets, This is a normalization factor for the weather-time feature dataset. For the first The factor node function for the subset of variables involved in each factor, the predicted value is expressed as:

[0133] .

[0134] Optionally, the energy storage constraints of charging piles include charging and discharging power constraints and energy conservation constraints; vehicle-pile matching constraints include: each vehicle can only be assigned one charging pile at the same time, each pile can only serve one vehicle, and the vehicle can only be charged after it arrives; electric vehicle charging behavior constraints include classifying electric vehicles into immediate charging type and waiting charging type; the charging behavior of immediate charging type is characterized by the shortest desired charging time, charging immediately when there is a charging pile available, and queuing when there is no available one, and charging at the rated power, which is not adjustable, and stopping charging and leaving after reaching the desired amount of power; the charging target is reached within the desired time period, and the charging power is adjustable during the charging period.

[0135] Optionally, a rolling optimization model for the low-carbon operation system of charging piles, taking into account the charging satisfaction of electric vehicles, can be established, including:

[0136] set up Indicates time period Carbon emission factors of grid power supply Indicates the power supply capacity of the charging station network. These are the weighting coefficients. Let represent the overall satisfaction of electric vehicle charging users during the optimization period. Then, the objective function of the optimization model is expressed as:

[0137] .

[0138] Optionally, establish system power balance constraints, charging pile energy storage constraints, vehicle-charging pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints, including:

[0139] set up for Predicted photovoltaic output at any given time for Time of the first The energy storage discharge power of each charging pile This is a collection of all charging stations. express Time of the first The charging power of each charging station for electric vehicles. express Time of the first The energy storage charging power of each charging pile is then expressed as follows:

[0140] ;

[0141] set up express Time of the first The energy storage capacity of each charging station Indicates the first The maximum energy storage capacity of each charging station Indicates the first Energy storage charging efficiency of each charging pile Indicates the first The energy storage discharge efficiency of a charging pile is then expressed as the energy storage constraint of the charging pile:

[0142] ;

[0143] set up Indicates the first The car is Charging power at any time For the collection of all charging vehicles, Indicates electric vehicle With charging piles exist The matching relation variable at time t is a Boolean variable, when Shike Electric Vehicle At the charging station When charging, ,otherwise ; Indicates electric vehicle If the arrival time is given, then the vehicle-pile matching constraint is expressed as:

[0144] ;

[0145] set up: This refers to a collection of electric vehicles that can be charged immediately. Let it be a Boolean variable representing an electric vehicle. The queuing status A value of 1 indicates that the user is in a queue. A value of 0 indicates that no queue has been formed; This indicates the maximum waiting time. After the maximum waiting time is exceeded, charging will be abandoned and the vehicle will leave. Let it be a Boolean variable representing an electric vehicle. The service status, A value of 1 indicates that the charging service has been accepted. A value of 0 indicates that charging has been abandoned and the vehicle has departed; It is an infinitely large constant. Indicates the rated charging power; Indicates electric vehicle The expected charging capacity; Indicates electric vehicle The initial charge; Indicates electric vehicle Maximum battery capacity; Indicates electric vehicle If the departure time is given, then the electric vehicle charging behavior constraint is expressed as:

[0146]

[0147] set up The satisfaction reward coefficient, This represents the waiting time penalty coefficient. Let represent the penalty coefficient for users not receiving charging services. Then, the user satisfaction constraint is expressed as:

[0148] .

[0149] Optionally, when solving the rolling optimization model of the low-carbon operation system of charging piles that considers the satisfaction of electric vehicle charging, based on the existing probability distribution model of grid power supply carbon emission factors and the probability distribution model of photovoltaic power output of charging stations, multiple random samplings are performed for each rolling optimization cycle to generate several typical scenarios; each scenario is transformed into a deterministic scheduling problem, and a commercial optimization solver is used to solve it one by one to obtain the optimal scheduling scheme of charging piles and energy storage systems under each scenario; based on the rolling optimization strategy obtained from the system rolling optimization model, the input data is updated in chronological order to iteratively advance the overall optimization process.

