Electric vehicle charging and discharging optimization method based on single vehicle level centralized EMPC strategy

Abstract

The invention discloses an electric vehicle charging and discharging optimization method based on a bicycle-level centralized EMPC strategy, and belongs to the technical field of electric vehicles. The invention aims to provide an electric vehicle charging and discharging optimization method based on a bicycle-level centralized EMPC strategy, which meets the charging demand of an individual user while ensuring the minimum power exchange between a micro-grid and a main grid through dynamic adjustment within a rolling range. The invention relates to a single-vehicle-level control target of electric vehicle charging and discharging control, which is to keep dynamic balance between charging and discharging power of each EV and wind energy and load. The invention relates to the field of EV and MG cooperative control, in particular to a single-vehicle-level CEMPC framework which is used for controlling EV charging and discharging power, is suitable for energy scheduling optimization in a V2G scene, improves system response speed and information consistency and enhances adaptive capacity and control precision of individual differences.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric vehicles. Background Art

[0002] With the intensification of global warming and the depletion of fossil fuel reserves, energy challenges have become the main concern. Hybrid renewable energy generation is becoming the key solution to efficiently meet energy demands. As a typical distributed power system, MG integrates local renewable energy and enhances power supply reliability. As a clean transportation carrier, EV has an idle time of more than 21 hours per day, and its battery energy storage potential can be converted into grid flexibility resources, providing a new path for renewable energy consumption.

[0003] V2G technology realizes auxiliary services such as frequency modulation and peak shaving through the bidirectional energy interaction between EV and the connected power grid. Its core lies in charging at a low price when the grid is redundant and sending energy back at a high price when there is a shortage, thus motivating users to participate in energy scheduling. Therefore, in order to alleviate the peak fluctuation of the power grid by integrating renewable energy into MG, it is crucial to manage the charging and discharging of a cluster of EVs.

[0004] At present, most of the existing research focuses on controlling the stable operation of MG with V2G participation. The research work focuses on designing the best scheduling strategy to enhance the power grid performance. Among them, the MPC (Model Predictive Control) method, due to its characteristics such as real-time rolling optimization, multi-constraint and multi-objective coordination, and flexibility and scalability, can effectively cope with the fluctuations of renewable energy and load uncertainty, and is widely used in the MG operation control strategy. When using MPC to formulate the control strategy, the hierarchical control architecture has received attention by decoupling functional requirements. The EV charging and discharging scheduling is divided into an upper-layer optimal scheduling layer and a lower-layer distributed control layer through the hierarchical architecture. The upper-layer optimal scheduling layer minimizes the power curve through MPC and determines the optimal operation strategy of the EV cluster; the lower-layer distributed control layer distributes the power based on the optimization results of the upper layer.

[0005] However, this hierarchical control architecture based on MPC has the following problems: 1. Inter-layer information transmission delay and inconsistency: The separation of the upper layer (global optimization) and the lower layer (single-vehicle control) in the hierarchical architecture leads to the need for cross-layer transmission of decision-making and state information. In dynamic scenarios (such as unstable distributed energy output or load fluctuation), communication delay or data update lag between layers will cause control deviation and reduce the dynamic response ability of the system.

[0006] 2. Limited individual difference adaptation ability: The access states of different EVs vary significantly. When the hierarchical control strategy only focuses on macroscopic goals at the upper layer, the lower layer can only perform local allocation and adjustment, lacking sufficient optimization freedom to handle special cases, which may lead to a decrease in resource utilization and system stability.

[0007] Generally speaking, although hierarchical control can simplify the system architecture and computational pressure in theory, there are still obvious deficiencies in real-time information interaction, vehicle collaboration, and adaptation to personalized needs. It is necessary to further improve and perfect the architecture design and implementation method to improve the efficiency and reliability of the coordinated operation of MG and EV. Summary of the Invention

[0008] The purpose of the present invention is an optimization method for electric vehicle charging and discharging based on a single-vehicle-level centralized EMPC strategy, which dynamically adjusts within a rolling range. This strategy can ensure the minimum power exchange between the microgrid and the main grid while meeting the charging needs of individual users.

