Intelligent charging management method and system based on cooperative control of charging pile cluster

Abstract

The invention relates to an intelligent charging management method and system based on cooperative control of a charging pile cluster, a charging request initiated by a user is regarded as a disturbance factor, the charging request moves in a dynamic mode in the charging pile cluster, and aggregation and decoupling of the request between nodes are realized through a response threshold value set by the charging pile nodes; forming a decentralized regulation and control starting point based on disturbance response; when a service saturation trend of a certain charging pile node appears, the node issues a task release statement, and peripheral idle nodes autonomously select to receive a migration task according to a local resource state; all the scheduling execution instructions form asynchronous and adjustable scheduling chain rhythms in the cluster through delay or advanced execution windows calculated by the charging pile nodes based on a local operation state; and each node evaluates energy potential energy according to the renewable energy access rate, the load redundancy degree and the idle period, releases node energy potential information through a user side interface, and guides a user to adjust reservation behaviors according to the situation.

Description

TECHNICAL FIELD

[0001] The present application relates to a smart charging management method and system for charging piles, in particular to a smart charging management method and system based on charging pile cluster collaborative control. BACKGROUND

[0002] In combination with the technical content disclosed in the patent document CN117858084A "A management method and device for group-managed group-controlled charging piles", there are still a series of deficiencies and drawbacks in the current smart charging management method based on charging pile cluster collaborative control, as follows: The patent scheme emphasizes that the head charging pile serves as the core node of group management, and the AMF network element only performs primary authentication process on the head pile, thereby reducing primary authentication overhead, reducing communication redundancy and improving efficiency in the cluster registration process. However, this structural head-slave architecture mode implies a number of problems. First, the centralization of core control introduces single-point failure risk. Once the head charging pile fails, communication is interrupted, or the load is too heavy to perform primary authentication and subordinate management responsibilities, the registration, scheduling and collaboration of the entire cluster will be paralyzed, lacking backup mechanisms and fault-tolerant self-recovery design. Secondly, this design assumes that the charging pile topology structure is relatively static, the head pile-slave pile relationship is explicitly defined in advance, and the pre-configuration is completed before registration. However, in the actual smart charging environment, nodes are frequently online and offline, geographical location changes greatly, vehicle migration frequency is high, and the status, energy consumption, load, availability of charging piles continue to evolve dynamically. If relying on a static head pile structure, not only is the maintenance complexity high in actual deployment, but also it is easy to cause scheduling imbalance. For example, when multiple user paths are concentrated to a high-energy head pile, if there is no adaptive dispersion mechanism, local congestion or resource depletion is easy to occur. In addition, from the technical details, the scheme mainly solves the simplification of the network layer registration and authentication process, and has not modeled the user charging request as a disturbance factor. It also does not consider how to realize true decentralization control according to the load dynamic walk request between nodes. The group management only stays in the logical division of the registration and access process, and does not have the substantive autonomous cooperation ability. The intelligent charging system needs to respond to multi-factor joint decision-making such as task congestion, user preference, energy storage state, power grid fluctuation, and path cost in real time. The existing method does not establish a multi-dimensional potential evaluation function, nor does it provide a user-side visual feedback mechanism and guidance strategy. It also lacks a dynamic learning and scheduling optimization mechanism for historical deviation behavior. For example, the scheme does not propose the influence calculation of the energy storage health state on the scheduling participation degree, nor does it distinguish the reduction of energy availability caused by electrochemical aging. Especially in the face of large-scale heterogeneous nodes, the scheme lacks node autonomous beat adjustment and local beat mutual exclusion window construction mechanism, and cannot suppress the instantaneous peak load or scheduling jitter that may be formed in the task release process. Furthermore, the scheme does not propose a potential density map construction method, which cannot identify the potential over-scheduling aggregation risk in the geographic area, nor can it adjust the user reservation behavior and path planning strategy through the energy induction mechanism, so as to realize true load balancing and regional coordination. More importantly, the scheme lacks modeling and guidance mechanism for user behavior parameters such as waiting tolerance time, path reconstruction cost, and task urgency, so that even if the node is successfully registered, the user behavior cannot be intervened by the system strategy, and it is difficult to avoid the congestion caused by the user's preference for a certain node. In terms of strategy flexibility, the scheme mainly uses network master authentication as the control point and fails to introduce local topology coordination mechanism and potential freezing and scheduling shielding mechanism between nodes. When a node is not suitable for further scheduling due to overcharging of energy storage or unstable power grid, the scheme does not set a broadcast mechanism to actively inform the surrounding nodes of its unavailable state. Therefore, it is impossible to build a scheduling shielding area during collaborative scheduling, which may cause risks such as misrecommendation and scheduling failure. Finally, the scheme also does not establish a multi-dimensional system stability monitoring mechanism. When the node response delay and system fluctuation indicators deviate from the stable interval, it does not set an automatic trigger reverse regulation mechanism (such as task migration activation, beat adjustment extension, and recommended potential down-regulation). Therefore, the anti-interference ability and recovery ability of the system in the face of user load impact, high-frequency fluctuation, and abnormal behavior have obvious shortcomings. SUMMARY

[0003] The purpose of the present application is to provide an intelligent charging management method and system based on charging pile cluster collaborative control, so as to solve some of the problems and deficiencies pointed out in the background art.

[0004] The present application solves the above technical problems by adopting the following technical solution: a smart charging management method and system based on charging pile cluster cooperative control, comprising: regarding a charging request initiated by a user as a disturbance factor, and making the disturbance factor wander in a dynamic manner in a charging pile cluster, and through a response threshold value set by a charging pile node, including a load critical value, a waiting tolerance time, and a power adjustable ratio condition, achieving aggregation and decoupling of the request between nodes, and forming a decentralized regulation starting point based on disturbance response; When a certain charging pile node service saturation trend appears, the node has an active release capability for part of the future to-be-executed charging tasks, and a task release declaration is issued, and a surrounding idle node autonomously selects to receive a migrated task according to a local resource state; all scheduling execution instructions are calculated by the charging pile node based on a delay or advance execution window of a local running state, and a non-synchronous and adjustable scheduling chain beat is formed in the cluster; Each node evaluates energy potential according to a renewable energy access rate, a load redundancy degree, and an idle period, releases node energy potential information through a user interface, and guides the user to adjust a reservation behavior according to the energy potential; the dynamic convergence of a cluster overall service quality curve is taken as a target, and a node idle rate variance, a response delay stability, and a recovery deviation index are continuously monitored, and when system fluctuation deviates from a stable interval, a task migration, a beat adjustment, and an energy induction mechanism are reversely activated.

