Optical storage direct flexible system power dispatching method based on cloud edge collaboration

Abstract

The invention is suitable for the technical field of power dispatching, and provides an optical storage direct flexible system power dispatching method based on cloud edge collaboration, which comprises the following steps: executing a dynamic partition operation at the beginning of each dispatching moment according to a communication network between intelligent terminals in a park, and dividing a plurality of collaborative dispatching areas; constructing a double-layer cloud edge collaborative architecture based on a partitioning result; each edge server performs distributed power dispatching optimization calculation in the corresponding cooperative dispatching area, solves an intelligent terminal dispatching instruction meeting power balance constraint and equipment operation constraint, and uploads an optimization result to a cloud computing center; and flexible load equipment in each cooperative scheduling region is used for cross-region secondary power scheduling, and a dynamic security constraint model is adopted in the scheduling process to protect the operation security of the equipment. And the cross-regional power regulation capability of the optical storage direct-flexible system is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power dispatching, in particular to a power dispatching method for a photovoltaic energy storage direct flexible system based on cloud-edge collaboration. BACKGROUND

[0002] In the process of power dispatching of the photovoltaic energy storage direct flexible system in the park, real-time collaboration of photovoltaic, energy storage and flexible load equipment is required for power distribution and optimization to maintain system power balance and reduce energy cost. For example, the current park power dispatching mainly adopts a centralized optimization mode, that is, all terminal data is uploaded to a cloud center server, and a central algorithm is used to solve the global optimal dispatching instruction. However, this mode has some problems. On the one hand, real-time uploading of a large amount of terminal data leads to a significant increase in communication bandwidth pressure and a significant delay in dispatching decision. In addition, due to data privacy barriers between different park management subjects, cross-park power collaboration requires the exchange of sensitive operation data such as energy storage state of charge and load power consumption curve, which has the risk of business secret leakage, making it difficult to implement cross-regional collaborative dispatching, and the local optimization effect is not good due to limited adjustment resources in a single park.

[0003] In view of this, a power dispatching method for a photovoltaic energy storage direct flexible system based on cloud-edge collaboration is proposed. SUMMARY

[0004] The present application provides a power dispatching method for a photovoltaic energy storage direct flexible system based on cloud-edge collaboration, which is used to solve the problem of poor optimization effect of system power dispatching.

[0005] The present application provides a power dispatching method for a photovoltaic energy storage direct flexible system based on cloud-edge collaboration, comprising: Performing a dynamic partitioning operation at the beginning of each dispatching time according to the communication network between intelligent terminals in the park to divide a plurality of collaborative dispatching regions; wherein the dynamic partitioning operation considers both rough balance of power in the region and power distribution willingness of intelligent terminal users; Based on the partitioning result, a double-layer cloud-edge collaboration architecture is constructed, wherein the cloud computing center is used to store global dispatching parameters and decompose the system cost function, and the edge servers deployed in each collaborative dispatching region are used to receive the decomposed system cost function; Each edge server performs distributed power dispatching optimization calculation in the corresponding collaborative dispatching region to solve intelligent terminal dispatching instructions that meet power balance constraints and device operation constraints, and uploads the optimization result to the cloud computing center; Flexible load devices in each collaborative dispatching region are used for cross-regional secondary power dispatching, and a dynamic safety constraint model is used to protect device operation safety during the dispatching process.

[0006] Furthermore, based on the communication network between intelligent terminals within the park, a dynamic partitioning operation is performed at the start of each scheduling moment to divide the area into multiple collaborative scheduling regions, including: A partition target loss function is constructed, which includes a modularity term and a power balance term. The modularity term is used to measure the tightness of the connection between smart terminals in the partition, and the power balance term is used to measure the deviation between the power balance and the target power in the region. The importance of the two terms is adjusted by weighting coefficients. Initialize the willingness weight of each smart terminal user to participate in multiple regions, and dynamically adjust the willingness weight according to the overall power balance target of the regions at subsequent times; A node selection strategy is adopted, which selects new nodes to be added to the existing region by calculating the change value of the local loss function. Among them, nodes that reduce the local loss function are selected to be added to the region until the partitioning is completed.

