Park cloud management side end collaborative supply protection method

Abstract

The invention discloses a park cloud management side end collaborative insurance and supply method. According to the method, the asynchronous alternating direction multiplier method is introduced to carry out iterative solution, so that the park cloud center can update the global optimization variable without waiting for calculation results returned by all side energy stations, and the overall calculation efficiency bottleneck caused by communication delay of individual nodes is effectively overcome; therefore, the cooperative supply guarantee strategy can be quickly generated under the extreme working condition, and the timeliness and reliability of the power supply response of the power grid in the park are remarkably improved.

Description

Technical Field

[0001] This invention relates to the technical field of power system optimization operation, specifically to a collaborative power supply guarantee method for cloud-edge-end systems in industrial parks. Background Technology

[0002] Industrial parks typically contain various types of loads, including industrial, residential, and commercial loads, as well as clean energy sources such as photovoltaics. Energy utilization is low, and extreme operating conditions often occur (such as fluctuations in clean energy output leading to supply and demand imbalances). Traditional industrial parks have small amounts of clean energy sources and numerous points of access. Their dispersed and distributed access to the park makes it difficult to quickly respond to park dispatch instructions to ensure supply. Furthermore, the dispersed use of various distributed resources leads to low energy utilization. At the same time, traditional distributed processing methods (such as the alternating direction multiplier method) are mainly based on synchronous calculation. In the case of communication delays, it is necessary to wait for the information of the remaining sub-regions to be transmitted before further processing can be carried out, resulting in slow calculation and poor adaptability.

[0003] In summary, among the relevant technologies, when the synchronous distributed optimization algorithm is used in the power grid of the park under extreme conditions, there is a technical problem that the overall computational efficiency is low due to the communication delay of individual edge nodes, which cannot meet the requirements for rapid generation of supply guarantee strategies. Summary of the Invention

[0004] The technical problem this invention aims to solve is that, in related technologies, when using synchronous distributed optimization algorithms for power grids in industrial parks under extreme operating conditions, the overall computational efficiency is low due to communication delays at individual edge nodes, failing to meet the requirements for rapid generation of supply guarantee strategies. The purpose is to provide a collaborative supply guarantee method for cloud-management-edge-end systems in industrial parks, solving the technical problem of failing to meet the requirements for rapid generation of supply guarantee strategies.

[0005] This invention is achieved through the following technical solution:

[0006] In a first aspect, the present invention provides a method for coordinated power supply between cloud, network, edge, and terminal systems in a park. The method is executed by a controller in the park's cloud center, which is communicatively connected to multiple edge-side energy stations. The method includes:

[0007] Acquire park operation data; wherein, the park operation data includes at least the operation status information of the plurality of side energy stations;

[0008] Based on the park's operational data, determine whether to activate the collaborative supply guarantee mode;

[0009] In response to the activation of the collaborative supply guarantee mode, based on the preset global optimization model of the park cloud center and the local optimization model of each side energy station, the asynchronous alternating direction multiplier method is used for iterative solution to obtain the park supply guarantee strategy; wherein, the global optimization model of the park cloud center is a model used to characterize the overall economic efficiency and reliability collaborative optimization of the park, the local optimization model of each side energy station is a model used to characterize the internal distributed resource operation optimization, and the asynchronous alternating direction multiplier method is a distributed optimization algorithm used to perform asynchronous collaborative calculation between the park cloud center and the multiple side energy stations.

[0010] Furthermore, the step of acquiring park operation data includes:

[0011] The operating status information of the multiple edge energy stations is obtained through the cloud-edge communication network. The operating status information is uploaded by the corresponding edge energy station to the controller in the park cloud center. The operating status information uploaded by any edge energy station is obtained by processing the raw operating data collected by the corresponding end-side device of the energy station through the edge server of the energy station. The cloud-edge communication network is a fiber optic communication network. The edge server and the end-side device communicate via the MQTT-SN protocol.

[0012] Furthermore, the step of determining whether to activate the collaborative supply guarantee mode based on the park's operational data includes:

[0013] Based on the operating status information of the multiple side energy stations, determine whether there is a power supply shortage;

[0014] If a power supply shortage is detected, the coordinated power supply mode is activated.

[0015] Furthermore, the steps for responding to the activation of the collaborative supply guarantee mode, based on the preset global optimization model of the park cloud center and the local optimization model of each side energy station, and using the asynchronous alternating direction multiplier method for iterative solution to obtain the park supply guarantee strategy, include:

[0016] The initialization step is configured to set the initial global coordination variables and convergence conditions required by the asynchronous alternating direction multiplier method;

[0017] The step of executing the iterative loop is configured to execute multiple iterative loops until the convergence condition is met; wherein, a single iterative loop includes: sending the global coordination variable in the current iteration to each side energy station; receiving the local optimization result returned by at least one side energy station; wherein, each side energy station solves its local optimization model in parallel based on the received global coordination variable to obtain its own local optimization result, and the controller executes the subsequent steps immediately after receiving at least one local optimization result without waiting for all side energy stations; updating the global coordination variable based on the received local optimization result, and determining whether the updated result meets the convergence condition;

[0018] The execution strategy generation step is configured to generate the park supply guarantee strategy based on the global optimization model of the park cloud center, the local optimization models of each side energy station, and the final global coordination variables and local optimization results when the convergence condition is met.

