Electric vehicle cluster distributed robust inertia scheduling method, system, device and medium

Abstract

The invention discloses an electric vehicle cluster distributed robust inertia scheduling method, system, device and medium, and belongs to the technical field of electric power system operation and control, and the method comprises the steps: constructing a control architecture of a double-layer time scale coupling mechanism, collecting the data of an electric vehicle user, and carrying out the control of the electric vehicle user; establishing an EV cluster probability model and opportunity constraint considering user behavior uncertainty, establishing a robust capacity boundary, establishing a V2G nonlinear depth loss model and a distributed robust inertia scheduling optimization model based on electrochemical dynamic stress, and establishing a distributed robust inertia scheduling optimization model based on a communication topological graph. And all electric vehicle aggregators interact marginal cost and power unbalance information with neighbor nodes, and an inertia support power instruction is iteratively updated by using a consistency algorithm. According to the method, the robustness and safety of the system are greatly improved, the service life of the battery of the electric vehicle is remarkably prolonged, and distributed decentralized control in a real sense is realized.

Description

A distributed robust inertia scheduling method, system, equipment, and medium for electric vehicle clusters. Technical Field

[0001] This invention relates to the field of advanced technology in power system operation and control, specifically to a distributed robust inertia scheduling method, system, equipment, and medium for electric vehicle clusters. Background Technology

[0002] Traditional power system frequency control relies on the physical characteristics of synchronous generators. When system load suddenly increases or generator tripping leads to power deficit, the rotor speed of the synchronous generator decreases due to the release of kinetic energy, thus suppressing the rate of change of system frequency in the first few seconds of the disturbance and buying time for primary frequency regulation. However, in low-inertia systems with a high proportion of renewable energy integration, this natural "buffer" is significantly weakened. The inertia support of distributed resources such as photovoltaics, energy storage, and electric vehicles exhibits "diversified and discrete" characteristics, resulting in a significant decrease in the system's frequency disturbance immunity. Once a large power deficit occurs, the frequency can easily drop below the safety threshold in a very short time, triggering the activation of low-frequency load shedding devices and even causing large-scale power outages.

[0003] To address the challenge of low inertia, exploring flexible demand-side resources to provide "virtual inertia" has become a research hotspot in both academia and industry. Electric vehicles, as large-scale, widely distributed mobile energy storage units with rapid response capabilities, are considered ideal frequency regulation resources. Through vehicle-to-grid (V2G) technology, EVs can not only function as controllable loads but also feed power back into the grid, simulating the inertial characteristics of synchronous generators.

[0004] Despite the theoretical foundation laid by the aforementioned research, inertia scheduling for large-scale EV clusters still faces three core challenges in practical engineering applications, which have not yet been effectively addressed in existing technical solutions: First, the fundamental contradiction between deterministic modeling and the uncertainty of user behavior. Existing frequency response models and control strategies are generally based on the "deterministic" assumption, which assumes that at the scheduling moment, the electric vehicles connected to the charging piles are stably online, and their adjustable capacity is a known and definite quantity. The model often directly calculates the inertia response power based on the current connection state. However, this assumption deviates significantly from the essential attributes of EVs as a means of transportation. User behavior is highly random and time-varying. Arrival time, departure time, initial state of charge, and temporary usage demand all follow a certain probability distribution, rather than deterministic values. In actual operation, once a large number of users leave the network early due to sudden needs, or refuse to respond to V2G commands due to concerns about battery life, the inertia support power calculated based on the deterministic model will instantly fail. This "power deficit" is fatal in low-inertia systems, potentially causing frequency drops far exceeding expectations and triggering a chain reaction of failures. Moving from "determinism" to "uncertainty," and introducing probabilistic models to describe the availability of EV clusters, is a key step in improving the practicality of scheduling strategies.

[0005] Second, there is a disconnect between linear cost models and the nonlinear deep loss of batteries. The power battery is the most expensive core component of an electric vehicle, and users are extremely sensitive to battery life loss caused by V2G. Existing scheduling strategies typically use simplified linear cost models or quadratic function models to describe regulation costs, only considering the impact of energy throughput on lifespan. However, the aging mechanism of batteries is highly nonlinear. Some scholars have proposed the concept of "inertia support stress" for stationary energy storage, pointing out that the short-term, high-frequency, high-rate current surges accompanying the inertia response process can cause irreversible damage to the internal electrochemical structure of the battery. This damage is in a complex coupling relationship with the current SOC level, ambient temperature, and charge / discharge rate. For example, high-power V2G under extremely high or low SOC conditions will result in an exponential increase in lifespan loss. Existing EV scheduling methods often ignore this crucial factor. Without an accurate V2G deep loss model, scheduling strategies may overdraw user battery life for the benefit of the grid, or forcibly activate the system when the battery is in a vulnerable state, leading to reduced user participation and even battery safety accidents.