[0009] The electric vehicle charging and discharging control method of the present invention The optimization problem of the single-vehicle level control target for electric vehicle charging and discharging control is defined as follows: min(f1(x)) Wherein, N current(i) represents the number of vehicles arriving at each moment; represents the charging and discharging power of each vehicle at each moment, if it is positive, it represents the vehicle is charging, if it is negative, it represents the vehicle is discharging; C load represents the difference between wind energy and load; The dynamic update formula of SOC is shown in formula (7): Wherein, η represents the charging and discharging power of the vehicle. If then η = η c ; if then η = η d ; h represents the sampling time step; The constraint condition of the EV charging power is shown in formula (8): Wherein, P rat is the rated charging and discharging power of the EV; The upper and lower limits of SOC of the EV before the last period in the rolling time range are obtained from formula (9): The upper and lower limits of the SOC of the EV in the last period of the rolling time range are obtained from Equation (10):

[0010] The beneficial effects of the present invention are as follows: 1. Improve the system response speed and information consistency. By constructing a single-layer unified optimization framework, the present invention eliminates the decision-making and state transfer links between the upper layer and the lower layer in the hierarchical control, avoiding the problems of inter-layer communication delay and information inconsistency. In dynamic operation scenarios (such as renewable energy fluctuations, load mutations, etc.), the overall control strategy can directly perform optimization and solution based on the global information of the system, thereby significantly improving the control response speed and the dynamic adaptability of the system; 2. Enhance the adaptability to individual differences and control accuracy. In the single-vehicle-level CEMPC optimization framework, the state variables and behavioral characteristics of each electric vehicle (such as access time, SOC level, battery degradation status) are explicitly incorporated into the optimization model, enabling the control strategy to fully consider individual differences and perform differential scheduling. Compared with the traditional hierarchical method that only performs static allocation in the lower layer, the overall control strategy can coordinate the behaviors of multiple vehicles during the solution stage, improve the resource utilization efficiency, and enhance the stability and robustness of the system operation. Description of the Drawings

[0011] Figure 1 is a microgrid system including wind energy, base load, and an electric vehicle cluster; Figure 2 is the main composition of the wind turbine; Figure 3 is the power change of wind energy and load energy during the daytime working hours; Figure 4 is the power change of wind energy and load energy in the night home charging scenario; Figure 5 is the comparison chart of the power tracking performance of different control strategies during the daytime working period; Figure 6 is the comparison chart of the power tracking performance of different control strategies during the typical night home charging period. Detailed Embodiments

[0012] The present invention proposes a single-layer MPC strategy for EV charging and discharging, which integrates the hierarchical structure and jointly optimizes the SOC (State of Charge) and grid load at the single-vehicle level. By dynamically adjusting within the rolling range, this strategy ensures the minimum power exchange between the microgrid and the main grid while meeting the charging needs of individual users. Compared with the traditional hierarchical strategy, this method reduces the information transfer between layers, predicts different electricity demands, and accurately matches the power.

[0013] The proposed bicycle-level CEMPC framework of the present invention mainly includes the following aspects: constructing a single-layer control framework based on economic model predictive control and solving it using an improved multi-objective particle swarm algorithm.

[0014] First, based on the hierarchical charging and discharging control strategy of MPC for cluster vehicles, the scheduling model of cluster EVs and the power allocation model for all EVs are integrated. Considering the charging demand of a single vehicle, a single-vehicle-level CEMPC control strategy is constructed. Secondly, a dynamic update formula for the charging state of the vehicle is constructed. Overall, these parts cooperate with each other to accurately control the charging power of different EVs and accurately regulate the energy fluctuation of the MG.

[0015] The implementation method of the present invention includes the following parts: The experimental part of the present invention is realized through Matlab simulation. The energy source of the MG involved is the energy provided by the wind turbine, which is consumed by the basic load and the cluster EVs. While achieving the purpose of stabilizing the energy fluctuation of the MG, the charging power of the EVs is allocated.

[0016] First, the energy data provided by the wind turbine comes from the wind resource database in the United States. The database provides the wind speed conditions generated based on the meteorological data set of the wind farm, including the cut-in wind speed, cut-out wind speed, rated wind speed, and the rated output power of the wind turbine. Finally, the output power of the wind turbine is simulated using common wind turbine types on the market. The load data comes from the official website of the New York power operator and is used for load electricity load prediction.

[0017] Secondly, the vehicle state is analyzed according to the vehicle arrival sampling situation. The states of the vehicle when accessing the power grid are mainly divided into the arrival time and the SOC. According to the analysis of the modeling results, the vehicle states can be divided into two scenarios: daytime - public charging area and nighttime - home charging station. Finally, the control strategy is simulated and analyzed according to different scenarios, and the effectiveness of the control model is verified.

[0018] The present invention conducts research on an MG system including wind energy, basic load, and cluster EVs. The MG is connected to the main power grid through a transmission line, as Figure 1 shown. By coordinating the power relationship between the cluster EVs and the wind energy and the basic load, the power exchange between the MG and the power grid is minimized, while meeting the charging needs of different EVs.

[0019] The goal of the present invention is to realize the charging and discharging control of EVs at the single-vehicle level on the basis of maintaining the stable operation of the MG, and to synergistically optimize the SOC target and the power grid load.