[0019] Furthermore, the global optimization model of the park cloud center includes a first objective function and a first constraint condition; wherein, the first objective function is configured to characterize the total operating cost of the park, and includes an energy purchase cost item, a mobile energy station scheduling cost item, and a load shedding cost item; wherein, the energy purchase cost item characterizes the cost of purchasing electricity from the external power grid, the mobile energy station scheduling cost item characterizes the cost of calling up mobile power generation equipment and mobile energy storage equipment, and the load shedding cost item characterizes the load outage loss caused by insufficient power supply;

[0020] The constraints include at least a power balance constraint; wherein the power balance constraint is configured to consider the power supply delay effect caused by the dispatching and moving process of mobile energy stations when balancing the total power supply and total load power of the park; wherein the power supply delay effect is incorporated into the power balance constraint through a time delay model; wherein the time delay model is configured to quantify the moving time of mobile energy stations using an up-rounding function to determine their effective power supply time in the power balance.

[0021] Furthermore, the local optimization model for each side energy station includes a second objective function and a second constraint; wherein the second objective function is configured to characterize the local operating cost of the corresponding energy station and includes at least one of the following:

[0022] The clean energy consumption cost item is used to characterize the penalty cost incurred due to the incomplete utilization of the photovoltaic forecast output;

[0023] The energy storage operating cost item is used to characterize the cost generated by the charging and discharging operations of the energy storage equipment in the energy station.

[0024] The power generation cost item is used to characterize the cost of power generation by the micro gas turbines within the energy station.

[0025] The local offload cost item is used to characterize the cost caused by the failure to meet load demand within the power supply range of the energy station;

[0026] The energy interaction cost item is used to characterize the cost incurred by the energy station in interacting with the park's cloud center for power.

[0027] The second constraint includes at least a local power balance constraint to ensure the instantaneous balance between power supply and power consumption within the energy station.

[0028] Further, the step of updating the global coordination variable based on the received local optimization results includes:

[0029] Based on the local optimization results of at least one side energy station that have been received so far, update the Lagrange multipliers in the global coordination variables;

[0030] Based on the updated Lagrange multipliers and the local optimization results of the at least one side energy station, the consensus interaction power in the global coordination variable is updated;

[0031] The step of generating the park supply guarantee strategy based on the global optimization model of the park cloud center, the local optimization models of each side energy station, and the final global coordination variables and local optimization results includes:

[0032] Based on the global optimization model of the park cloud center and the local optimization model of each side energy station, the final global coordination variables and local optimization results determine at least one of the following control commands: the output plan command for distributed resources within each traditional energy station; the scheduling time and output command for mobile energy stations; and the adjustment command for adjustable loads within the park.

[0033] The control command is sent to the corresponding execution unit.

[0034] Secondly, the present invention provides a park cloud-edge-end collaborative power supply device, wherein the device is configured in the controller of the park cloud center, the controller of the park cloud center is communicatively connected to multiple edge-side energy stations, and the device includes:

[0035] The acquisition module is used to acquire park operation data; wherein, the park operation data includes at least the operation status information of the plurality of side energy stations;

[0036] The determination module is used to determine whether to activate the collaborative supply guarantee mode based on the park's operational data.

[0037] An optimization module, in response to the activation of the collaborative supply guarantee mode, uses an asynchronous alternating direction multiplier method to iteratively solve a solution based on a preset global optimization model of the park cloud center and local optimization models of each side energy station to obtain the park supply guarantee strategy. The global optimization model of the park cloud center is a model representing the overall economic efficiency and reliability of the park in a coordinated manner; the local optimization models of each side energy station are models representing the internal distributed resource operation optimization; and the asynchronous alternating direction multiplier method is a distributed optimization algorithm used for asynchronous collaborative computation between the park cloud center and the multiple side energy stations.

[0038] Thirdly, the present invention provides an electronic device, comprising: a memory, and one or more processors communicatively connected to the memory; the memory stores instructions executable by the one or more processors, the instructions being executed by the one or more processors to cause the one or more processors to implement the method described above.

[0039] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0040] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0041] This invention introduces an asynchronous alternating direction multiplier method for iterative solution, enabling the park cloud center to update global optimization variables without waiting for calculation results from all side energy stations. This effectively overcomes the overall computational efficiency bottleneck caused by communication delays of individual nodes, thereby enabling the rapid generation of collaborative power supply strategies under extreme conditions and significantly improving the timeliness and reliability of the park's power grid response. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0043] Figure 1 A flowchart illustrating a cloud-edge-device collaborative supply guarantee method for a park, as provided in the embodiments of this specification;

[0044] Figure 2 This is a block diagram of a cloud-edge-end collaborative supply guarantee device for a park, as provided in the embodiments of this specification.