[0005] Further, the disturbance factor adjusts a propagation path priority according to a response threshold value change trend of a node in a wandering process, so as to select a node with a response threshold value downward trend for next hop propagation; and an upper limit of a number of disturbance factors received by each node within a time window is dynamically adjusted according to a recent scheduling response success rate.

[0006] Further, the task release operation is triggered only when the original node determines that a current task remaining execution time is higher than a system set migration threshold value; and a broadcast range of task release information is limited in a near neighbor topology of the original node, so as to reduce redundant propagation control communication load.

[0007] Further, the scheduling chain beat is determined by the node according to a current average load change rate, and a beat interval time is correspondingly increased when the load rapidly rises; and after the node performs beat adjustment, adjustment information is broadcast to adjacent nodes, so as to establish a temporary beat mutual exclusion window.

[0008] Further, the energy potential of the node takes a local energy storage remaining amount as a correction factor when being evaluated, and preferentially guides the user to select an energy storage sufficient node; when a node selected by the user is inconsistent with a system recommended node, the system records a user selection deviation degree and dynamically adjusts a subsequent recommendation strategy; and the energy potential is a dynamic adjustable power capacity of the node that can be used for distribution to a new charging task in a unit time. By evaluating the energy potential function value of each node, the user is recommended and guided based on the consideration of energy storage system, power grid power supply capacity and node response capacity, and the subsequent recommendation strategy is dynamically adjusted according to the user behavior deviation; the energy potential is not only used to measure the task carrying capacity of the current node, but also used as a control factor for user path decision guidance; the core technical scheme is as follows: The energy potential value of each node is dynamically calculated, and the following definition function is used: Wherein: is the energy potential value of the node ; is the current adjustable power of the node from the power grid; is the current releasable remaining power of the node of the energy storage system; is the energy storage participation coefficient (dynamically changes according to the battery health state, 0-1); is the current task response queue length (task number) of the node; is the node task response change rate (response load slope); is the response sensitivity coefficient (used to adjust the punishment degree of load growth); is the historical deviation weight of the user side to the node (representing the long-term preference deviation degree of the user); user feedback influence coefficient (determines whether the user deviation affects the current scheduling); The numerator part of the potential function represents the energy supply capacity of the node, which is composed of the real-time adjustable power and the energy storage adjustable power , and the latter is weighted and corrected by the energy storage participation ; the denominator part introduces two types of penalty factors: one is the change rate of node response capacity , which is used to weaken the potential in the state of rapid load rise; the other is the historical user deviation behavior influence factor , which avoids the node that is continuously rejected by the user from being recommended; The derivation process of the function includes: Starting from the instantaneous power supply capacity that the node can provide, the grid-side adjustable power is defined as , which represents the instantaneous power accessed from the power grid at the current time point which is not occupied by the task; considering that the node usually also has an energy storage unit, the energy storage adjustable power ​​represents its current available residual energy capacity (in kWh), but due to the schedulability of the energy storage and its electrochemical state, cycle life factors, it cannot be directly added up, so a weight parameter is introduced , as an energy storage participation coefficient, is used to dynamically scale the contribution of the energy storage value, thereby constructing the molecular part of the energy supply potential: ; However, it is incomplete to evaluate the node potential only in terms of energy supply capacity; two weakening terms must be considered to reflect the dynamic response characteristics of the system and the historical feedback of user behavior on system scheduling; First, the current load fluctuation of the node directly affects its scheduling stability, so define represents the current node task response queue length, and its derivative with respect to time represents the rate of change of the task load; if the node is in a rapid load rising phase, its scheduling stability is poor, and it is not suitable for allocating new tasks; therefore, the absolute value of this derivative is multiplied by the response sensitivity coefficient of the node (representing the sensitivity of the node to task fluctuations) as the first weakening factor introduced into the denominator of the potential function; Second, considering that the goal of scheduling is not only to balance the system load, but also to improve user acceptance and path fit, introduce the user behavior feedback quantity , which represents the deviation rate of the user's historical behavior from the recommendation of the node: the higher the deviation frequency, the more the node is not accepted by the user, so its recommendation priority should be reduced; for this purpose, set the parameter as the user feedback influence coefficient to adjust the suppression strength of this factor on the overall potential value; finally, these two weakening terms are combined in the form of addition, together with 1 to form the denominator term of the potential function , thereby constructing a complete potential evaluation function; This function takes into account four key dimensions in its design: current power schedulability, energy storage elastic energy supply capacity, task load dynamic risk, and user behavior preference feedback; its structure has nonlinear coupling characteristics, real-time dynamic adjustment characteristics, and behavior feedback closed-loop characteristics, and is suitable for deployment on any edge node with basic collection functions, while supporting scheduling sorting, user guidance, and system behavior backtracking learning multiple control functions.

[0009] Further, the calculation of the dynamic adjustable power capacity includes the voltage fluctuation frequency of the node connected to the power grid as a real-time correction factor, which reduces the upper limit of the potential when power supply instability is detected; after the node calculates the energy potential, it broadcasts the energy potential value to neighboring nodes to form a local potential density map, which is used to limit the risk of centralized scheduling of tasks within the same geographic area.

[0010] Further, the node integrates the cycle life and the current electrochemical state of the energy storage battery in the energy storage correction process, if the life is insufficient or in the inefficient charging and discharging state, the correction factor is dynamically set to zero, if the energy storage system of a node is in the overcharge protection state, the energy potential is set to a frozen value by the system, and the ice frozen state is transmitted to the adjacent nodes by a broadcast control signal to form a scheduling shield.