[0007] Furthermore, the expression for the partitioning objective loss function is: in: for The loss function value at time t. Let be the total number of edges in the communication network. For nodes With nodes exist Edge weights at time step and They are nodes With nodes exist The degree of time, For indicator functions, when node and nodes The value is 1 if the members belong to the same community, and 0 otherwise. These are the weighting coefficients for the power balance term. This represents the total number of regions. For the region The set of nodes, For nodes exist Time allocation to region Willingness weight, For nodes exist Available power at any given moment.

[0008] Furthermore, the expression for modifying the willingness weight is: in: For nodes The generated new weight for the next moment , The intention weight of the node At the moment, The weight update coefficient, The absolute value of the disposable power of the node At the moment, The overall power balance target value of the system, The total number of intelligent terminals, The absolute value of the mean target power of a single node, The corrected weight will be In the partition target loss function of the next moment .

[0009] Further, the construction of the double-layer cloud-edge collaborative architecture based on the partition result comprises: The global scheduling parameters stored in the cloud computing center include the node topology relationship of each collaborative scheduling region, the power transmission constraint between regions, the real-time electricity price of the power grid, the overall power balance target value of the system, and the partition weight coefficient; The cloud computing center decomposes the system total cost function into a global optimization item and a regional local optimization item according to the collaborative scheduling region, wherein the global optimization item contains the cross-regional power interaction cost and the power grid purchase cost, and the regional local optimization item contains the operation cost of the optical storage equipment in each region and the load adjustment cost; The decomposed cost function and the regional constraint parameters associated with the cost function are distributed to the edge server of the corresponding collaborative scheduling region.

[0010] Further, the decomposition formula of the system total cost function is: Wherein: The global optimization item contains the power grid interaction cost and the system power balance penalty, The local optimization item of the region only depends on the device data in the region, The power grid interaction cost coefficient of the region is dynamically allocated by the cloud computing center according to the partition result, The power interaction cost function of the region and the power grid, The system overall power balance penalty coefficient, The absolute value of the disposable power of the node At Scheduled power at any given time , , For nodes The local operating cost coefficient of photovoltaic, energy storage, and load equipment. , , For nodes The corresponding sub-function for the operating cost of the equipment.

[0011] Furthermore, the constraints that the distributed power dispatch optimization calculation must satisfy include power balance constraints and equipment operation constraints, which are mathematically expressed as follows: Regional power balance constraints: in: For the region The power exchanged with the power grid has a positive input and a negative output. For the region Power interaction with other regions, with positive input and negative output; Equipment operating constraints: in: For photovoltaic nodes exist Maximum output at any moment , For energy storage nodes exist The charging and discharging power limit at any time, , For energy storage nodes exist Time and The state of charge of the energy storage at any given time. For rated capacity, , For energy storage nodes exist The charging and discharging power at any given time , For charging and discharging efficiency, For time intervals, , For load nodes exist Adjustable load power adjustment range at any time.

[0012] Furthermore, the distributed power dispatch optimization calculation includes: The edge server constructs a regional optimization target based on the received decomposition cost function and global parameters; The optimization problem is decomposed into local device optimization calculation of the intelligent terminal and regional power balance coordination calculation of the edge server, and a scheduling instruction is generated through a distributed iterative algorithm; The optimized regional total power, cross-region interaction demand and device safety state are uploaded to the cloud computing center.

[0013] Further, the flexible load device in each of the collaborative scheduling regions is used for cross-region secondary power scheduling, which comprises: Under the guidance of the cloud computing center, the remaining power regulation range of the flexible load device in each collaborative scheduling region is identified, and a cross-region power support instruction is issued to the edge server of the target region according to the system overall power shortage or surplus state; The edge server calculates the power adjustment amount of the flexible load based on the dynamic safety constraint model, generates a device-level scheduling instruction and executes it.