[0006] Third, there are inherent vulnerabilities in centralized architectures and a lack of robustness in distributed algorithms. Faced with a massive number of dispersed electric vehicles, traditional centralized scheduling not only suffers from communication bandwidth bottlenecks, the curse of computational dimensionality, and single-point-of-failure risks, but also poses serious privacy risks due to the centralized collection of large-scale data. Although distributed algorithms offer a path to address these issues, existing research is largely limited to deterministic optimization. With the introduction of user behavior uncertainty, existing distributed iterative mechanisms struggle to balance algorithm convergence with grid security constraints under the "worst-case scenario." Existing literature has yet to provide a systematic solution to this critical challenge of "distributed robustness."

[0007] In summary, there is an urgent need to develop a novel inertia scheduling method that can simultaneously address user behavior uncertainty, battery nonlinearity and deep loss, and the robustness of the computing architecture. Summary of the Invention

[0008] In view of the above-mentioned problems, the present invention is proposed.

[0009] Therefore, this invention aims to address the frequency security risks caused by random fluctuations in the available capacity of EV clusters through probabilistic modeling and robust optimization techniques; to solve the nonlinear damage to battery life caused by V2G scheduling through a refined electrochemical stress model; and to achieve collaborative optimization and privacy protection of large-scale distributed resources through a distributed consensus architecture.

[0010] To address the aforementioned technical problems, this invention provides the following technical solution: a distributed robust inertia scheduling method for electric vehicle clusters, comprising: constructing a control architecture with a two-layer time-scale coupling mechanism; collecting data from electric vehicle users; constructing an EV cluster probability model and chance constraints that take into account the uncertainty of user behavior; constructing a robust capacity boundary based on the random distribution of available capacity of the electric vehicle cluster using chance-constrained programming theory; establishing a V2G nonlinear deep loss model based on electrochemical dynamic stress; and constructing a distributed robust inertia scheduling optimization model; based on the communication topology graph, all electric vehicle aggregators interact with neighboring nodes to exchange marginal cost and power imbalance information, and iteratively updating the inertia support power command using a consensus algorithm.

[0011] As a preferred embodiment of the distributed robust inertia scheduling method for electric vehicle clusters described in this invention, the control architecture for constructing a two-layer time-scale coupling mechanism includes: a first layer being a millisecond-level physical response layer, where each electric vehicle converter performs local droop control and responds to frequency deviations; and a second-level economic optimization layer, which executes a consensus algorithm to dynamically adjust the power reference point and droop coefficient of the first-layer controller.

[0012] As a preferred embodiment of the distributed robust inertia scheduling method for electric vehicle clusters described in this invention, the EV cluster probability model includes: collecting historical travel data of electric vehicle users and establishing a joint probability density function of user arrival time, departure time, and expected state of charge; wherein, for any electric vehicle managed by the aggregator, availability is determined by the user's arrival time and departure time, and a truncated normal distribution is fitted; simultaneously, EVs participate in inertia response to meet the user's travel energy needs.

[0013] As a preferred embodiment of the distributed robust inertia scheduling method for electric vehicle clusters described in this invention, the opportunity constraint includes, for the aggregator, requiring that the probability that the total inertia power of the corresponding actual scheduling does not exceed the physical limit is not lower than the confidence level; using the Gaussian approximation method, assuming that the total maximum available power of the cluster follows a normal distribution, introducing a confidence level parameter, the random distribution of the available capacity of the electric vehicle cluster is transformed into a deterministic robust upper and lower bound.

[0014] As a preferred embodiment of the distributed robust inertia scheduling method for electric vehicle clusters described in this invention, the V2G nonlinear deep loss model includes: constructing a V2G deep loss model suitable for EVs based on energy storage stress theory; and defining dynamic stress coefficients. Functions belonging to real-time operating conditions: in, Power is supported by inertia. For the battery's rated capacity, The state of charge at time t The system frequency change rate; , , For different weighting coefficients; m is the multiplier effect index ( ); It is a monotonically increasing function; based on the dynamic stress coefficient, the instantaneous operating cost function of the i-th EV aggregator is constructed. : in, For user expectations, These are quadratic cost coefficients, used to ensure the strong convexity of the cost function. This serves as the basic cost coefficient for electric vehicles to participate in regulation. The linear cost coefficient for instantaneous operation. These are the weighting coefficients for the SOC maintenance term based on the potential energy field, used to pull the charge back to the expected value. This represents the real-time state of charge of the i-th electric vehicle aggregator.