[0011] Further, the priority guiding user strategy comprises: when the system detects that multiple users simultaneously select the same high potential node, starting a reservation allocation mechanism, and based on the user path reconstruction cost and the waiting tolerance time, the order is sorted; and in the node recommendation sent by the system to the user, the recommendation is additionally offset by the recommended distance according to the historical task urgency of the user; The system of the intelligent charging management method based on charging pile cluster collaborative control comprises: The charging pile cluster is composed of multiple intelligent charging pile nodes with local load sensing, task scheduling and communication capabilities, and a local interconnection topology is formed between the nodes; a disturbance management unit is used for receiving and modeling user charging requests as disturbance factors, and dynamically adjusting the propagation path according to the node response threshold; a task scheduling unit is used for the node to actively release future tasks when the load tends to be saturated, and to realize task migration through local broadcast; a beat control unit is used for adjusting the scheduling beat according to the load change rate, and establishing a beat mutual exclusion mechanism between the nodes; a potential evaluation unit is used for evaluating the available power of the node in combination with the energy storage state and the power grid fluctuation factor, and issuing induction information to guide user behavior; a user recommendation unit is used for node recommendation and reservation sorting according to the user path cost and task urgency parameters; and a stability monitoring unit is used for monitoring system operation indexes, and triggering corresponding regulation mechanisms when deviating from the stable interval.

[0012] The intelligent charging management method based on charging pile cluster collaborative control has the advantages that: the method constructs a mechanism for the disturbance factor to walk between nodes, converts the charging request initiated by the user from centralization to node autonomous response, so that the system has high decentralized regulation and control capability. By introducing the node response threshold (such as the load threshold, the waiting tolerance time, etc.) as the screening condition for disturbance propagation, the request can be dynamically decoupled and reorganized in the charging pile cluster, thereby effectively relieving the queuing congestion problem of popular nodes. At the same time, when the node load tends to be saturated, the method supports the node to actively release future to-be-executed tasks, and the surrounding idle nodes receive the tasks based on the local resource state, realizing a task migration collaborative mechanism of "node spontaneous - surrounding self-selection", which enhances the system flexibility and the adaptability of scheduling.

[0013] Further, the application introduces the concept of node energy potential, integrates multi-source parameters such as energy storage power, grid voltage fluctuation, and renewable energy access rate into the evaluation model, and actively publishes adjustable resource information through the user interface to guide the user reservation behavior to dynamically tilt towards nodes with more abundant energy. Combined with user behavior deviation records and historical emergency parameters, the system can dynamically adjust the recommendation strategy to adapt to the trade-off between path cost and service preference of different users, and improve the balance of overall reservation distribution. During system operation, the system effectively limits the concentration of scheduling risks in high potential areas through mechanisms such as beat control and potential shielding, and ultimately realizes the beneficial effects of dynamic convergence of service quality and collaborative evolution of system stability. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The charging pile cluster intelligent charging management core flowchart of the application.

[0015] Figure 2 The charging request disturbance adaptive distribution and scheduling chain beat control curve functional relationship of the application.

[0016] Figure 3 The charging pile cluster dynamic potential scheduling and user guidance functional relationship diagram of the application.

[0017] Figure 4 The new energy parking lot intelligent charging pile cluster scheduling core schematic diagram of embodiment 1 of the application.

[0018] Figure 5 The intelligent charging pile cluster dynamic recommendation and scheduling function schematic diagram of embodiment 2 of the application. DETAILED DESCRIPTION

[0019] The specific embodiments of the application will be described in detail below with reference to the accompanying drawings.

[0020] The specific embodiments of the application will be described in detail below with reference to the accompanying drawings. Figure 1The present application is based on a smart charging management method based on charging pile cluster cooperative control. A charging request initiated by a user is modeled as a system disturbance factor. The disturbance factor is not immediately statically bound to a certain node after entering the charging pile cluster, but is dynamically propagated in the cluster in a wandering manner. The propagation mechanism is limited by the response threshold conditions set by the node locally. Specifically, it includes the current load critical value of the node, the maximum waiting tolerance time allowed by the user, and the power adjustment ratio (i.e., power adjustable ratio) that the node can adjust for new charging tasks within the current period. The load critical value is used to represent the maximum task load boundary that the node can carry within the current scheduling period. If the current node load has approached this boundary, it is considered not to meet the response condition, and the node will refuse to accept the disturbance factor and interrupt its wandering path. The waiting tolerance time is declared synchronously when the user side uploads the request. The system estimates the node service queue length and expected waiting time when the disturbance is propagated. If it exceeds the user-declared threshold, the node does not enter the candidate response set. In addition, the power adjustable ratio represents the proportion of dynamic power capacity that the current node can schedule per unit time. It is estimated by the node in combination with its energy storage state and real-time scheduling ability of the power grid. If the adjustable ratio is lower than the system set reference value, it is also considered as a node that cannot accept the disturbance factor. During the propagation process of the disturbance factor, each jump is prioritized according to the response threshold matching degree of the surrounding nodes, and is preferentially jumped to the target node with a significant response ability enhancement trend. During the multi-hop propagation process, if there are multiple nodes that meet the response condition, the system will select them according to the current load redundancy of the node and the moving cost from the user's location.

[0021] When a charging pile node in the cluster is in a service saturation trend, i.e., the current queued task amount of the node has approached the upper limit of the node scheduling processing capability, or the task response load growth rate exceeds a threshold within a certain prediction window, the node can autonomously trigger an active release mechanism, i.e., screening the future charging tasks that have been allocated but have not been executed or have not reached the scheduling window, and selecting a part of the tasks that have transferable conditions to release. The transferable tasks include but are not limited to low-priority tasks with later scheduled execution time, higher user waiting tolerance, or less sensitive to path migration. After completing the task screening, the node broadcasts a task release declaration to the neighboring nodes or the local sub-cluster, which contains necessary information such as task identification, estimated execution time, required power, user location, and constraint conditions. After receiving the task release declaration, the surrounding idle nodes calculate and judge whether the task receiving conditions are met according to their local resource states, including the current task queue state, power schedulable ability, reserved time window, and geographical location matching degree. If the conditions are met, the idle nodes initiate a task receiving request, and the original node completes the task migration according to the receiving request priority. In addition, to realize the scheduling coordination in the time domain within the cluster, a scheduling rhythm mechanism is introduced, i.e., all charging tasks are not executed at the same time in the whole network, but each node independently calculates the delay time or advance time of scheduling execution according to its local running state. The time window is calculated by weighting the average load change rate of the node, task processing efficiency, and power adjustment response time lag, thereby forming a non-synchronous scheduling rhythm chain between nodes. This rhythm chain can be periodically self-adjusted according to the cluster load, so that the task scheduling has distribution and flexibility in the time dimension, and the execution conflicts and current surges caused by centralized scheduling are reduced.