[0014] Further, the dynamic safety constraint model limits the power change rate and regulation range of the flexible load device in real time, and ensures the operation safety and service life of the device in the cross-region power scheduling process.

[0015] From the above technical solutions, the present application has the following advantages: The present application divides the system into multiple collaborative scheduling regions at the beginning of each scheduling moment through a dynamic partitioning algorithm, which quantifies the terminal communication density in the region by using a modularity index, reduces the collaborative delay, introduces a power balance term and a user willingness weight, and ensures rough power balance in the partition and respects user participation preferences; after partitioning, the region is autonomous, reducing direct data interaction between the massive terminals and the cloud center; secondly, a double-layer cloud-edge collaborative architecture is constructed based on the partitioning results; sensitive data is not taken out of the region, and global optimization only needs regional power aggregation values; finally, resource collaboration is achieved through two-stage scheduling. The present application breaks through the limitation of single park resources by using cross-region secondary scheduling of flexible load, effectively improves the system regulation capability; through cost function decoupling, the distributed optimization based on device operation constraints and the dynamic safety protection model can greatly reduce the total energy consumption cost of the system and reduce resource waste. BRIEF DESCRIPTION OF DRAWINGS

[0016] Fig. 1 An embodiment flowchart of a kind of photovoltaic storage direct flexible system power dispatching method based on cloud edge collaboration in the present application is shown in the figure; Fig. 2 Another embodiment flowchart of a kind of photovoltaic storage direct flexible system power dispatching method based on cloud edge collaboration in the present application is shown in the figure; Fig. 3Figure 2 is a flowchart illustrating another embodiment of a method for power dispatch of a photovoltaic storage direct flexible system based on cloud-edge collaboration according to the present application. DETAILED DESCRIPTION

[0017] The terms "first", "second", "third", "fourth" and the like in the description of the application and in the claims, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the use of the terms so termed herein is solely for descriptive purposes and not for pronouncing the limitations of the application described herein. Embodiments of the application described herein can be carried out in other sequences than the one described, unless otherwise specified herein. Furthermore, the terms "comprise" and "comprising", and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, product or apparatus that comprises a list of elements is not necessarily limited to those elements, but can include other elements not expressly listed or inherent to such process, method, system, product or apparatus.

[0018] Embodiment one The method implemented in this embodiment can be implemented in a system, and can be implemented in a server or in a terminal, and the specific implementation is not limited. From the perspective of system implementation, the method for power dispatch of a photovoltaic storage direct flexible system based on cloud-edge collaboration according to the present application will be introduced below. Please refer to Figs. 1 to 3 The method provided in the embodiment of the present application comprises the following steps: S1. At the beginning of each dispatch time, performing a dynamic partitioning operation according to the communication network between intelligent terminals in the park, and dividing a plurality of collaborative dispatching areas; wherein the dynamic partitioning operation considers both the rough balance of power in the area and the power allocation willingness of the intelligent terminal users; In this embodiment, the partitioning process takes double-target optimization as the core, on the one hand, maximizes the tightness of terminal communication connection in the area to reduce the collaborative delay, and on the other hand, pursues the rough balance of power within the area to reduce the cross-area dispatching pressure. The partitioning result directly affects the efficiency and safety of the subsequent cloud-edge collaborative dispatching, and the specific implementation process is as follows: S11. Constructing a partitioning target loss function, the loss function comprising a modularity term and a power balance term, wherein the modularity term is used to measure the tightness of intelligent terminal connection within the partitioning, and the power balance term is used to measure the deviation of power balance within the area from the target power, and the importance of the two terms is adjusted through a weight coefficient; The partitioning target loss function is constructed, and the expression is as follows: Wherein: is the loss function value at the time t, is the loss function value at the time t, is the total number of edges in the communication network, is the node and the node In the edge weight of time t, And The degree of node and node In time t, is an indicator function, which takes the value 1 when node and node belongs to the same community, otherwise 0, is the weight coefficient of the power balance term, is the total number of regions, is the node set of region , is the willingness weight of node assigned to region at time t, is the disposable power of node at time t. The partition target loss function compares the actual edge weight with the random continuous expectation value