[0045] Figure 3This is a block diagram of an electronic device provided in the embodiments of this specification. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0047] In related technologies, with the rapid development of energy internet technology, modern industrial park power systems often adopt a collaborative architecture that includes a cloud center, edge energy stations, and terminal devices to achieve optimized energy management. In this system architecture, the park dispatch center acts as the cloud core, responsible for overall control. Edge energy stations (e.g., traditional energy stations and mobile energy stations) distributed throughout the park are connected via power lines, responsible for supplying energy within their respective areas. Various energy devices act as terminals, collecting operational data in real time. This hierarchical distributed structure provides fundamental support for achieving precise energy supply within the park.

[0048] In related technologies, to meet the supply guarantee requirements under extreme operating conditions such as fluctuations in photovoltaic output, the industrial park can adopt a distributed optimization algorithm based on the synchronous alternating direction multiplier method. The workflow of this method is as follows: after the cloud center broadcasts the global coordination variables to all edge energy stations, it needs to wait for all nodes to complete their local optimization calculations and return the results before proceeding to the next round of iterative updates.

[0049] The aforementioned full-node synchronization mechanism revealed significant shortcomings in actual operation: due to the complex environment of the park and the strong heterogeneity of nodes, some edge energy stations often failed to respond in a timely manner due to communication network latency or limited computing resources. Once a single slow node appeared, the cloud center optimization process was blocked, and the entire system fell into a waiting state. This computational efficiency bottleneck caused by the "weakest link" effect resulted in a severe lag in the generation speed of optimization strategies in extreme, time-sensitive supply guarantee scenarios, making it difficult to meet the actual needs of rapid response.

[0050] At its root, the aforementioned technical problems stem from the inherent contradiction between the strong coupling characteristics of synchronization algorithms and the uncertainties of real-world distributed systems. Synchronization mechanisms require all nodes to coordinate within a strict time window, but uncertainties such as latency fluctuations in the campus power communication network and differences in node computing capabilities directly undermine the preconditions of this mechanism, causing the overall system performance to be constrained by the slowest node.

[0051] To address this technical bottleneck, the inventive concept of this invention lies in replacing the traditional synchronous mechanism with an asynchronous alternating directional multiplier method. This allows the cloud center to immediately perform a global update by receiving only the optimization results from a portion of the side energy stations during the iteration process, thereby completely eliminating the waiting time for slow nodes and ultimately achieving efficient and rapid generation of the park's supply guarantee strategy.

[0052] This embodiment provides a cloud-edge-device collaborative power supply method for industrial parks. The method is executed by a controller in the park's cloud center, which communicates with multiple edge energy stations. Specifically, the controller in the park's cloud center can be an industrial server deployed in the park's data center, an industrial computer, a cloud server, or a server cluster. The controller in the park's cloud center can communicate with an edge server within each edge energy station. This edge server can be the local control core of the edge energy station, used to aggregate and process raw operating data collected by its subordinate edge devices (e.g., photovoltaic inverters, energy storage converters, gas turbine controllers, load monitoring units, etc.), and to represent the energy station in data interaction and coordinated control with the park's cloud center.

[0053] like Figure 1 As shown, the method may include:

[0054] Step S12: Obtain park operation data; wherein, the park operation data includes at least the operation status information of the plurality of side energy stations;

[0055] In this embodiment, obtaining park operation data can be represented as the controller of the park cloud center obtaining the corresponding park operation data from each side energy station (edge ​​server) with which it is communicatively connected.

[0056] Specifically, the park operation data includes at least the operation status information of the multiple side energy stations. The operation status information may include: the real-time output power of the photovoltaic modules inside the side energy station, the current state of charge and charging / discharging power of the energy storage equipment, the current output power of the micro gas turbine, and the total real-time power of the load under the jurisdiction of the energy station.

[0057] In this embodiment, the controller can periodically or event-triggeredly retrieve or receive the operating status information pushed by the edge servers of each side energy station through a preset communication protocol interface.

[0058] In this embodiment, the park operation data may include not only the operation status information of the multiple side-side energy stations, but also external power grid data. Specifically, the external power grid data can represent the operation status and market information of the upstream power grid connected to the park. For example, it may include the power purchase limit from the external power grid, i.e., the upper and lower limits of the power that the park can obtain from the external power grid at any given time. It may also include time-of-use electricity price information, i.e., the sequence of changes in the unit price of electricity purchased from the external power grid over a future period. It may also include dispatch instructions or frequency support requirements from the external power grid, for example, when the main power grid frequency is abnormal, the park is required to reduce load or increase power generation output.

[0059] In this embodiment, the park operation data may include not only the operational status information of the multiple side energy stations, but also environmental and meteorological forecast data. Specifically, it may be solar irradiance prediction data for a future time period, used to predict the power generation capacity of the photovoltaic power generation system. It may also be ambient temperature, humidity, and wind speed prediction data.

[0060] In this embodiment, the park operation data may include not only the operation status information of the multiple side-side energy stations, but also park load-side data. For example, this includes classification information and distribution of important and general loads, the adjustment capacity and characteristic parameters of adjustable loads, and load forecast data.

[0061] Step S14: Based on the park operation data, determine whether to activate the collaborative supply guarantee mode.

[0062] In this embodiment, determining whether to activate the collaborative supply guarantee mode can be expressed as determining whether the park's power grid has entered an emergency operation state that requires the activation of high-level collaborative optimization functions.