[0022] Each charging pile node periodically performs multi-parameter comprehensive evaluation on the local running state to calculate the energy potential value of the current node, which is used to represent the power release capability of the node that can be allocated to the new charging task in unit time. The core parameters for calculation include but are not limited to: first, the renewable energy access rate, which refers to the proportion of renewable energy such as photovoltaic and wind energy in the total power supply connected to the node. The higher the value, the stronger the green power utilization capability of the node; second, the load redundancy degree, which refers to the margin between the currently allocated task load of the node and its theoretical maximum processing capability. The larger the margin, the more relaxed the node scheduling space; third, the idle time period ratio, which refers to the time ratio of the low load period of the node in the current or predicted period, which is used to reflect the future schedulability trend. After the node obtains the comprehensive energy potential value, it publishes the potential information to the user in a graphical way through the user interface. The published content includes information such as recommendation level, estimated waiting time, price advantage, and green energy ratio, so as to guide users to actively choose high potential nodes when initiating charging reservations, realize the matching optimization of charging path and node resource state, and adjust the user's reservation behavior according to the change of node potential, forming a positive guiding closed loop mechanism among energy state-user behavior-load distribution. In the system operation process, in order to ensure the scheduling efficiency and service stability of the whole cluster, the system collects the running indexes of each node in real time, and takes the dynamic convergence of the overall service quality curve of the cluster as the system stability target. The specific monitoring parameters include the variance of node idle rate (used to reflect the resource distribution balance), the stability of response delay (used to measure the service response volatility), and the service recovery deviation (used to describe the deviation between the node load recovery capability and the system prediction). When any of the above indexes exceeds the set stability threshold, the system determines that the overall regulation deviates, and triggers the reverse activation mechanism, that is, restores part of the regulation strategy to re-converge the scheduling state. The reverse activation includes restarting the aforementioned task release and migration mechanism, adjusting the scheduling rhythm strategy of local nodes, and rebroadcasting the node energy potential information to re-induce user selection behavior.

[0023] Combination with the accompanying drawings Figure 2, through the walk process of simulating disturbance factors, adaptive distribution of user charging requests within the cluster is realized, specifically, when a user initiates a charging request, the request is modeled as a disturbance factor by the system, and is propagated according to certain path rules in the topology network composed of charging piles, the disturbance walk process is not random or static control, but according to the dynamic change trend of the response threshold of each node to select the path, wherein the change trend of the response threshold refers to the evolution direction of the load change curve, the power adjustable ratio change and the queuing waiting time of the node in the recent scheduling period, for example, when the load critical value of a certain node shows a downward trend, the waiting queue shortens, and the power adjustment ability strengthens, the system judges that the response ability of the node is being enhanced, that is, the response threshold of the node is in the downward trend, at this time, the node will be preferentially selected as the target node of the next hop disturbance factor, so as to improve the convergence efficiency of the disturbance factor in the propagation process, and avoid the request to stay or stay in the node with heavy load or weakened response ability; in addition, in order to prevent the local overload or scheduling imbalance of the high response tendency node due to attracting too many disturbance factors, a disturbance receiving upper limit mechanism is introduced on each node side, that is, within the set scheduling time window, the number of disturbance factors that the node can receive has a dynamic upper limit, which is not a fixed value, but is adjusted in real time according to the scheduling response success rate of the node in the recent period, specifically: if the node has a high scheduling success rate after receiving the request in the recent several scheduling periods (i.e. the request can be executed on time), the system appropriately increases the disturbance receiving upper limit of the node, otherwise if the success rate is continuously low, the receiving ability limit of the node is automatically lowered, so as to realize the dynamic load control based on the actual execution ability, so as to ensure that the disturbance factor can quickly tend to the node with enhanced response ability in the walk process, and can also avoid excessive concentration.

[0024] The task release mechanism has strict triggering judgment conditions and propagation range constraints to balance the scheduling efficiency and communication cost at the system level. Specifically, when the charging pile node needs to transfer the load due to task congestion or resource shortage, instead of immediately releasing and broadcasting all allocated tasks, the remaining execution time of the future to-be-executed tasks in the local task queue is first evaluated. If the difference between the estimated start time of a task and the current time is higher than the migration threshold preset by the system, that is, there is enough time for inter-node migration and reallocation, the task is considered as a releasable task. Otherwise, it is considered as a short-time high-priority task and is not transferred, thereby ensuring that the transferred task can still be completed on time after landing on the new node and will not damage the user service expectation. The migration threshold can be dynamically set by the system according to the running state of the cluster, for example, the threshold can be appropriately shortened to enhance the release ability when the task congestion is serious, and vice versa to maintain stability. In the broadcasting link of the task release information, in order to avoid the task release message being widely propagated in the whole cluster range and causing redundant response and waste of communication resources, a near-neighbor broadcast limiting mechanism is adopted, that is, the task release declaration is only propagated in the near-neighbor topology of the original node. The near-neighbor topology refers to the node set in the same sub-cluster through one physical hop, and the original node only broadcasts the release information to the nodes in the set. These candidate nodes autonomously evaluate whether to respond to the transfer request according to the local resource situation.