[0019] , uses function to determine whether node and node are in the same region, and the same region Quantifies the degree of aggregation of communication network; adds the power weighted sum of nodes in each region , the smaller the value, the more balanced the regional power; the weight coefficient dynamically adjusts the importance of the two terms, Increase to give priority to power balance. S12. Initialize the willingness weight of each intelligent terminal user participating in multiple regions, and dynamically correct the willingness weight according to the overall power balance target of the region at the next time; Initialize the willingness weight of each terminal to the region

[0020] , which is equally allocated by default, and is corrected at the next time according to the following expression: Where: is the new weight generated by node for the next time , is the willingness weight of node at time t, is the weight update coefficient, is the willingness weight of node at time t, is the weight update coefficient, is the willingness weight of node at time t, is the weight update coefficient, The absolute value of available power at any given moment. The target value for overall system power balance. This represents the total number of smart terminals. The weighted average is the absolute value of the average target power of a single node. Will be the next moment In the partitioned objective loss function .

[0021] When node Actual power contribution at the previous moment Above the target average When contributing to the regional target mean, reduce its weight to alleviate pressure; when contributing to the regional target mean, increase its weight to incentivize participation and update the coefficients. Control the adjustment range to avoid weight oscillations.

[0022] S13. A node filtering strategy is adopted, which selects new nodes to be added to the existing region by calculating the change value of the local loss function. Among them, nodes that reduce the local loss function are selected to be added to the region until the partitioning is completed.

[0023] A greedy strategy is used to select nodes one by one: 1. The communication hub node is used as the initial region center; 2. The change in local loss of unassigned nodes when added to each region is calculated; 3. The node that maximizes the change in loss is selected to join the corresponding region, i.e., the loss function is minimized; 4. This process is repeated iteratively until the assignment of all nodes is determined, and the final partitioning scheme is output. .

[0024] S2. Construct a two-layer cloud-edge collaborative architecture based on the partitioning results. The cloud computing center is used to store global scheduling parameters and decompose the system cost function, while the edge servers deployed in each collaborative scheduling area are used to receive the decomposed system cost function. A set of collaborative scheduling regions generated based on dynamic partitioning A two-tier collaborative architecture of cloud computing center and edge server is constructed. The core principle of this architecture is: the cloud computing center stores cross-regional coupling parameters and decomposes system-level optimization problems; the edge server solves sub-problems based on local data to avoid cross-domain transmission of sensitive data; the partition weight coefficient directly affects the weight of cost function decomposition.

[0025] The detailed implementation of the specific architecture is as follows: 1. The global scheduling parameters stored in the cloud computing center include the node topology relationships of each collaborative scheduling region, inter-regional power transmission constraints, real-time electricity prices of the power grid, the overall power balance target value of the system, and the partition weight coefficients; 2The computing center decomposes the system total cost function into global optimization items and regional local optimization items according to the collaborative scheduling region, wherein the global optimization items include cross-regional power interaction cost and grid power purchase cost, and the regional local optimization items include light storage equipment operation cost and load adjustment cost in each region; 3The decomposed cost function and the regional constraint parameters associated with the cost function are distributed to the edge server of the corresponding collaborative scheduling region.

[0026] The regional node topology relationship is a node-region mapping table derived from the partition result , recording the region of each intelligent terminal; the cross-region transmission constraint is the power interaction amount between regions , determined by the physical line capacity; the real-time electricity price is the time-of-use grid power purchase price and the power selling price ; the balance target value is the system overall power target, generated according to the load prediction; the dynamic weight coefficient is the weight coefficient of the power balance item in the partition loss function, used to adjust the weight of the power balance item in the cost function.