[0015] As a preferred embodiment of the distributed robust inertia scheduling method for electric vehicle clusters described in this invention, the construction of the distributed robust inertia scheduling optimization model includes setting the overall objective as minimizing the generalized operating cost of the entire network while satisfying frequency security and user needs. : The set of constraints includes system power balance constraints, frequency security constraints, and distributed robust capacity constraints; the system power balance constraints are as follows: The frequency security constraint is: The distributed robust capacity constraint is: in, For the total power of the electric vehicle cluster, Reduce capacity for electric vehicles To increase capacity for electric vehicles, For the total scheduling cycle, For time variables, Let i be a set of electric vehicle clusters, and j, k be variable indices. Let i be the operating cost of the i-th electric vehicle cluster. Let be the power of the i-th cluster at time t. A collection of energy storage systems, Let the operating cost of the j-th energy storage system be... Let be the power of the j-th energy storage system at time t. For distributed photovoltaic arrays, Let the operating cost be the cost of the k-th photovoltaic unit. Let be the power of the k-th photovoltaic unit at time t. Let be the power of electric vehicle cluster i at time t. Let j be the power of the energy storage system at time t. Let be the power of photovoltaic unit k at time t. Total load power, Where D is the power generation capacity of a conventional generating unit, and D is the damping coefficient. For system frequency deviation, The system's equivalent inertia constant, The maximum safe threshold for the system frequency change rate. Let be the lower bound of the robust available capacity of the i-th cluster at time t. Let be the upper bound of the robust available capacity of the i-th cluster at time t.

[0016] As a preferred embodiment of the distributed robust inertia scheduling method for electric vehicle clusters described in this invention, the iterative update of inertia support power commands using a consensus algorithm includes designing a discrete consensus algorithm based on projection operators; according to the Lagrange multiplier method, when there are no constraints, the optimal solution satisfies the marginal cost of all participating units. equal: in, For cost function, Let this be the power adjustment amount; Let $\mathbf{k}$ be the marginal cost variable at the k-th iteration. To estimate the local power imbalance, the state variables are initialized, and each node sets its parameters based on its local load and initial output. and Node i communicates with its neighboring nodes in the communication topology. send and receive It performs neighbor information exchange, and after the exchange, it performs a consistency update, updating the marginal cost variable to tend towards global consistency: in, The elements are the random weight matrix. To converge the step size, For the set of neighboring nodes, The marginal cost of the neighboring nodes; based on the updated... Calculate the unconstrained optimal power : Applying projection operators Will Mapping to robust feasible region Internally; it utilizes dynamic consistency to track global power deficits and updates power imbalances. At that time, the algorithm converges, and all EV aggregators follow the final... Commands control the operation of the charging station.

[0017] Another objective of this invention is to provide a distributed robust inertia scheduling system for electric vehicle clusters.

[0018] To address the aforementioned technical problems, this invention provides the following technical solution: a distributed robust inertia scheduling system for electric vehicle clusters, comprising: a control module, which constructs a control architecture with a two-layer time-scale coupling mechanism, collects data from electric vehicle users, and constructs an EV cluster probability model and chance constraints that take into account the uncertainty of user behavior; a boundary determination module, which constructs a robust capacity boundary based on chance-constrained programming theory for the random distribution of available capacity of the electric vehicle cluster; a loss calculation module, which establishes a V2G nonlinear deep loss model based on electrochemical dynamic stress and constructs a distributed robust inertia scheduling optimization model; and an update module, which, based on the communication topology graph, allows all electric vehicle aggregators to interact with neighboring nodes regarding marginal cost and power imbalance information, and uses a consensus algorithm to iteratively update the inertia support power command.

[0019] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the distributed robust inertia scheduling method for electric vehicle clusters.

[0020] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of the distributed robust inertia scheduling method for electric vehicle clusters.