[0025] A control mechanism of scheduling chain rhythm is introduced to coordinate the task scheduling rhythm of each node in the cluster, so as to avoid resource contention, current impact or communication congestion caused by simultaneous response or processing of charging tasks. The scheduling chain rhythm is a scheduling execution time interval generated by the node adaptively, which is dynamically calculated by the node based on the current average load change rate. The average load change rate represents the growth rate of the task load of the node in the recent several scheduling periods. The greater the value is, the faster the node load is in the rapid rising stage. In order to prevent the resource from being occupied excessively or the task processing from exceeding the upper limit of the physical capacity, the node will prolong its scheduling rhythm interval time, that is, reduce the task response frequency, so that the node obtains longer state adjustment time. On the contrary, when the node load grows slowly or is in a downward trend, the rhythm interval can be shortened appropriately to improve the scheduling responsiveness. This mechanism ensures that different nodes participate in the scheduling operation in a non-synchronous way according to their own running state, thereby forming a scheduling chain with distributed characteristics. In order to avoid scheduling collision between adjacent nodes due to rhythm overlap, after the node completes the rhythm interval time adjustment, it will immediately broadcast a rhythm adjustment information to its physical or logical neighbor nodes. The information contains the adjusted rhythm starting time and duration. After receiving the broadcast, the adjacent nodes will generate a temporary rhythm mutual exclusion window according to their current scheduling plan, and suspend the scheduling request or task receiving operation of the same resource in the window, so as to effectively avoid the conflict or system jitter caused by the overlap of scheduling behavior. This rhythm mutual exclusion mechanism is a lightweight soft coordination method, which does not rely on central control or forced synchronization, but realizes the spatial separation and time peak shifting of the cluster scheduling rhythm through local coordination, so as to realize the elastic self-consistency and load self-balancing of the whole cluster running rhythm on the basis of ensuring the stability of high-concurrency task scheduling.

[0026] The accompanying drawings are incorporated Figure 3 . A dynamic potential model is constructed around the energy schedulable capacity of the charging pile node to realize the coordinated regulation of the optimal allocation of charging tasks in the cluster range and the user path guiding strategy. Specifically, the energy potential value of each charging pile node is defined as: ; Wherein, represents the energy potential of the node , which represents the dynamic available power capacity that can be scheduled and allocated by the node in unit time; is the real-time power that can be allocated by the node from the power grid at present; is the remaining amount of electricity that can be released in the energy storage system of the node at present; is the energy storage participation coefficient, which ranges from , and is used to adjust the effective contribution of the energy storage according to the battery health state, chemical state and the like; This represents the length of the current task response queue (in terms of the number of tasks), and its time derivative. This indicates the task growth rate, used to characterize the trend of dynamic changes in the current load of a node; The response sensitivity coefficient reflects the penalty weight of a node for load growth; As a feedback factor representing the degree of deviation from the user's historical behavior, the trend of the proportion of users who did not select the node multiple times is recorded; The user feedback impact coefficient represents the degree of influence of user deviation behavior on the recommendation strategy. The numerator in the formula expresses the comprehensive value of the node's real-time power supply capacity, encompassing the power supply capacity of the basic grid and the energy storage system. The energy storage component consists of… The weighted adjustment reflects its uncertainty and decay characteristics; the denominator introduces two penalty terms, one being the node task load volatility. Multiply by the response sensitivity coefficient First, it is used to suppress the potential energy of nodes with rapidly increasing load, preventing the accumulation of scheduling risks at system edge nodes; second, it includes user behavior feedback items. and The product of these factors is used to dynamically reduce the recommendation weight of nodes that users have repeatedly rejected in the scheduling order, ensuring consistency between user path preferences and system scheduling priorities.

[0027] The evaluation process for the dynamically adjustable power capacity of each node not only considers the node's own power resource dispatchability and energy storage status, but also further introduces the voltage fluctuation frequency of the connected grid as a real-time correction factor to improve the responsiveness of potential energy calculation to changes in power supply stability. Specifically, when evaluating its energy potential, the node monitors the voltage change frequency of the connected grid in real time. When continuous excessive voltage fluctuations are detected within a short period of time, i.e. Exceeding the system's set stable frequency threshold If the node is in an unstable power supply state, the system will automatically apply a suppression factor to the originally calculated upper limit of the potential energy of that node. Its current energy potential value Revised to , so as to avoid unstable power supply nodes being preferentially selected in task scheduling, affecting the overall robustness of system scheduling; after the nodes complete the dynamic calculation of the energy potential, the current potential value of the node is broadcast to the adjacent nodes through a local broadcast protocol, and a local potential density map is constructed among the adjacent nodes based on the received multi-node potential values. The density map takes geographical proximity as the boundary and the modified potential value of each node as the weight, forming a spatial distribution map of power scheduling capability. When receiving user charging requests and recommending paths, the scheduling system combines the density map to evaluate the scheduling concentration risk in a certain area. When there is a potential peak area in the node density map of a certain geographic area and continuous scheduling requests are received, the system will actively limit the scheduling strategy or disperse the scheduling guidance to prevent scheduling congestion, power burst load or service delay in the area caused by the selection of high potential nodes.

[0028] In the dynamic calculation process of node energy potential, a multi-dimensional correction mechanism of energy storage system state is further introduced to improve the accuracy of potential evaluation and the stability of system scheduling. Specifically, when calculating the contribution of the energy storage part to the energy potential, the node not only uses the remaining power as a weighting item, but also evaluates the cycle life parameters and current electrochemical state of the energy storage battery, including charge and discharge efficiency, voltage platform stability and temperature characteristics, etc. If the system detects that the node energy storage unit has approached the lower limit of its rated cycle life, or is currently in a state of low conversion efficiency and cannot achieve high-quality power release, it is determined that the energy storage unit does not have the ability to effectively participate in scheduling support. At this time, the corresponding energy storage correction factor will be dynamically set to zero by the system, so that the node no longer includes the energy storage item in the potential function to avoid scheduling errors caused by unreliable energy storage; in addition, to avoid system safety hazards caused by overcharging risk, when the node energy storage system is in an overcharging protection state, i.e. its current energy storage voltage or capacity exceeds the preset threshold and triggers the protection mechanism, the system will freeze the energy potential value of the node , usually with an identifier state (such as zero or negative) that cannot participate in scheduling, and sends a freeze state control signal to the adjacent nodes of the node through local broadcast, informing them that the frozen node cannot be used as a task scheduling candidate target in the current scheduling period, thereby establishing a scheduling shielding mechanism at the cluster scheduling level, effectively avoiding the problem of overcharging nodes being mistakenly selected due to apparent resource abundance, while avoiding risks such as accelerated degradation of energy storage systems, unstable power burst release, and scheduling chain rupture.