[0027] The cloud computing center decouples the total cost function into: Wherein: is the global optimization item, including the grid interaction cost and the system power balance penalty, is the local optimization item of region , only depending on the device data in the region, is the grid interaction cost coefficient of region , dynamically allocated by the cloud computing center according to the partition result, is the power interaction cost function of region and the grid, is the system overall power balance penalty coefficient, is the schedulable power of node at time , , , is the local operation cost coefficient of the photovoltaic, energy storage and load devices of node , , , is the operation cost sub-function of the corresponding device of node .

[0028] S3. Each edge server performs distributed power dispatch optimization calculation in the corresponding collaborative dispatch area, solves the intelligent terminal dispatch instruction meeting the power balance constraint and device operation constraint, and uploads the optimization result to the cloud computing center; Each edge server performs distributed optimization calculation in the collaborative dispatch area, solves the dispatch instruction meeting the power balance constraint and device safety constraint, and realizes regional autonomous optimization. The technical principle is based on: using edge computing capability, relying only on regional node data (photovoltaic output, energy storage SOC, load power) for optimization, avoiding cross-domain transmission of raw data; the global power balance constraint is decomposed to the regional layer through the Lagrange multiplier, ensuring the consistency of local optimization and system target; the device operation constraint (such as energy storage SOC limit) is directly injected into the optimization model as a hard constraint to eliminate the risk of exceeding the limit. The implementation process is as follows: S31. The edge server constructs the regional optimization target based on the received decomposition cost function and global parameters; S32. The optimization problem is decomposed into local device optimization calculation of intelligent terminal and regional power balance coordination calculation of edge server, and the dispatch instruction is generated through distributed iterative algorithm; S33. The optimized regional total power, cross-region interaction demand and device safety state are uploaded to the cloud computing center.

[0029] The edge server constructs the local optimization problem based on the received S2 parameter package: The objective function is: Wherein: is the regional device cost, , is the global parameter issued by the cloud center.

[0030] The constraint conditions that the distributed power dispatch optimization calculation needs to meet include power balance constraint and device operation constraint, and the mathematical expression is: Regional power balance constraint: Wherein: is the interaction power of the regional with the power grid, the input is positive and the output is negative, is the power interaction amount of the regional with other regions, the input is positive and the output is negative; Device operation constraint: Wherein: is the photovoltaic node exist Maximum output at any moment , For energy storage nodes exist The charging and discharging power limit at any time, , For energy storage nodes exist Time and The state of charge of the energy storage at any given time. For rated capacity, , For energy storage nodes exist The charging and discharging power at any given time , For charging and discharging efficiency, For time intervals, , For load nodes exist Adjustable load power adjustment range at any time.

[0031] The alternating direction multiplier method is used for hierarchical iterative solution: each smart terminal Receive power reference value sent by the edge server and Lagrange multipliers Please solve: Edge servers collect all terminal power. Update the total power deviation in the region Update Lagrange multipliers Update reference power The termination condition is when (Tolerance threshold), or stop when iteration exceeds the limit. Finally, the edge server uploads an encrypted data packet to the cloud computing center, which includes the total dispatchable power of the region, the power exchanged with the grid, the cross-regional exchange demand; the average SOC of energy storage, the load adjustment margin; timestamp, digital signature; the upload protocol is transmitted via MQTT protocol through a TLS encrypted channel, and the data packet format is JSON-LD.

[0032] S4. Use flexible load devices in each collaborative scheduling area for cross-regional secondary power scheduling, and adopt a dynamic safety constraint model to protect the safe operation of the devices during the scheduling process.

[0033] After the regional autonomy optimization is completed, the power regulation margin of the flexible load device is used to implement cross-regional secondary scheduling to cope with global power fluctuations and improve system economy. The technical principle is based on: the remaining regulation capacity of the flexible load is used as a "power buffer pool" at the system level to solve the problem of short-term power shortage / surplus between regions; the dynamic security constraint model is used to limit the load power change rate and regulation range in real time to prevent equipment overload and damage; the cloud computing center coordinates globally and the edge server executes locally. The specific implementation process of cross-regional secondary power scheduling is as follows: 1. Under the guidance of the cloud computing center, the remaining power regulation range of the flexible load device in each collaborative scheduling region is identified. According to the system overall power shortage or surplus state, cross-regional power support instructions are issued to the edge server of the target region; 2. The edge server calculates the power adjustment amount of the flexible load based on the dynamic security constraint model, generates device-level scheduling instructions, and executes.