[0020] The specific implementation steps taken by the present invention are as follows: 1. System Modeling and Scenario Description 1.1 Wind Turbine Overview A wind turbine is an electrical device that can convert wind energy into electrical energy. It generally consists of a wind wheel, a generator, and a tower. As shown Figure 2 in the figure, its working principle is to use the wind force to drive the blades to rotate, then drive the generator to rotate, and then convert the kinetic energy generated by the rotation into electrical energy. Finally, the electrical energy is transmitted to the power grid through an inverter.

[0021] Due to the uncertainty of wind speed, the actual output power of the wind turbine also changes with time. The calculation method of its output power characteristic is as shown in formula (1): Among them, \(P_{WT}\) represents the actual output power of the wind turbine, \(v\) represents the actual wind speed obtained by the wind turbine, \(v_{ci}\) represents the cut-in wind speed, \(v_{co}\) represents the cut-out wind speed, \(v_{r}\) represents the rated wind speed, and \(P_{WT}\) rated represents the rated output power of the wind turbine.

[0022] Since the output of wind energy and the consumption of load have certain fluctuations, the simulation of the input power situation of the MG in the present invention is divided into two types: daytime - public charging area and nighttime - home charging station.

[0023] Figure 3 It shows the power change situation of wind energy and load energy during the daytime working hours. The load demand shows significant periodic fluctuations. The wind energy fluctuates greatly in the early stage, then tends to be stable, and the overall power is relatively low, only approaching the load demand in a few periods. In some periods, the power provided by the wind energy is difficult to meet the load demand, and it is necessary to optimize the dispatching strategy and use the energy stored in the EV to adjust the MG.

[0024] Figure 4 It reflects the change trend of wind energy and load energy over time at night. The wind energy power fluctuates frequently and is higher than the load demand, while the load energy demand shows relatively stable periodic fluctuations. By effectively controlling the charging and discharging of the EV, the wind energy fluctuations can be effectively smoothed and the wind energy utilization efficiency can be improved, avoiding the waste of excess wind energy.

[0025] 1.2 EV charging demand characteristics The EV charging demand characteristics mainly include the time of connecting to the power grid and the initial SOC at the arrival time, which can be described by probability distribution.

[0026] To accurately model the access behavior of electric vehicles in different scenarios, the present invention adopts a truncated normal distribution function to define the probability density of the vehicle arrival time.

[0027] In the public charging mode, electric vehicle owners usually connect the EV to the MG when they arrive at the workplace at about 8:30 and leave the workplace after work at about 17:30. The probability density distribution function is as shown in formula (2): where f(t 2c ) corresponds to the value of the probability density function of the electric vehicle arriving at the charging station at time t 2c , t 2c is the actual arrival time, and represent the expected value and standard deviation of the time to arrive at the public charging station in the public charging mode respectively, and exp(·) represents the natural exponential function, which is used to describe the exponential term of the probability density.

[0028] In the home charging mode, the electric vehicle owner returns home at about 18:00 and goes out to work at about 8:00. The arrival time t 1c of the electric vehicle follows a normal distribution, as shown in formula (3): where f(t 1c ) corresponds to the value of the probability density function of the electric vehicle arriving at the charging station at time t 1c , t 1c is the actual arrival time, and represent the expected value and standard deviation of the arrival time respectively, and exp(·) represents the natural exponential function, which is used to describe the exponential term of the probability density.

[0029] In the modeling of the EV charging demand, when the EV starts charging, the remaining energy of the EV depends on the driving distance. The daily driving distance of the EV can be modeled by a lognormal distribution, and its distribution is given by formula (4): where f d corresponds to the value of the probability density function of the daily driving distance d of the electric vehicle; d represents the driving distance of the EV on that day, with the unit of kilometer (km); μ2 represents the expected value of the variable lnd; σ2 represents the standard deviation of the variable lnd; lnd represents the natural logarithm value of the daily driving distance d, and exp(·) represents the natural exponential function, which is used to describe the exponential term of the probability density.

[0030] Based on the driving distance, the initial SOC of the EV can be calculated by formula (5): where: C EV represents the rated capacity of the EV battery, d represents the driving distance of the EV on that day, and P kM represents the energy consumption per unit distance of the EV.

[0031] 2. Electric Vehicle Charging and Discharging Control Strategy The single-vehicle level control objective for electric vehicle charging and discharging is to maintain a dynamic balance between the charging and discharging power of each EV and the wind energy and load. The optimization problem is defined as follows: Among them, N current(i) represents the number of vehicles arriving at each moment; represents the charging and discharging power of each vehicle at each moment, then represents that the vehicle is charging, then represents that the vehicle is discharging; C load represents the difference between wind energy and load.