[0021] The beneficial effects of this invention are as follows: This invention greatly improves the robustness and security of the system: By shifting from a "deterministic" to an "uncertain" mindset, and utilizing probabilistic models and chance constraints, the system can tolerate a certain proportion of random user behavior. Even in severe scenarios involving large-scale unexpected user disconnection, the reserved "robust margin" can ensure that the power grid frequency does not collapse, filling the gap in existing scheduling strategies for handling uncertainty.

[0022] This significantly extends the lifespan of electric vehicle batteries. By introducing a refined nonlinear V2G loss model, the algorithm can "sense" the battery's health status and real-time operating conditions. When the frequency fluctuates drastically or the battery is in a vulnerable state, the algorithm automatically increases the marginal cost of that node, thereby reducing its usage and avoiding irreversible damage to the battery caused by coarse scheduling, thus improving the long-term economic benefits of user participation in V2G.

[0023] This system achieves truly distributed decentralized control. The proposed robust consensus algorithm does not require a central controller to collect user privacy data; it only exchanges anonymized cost gradient information. This not only protects user privacy but also eliminates the risk of single points of failure, supports massive EV access and plug-and-play functionality, and has extremely high engineering application value. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 is a flowchart of a distributed robust inertia scheduling method for electric vehicle clusters provided by an embodiment of the present invention. Detailed Implementation

[0026] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0027] Example 1, referring to Figure 1, is an embodiment of the present invention. This embodiment provides a distributed robust inertia scheduling method for electric vehicle clusters, including: S100, constructing a control architecture with a two-layer time-scale coupling mechanism, collecting data from electric vehicle users, and constructing an EV cluster probability model and chance constraints that take into account the uncertainty of user behavior; S200, constructing a robust capacity boundary for the random distribution of available capacity of the electric vehicle cluster based on chance-constrained programming theory; S300, establishing a V2G nonlinear deep loss model based on electrochemical dynamic stress, and constructing a distributed robust inertia scheduling optimization model; S400, based on the communication topology graph, all electric vehicle aggregators and neighboring nodes interact with marginal cost and power imbalance information, and iteratively updating the inertia support power command using a consensus algorithm; It should be noted that the existing technology suffers from the fundamental contradiction between deterministic modeling and user behavior uncertainty, the disconnect between the linear cost model and the nonlinear deep loss of the battery, the inherent vulnerability of the centralized architecture, and the lack of robustness of the distributed algorithm.

[0028] Therefore, addressing the aforementioned problems, through steps S100-S400, this invention relates to a scheduling method for large-scale distributed electric vehicle clusters. This method considers the complex nonlinear electrochemical loss characteristics of power batteries and leverages a distributed computing architecture to achieve robust virtual inertia support in environments with highly random user travel behavior and uncertainties such as communication network delays or packet loss. The invention is applicable to electric vehicle aggregators, virtual power plant control centers, and microgrid energy management systems on the distribution network side. It aims to solve the frequency stability problem of low-inertia power systems under high-power disturbances, while simultaneously ensuring the travel needs of electric vehicle users and optimizing the economic efficiency throughout the battery's lifecycle.

[0029] Example 2, referring to Figure 1, is an embodiment of the present invention. This embodiment provides a distributed robust inertia scheduling method for electric vehicle clusters, including: In this embodiment, in S100, a control architecture with a two-layer time-scale coupling mechanism is constructed, data of electric vehicle users is collected, and an EV cluster probability model and chance constraints considering user behavior uncertainty are constructed, including the following steps S101-S102: S101, The present invention is applicable to distribution networks or microgrid systems containing multiple types of distributed resources. The system physical layer includes distributed photovoltaic (PV), battery energy storage systems (BESS), electric vehicle charging stations (EVCS), and local loads. The information layer is based on a communication topology graph described by graph theory. , where V is the set of nodes and E is the set of communication links.

[0030] The control architecture employs a two-layer time-scale coupling mechanism, including a millisecond-level physical response layer and a second-level economic optimization layer. The millisecond-level physical response layer is the fast path of the physical layer: each EV converter executes local droop control based on the real-time frequency deviation. and rate of change of frequency Output basic inertia power : in, The droop coefficient is... Here, t represents the virtual inertia coefficient, and t is the time variable.

[0031] The second-level economic optimization layer is the slow channel of the economic layer (seconds / minutes): a distributed robust consensus algorithm runs periodically, dynamically optimizing and adjusting the base power of each node based on the global supply and demand status and battery health. and control coefficient To achieve optimal cost and robust security across the entire network.