[0034] Among them, the dynamic security constraint model limits the power change rate and regulation range of the flexible load device in real time to ensure the operation safety and service life of the device during cross-regional power scheduling.

[0035] First, the cloud computing center generates a global flexible load resource map based on the uploaded optimization results, calculates the real-time regulation upper limit of the load node in each region, and forms a cross-regional scheduling resource pool. The formula is: the first term is the static power margin, the second term is the dynamic change rate limit ( is the safety factor); the adjustable power of each region is summarized to form a cross-regional scheduling resource pool. Second, the cloud computing center performs global secondary optimization, and the optimization goal is to minimize the total scheduling cost of the system: , is the regional regulation cost coefficient, is the global power deviation, and finally sends the scheduling instructions to the target edge server, so that the edge server executes the instructions based on the dynamic security constraint model, which includes converting the scheduling instructions into device-level safety instructions ; including real-time monitoring of load temperature and current harmonic distortion rate , if it exceeds the limit, it will trigger a rollback: ; upload the actual regulation amount and safety status code to the cloud computing center, where 0=normal, 1=partial rollback, and 2=complete failure.

[0036] ​​The above embodiment breaks through the limitation of single park resources by cross-region secondary scheduling of flexible load, and effectively improves system regulation capability; the distributed optimization and dynamic security protection model based on equipment operation constraints can greatly reduce the total energy cost of the system and reduce the waste of resources through cost function decoupling.

[0037] It can be understood that those skilled in the art can combine various embodiments in the above embodiments under the guidance of the above embodiments to obtain various embodiments of the technical solutions.

[0038] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A cloud-edge collaborative photovoltaic storage direct flexible system power dispatching method, characterized in that, The method comprises the following steps of: performing a dynamic partition operation at the beginning of each scheduling time according to a communication network among intelligent terminals in the park to divide a plurality of collaborative scheduling areas; wherein the dynamic partition operation simultaneously considers rough balance of power in the area and power distribution willingness of the intelligent terminal user; constructing a double-layer cloud-edge collaborative architecture based on the partition result, wherein a cloud computing center is used to store global scheduling parameters and decompose a system cost function, and an edge server deployed in each collaborative scheduling area is used to receive the decomposed system cost function; each edge server performs distributed power scheduling optimization calculation in the corresponding collaborative scheduling area to solve intelligent terminal scheduling instructions meeting power balance constraints and device operation constraints, and uploads the optimization result to the cloud computing center; flexible load devices in each collaborative scheduling area are used for cross-area secondary power scheduling, and a dynamic safety constraint model is used to protect device operation safety during the scheduling process.

2. The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 1, characterized in that, The method of performing a dynamic partition operation at the beginning of each scheduling time according to a communication network among intelligent terminals in the park to divide a plurality of collaborative scheduling areas comprises the following steps of: constructing a partition target loss function, wherein the partition target loss function comprises a modularity term and a power balance term, the modularity term is used to measure the closeness of intelligent terminal connection in the partition, and the power balance term is used to measure the deviation of power balance in the area from the target power, and the importance of the two terms is adjusted by a weight coefficient; initializing the willingness weight of each intelligent terminal user to participate in a plurality of areas, and dynamically correcting the willingness weight according to the overall power balance target of the area at a subsequent time; adopting a node screening strategy to select a new node to join an existing area by calculating the change value of a local loss function, wherein a node that reduces the local loss function is selected to join the area, and the partition is completed until the partition is completed. 3.The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 2, characterized in that, The expression of the partition target loss function is as follows: in: for The loss function value at time t. Let be the total number of edges in the communication network. For nodes With nodes exist Edge weights at time step and They are nodes With nodes exist The degree of time, For indicator functions, when node and nodes The value is 1 if the members belong to the same community, and 0 otherwise. These are the weighting coefficients for the power balance term. This represents the total number of regions. For the region The set of nodes, For nodes exist Time allocation to region Willingness weight, For nodes exist Available power at any given moment.