[0029] A set of guiding mechanisms with user preference perception and scheduling conflict resolution capability is used to dynamically recommend paths and order reservations for users to improve scheduling fairness and resource utilization. Specifically, when the system detects that multiple users almost simultaneously select the same high potential node as the reservation target, the system will automatically trigger the reservation allocation mechanism to reconfigure the priority of the users entering the conflict node. The priority is sorted according to two key parameters: one is the user path reconstruction cost, that is, if a user is guided to an alternative node, the cost of the newly added travel distance, time consumption or deviation from the path, and the lower the cost, the easier it is to be guided; the second is the user waiting tolerance time, which is set by the user or learned from historical behavior, indicating the acceptable degree of charging start delay, and the lower the tolerance, the higher the sorting. Finally, the above two indicators are combined to form a guiding priority list for the conflict node at the current time, so as to guide the users to arrive at different times and avoid the instantaneous overload of the node caused by the swarm scheduling; at the same time, in order to enhance the individualized adaptability of scheduling, the system will also add a recommended offset distance to each recommended node when pushing the candidate node recommendation list to the user. This parameter is dynamically generated based on the urgency of the user's historical tasks (such as initiating a charging request at a critical battery level multiple times), and is used to adjust the range of the shortest distance between the recommended node and the user's current location. Urgent task users will be recommended to be closer to the node, and non-urgent users will be recommended to be slightly farther away to help the system to shunt. The above method relies on the multi-module collaborative architecture built by the system, in which the charging pile cluster is composed of multiple intelligent nodes with local load perception, task scheduling and communication capabilities. These nodes form a local interconnection topology and cooperate with each function unit: the disturbance management unit models the user's charging request as a disturbance factor and dynamically adjusts its propagation path according to the node response threshold; the task scheduling unit makes the node actively release future tasks when the load tends to be saturated and realizes task migration through local broadcast; the beat control unit adjusts the scheduling beat according to the current load change rate of the node and establishes a beat mutual exclusion mechanism between nodes to avoid conflicts; the potential evaluation unit evaluates the energy potential value of the node according to the node energy storage state, power grid fluctuation, etc. and publishes the induction information to the user end; the user recommendation unit calculates the user path cost and task urgency to complete the scheduling sorting and guiding allocation; the stability monitoring unit continuously monitors the cluster operation state indicators such as response delay, load variance, etc., and automatically triggers the corresponding control mechanism once it deviates from the set stable interval. Embodiment 1

[0030] Combined with the Figure 4 In a new energy parking lot in a city, a cluster system composed of 9 intelligent charging piles is deployed, numbered to , where each node is equipped with local load awareness, response threshold calculation, task scheduling and communication capabilities, and forms a local interconnected topology defined by physical distance and gateway access method. The current time of the system is 10:00, user A drives an electric car to the park entrance with 12% power left, initiates a charging request, and the system models the request as a disturbance factor , and starts to spread among nodes to determine the optimal charging service target node.

[0031] The propagation path of the disturbance factor is not a traditional shortest path routing, but a priority decision based on the real-time calculated response threshold change trend of each node. First, the system performs the first hop from the nearest node at the entrance, and the current response threshold of this node is composed of the following three parts: load critical value , waiting tolerance time , and power adjustable ratio , compared with 5 minutes ago to now, rose by 12%, rose by 2min, decreased by 0.1, and the comprehensive judgment response threshold tends to deteriorate, so it is not recommended to continue to spread to in the next hop.

[0032] At the same time, the adjacent nodes are , and , and the system detects that the response threshold change trend of is: from 78% to 65%, from 7min to 4min, from 0.68 to 0.82, indicating current load tends to ease, adjustable power rises, which is a typical response threshold down trend node, so the system will next hop propagation path of .

[0033] Before propagating to , the system also needs to determine the upper limit of the number of disturbance factors that the node can receive in the unit time window (such as 10 minutes) at present. Assuming that the historical task scheduling response success rate in the current time window is , and the basic default receiving upper limit is , then according to the dynamic adjustment rule:

[0034] means There are at most 4 disturbance factors that can be received in the current window. The node has received 2 disturbance tasks that are still in the scheduling chain 、 Therefore, there are still 2 quotas, allowing to enter and make a response judgment.

[0035] After entering , the node reads its own resource state and finds that the current queue remaining time is 9 min, the estimated acceptable task insertion time is 3 min, and the estimated arrival time of user A is 4 min later, there is a 1 min gap, the system judges that it can still schedule the task, so it returns as a candidate node recommendation to user A.

[0036] Suppose user A does not directly accept the system recommendation, but manually selects node , the system will record the user's behavior deviation value in the background: Δd=|recommended distance-actual distance|=|45m-62m|=17m and put it into the user deviation model to fine-tune the node recommendation priority in the subsequent scheduling strategy, if the deviation occurs frequently, the recommendation weight of high potential energy nodes for this user will be reduced to balance the resources.

[0037] Suppose the current time has entered 10:15, node successfully receives the disturbance factor of user A and completes the scheduling recommendation, but at this time, node has 4 charging tasks in the execution phase in its task queue , and the new user task is placed in the execution task pool, as the actual load of the node grows, its average load change rate increases from 0.12 to 0.37 (unit: task / min) in the past 5 minutes, the system preset task migration trigger threshold is 12 minutes, that is, when any execution task remaining waiting time is greater than the threshold, it can be a migration candidate, the current waiting queue estimated time is 15 minutes, which has exceeded , so it meets the trigger condition of task release operation, node initiates a task release declaration actively under the condition that the local future scheduling pressure continues to rise, releases and sends the task migration information to its adjacent nodes 、 and through broadcast, avoiding the communication burden brought by broadcasting to the whole network.

[0038] After receiving the release declaration, each adjacent node autonomously assesses whether its local resources are capable of taking over. Take node for example, which currently has only one queued task, a current response threshold of 48% of the load threshold, a waiting tolerance time of 5 minutes, and an adjustable power ratio of 0.82. After assessment by the local scheduling module, it is confirmed that it can accept , and it returns a task migration response to , synchronously updating the local scheduling state; After successful migration to , , the task queue of returns to the normal load interval, and the system avoids the spread of fluctuations caused by resource overload.