[0032] S102. EV Cluster Probabilistic Model Considering User Behavior Uncertainty S1021. Stochastic Availability Description of Individual EVs For the k-th electric vehicle managed by the i-th aggregator, its availability is determined by the user's arrival time. and departure time Therefore, this invention introduces a truncated normal distribution to fit these random variables: in, The standard deviation of arrival time, The expected arrival time, / To truncate the upper / lower time bounds of the normal distribution; similarly, departure time. It also follows the corresponding probability distribution.

[0033] Define binary state variables Indicates the vehicle's online status: in, / Let be the arrival / departure time of vehicle k; and at any scheduling time t, the total number of vehicles online for aggregator i. It is a random variable; according to the central limit theorem, when the cluster size is large, It approximately follows a normal distribution ,in, The expected value of the total number of online vehicles. The standard deviation of the total number of online vehicles.

[0034] S1022, Energy Boundary and SOC Evolution: EV participation in inertial response must meet the user's travel energy requirements, i.e., satisfy the following upon departure: The battery energy state evolution equation is: in, In a charged state, For the user's desired off-grid state of charge, Let t be the battery energy state. For charging efficiency, For charging power, For discharge efficiency, This is the discharge power; due to It is random, remaining stay time It is also random; to ensure that the battery level is sufficient when the user leaves, the maximum discharge power at time t is determined. This means that the maximum V2G inertia support capability is strictly limited: in, For the remaining stay, This represents the energy demand for travel when the user leaves; due to the denominator It has uncertainty, leading to It becomes a random variable.

[0035] In this embodiment of the invention, S200 constructs a robust capacity boundary for the random distribution of available capacity of electric vehicle clusters based on chance-constrained programming theory, including the following steps S201: S201, the construction and transformation of chance constraints to ensure the robustness of scheduling, this invention introduces chance constraints; for aggregator i, the total inertia power of its actual scheduling is required. The probability of not exceeding the physical limit is not less than the confidence level. : in, Represents an uncertain parameter vector. A collection of electric vehicles managed by an aggregator. Let be a probability function. Let be the actual total inertia power of aggregator j.

[0036] Using the Gaussian approximation method, let the total maximum available power of the cluster be... Follows a normal distribution The above opportunity constraints can be transformed into deterministic linear inequality constraints: make This value is the upper bound of the robust capacity.

[0037] in, For the time calculated based on the probability model The mathematical expectation of the maximum available power of the cluster. Standard deviation The preset upper limit of the allowable probability of default, with a confidence level of ; It is the inverse cumulative distribution function of the standard normal distribution.

[0038] The inequality constraint indicates that, in order to mitigate the risk of random user churn, the dispatch center believes that available capacity must be reduced by a safety margin from the average capacity; uncertainty. The larger, or the greater the safety requirements The higher the value, the greater the margin, and the more conservative and secure the system scheduling becomes.

[0039] In an embodiment of the present invention, a V2G nonlinear deep loss model based on electrochemical dynamic stress is established in S300, and a distributed robust inertia scheduling optimization model is constructed, including the following steps S301-S302: S301, V2G nonlinear deep loss model based on electrochemical stress. In order to overcome the defects of the linear cost model in the prior art, the present invention constructs a V2G deep loss model suitable for EV based on the energy storage stress theory.

[0040] First, the dynamic inertia stress coefficient is defined: battery aging mainly consists of cycle aging and calendar aging. Under high-frequency power fluctuations in inertia response, the growth of the solid electrolyte interface film on the electrode surface and the mechanical fatigue of active material particles are the main failure mechanisms; the dynamic stress coefficient is defined. It is a function of real-time operating conditions: in, The rate of change of frequency, Let represent the state of charge of the i-th battery; the specific expression is constructed as follows: in, The inertia support power at time t For the battery's rated capacity, For real-time state of charge, The system frequency change rate; , , Here, m is the weighting coefficient, and m is the multiplier effect index. ); It is a monotonically increasing function used to characterize the amplification effect of the severity of power grid frequency fluctuations on the unit regulation cost of batteries.

[0041] Among them, the magnification stress term Representing the charge / discharge rate, the exponent m is usually taken as 1.5-2.0, indicating that the loss increases superlinearly with power, which reflects the damage to the SEI film caused by the heat generation and polarization effect of the large current.

[0042] State stress term Using an exponential function, when the SOC deviates from the 50% central interval and approaches 0% or 100%, The value rises sharply. This forces the algorithm to avoid calling V2G when the battery is over-saturated or under-saturated, preventing lithium plating or over-discharge.