[0150] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A low-carbon operation method for an integrated charging and storage charging pile, characterized in that, include: Uncertainty modeling of grid-connected power supply carbon emission factor and photovoltaic output is carried out, including establishing a probability distribution model of grid-connected power supply carbon emission factor based on variational autoencoder and establishing a probability distribution model of charging station photovoltaic output based on weather-time joint factor graph; A probability distribution model for grid-connected power supply carbon emission factors based on variational autoencoders is established, including: constructing historical training samples using grid-connected power supply carbon emission factors from several time periods; constructing an encoder to perform latent representation on the input data, mapping the historical observation data to the parameter space of latent variables; and constructing a decoder to transform the given latent variables back into the original data space, generating predicted values ​​of grid-connected power supply carbon emission factors at each time period. The number of time periods is set to... , for Historical training sample sets for each time period for The carbon emission factor of grid power supply at any given time, The historical training sample set for each time period is represented as follows: ;set up Let N be a latent variable, representing a Gaussian distribution. Represents the mean. Represents variance. This indicates that historical observation data will be used. The parameter space mapped to latent variables. Indicates that given input data Later on latent variables The approximate posterior distribution is represented by a Gaussian distribution: ;set up Representing latent variables The prior distribution, Let represent a square matrix with 1s on the diagonal and 0s elsewhere, indicating that each latent variable dimension is independent of the others and has a variance of 1. express Predicted carbon emission factor for electricity supplied by the time network. Indicates latent variables Convert back to the original data space. Indicates that given latent variables Under the condition of the generation probability distribution of emission factors, For parameters The decoder neural network assumes latent variables The prior distribution is a multivariate Gaussian distribution with a mean of 0 and a covariance of the identity matrix. The decoder will take the given latent variable... Transforming back to the original data space, the formula for the latent variables becomes: ; The predicted carbon emission factor for electricity supply from the TimeNet network is: This study establishes a probability distribution model for photovoltaic (PV) output at charging stations based on a weather-time joint factor graph. This includes: constructing a weather-time feature dataset; designing the factor graph structure and establishing a joint probability model; learning and training the factor graph using historical sample datasets to determine the factor graph structure and factor node functions; and using Gibbs sampling to obtain the predicted distribution. The weather-time feature dataset is constructed by: collecting historical data from the charging station's location; constructing a multi-dimensional input dataset in time-series format, including PV output datasets for several historical time periods and weather feature datasets for several historical time periods; and setting... Representing history Temperature at different times Representing history Cloud cover for a given period of time. Representing history Relative humidity at different times Representing history Rainfall in a given period Representing history Solar irradiance for each time period, the number of time periods being: ,history The photovoltaic output data for each time period is as follows ,history The weather feature dataset for each time period is as follows: ,but: Design a factor graph structure and establish a joint probability model, including: incorporating historical data... Temperature and historical data for each period Cloud cover for a specific time period, historical data Relative humidity for each time period, historical Rainfall in each period and historical data A two-level factor graph is constructed using solar irradiance over a given time period as variables. Variable nodes include all input and output variables, and factor nodes represent local dependencies between variables. Factors include joint factors between weather variables, moderating factors between time and weather variables, mapping factors between photovoltaic output and weather variables, and smoothing factors between outputs at adjacent times. The factor graph is used to express the overall joint probability distribution. This is the partition function, used to normalize the entire distribution. Indicates the first The subset of variables involved in each factor Let the factor node function be a factor node function, and the factor node function reflects local