[0063] Specifically, the determination process may include: First, calculating the total power generation and total load power of the park in real time based on the operating status information of the multiple side-side energy stations. Then, determining whether there is a power supply deficit. The power supply deficit can be represented as the situation where the total load power exceeds the sum of the total power generation and the maximum power that the park can purchase from the external power grid. Specific determination conditions may include: if the calculated power deficit continuously exceeds a preset power threshold for a preset time period, then a stable power supply deficit is determined to exist. Finally, if a power supply deficit is determined to exist, a start command is generated to activate the collaborative power supply guarantee mode.

[0064] Step S16: In response to the activation of the collaborative supply guarantee mode, based on the preset global optimization model of the park cloud center and the local optimization model of each side energy station, the asynchronous alternating direction multiplier method is used for iterative solution to obtain the park supply guarantee strategy; wherein, the global optimization model of the park cloud center is a model used to characterize the overall economic efficiency and reliability collaborative optimization of the park, the local optimization model of each side energy station is a model used to characterize the internal distributed resource operation optimization, and the asynchronous alternating direction multiplier method is a distributed optimization algorithm used to perform asynchronous collaborative calculation between the park cloud center and the multiple side energy stations.

[0065] In this embodiment, the global optimization model of the park cloud center can be an optimization model targeting the entire park system, with the goal of achieving synergistic optimization of the park's overall economy and reliability. Specifically, the economic optimization can be expressed as minimizing the park's total operating cost, which may include: the cost of purchasing electricity from the external grid, the fuel and maintenance costs of deploying mobile generators and mobile energy storage, and the compensation costs paid for calling up interruptible loads within the park. The reliability optimization can be expressed as minimizing load losses caused by insufficient power supply. For example, a high load loss penalty cost can be introduced into the objective function, or a minimum power supply reliability index that must be met can be set in the constraints.

[0066] In this embodiment, the local optimization model of each side energy station can be represented as a sub-problem where each energy station optimizes resources within its own jurisdiction, with the goal of optimizing the operation of its internal distributed resources. Specifically, the optimization objectives can be: minimizing the station's operating costs (e.g., fuel costs, maintenance costs), maximizing the local consumption of clean energy (e.g., photovoltaic) to reduce curtailment penalties, and meeting the interactive power commands issued by the cloud center. The optimization variables can be: the actual output setpoint of the photovoltaic system within the station, the charging and discharging power of the energy storage, the output of the micro gas turbine, and the interactive power between the station and the park's main grid.

[0067] In this embodiment, the asynchronous alternating direction multiplier method can be represented as an iterative algorithm for solving distributed optimization problems. It enables asynchronous collaborative computation, meaning the computation processes between the park cloud center and the multiple side energy stations are not strictly synchronized. Specifically, in each iteration: the controller can broadcast the global coordination variables of the current iteration to each side energy station. Each side energy station can solve its local optimization model in parallel and return the local optimization result to the controller. It should be noted that the controller does not need to wait for all side energy stations to return results; as long as it receives the local optimization result from at least one side energy station, it can update the global coordination variables based on the received result and proceed to the next judgment or iteration. This method effectively overcomes the problem of low overall computational efficiency caused by communication delays of individual nodes. The iterative process continues until the preset convergence condition is met, ultimately outputting the park's supply guarantee strategy. The supply guarantee strategy may include output planning instructions for distributed resources within each energy station, scheduling instructions for mobile energy stations, etc.

[0068] This implementation method introduces an asynchronous alternating direction multiplier method for iterative solution, enabling the park cloud center to update global optimization variables without waiting for all side energy stations to return calculation results. This effectively overcomes the overall computational efficiency bottleneck caused by communication delays of individual nodes, thereby enabling the rapid generation of collaborative power supply strategies under extreme conditions and significantly improving the timeliness and reliability of the park's power grid power supply response.

[0069] In some implementations, the step of acquiring park operation data includes:

[0070] Step S122: Obtain the operating status information of the multiple edge energy stations through the cloud-edge communication network; wherein, the operating status information is uploaded by the corresponding edge energy station to the controller of the park cloud center, and the operating status information uploaded by any edge energy station is obtained by processing the original operating data collected by the corresponding end-side device of the energy station through the edge server of the energy station; wherein, the cloud-edge communication network is an optical fiber communication network; the edge server and the end-side device communicate through the MQTT-SN protocol.

[0071] In this embodiment, the park cloud central controller can establish a stable data transmission channel with the edge energy station and reliably collect operational status information. Specifically, the cloud-edge communication network can be an optical fiber communication network.

[0072] In this embodiment, the end-side device can be used to collect raw operating data. The end-side device may include: a smart meter installed on the photovoltaic array, a battery management system for the energy storage system, a controller for the micro gas turbine, and a smart monitoring terminal on the load side, etc. This end-side device can collect raw operating data in real time, such as voltage, current, power, energy storage state of charge (SOC), and device start-up / shutdown status.