[0043] Frequency coupling stress term It is a monotonically increasing function of the rate of change of frequency, when the power grid disturbance is severe ( When (large), An increase means that the adjustment cost per unit of power is higher, which reflects a strategy to protect the battery under extreme operating conditions.

[0044] Secondly, based on the aforementioned stress coefficients, the instantaneous operating cost function of the i-th EV aggregator is constructed. : in, For user expectations, These are quadratic cost coefficients, used to ensure the strong convexity of the cost function. This serves as the basic cost coefficient for electric vehicles to participate in regulation. The linear cost coefficient for instantaneous operation. These are the weighting coefficients for the SOC maintenance term based on the potential energy field, used to pull the charge back to the expected value. This represents the real-time state of charge of the i-th electric vehicle aggregator.

[0045] This function is strongly convex, guaranteeing the existence and uniqueness of the solution to the optimization problem. The third term is a potential energy field penalty term, used to pull the state of charge (SOC) back to the user's expected value. This prevents V2G from causing the battery to run out.

[0046] The third term, the potential energy field penalty term, is a SOC maintenance term based on the potential energy field. This term serves as a penalty factor in the objective function, forcing the electric vehicle to consistently approach the user's desired energy state while providing inertia support. This is to prevent excessive regulation from causing the battery to run out and making it unusable for travel.

[0047] S302. Solving the distributed robust inertia scheduling optimization problem, firstly, setting a global optimization objective: the overall system objective is to minimize the generalized operating cost of the entire network while satisfying frequency security and user requirements. : The set of constraints includes system power balance constraints, frequency security constraints, and distributed robust capacity constraints; system power balance constraints (including inertia requirements): Frequency security constraints: Distributed robust capacity constraints: in, For the total power of the electric vehicle cluster, Reduce capacity for electric vehicles To increase capacity for electric vehicles, For the total scheduling cycle, For time variables, Let i be a set of electric vehicle clusters, and j, k be variable indices. Let i be the operating cost of the i-th electric vehicle cluster. Let be the power of the i-th cluster at time t. A collection of energy storage systems, Let the operating cost of the j-th energy storage system be... Let be the power of the j-th energy storage system at time t. For distributed photovoltaic arrays, Let the operating cost be the cost of the k-th photovoltaic unit. Let be the power of the k-th photovoltaic unit at time t. Let be the power of electric vehicle cluster i at time t. Let j be the power of the energy storage system at time t. Let be the power of photovoltaic unit k at time t. Total load power, Where D is the power generation capacity of a conventional generating unit, and D is the damping coefficient. For system frequency deviation, The system's equivalent inertia constant, The maximum safe threshold for the system frequency change rate. Let be the lower bound of the robust available capacity of the i-th cluster at time t. Let be the upper bound of the robust available capacity of the i-th cluster at time t.

[0048] In an embodiment of the present invention, in step S400, based on the communication topology graph, all electric vehicle aggregators interact with their neighboring nodes to exchange marginal cost and power imbalance information, and iteratively update the inertia support power command using a consensus algorithm, including the following steps S401-S402: S401, according to the Lagrange multiplier method, when there are no constraints, the optimal solution satisfies the marginal cost of all participating units. (Marginal Cost) equal: in, For cost function, This is a power adjustment amount; the present invention extends it to scenarios with nonlinear losses and probabilistic constraints.

[0049] set up Let be the marginal cost variable of node i in the k-th iteration. This is an estimate of the local power imbalance.

[0050] S4011. Initialize the state variables. Each node sets the state variables according to its local load and initial output. and .

[0051] S4012, Neighbor Information Interaction: Node i communicates with neighboring nodes in the communication topology. send and receive .

[0052] S4013, Consistent Update: Update the marginal cost variable to make it globally consistent. in, For the set of neighboring nodes, The marginal cost of the neighboring nodes; For marginal cost, This is the power imbalance quantity. The elements are the random weight matrix. To achieve convergence, this step not only achieves cost equilibrium but also introduces a supply-demand imbalance. As a feedback correction item.

[0053] S4014, Projection-based power calculation: based on the updated... Calculate the unconstrained optimal power : Applying projection operators Will Mapping to robust feasible region Inside: By truncation, it is ensured that no node's output command will exceed its probabilistic safety boundary, even during iteration.