dependencies. Then: ;set up for Predicted photovoltaic output at any given time for The random variable of photovoltaic power output at any given time. For weather-time feature datasets, This is a normalization factor for the weather-time feature dataset. For the first The factor node function for the subset of variables involved in each factor, the predicted value is expressed as: ; With the goal of minimizing carbon emissions within the optimization cycle, this paper establishes system power balance constraints, charging pile energy storage constraints, vehicle-pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints. A rolling optimization model for the low-carbon operation system of charging piles, considering electric vehicle charging satisfaction, is then established. Charging pile energy storage constraints include charging and discharging power constraints and energy conservation constraints. Vehicle-pile matching constraints include: each vehicle can only be assigned one charging pile at any given time, each pile can only serve one vehicle, and the vehicle can only charge upon arrival. Electric vehicle charging behavior constraints include classifying electric vehicles into immediate charging and waiting-to-charge types. Immediate charging is characterized by the shortest desired charging time; charging occurs immediately when a charging pile is available, and waiting in line when no pile is available. Charging power is fixed at the rated power and cannot be adjusted. Charging stops and the vehicle departs once the desired charge level is reached. The charging target is achieved within the desired time period, and the charging power is adjustable during charging. A rolling optimization model for the low-carbon operation system of charging piles, considering electric vehicle charging satisfaction, is established, including: setting... Indicates time period Carbon emission factors of grid power supply Indicates the power supply capacity of the charging station network. These are the weighting coefficients. Let represent the overall satisfaction of electric vehicle charging users during the optimization period. Then, the objective function of the optimization model is expressed as: Establish system power balance constraints, charging pile energy storage constraints, vehicle-charging pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints, including: setting... for Predicted photovoltaic output at any given time for Time of the first The energy storage discharge power of each charging pile This is a collection of all charging stations. express Time of the first The charging power of each charging station for electric vehicles. express Time of the first The energy storage charging power of each charging pile is then expressed as follows: ;set up express Time of the first The energy storage capacity of each charging station Indicates the first The maximum energy storage capacity of each charging station Indicates the first Energy storage charging efficiency of each charging pile Indicates the first The energy storage discharge efficiency of a charging pile is then expressed as the energy storage constraint of the charging pile: ;set up Indicates the first The car is Charging power at any time For the collection of all charging vehicles, Indicates electric vehicle With charging piles exist The matching relation variable at time t is a Boolean variable, when Shike Electric Vehicle At the charging station When charging, ,otherwise ; Indicates electric vehicle If the arrival time is given, then the vehicle-pile matching constraint is expressed as: ;set up: This refers to a collection of electric vehicles that can be charged immediately. Let it be a Boolean variable representing an electric vehicle. The queuing status A value of 1 indicates that the user is in a queue. A value of 0 indicates that no queue has been formed; This indicates the maximum waiting time. After the maximum waiting time is exceeded, charging will be abandoned and the vehicle will leave. Let it be a Boolean variable representing an electric vehicle. The service status, A value of 1 indicates that the charging service has been accepted. A value of 0 indicates that charging has been abandoned and the vehicle has departed; It is an infinitely large constant. Indicates the rated charging power; Indicates electric vehicle Expected charging capacity; Indicates electric vehicle The initial charge; Indicates electric vehicle Maximum battery capacity; Indicates electric vehicle If the departure time is given, then the electric vehicle charging behavior constraint is expressed as: ； set up The satisfaction reward coefficient, This represents the waiting time penalty coefficient. Let represent the penalty coefficient for users not receiving charging services. Then, the user satisfaction constraint is expressed as: ； Solve the rolling optimization model of the low-carbon operation system of charging piles that takes into account the charging satisfaction of electric vehicles, and obtain the optimal low-carbon operation strategy of integrated charging and storage charging piles under multi-source uncertainty.