[0073] In this embodiment, the edge server can receive raw operational data from end-side devices. It can also perform preliminary integration, cleaning, and formatting of the raw operational data. For example, it can perform data quality checks to remove outliers, apply moving average filtering to high-frequency collected data to smooth fluctuations, convert data from different protocols of different devices into data packets of a unified format, and perform threshold analysis to determine whether the devices are operating normally. In this embodiment, after the above processing, standardized operational status information that can be used for optimization calculations can be formed.

[0074] In this embodiment, the operational status information processed by the edge server can be actively uploaded by the edge energy station to the controller of the park cloud center via the fiber optic communication network or uploaded in response to a request from the park cloud center. The MQTT-SN (MQTT for Sensor Networks) protocol is a lightweight message transmission protocol specifically designed for IoT sensor networks. This protocol is suitable for the communication needs of embedded devices such as photovoltaic inverters and sensors, and can achieve efficient aggregation of data from edge devices to the edge server.

[0075] In some implementations, the step of determining whether to activate the collaborative supply guarantee mode based on the park's operational data includes:

[0076] Step S142: Based on the operating status information of the multiple side-side energy stations, determine whether there is a power supply shortage.

[0077] In this embodiment, the controller of the park cloud center can perform real-time analysis on the collected power data of the entire network to detect whether there is an emergency situation where the total power supply is insufficient to meet the total power demand.

[0078] In this embodiment, the power supply deficit can be a state quantity, that is, the state in which the total load demand of the park's power grid exceeds the sum of the total power that can be provided by all currently available power sources (including local power generation from side energy stations, energy storage discharge, and available capacity of power purchased from external power grids).

[0079] In one possible and specific implementation, the existence of a power supply shortage can be determined based on the operating status information of the multiple side-side energy stations in the following way:

[0080] First, the controller can aggregate the operating status information uploaded by each side energy station. For example, the net power provided by each traditional energy station (industrial, residential, and commercial energy station) to its load, the discharge power of the energy storage devices in each station, the standby status and callable capacity of mobile energy stations, and the actual power purchased by the park from the external power grid and the maximum power purchase limit.

[0081] The controller can then calculate the current total load power of the park. The total load power can be the sum of the power supplied by each side energy station, or it can be the total power value obtained directly from the total inlet and outlet metering points of the park.

[0082] Next, the controller can calculate the current maximum available power supply for the park. This maximum available power supply can be the sum of the available power limits of all the aforementioned power sources, including: the maximum technical output of distributed generators within each traditional energy station, the maximum allowable discharge power of energy storage devices, and the maximum purchasable power limit of the external power grid. In extreme conditions, only the minimum power source combination necessary for ensuring power supply can be considered.

[0083] Finally, the controller can compare the total load power with the maximum available power in real time. This real-time comparison can be instantaneous or a power deficit threshold can be set. Only when the calculated power deficit continuously exceeds the threshold for a preset duration (e.g., two consecutive sampling periods) is it finally determined that there is a stable power supply deficit that requires the activation of the collaborative power supply guarantee mode.

[0084] Step S144: If it is determined that there is a power supply shortage, then the coordinated power supply mode is activated.

[0085] In this embodiment, if the determination in step S142 indicates a power supply shortage, the controller can generate a command signal to activate the collaborative power supply mode.

[0086] In some implementations, the step of responding to the activation of the collaborative supply guarantee mode by iteratively solving the problem using the asynchronous alternating direction multiplier method based on a preset global optimization model of the park cloud center and the local optimization models of each side energy station to obtain the park supply guarantee strategy includes:

[0087] Step S162: Perform the initialization step, which is configured to set the initial global coordination variables and convergence conditions required for the asynchronous alternating direction multiplier method.

[0088] In this embodiment, the initial global coordination variable can be represented as a variable maintained by the park controller and used to transmit optimization coordination information between the park controller and the side energy station.

[0089] In this embodiment, the initial global coordination variable may include a consensus variable and Lagrange multipliers. Specifically, the consensus variable can be used to characterize the intermediate value of the optimization decision that is expected to be agreed upon between the cloud center and the edge energy stations. It can be a temporary consensus value representing the planned interaction power between each edge energy station and the park's cloud center. For example, for a park with three energy stations, this variable can be a vector with three components, each component corresponding to the initial consensus value of the interaction power of one energy station. The Lagrange multipliers can be dual variables used to enforce consistency constraints in distributed optimization, and their values ​​can characterize the degree of difference between the global objective and the local decision.

[0090] In one possible and specific implementation, the initial values ​​of all consensus variables can be set to zero, indicating that at the initial moment, it is assumed that there is no power interaction between each energy station and the cloud center, and the initial values ​​of all Lagrange multipliers are also set to zero.

[0091] In one possible and specific implementation, initialization can be based on historical operating data or preset rules. For example, the initial value of the consensus variable can be set to the power value allocated according to the load forecast ratio of each station, or the initial value of the Lagrange multiplier can be set to a small constant.

[0092] In this embodiment, the initialization process may further include setting a penalty factor parameter, which is used to adjust the penalty intensity for constraint violations in the algorithm. The initial value of the penalty factor parameter can be empirically set according to the scale of the problem, for example, set to 100.