[0039] Next, to avoid repeated or conflicting scheduling between nodes, the scheduling link beat mechanism comes into effect. Since the average load change rate of increases significantly within 5 minutes, the system takes the basic beat period as a reference and adjusts the beat interval according to the following beat delay factor calculation method:

[0040] That is, node will update every 82.2 seconds in subsequent rounds of scheduling commands, rather than the fixed 60-second period. To prevent adjacent nodes from activating repeated scheduling operations immediately after releases tasks, node sends beat adjustment broadcast signals to adjacent nodes , , through the control communication interface, establishing a beat mutual exclusion window: =[10:15:00,10:17:00], during which the surrounding nodes suspend scheduling strategy updates, thus avoiding the phenomenon of multiple nodes in the cluster competing for the same type of resources within a short period of time. Embodiment 2

[0041] In combination with the attached Figure 5 , a group of charging pile clusters is deployed in a new energy community in a city, including nodes to , each equipped with a local energy storage unit and an intelligent scheduling module. User B plans to initiate a 7kWh charging request at 18:30, and the system needs to intelligently recommend appropriate nodes based on the current energy potential evaluation results of each node. At this time, the system collects and calculates the energy potential of each node in real time , using the following defined formula: ; where is the adjustable power from the grid side (in kW), is the current remaining energy storage (in kWh), is the energy storage engagement factor adjusted dynamically according to the battery health state; is the node task response rate, is the response sensitivity factor; is the user deviation history, is the user feedback influence factor.

[0042] Take node N3 as an example, its current state parameters are as follows: (green grid power supply margin); (releasable energy storage); (battery health is good); (task accumulation rate); ; (past user has partially deviated from the recommended record); ; Substitute the formula to calculate the energy potential of this node as follows: The numerator part is: ; The denominator part is: ; The final potential value is: ; At the same time, other nodes such as , , , are also calculated in the same way, resulting in: ; ; ; ; The system recommendation ranking is: > > > > , preferentially guiding user B to charge. But user B chooses due to geographical location or queuing preference, this behavior is recorded by the system as a deviation weight increased to 0.35, affecting the subsequent recommendation weight. In the future under similar conditions, if the deviation continues, The recommendation weight will be gradually reduced in order to improve matching efficiency.

[0043] The current time is set to 18:40, the evening peak, with a large electricity load and the system directly connected to the node. to Real-time potential energy correction and local density control are performed. Node In the previous round of potential energy calculation, it was selected by user B. In this round, its adjustable power capacity needs to be reassessed, and the risk of scheduling anomalies needs to be detected.

[0044] First, the grid-side voltage fluctuation frequency is introduced. This serves as a correction factor for the upper limit of the node's potential energy. It sets the system's parameters for the nodes. The voltage fluctuations at the grid connection were frequently found to exceed ±3% within the past 5 minutes, with the fluctuation frequency being: ; Set the system's allowed upper limit for stable power supply fluctuation frequency as follows: Then, a decay correction factor is set: ; The original potential energy value was calculated in the previous round. Then the corrected node Dynamically adjustable power capacity: ; The system broadcasts this value to neighboring nodes. , A local potential energy density map is generated, showing the corrected potential energy values ​​of each node within this geographic area: ; ; ; At this time, the system detected... The total adjustable power capacity of the three nearby nodes is less than 12kW, and nine consecutive tasks have been connected in the past 10 minutes, exceeding the set density risk threshold. Therefore, the system activates the density diffusion mechanism, temporarily suspending the recommendation of new tasks for this region and prioritizing scheduling tasks for other regions (such as...). , ).

[0045] Meanwhile, nodes The energy storage system malfunctioned. Its battery cycle life was detected as <15%, and its SOC (state of charge) reached 98%, placing it in the overcharge protection zone. The system automatically adjusted its energy storage correction factor. And freeze the potential energy as: ; The frozen state is marked as a shield node in the system, and it is informed by broadcasting With , avoid scheduling pressure continues to accumulate. At the same time, the system interface hides the N2 node reservation entrance from the user and recommends switching to the With the highest potential value.

[0046] The system time advances to 19:10, and multiple users arrive at the community south parking area for charging. The current node The potential is the highest ( ), and the system determines that the node is the charging concentration point. At this time, the system detects that users C, D, and E have submitted charging reservation requests almost simultaneously, and all have selected . To avoid scheduling congestion caused by users selecting, the system starts the "reservation allocation mechanism" and sorts by calculating the path reconstruction cost and waiting tolerance time of each user.

[0047] The straight-line distance of user C's current path to is 1.2km, and the reconstructed path switches to , increasing by 600m. The user's waiting tolerance time (the maximum acceptable waiting time for charging) is set to 10min; user D starts from the north gate, and the path to is 900m, and switching to increases by 300m, but its waiting tolerance time is only 4min; user E comes from the west side, which is only 600m from , and changes to , increasing the distance by 1.2km, and the tolerance time is 8min. The system prioritizes the three users according to the following sorting function: ; Among them: , , . Substituting the parameters of the three users: User C: ; User D: ; User E: .

[0048] The final sorting is: user C> user D> user E. The system allocation result is: user C's reservation at is kept, user D is recommended to , and user E is recommended to , and a prompt is popped up that the current node is congested and the recommended node has a shorter waiting time.

[0049] At the same time, the user recommendation unit in the system also introduces a bias recommendation distance for the task urgency of different users. For example, user F sets the urgency level of the task as high (the remaining power is only 8%, and the driving endurance is less than 2km) when submitting the task, and the system historical scheduling also records that 4 out of 5 tasks of user F are high urgency, so the system sets the bias recommendation maximum as 200m in the recommendation process, and tries to recommend the nearest available node; while user G's task is not urgent (the remaining power is 50%), the system allows the bias range to expand to 1km to reduce the scheduling concentration. In the recommendation list, user F is only recommended to the nodes within 200m from his path With ; user G can also see ~ node sorting.