2. The low-carbon operation method of an integrated charging and storage charging pile according to claim 1, characterized in that, When solving the rolling optimization model of the low-carbon operation system of charging piles that takes into account the satisfaction of electric vehicle charging, based on the existing probability distribution model of grid power supply carbon emission factor and the probability distribution model of photovoltaic power output of charging station, multiple random samplings are performed for each rolling optimization cycle to generate several typical scenarios; each scenario is transformed into a deterministic scheduling problem, and a commercial optimization solver is used to solve it one by one to obtain the optimal scheduling scheme of charging piles and energy storage system under each scenario. The rolling optimization strategy, derived from the system's rolling optimization model, updates the input data in chronological order, iteratively advancing the overall optimization process.

3. A low-carbon operation system for an integrated charging and storage charging pile, characterized in that, It includes model building units, constraint building units, and model solving units; The model building unit is used to model the uncertainty of grid power supply carbon emission factors and photovoltaic output, including establishing a probability distribution model of grid power supply carbon emission factors based on variational autoencoders and establishing a probability distribution model of charging station photovoltaic output based on weather-time joint factor graphs. A probability distribution model for grid-connected power supply carbon emission factors based on variational autoencoders is established, including: constructing historical training samples using grid-connected power supply carbon emission factors from several time periods; constructing an encoder to perform latent representation on the input data, mapping the historical observation data to the parameter space of latent variables; and constructing a decoder to transform the given latent variables back into the original data space, generating predicted values ​​of grid-connected power supply carbon emission factors at each time period. The number of time periods is set to... , for Historical training sample sets for each time period for The carbon emission factor of grid power supply at any given time, The historical training sample set for each time period is represented as follows: ;set up Let N be a latent variable, representing a Gaussian distribution. Represents the mean. Represents variance. This indicates that historical observation data will be used. The parameter space mapped to latent variables. Indicates that given input data Later on latent variables The approximate posterior distribution is represented by a Gaussian distribution: ;set up Representing latent variables The prior distribution, Let represent a square matrix with 1s on the diagonal and 0s elsewhere, indicating that each latent variable dimension is independent of the others and has a variance of 1. express Predicted carbon emission factor for electricity supplied by the time network. Indicates latent variables Convert back to the original data space. Indicates that given latent variables Under the condition of the generation probability distribution of emission factors, For parameters The decoder neural network assumes latent variables The prior distribution is a multivariate Gaussian distribution with a mean of 0 and a covariance of the identity matrix. The decoder will take the given latent variable... Transforming back to the original data space, the formula for the latent variables becomes: ; The predicted carbon emission factor for electricity supply from the TimeNet network is: A probability distribution model for photovoltaic power output at charging stations based on weather-time joint factor graphs is established, including: constructing a weather-time feature dataset; designing the factor graph structure and establishing a joint probability model; learning and training the factor graph using historical sample datasets to determine the factor graph structure and factor node functions, and using Gibbs sampling to obtain the predicted distribution; constructing the weather-time feature dataset, including: Historical data of the area where the charging station is located was collected to construct a multidimensional input dataset in time series form, including power output datasets for several historical photovoltaic periods and weather feature datasets for several historical periods; [The remaining text appears to be a fragmented and incomplete sentence, possibly due to OCR errors. A more accurate translation would require the full context.] Representing history Temperature at different times Representing history Cloud cover for a given period of time. Representing history Relative humidity at different times Representing history Rainfall in a given period Representing history Solar irradiance for each time period, the number of time periods being: ,history The photovoltaic output data for each time period is as follows ,history The weather feature dataset for each time period is as follows: ,but: Design a factor graph structure and establish a joint probability model, including: incorporating historical data... Temperature and historical data for each period Cloud cover for a specific time period, historical data Relative humidity for each time period, historical Rainfall in each period and historical data A two-level factor graph is constructed using solar irradiance over a given time period as variables. Variable nodes include all input and output variables, and factor nodes represent local dependencies between variables. Factors include joint