[0093] In this embodiment, the convergence condition can be a residual convergence condition. For example, it can be checked whether the original residual (i.e., the difference norm between the interaction power calculated locally by each side station and the current consensus variable) and the dual residual (i.e., the norm of the change in the consensus variable in two consecutive iterations) are both less than a preset absolute threshold (e.g., 0.01) or a relative threshold.

[0094] In this embodiment, the convergence condition can be a variable change convergence condition. For example, it can be checked whether the maximum or average change in the consensus interaction power vector of all energy stations in two consecutive iterations is lower than a set precision threshold (e.g., 0.05).

[0095] Step S164: The step of executing an iterative loop, which is configured to execute multiple iterative loops until the convergence condition is met; wherein, a single iterative loop includes:

[0096] Step S1641: Send the global coordination variables in the current iteration to each side energy station.

[0097] In this embodiment, the controller of the park cloud center can proactively distribute the latest coordination information to all side energy stations.

[0098] Specifically, the controller in the park's cloud center can use its communication interface to assemble the global coordination variables (i.e., the updated consensus variables and Lagrange multipliers) of the current iteration into a data packet. The transmission can be done in broadcast mode, sending the same data packet to all side stations at once, or in unicast or multicast mode, sending data to each side station separately. Data transmission can be based on the TCP / IP protocol to ensure reliability.

[0099] Step S1642: Receive local optimization results returned by at least one side energy station; wherein, each side energy station solves its local optimization model in parallel based on the received global coordination variables to obtain its own local optimization results, and the controller executes subsequent steps immediately after receiving at least one local optimization result without waiting for all side energy stations.

[0100] In this embodiment, the local optimization result can be represented as the output obtained by each side energy station independently solving its local optimization model based on the received global coordination variables. The parallel solution of the local optimization model by each side energy station based on the received global coordination variables can be represented as each side station simultaneously and independently starting its local computation process after receiving the variables broadcast by the cloud center. The solver used by each station can be the same, or different optimization algorithms can be used due to the different complexity of their local models.

[0101] In this embodiment, the controller executes subsequent steps immediately upon receiving at least one local optimization result, without waiting for all side-side energy stations (specifically, the controller of the park cloud center can set a receiving window, but does not block waiting for responses from all side-side stations). Once any one or more side-side stations have completed their calculations and returned their local optimization results to the controller, the controller, upon recognizing the arrival of a result (e.g., by listening to data packets on the corresponding port or checking the message queue), immediately terminates the waiting state of the current round and proceeds to execute subsequent update steps. The state of side-side stations that have not yet returned results will be temporarily suspended in this iteration.

[0102] Step S1643: Update the global coordination variable based on the received local optimization results, and determine whether the updated result satisfies the convergence condition.

[0103] In this embodiment, updating the global coordination variable based on the received local optimization results can be represented as the controller of the park cloud center using only the local optimization results of the side stations successfully received in the current iteration to perform the update calculation. For side stations that have returned results, the controller of the park cloud center can calculate the new values ​​of the consensus variable and Lagrange multiplier corresponding to the side station according to the update rules of the asynchronous alternating direction multiplier method. For side stations that have not returned results in this iteration, their corresponding global coordination variables (consensus variable and Lagrange multiplier) can remain unchanged from the previous iteration or be interpolated or extrapolated. It should be noted that their update does not depend on new information not yet received in this round.

[0104] In this implementation, determining whether the updated result meets the convergence condition can specifically involve checking whether the maximum value or norm of the change in the consensus variable of all side stations (including those that have been updated and those that have not) in two consecutive iterations is less than the convergence precision set during initialization. Alternatively, it can involve checking whether both the original residual and the dual residual are sufficiently small.

[0105] If the calculation results indicate that the convergence condition is met, the loop terminates and the process proceeds to step S166. Otherwise, the iteration counter is incremented, the process returns to the beginning of this step, and a new round of looping begins, broadcasting the updated global coordination variables again.

[0106] Step S166: Execute the strategy generation step, which is configured to generate the park supply guarantee strategy based on the global optimization model of the park cloud center, the local optimization model of each side energy station, and the final global coordination variables and local optimization results when the convergence condition is met.

[0107] In this embodiment, the converged solution achieved by the asynchronous iterative solution process can be transformed and integrated into a set of executable control instructions. The park supply guarantee strategy can be represented as a coordinated control command used to eliminate power shortages, achieve energy supply and demand balance in the park, and take into account both economy and reliability.

[0108] In this embodiment, when the iterative loop terminates due to the convergence condition, the controller can use the final global coordination variables and the final local optimization results returned by each side station, combined with the optimization objectives and constraints defined by the global optimization model of the park cloud center and the local optimization models of each side energy station, to generate the park supply guarantee strategy.

[0109] In one possible and specific implementation, the park's supply guarantee strategy may include: issuing output planning instructions to distributed resources (e.g., photovoltaic, energy storage, gas turbines) within each traditional energy station; issuing scheduling time, target location, and output instructions to mobile energy stations; and issuing adjustment instructions (e.g., load reduction instructions) to adjustable loads within the park.

[0110] In this embodiment, the controller can send the specific control commands mentioned above to the corresponding execution units (e.g., the control system of the energy station, the dispatch center of the mobile generator, and the load control terminal) through the communication network, thereby completing the deployment and execution of the supply guarantee strategy.