4. The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 3, characterized in that, The expression of the correction of the willingness weight is as follows: wherein: is the node generated for the next moment of new weights, is the node at the moment of will weights, is the weight update coefficient, is the node at the moment of absolute values of disposable power, is the overall power balance target value of the system, is the total number of intelligent terminals, is the absolute value of the mean target power of a single node, the corrected weight will be in the partition target loss function of the next moment .​​ 5. The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 1 or 2, characterized in that, The method of constructing a double-layer cloud-edge collaborative architecture based on the partition result comprises the following steps of: the global scheduling parameters stored in the cloud computing center comprise node topological relations of each collaborative scheduling area, power transmission constraints between areas, real-time power price of the power grid, overall power balance target value of the system and partition weight coefficients; the cloud computing center decomposes a system total cost function into a global optimization term and a regional local optimization term according to the collaborative scheduling area, wherein the global optimization term contains cross-area power interaction cost and power grid purchase cost, and the regional local optimization term contains light storage device operation cost and load adjustment cost in each area; the decomposed cost function and the area constraint parameters associated with the cost function are sent to the edge server of the corresponding collaborative scheduling area.

6. The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 5, characterized in that, The system total cost function The decomposition formula is: Wherein: is the global optimization item, including grid interaction cost and system power balance penalty, is the local optimization item of the region , only relying on the local device data of the region, is the grid interaction cost coefficient of the region , dynamically assigned by the cloud computing center according to the partition result, is the power interaction cost function of the region with the grid, is the system overall power balance penalty coefficient, is the schedulable power of the node at the time , , , is the local operation cost coefficient of the photovoltaic, energy storage, load device of the node , , , is the operation cost sub-function of the corresponding device of the node .

7. The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 6, characterized in that, The constraint conditions that need to be met by the distributed power scheduling optimization calculation include power balance constraints and device operation constraints, and the mathematical expression is as follows: area power balance constraint: where: is the region interaction power with the grid, input positive, output negative, is the region interaction power with other regions, input positive, output negative; device operation constraint: wherein: is the photovoltaic node at the maximum output power at the time instant, , is the energy storage node at the charge and discharge power limit at the time instant, , is the energy storage node at the energy storage state of charge at the time instant and at the time instant, is the rated capacity, , is the energy storage node at the charge and discharge power at the time instant, , is the charge and discharge efficiency, is the time interval, , is the load node at the adjustable load power regulation range at the time instant. 8.The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 7, characterized in that, The distributed power scheduling optimization calculation comprises the following steps of: the edge server constructs a regional optimization target based on the received decomposed cost function and global parameters; the optimization problem is decomposed into local device optimization calculation of the intelligent terminal and regional power balance coordination calculation of the edge server, and a scheduling instruction is generated through a distributed iteration algorithm; The optimized regional total power, cross-region interaction demand and equipment safety state are uploaded to a cloud computing center. 9.The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 1, wherein, The flexible load equipment in each of the collaborative scheduling regions is used for cross-region secondary power scheduling, including: Under the guidance of the cloud computing center, the remaining power regulation range of the flexible load equipment in each collaborative scheduling region is identified, and a cross-region power support instruction is issued to the edge server of the target region according to the overall power shortage or surplus state of the system; The edge server calculates the power adjustment amount of the flexible load based on a dynamic safety constraint model, generates a device-level scheduling instruction, and executes it.

10. The cloud-edge collaboration based power dispatching method for optical storage and direct flexible system according to claim 9, characterized in that, The dynamic safety constraint model ensures the operation safety and service life of the equipment in the cross-region power scheduling process by limiting the power change rate and regulation range of the flexible load equipment in real time.

Images