[0054] S4015, Power Imbalance Update Dynamic consistency tracking of global power deficit is used to ensure that the system supply and demand are balanced when the algorithm converges.

[0055] S4016. Convergence Judgment and Execution: When At that time, the algorithm converges. Each EV aggregator follows the final... Commands control the operation of the charging station.

[0056] Example 3 is an embodiment of the present invention, illustrating a distributed robust inertia scheduling method for electric vehicle clusters. It should be noted that the technical solution of a distributed robust inertia scheduling system for electric vehicle clusters and the technical solution of the distributed robust inertia scheduling method for electric vehicle clusters described above belong to the same concept. Details not described in detail in the technical solution of the distributed robust inertia scheduling system for electric vehicle clusters in this embodiment can be found in the description of the technical solution of the distributed robust inertia scheduling method for electric vehicle clusters described above.

[0057] This embodiment provides a distributed robust inertia scheduling system for electric vehicle clusters, comprising: a control module, which constructs a control architecture with a two-layer time-scale coupling mechanism, collects data from electric vehicle users, and constructs an EV cluster probability model and chance constraints that take into account the uncertainty of user behavior; a boundary determination module, which constructs a robust capacity boundary based on the random distribution of available capacity of the electric vehicle cluster according to chance-constrained programming theory; a loss calculation module, which establishes a V2G nonlinear deep loss model based on electrochemical dynamic stress and constructs a distributed robust inertia scheduling optimization model; and an update module, which, based on the communication topology graph, allows all electric vehicle aggregators to interact with neighboring nodes on marginal cost and power imbalance information, and uses a consensus algorithm to iteratively update the inertia support power command.

[0058] This embodiment also provides an electronic device applicable to a distributed robust inertia scheduling method for electric vehicle clusters, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the distributed robust inertia scheduling method for electric vehicle clusters as proposed in the above embodiment.

[0059] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements a distributed robust inertia scheduling method for electric vehicle clusters as proposed in the above embodiment.

[0060] The storage medium proposed in this embodiment and the method for implementing a distributed robust inertia scheduling method for electric vehicle clusters proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0061] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0062] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A distributed robust inertia scheduling method for electric vehicle clusters, characterized in that: This includes constructing a control architecture with a two-layer time-scale coupling mechanism, collecting data from electric vehicle users, and constructing an EV cluster probability model and opportunity constraints that take into account the uncertainty of user behavior. Based on the chance-constrained programming theory, a robust capacity boundary is constructed for the stochastic distribution of available capacity of electric vehicle clusters; a V2G nonlinear deep loss model based on electrochemical dynamic stress is established, and a distributed robust inertia scheduling optimization model is constructed. Based on the communication topology graph, all electric vehicle aggregators interact with neighboring nodes to exchange marginal cost and power imbalance information, and use a consensus algorithm to iteratively update the inertia support power command.

2. The distributed robust inertia scheduling method for electric vehicle clusters as described in claim 1, characterized in that: The control architecture that constructs a two-layer time-scale coupling mechanism includes a first layer that is a millisecond-level physical response layer, where each electric vehicle converter performs local droop control and responds to frequency deviations; and a second-level economic optimization layer that executes a consensus algorithm to dynamically adjust the power reference point and droop coefficient of the first-layer controller.

3. The distributed robust inertia scheduling method for electric vehicle clusters as described in claim 2, characterized in that: The EV cluster probability model includes collecting historical travel data of electric vehicle users and establishing a joint probability density function of user arrival time, departure time, and expected state of charge. For any electric vehicle managed by the aggregator, availability is determined by the user's arrival time and departure time, and a truncated normal distribution is fitted. At the same time, the EV participates in inertial response to meet the user's travel energy needs.

4. The distributed robust inertia scheduling method for electric vehicle clusters as described in claim 3, characterized in that: The opportunity constraints include, for aggregators, requiring that the probability that the total inertial power of the actual scheduling does not exceed the physical limit is not lower than the confidence level; using the Gaussian approximation method, assuming that the total maximum available power of the cluster follows a normal distribution, introducing a confidence level parameter, the random distribution of the available capacity of the electric vehicle cluster is transformed into deterministic robust upper and lower bounds.