[0111] This implementation employs an asynchronous alternating direction multiplier method for iterative solution. In each iteration, the park's cloud center does not need to wait for calculation results from all edge energy stations; it only needs to receive the local optimization result from at least one node to update global variables and advance the calculation process. This asynchronous mechanism effectively overcomes the overall computational blocking problem caused by communication delays of individual edge nodes, thereby significantly improving the efficiency and speed of optimization solutions. This ensures the rapid generation of reliable power supply strategies under extreme conditions of the park's power grid, greatly enhancing the timeliness of emergency power supply response.

[0112] In some implementations, the global optimization model of the park cloud center includes a first objective function and a first constraint; wherein, the first objective function is configured to characterize the total operating cost of the park and includes an energy purchase cost item, a mobile energy station scheduling cost item, and a load shedding cost item; wherein, the energy purchase cost item characterizes the cost of purchasing electricity from the external power grid, the mobile energy station scheduling cost item characterizes the cost of calling up mobile power generation equipment and mobile energy storage equipment, and the load shedding cost item characterizes the load outage loss caused by insufficient power supply.

[0113] In this embodiment, the first objective function can be a scalar function, which can be quantified and optimized through a mathematical minimization process. Specifically, the total operating cost is constructed as a weighted sum of three main cost items.

[0114] The energy purchase cost item can be used to measure the economic cost incurred by the park in purchasing electricity from the external main power grid. Specifically, this cost can be expressed as the sum of the products of the purchased power and the time-of-use electricity price for each time period within the dispatch cycle.

[0115] The mobile energy station dispatch cost item measures all expenses incurred in calling up emergency resources such as mobile generators and mobile energy storage vehicles, specifically including fuel costs, transportation costs, and equipment depreciation. This cost item links the output power of the mobile energy station to the unit dispatch cost.

[0116] The offload cost quantifies the economic losses caused by power outages due to insufficient power supply capacity, and this cost is directly related to power supply reliability. This cost can be expressed as the sum of the products of the offload power of different priority loads (e.g., industrial, commercial, and residential loads) and their corresponding unit outage loss value. Minimizing this cost term means maximizing power supply reliability.

[0117] The constraints include at least a power balance constraint; wherein the power balance constraint is configured to consider the power supply delay effect caused by the dispatching and moving process of mobile energy stations when balancing the total power supply and total load power of the park; wherein the power supply delay effect is incorporated into the power balance constraint through a time delay model; wherein the time delay model is configured to quantify the moving time of mobile energy stations using an up-rounding function to determine their effective power supply time in the power balance.

[0118] In this embodiment, the first constraint can be represented as the physical laws and system operation limitations that must be satisfied during the optimization process. Among them, the power balance constraint is the most critical physical constraint, which states that at any given time, the total power supply of the park must equal the total load power. In this embodiment, the refined modeling of this constraint specifically considers the power supply time delay effect of mobile energy stations.

[0119] Specifically, the power supply time delay effect can be expressed as the time delay between the mobile energy station receiving the dispatch instruction and arriving at the site to actually provide effective power support. To account for this time delay in the optimization model, a time delay model is introduced.

[0120] In one possible and specific implementation, the time-delay model can be configured to quantify the travel time of mobile power stations using an up-rounding function. Specifically, first, the estimated travel time (e.g., K minutes) of the mobile power station from its standby location to the target power supply location can be obtained. Then, this travel time can be divided by the basic scheduling interval used by the optimization model (e.g., 15 minutes) to obtain a quotient. Finally, the up-rounding function can be applied to this quotient, that is, rounding up non-integer multiples of the travel time to integer multiples of the basic scheduling interval. The result of this operation is the power supply time delay step of the mobile power station on the time scale of the optimization model. When establishing power balance constraints, in the power balance equation at time t, the contribution power of the mobile power station is not taken from time t, but from a future time interval separated by the power supply time delay step (i.e., t + up-rounded (K / 15) time). This modeling approach enables the optimization model to anticipate the scheduling of mobile power stations, thereby generating a more time-accurate and practically feasible power supply guarantee strategy.

[0121] In some implementations, the local optimization model of each side energy station includes a second objective function and a second constraint; wherein the second objective function is configured to characterize the local operating cost of the corresponding energy station.

[0122] Specifically, the second objective function can be a predefined mathematical expression used to minimize the total local operating cost of the side energy station itself, and it includes at least one of the following:

[0123] The clean energy consumption cost item is used to characterize the penalty cost incurred due to the incomplete utilization of the photovoltaic forecast output.

[0124] In this embodiment, the clean energy consumption cost item is used to incentivize the maximization of local consumption of clean energy such as photovoltaic power, thereby avoiding energy waste. This cost item can be configured as a penalty mechanism. Specifically, when the actual output of photovoltaic equipment is lower than its predicted maximum power generation capacity, curtailment occurs. This cost item can be calculated based on the amount of curtailed power and a preset unit penalty price, that is, minimizing the amount of curtailment through economic leverage.

[0125] The energy storage operating cost item is used to characterize the cost generated by the charging and discharging operations of the energy storage equipment in the energy station.