[0032] The SOC dynamic update formula is shown in formula (7): Among them, η represents the charging and discharging power of the vehicle. If then η = η c , if then η = η d ; h represents the sampling time step.

[0033] The constraint condition of the EV charging power is shown in formula (8): Among them, P rat is the rated charging and discharging power of the EV.

[0034] The upper and lower limits of the SOC of the EV before the last period in the rolling time range are obtained from formula (9); formula (10) represents the upper and lower limits of the SOC of the EV in the last period of the rolling time range:

[0035] Simulation analysis The present invention mainly aims at a wind turbine, a basic load and 150 EVs existing in a certain area. All parameters involved in the experiment are shown in Table 1.

[0036] Table 1 Simulation-related parameters

[0037] Such as Figure 5As shown, it is a comparison chart of power tracking performance under different control strategies during a typical working period in the daytime. Curve (a) represents the target power curve, i.e., the power difference between wind energy and the load; curve (b) represents the power response under the proposed single-layer single-vehicle-level CEMPC strategy; curve (c) represents the power response under the traditional hierarchical cluster vehicle model predictive control strategy. It can be seen from the figure that the single-vehicle-level CEMPC strategy (b) can more closely follow the target power curve (a) in most periods, and its output curve is basically consistent with the wind energy fluctuation trend, showing good tracking consistency and dynamic stability. In contrast, there are obvious severe fluctuations and high-frequency oscillations in the hierarchical control strategy (c), especially at 9:00, 14:30, 16:30 and other periods, and its power response deviates significantly from the target power.

[0038] As Figure 6 shown, it is a comparison chart of power tracking performance under different control strategies during a typical home charging period at night. In the figure: Curve (a) represents the target power curve, i.e., the power difference between wind energy and the load; curve (b) represents the power response under the proposed single-layer single-vehicle-level CEMPC strategy; curve (c) represents the power response under the traditional hierarchical cluster vehicle model predictive control strategy. It can be observed from the figure that the single-vehicle-level CEMPC strategy (b) can track the target power curve (a) more smoothly and closely, with less overall volatility and a smoother change trend. In contrast, there are obvious severe fluctuations in the hierarchical control strategy (c), especially in the period from 00:00 to 04:00, where there is a large deviation from the target power, indicating its poor adaptability during periods of large wind energy fluctuations.

[0039] The differences between the two control strategies mainly stem from the fact that the overall control strategy proposed in the present invention can fully consider the state information of each electric vehicle during the optimization process and coordinate its charging and discharging behaviors in real time, thereby achieving better system-level power regulation, improving the coordination and control accuracy of the response, effectively suppressing frequent power mutations, and thus improving the dynamic stability of power exchange. In traditional hierarchical control, due to the delay in information transmission between the upper and lower layers and uneven control distribution, the response is not timely and the power deviation is large.

[0040] In summary, the proposed single-layer control architecture is significantly superior to the traditional hierarchical control strategy in terms of power tracking performance, effectively improving the smoothness and operating reliability of the microgrid system, and verifying the technical advantages of the present invention in practical applications.

[0041] Generally speaking, these advantages highlight the economic effectiveness of the single-vehicle-level CEMPC control strategy, which ensures smoother power regulation while maintaining grid stability and improves the economic performance of users participating in V2G. Therefore, the strategy proposed by the present invention has significant innovation and engineering application value in EV aggregation scheduling and MG cooperative control.

[0042] The present invention relates to the field of cooperative control of EV (Electric Vehicle) and MG (Microgrid), specifically a single-vehicle-level CEMPC (Centralized Economic Model Predictive Control) framework for controlling the charging and discharging power of EVs, which is applicable to the energy scheduling optimization in the V2G (Vehicle-to-Grid) scenario.

Claims

1. An optimization method for electric vehicle charging and discharging based on a bicycle-level centralized EMPC strategy, characterized in that: The electric vehicle charging and discharging control method is as follows: The optimization problem definition of the single-vehicle level control objective for electric vehicle charging and discharging control is as follows: min(f1(x)) Among them, N curren t(i) represents the number of vehicles arriving at each moment; represents the charging and discharging power of each vehicle at each moment, indicates that the vehicle is charging, indicates that the vehicle is discharging; C load represents the difference between wind energy and load; The SOC dynamic update formula is shown in Formula (7): Among them, η represents the charging and discharging power of the vehicle. If then η = η c , if then η = η d ; h represents the time step of sampling. The constraint condition of the EV charging power is shown in Formula (8): Among them, P rat is the rated charge and discharge power of the EV; The upper and lower limits of the SOC of the EV before the last period in the rolling time range are obtained from Formula (9): The upper and lower limits of the SOC of the EV in the last period of the rolling time range are obtained from Formula (10):

Images