5. The distributed robust inertia scheduling method for electric vehicle clusters as described in claim 4, characterized in that: The V2G nonlinear deep loss model includes, based on energy storage stress theory, constructing a V2G deep loss model suitable for EVs; and defining dynamic stress coefficients. Functions belonging to real-time operating conditions: in, Power is supported by inertia. For the battery's rated capacity, The state of charge at time t The system frequency change rate; 、 、 For different weighting coefficients; m is the multiplier effect index ( ）； It is a monotonically increasing function; based on the dynamic stress coefficient, the instantaneous operating cost function of the i-th EV aggregator is constructed. ： in, For user expectations, These are quadratic cost coefficients, used to ensure the strong convexity of the cost function. This serves as the basic cost coefficient for electric vehicles to participate in regulation. The linear cost coefficient for instantaneous operation. These are the weighting coefficients for the SOC maintenance term based on the potential energy field, used to pull the charge back to the expected value. This represents the real-time state of charge of the i-th electric vehicle aggregator.

6. The distributed robust inertia scheduling method for electric vehicle clusters as described in claim 5, characterized in that: The construction of the distributed robust inertia scheduling optimization model includes setting the overall objective as minimizing the generalized operating cost of the entire network while satisfying frequency security and user needs. ： The set of constraints includes system power balance constraints, frequency security constraints, and distributed robust capacity constraints; the system power balance constraints are as follows: The frequency security constraint is: The distributed robust capacity constraint is: in, The total power of the electric vehicle cluster Reduce capacity for electric vehicles To increase capacity for electric vehicles, For the total scheduling cycle, For time variables, Let i be a set of electric vehicle clusters, and j, k be variable indices. Let i be the operating cost of the i-th electric vehicle cluster. Let be the power of the i-th cluster at time t. A collection of energy storage systems, Let the operating cost of the j-th energy storage system be... Let be the power of the j-th energy storage system at time t. For distributed photovoltaic arrays, Let the operating cost be the cost of the k-th photovoltaic unit. Let be the power of the k-th photovoltaic unit at time t. Let be the power of electric vehicle cluster i at time t. Let j be the power of the energy storage system at time t. Let be the power of photovoltaic unit k at time t. Total load power, Where D is the power generation capacity of a conventional generating unit, and D is the damping coefficient. For system frequency deviation, The system's equivalent inertia constant, The maximum safe threshold for the system frequency change rate. Let be the lower bound of the robust available capacity of the i-th cluster at time t. Let be the upper bound of the robust available capacity of the i-th cluster at time t.

7. The distributed robust inertia scheduling method for electric vehicle clusters as described in claim 6, characterized in that: The iterative update of the inertia support power command using a consensus algorithm includes designing a discrete consensus algorithm based on projection operators; according to the Lagrange multiplier method, when there are no constraints, the optimal solution satisfies the marginal cost of all participating units. equal: in, For cost function, Let this be the power adjustment amount; Let $\mathbf{k}$ be the marginal cost variable at the k-th iteration. To estimate the local power imbalance, the state variables are initialized, and each node sets its parameters based on its local load and initial output. and Node i communicates with its neighboring nodes in the communication topology. send and receive It performs neighbor information exchange, and after the exchange, it performs a consistency update, updating the marginal cost variable to tend towards global consistency: in, The elements are the random weight matrix. To converge the step size, For the set of neighboring nodes, The marginal cost of the neighboring nodes; based on the updated... Calculate the unconstrained optimal power ： Applying projection operators Will Mapping to robust feasible region Internally; it utilizes dynamic consistency to track global power deficits and updates power imbalances. At that time, the algorithm converges, and all EV aggregators follow the final... Commands control the operation of the charging pile.

8. A distributed robust inertia scheduling system for electric vehicle clusters, employing the distributed robust inertia scheduling method for electric vehicle clusters as described in any one of claims 1 to 7, characterized in that, include: The control module constructs a control architecture with a two-layer time-scale coupling mechanism, collects data from electric vehicle users, and builds an EV cluster probability model and chance constraints that take into account the uncertainty of user behavior. The boundary determination module constructs a robust capacity boundary for the random distribution of available capacity of the electric vehicle cluster based on chance-constrained programming theory. The loss calculation module establishes a V2G nonlinear deep loss model based on electrochemical dynamic stress and constructs a distributed robust inertia scheduling optimization model. The update module, based on the communication topology graph, allows all electric vehicle aggregators to interact with neighboring nodes to exchange marginal cost and power imbalance information, and uses a consensus algorithm to iteratively update the inertia support power command.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the distributed robust inertia scheduling method for electric vehicle clusters as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the distributed robust inertia scheduling method for electric vehicle clusters as described in any one of claims 1 to 7.

Images