factors between weather variables, moderating factors between time and weather variables, mapping factors between photovoltaic output and weather variables, and smoothing factors between outputs at adjacent times. The factor graph is used to express the overall joint probability distribution. This is the partition function, used to normalize the entire distribution. Indicates the first The subset of variables involved in each factor Let the factor node function be a factor node function, and the factor node function reflects local dependencies. Then: ;set up for Predicted photovoltaic output at any given time for The random variable of photovoltaic power output at any given time. For weather-time feature datasets, This is a normalization factor for the weather-time feature dataset. For the first The factor node function for the subset of variables involved in each factor, the predicted value is expressed as: ; The constraint establishment unit is used to establish system power balance constraints, charging pile energy storage constraints, vehicle-pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints with the goal of minimizing carbon emissions within the optimization cycle. It establishes a rolling optimization model for the low-carbon operation system of charging piles that considers electric vehicle charging satisfaction. Charging pile energy storage constraints include charging and discharging power constraints and energy conservation constraints. Vehicle-pile matching constraints include: each vehicle can only be assigned one charging pile at a time, each pile can only serve one vehicle, and the vehicle can only charge after arriving. Electric vehicle charging behavior constraints include classifying electric vehicles into immediate charging and waiting-to-charge types. Immediate charging behavior is characterized by the shortest desired charging time; charging occurs immediately when a charging pile is available, and waiting in line when no pile is available. Charging power is fixed at the rated power and cannot be adjusted. Charging stops and the vehicle leaves once the desired charge level is reached. The charging target is achieved within the desired time period, and the charging power is adjustable during charging. The rolling optimization model for the low-carbon operation system of charging piles that considers electric vehicle charging satisfaction includes: setting... Indicates time period Carbon emission factors of grid power supply Indicates the power supply capacity of the charging station network. These are the weighting coefficients. Let represent the overall satisfaction of electric vehicle charging users during the optimization period. Then, the objective function of the optimization model is expressed as: Establish system power balance constraints, charging pile energy storage constraints, vehicle-charging pile matching constraints, electric vehicle charging behavior constraints, and user satisfaction constraints, including: setting... for Predicted photovoltaic output at any given time for Time of the first The energy storage discharge power of each charging pile This is a collection of all charging stations. express Time of the first The charging power of each charging station for electric vehicles. express Time of the first The energy storage charging power of each charging pile is then expressed as follows: ;set up express Time of the first The energy storage capacity of each charging station Indicates the first The maximum energy storage capacity of each charging station Indicates the first Energy storage charging efficiency of each charging pile Indicates the first The energy storage discharge efficiency of a charging pile is then expressed as the energy storage constraint of the charging pile: ;set up Indicates the first The car is Charging power at any time For the collection of all charging vehicles, Indicates electric vehicle With charging piles exist The matching relation variable at time t is a Boolean variable, when Shike Electric Vehicle At the charging station When charging, ,otherwise ; Indicates electric vehicle If the arrival time is given, then the vehicle-pile matching constraint is expressed as: ;set up: This refers to a collection of electric vehicles that can be charged immediately. Let it be a Boolean variable representing an electric vehicle. The queuing status A value of 1 indicates that the user is in a queue. A value of 0 indicates that no queue has been formed; This indicates the maximum waiting time. After the maximum waiting time is exceeded, charging will be abandoned and the vehicle will leave. Let it be a Boolean variable representing an electric vehicle. The service status, A value of 1 indicates that the charging service has been accepted. A value of 0 indicates that charging has been abandoned and the vehicle has departed; It is an infinitely large constant. Indicates the rated charging power; Indicates electric vehicle Expected charging capacity; Indicates electric vehicle The initial charge; Indicates electric vehicle Maximum battery capacity; Indicates electric vehicle If the departure time is given, then the electric vehicle charging behavior constraint is expressed as: ； set up The satisfaction reward coefficient, This represents the waiting time penalty coefficient. Let represent the penalty coefficient for users not receiving charging services. Then, the user satisfaction constraint is expressed as: ； The model solving unit is used to solve the rolling optimization model of the low-carbon operation system of charging piles that takes into account the charging satisfaction of electric vehicles, and obtain the optimal low-carbon operation strategy of charging and storage integrated charging piles under the consideration of multi-source uncertainties.

Images