[0050] From the perspective of system composition, the intelligent charging management system is supported by the following six functional units: 1) charging pile cluster: each intelligent pile has the ability of power perception, state uploading and autonomous response, and a local topology is constructed through an edge communication network; 2) disturbance management unit: receiving user requests and converting them into disturbance packages to respond to thresholds to guide their wandering paths, realizing decentralized regulation and control of charging requests; 3) task scheduling unit: when the node detects task accumulation or capacity shortage, future tasks are released to attract surrounding idle nodes to undertake through local broadcast; 4) beat control unit: adjusting the scheduling beat according to the average load slope of the node in real time, and notifying the adjacent nodes to form mutually exclusive scheduling rhythm, avoiding load synchronous oscillation; 5) potential evaluation unit: generating potential values based on multiple indicators such as grid adjustable power, energy storage correction, load change and user behavior, for node sorting and user guidance; 6) stability monitoring unit: real-time tracking of stability indicators such as response delay fluctuation and node idle rate variance of the global system, and activating task migration and scheduling smoothing mechanism when the threshold is exceeded.

[0051] The system not only can accurately identify the overload risk of peak nodes, but also can realize dynamic guidance and scheduling under the balance of user experience in the case of multi-user concurrency and task differentiation, and has high distributed self-organizing ability, ensuring the efficient operation and service stability of the cluster in complex scenarios.

[0052] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above examples, and the above examples and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A smart charging management method based on charging pile cluster cooperative control, characterized in that The application relates to a charging request response method based on a disturbance factor. The user-initiated charging request is regarded as a disturbance factor, and the charging pile cluster is dynamically walked through a response threshold set by a charging pile node, including a load critical value, a waiting tolerance time and a power adjustable ratio condition, so that the request is aggregated and decoupled between nodes, a decentralized control starting point based on disturbance response is formed, and when a certain charging pile node service saturation trend appears, the node has an active release capability for part of future to-be-executed charging tasks, a task release declaration is issued, and a peripheral idle node autonomously selects a received migration task according to a local resource state. All scheduling execution instructions are formed into a non-synchronous and adjustable scheduling chain beat in the cluster by a delay or advance execution window calculated by the charging pile node based on a local running state. Each node evaluates energy potential according to a renewable energy access rate, a load redundancy degree and an idle period, releases node energy potential information through a user interface, and guides the user to adjust the reservation behavior according to the energy potential. The dynamic convergence of the overall service quality curve of the cluster is taken as a target, and the node idle rate variance, the response delay stability and the recovery deviation index are continuously monitored, and when the system fluctuation deviates from the stable interval, the task migration, the beat adjustment and the energy induction mechanism are reversely activated. The disturbance factor adjusts a propagation path priority according to a response threshold change trend of the node in the walking process, so that the node with a response threshold downward trend is selected for next hop propagation, and the number of disturbance factors received by each node within a time window is dynamically adjusted according to a recent scheduling response success rate. 2.The smart charging management method based on charging pile cluster cooperative control according to claim 1, characterized in that The task release operation is triggered only when the original node determines that the remaining execution time of the current task is higher than a system set migration threshold value, and the broadcast range of the task release information is limited in the near neighbor topology of the original node, so that the communication load is controlled by reducing redundant propagation. 3.The smart charging management method based on charging pile cluster cooperative control according to claim 1, characterized in that The scheduling chain beat is determined by the node according to a current average load change rate, and the beat interval time is correspondingly increased when the load rapidly rises. 4.The smart charging management method based on charging pile cluster cooperative control according to claim 1, characterized in that The energy potential of the node takes the remaining power of the local energy storage as a correction factor during evaluation, and preferentially guides the user to select the energy storage sufficient node. 5.The smart charging management method based on charging pile cluster cooperative control according to claim 1, characterized in that When the user-selected node is inconsistent with the system-recommended node, the system records the user selection deviation degree and dynamically adjusts the subsequent recommendation strategy. 6.The smart charging management method based on charging pile cluster cooperative control according to claim 5, characterized in that The calculation of the dynamic adjustable power capacity includes the voltage fluctuation frequency of the grid connected to the node as a real-time correction factor, and the potential upper limit is reduced when unstable power supply is detected. After the node calculates the energy potential, the energy potential value is broadcast to the adjacent nodes to form a local potential density map, and the density map is used to limit the centralized scheduling risk of the task in the same geographic area.

7. The intelligent charging management method based on charging pile cluster cooperative control according to claim 6, characterized in that During the energy storage correction process, the cycle life and the current electrochemical state of the energy storage battery are comprehensively considered, and if the life is insufficient or in an inefficient charging and discharging state, the correction factor is dynamically set to zero. If a node's energy storage system is in overcharge protection state, the energy potential will be set to a frozen value by the system, and the ice will pass the frozen state to the adjacent nodes through a broadcast control signal to form a scheduling shield. 8.The smart charging management method based on charging pile cluster cooperative control according to claim 7, characterized in that The priority guiding user policy includes: when the system detects that multiple users simultaneously select the same high potential node, starting the reservation allocation mechanism, and sorting the guiding order based on the user path reconstruction cost and the waiting tolerance time; and in the node recommendation issued by the system to the user, the recommendation is attached with a bias suggestion distance according to the user's historical task urgency.

9. A system for implementing the intelligent charging management method based on the coordinated control of charging pile clusters according to any one of claims 1 to 8, characterized in that, It comprises: A charging pile cluster composed of multiple intelligent charging pile nodes with local load sensing, task scheduling and communication capabilities, forming a local interconnection topology between nodes; A disturbance management unit for receiving and modeling user charging requests as disturbance factors and dynamically adjusting the propagation path according to the node response threshold; A task scheduling unit for nodes to actively release future tasks when the load tends to be saturated, and to realize task migration through local broadcast; A beat control unit for adjusting the scheduling beat according to the load change rate and establishing a beat exclusion mechanism between nodes; A potential evaluation unit for evaluating the available power of the node in combination with the energy storage state and power grid fluctuation factors, and issuing induction information to guide user behavior; A user recommendation unit for node recommendation and reservation sorting according to user path cost and task urgency parameters; A stability monitoring unit for monitoring system operation indicators and triggering corresponding control mechanisms when deviating from the stable interval.

Images