[0126] In this embodiment, the energy storage operating cost item can be the losses and maintenance costs of the energy storage device during the charge and discharge cycle. This energy storage operating cost item can be proportional to the charge and discharge power of the energy storage device, and is used to balance the number and depth of energy storage use in optimization to avoid unnecessary equipment losses.

[0127] The power generation cost item is used to characterize the cost of power generation by the micro gas turbines within the energy station.

[0128] In this embodiment, the power generation cost item can be the cost of fuel consumed by conventional power generation equipment such as micro gas turbines. This power generation cost item can be expressed as a function of the gas turbine's output power (e.g., a linear or quadratic function) to prioritize the scheduling of low-cost resources in optimization.

[0129] The local offload cost item represents the cost incurred due to unmet load demand within the power supply range of the energy station.

[0130] In this embodiment, the local off-load cost item can be used to quantify the losses caused by power outages within the power supply range due to insufficient power supply capacity of the energy station. This local off-load cost item is directly related to the power supply reliability of the station, and its calculation is the product of the off-load power within the energy station's range and a preset, relatively high unit off-load penalty price. This unit price can be differentiated according to load type (e.g., industrial, commercial, residential).

[0131] The energy interaction cost item is used to characterize the cost incurred by the energy station in interacting with the park's cloud center.

[0132] In this embodiment, the energy interaction cost can be represented as the product of the interaction power and the interaction electricity price (penalty price). When the energy station absorbs power from the outside (interaction power is positive), a cost is incurred; when it outputs power to the outside (interaction power is negative), a revenue is generated (i.e., a negative cost).

[0133] The second constraint includes at least a local power balance constraint to ensure the instantaneous balance between power supply and power consumption within the energy station.

[0134] In this embodiment, the most fundamental constraint in the second constraint condition can be the local power balance constraint. This local power balance constraint ensures that at any given time, the total power generated by all power sources within the energy station equals the algebraic sum of the power consumed by all loads and the power exchanged with external systems. Specifically, this local power balance constraint can be configured as follows: the sum of photovoltaic output, energy storage discharge power, and micro gas turbine power generation within the energy station, minus the energy storage charging power, plus the power exchanged with external systems (or minus the power exchanged with external systems), must equal the power supplied to the loads by the station (including satisfied loads and unloaded power). This constraint is fundamental to ensuring the real-time stable operation of the energy station.

[0135] In this embodiment, the second constraint may further include upper limit constraints on photovoltaic output, upper and lower limits constraints on the state of charge and charging / discharging power of energy storage devices, upper and lower limits constraints on the output of micro gas turbines, and constraints on interactive power transmission capacity, etc. By solving this local optimization model, each side-side energy station can optimally schedule its internal resources while meeting local operating constraints, and feed its optimization results back to the cloud center to jointly achieve global optimization.

[0136] In some implementations, the step of updating the global coordination variable based on the received local optimization results includes:

[0137] Step S16431: Based on the local optimization results of at least one side energy station that has been received, update the Lagrange multipliers in the global coordination variables.

[0138] In this implementation, the dual variable (i.e., the Lagrange multiplier) representing the degree of violation of coupling constraints can be adjusted based on the local optimization information returned in the current iteration. The purpose is to guide the local optimization results of each side energy station to converge towards satisfying the global consistency constraint through an economic penalty signal.

[0139] Specifically, the Lagrange multiplier can be the dual variable in the asynchronous alternating direction multiplier method, which can be expressed as the unit penalty price corresponding to the degree of violation of the coupling constraint (which can be power balance or interactive power consistency constraint) between the optimization objective of the park cloud center and the local optimization objective of the side energy station.

[0140] In this embodiment, the Lagrange multipliers in the global coordination variable can be updated using gradient ascent or dual ascent. That is, the Lagrange multipliers are modified based on the direction of the difference between the current local decision and the global consensus.

[0141] In one possible and specific implementation, firstly, for each side-side energy station that has returned a local optimization result in this iteration, the controller can calculate the difference between the interaction power value in its local optimization result and the consensus interaction power value corresponding to that station in the current iteration. Then, the controller can scale this difference using a preset penalty factor parameter. Next, this scaled difference can be added to the Lagrange multiplier value of the side-side station in the previous iteration to obtain the updated Lagrange multiplier value for the next iteration. For side-side energy stations that have not returned a local optimization result in this iteration, their corresponding Lagrange multipliers remain unchanged in this update, i.e., they use their values ​​from the previous iteration. Through the above operations, the Lagrange multipliers of side-side stations that provided local results in this iteration are adjusted in real time according to the deviation between their local decision and the global consensus. If the local interaction power is higher than the consensus value, the multipliers tend to increase, thereby increasing the cost of generating power beyond the consensus in the next iteration and suppressing its behavior, and vice versa.

[0142] Step S16432: Based on the updated Lagrange multipliers and the local optimization results of the at least one side energy station, update the consensus interaction power in the global coordination variable.

[0143] In this implementation, the global consensus value (i.e., the interaction power reference value) used to coordinate the optimization behavior of each side station can be recalculated and updated under the guidance of the new penalty signal (updated Lagrange multiplier) in order to gradually narrow the gap between the global objective and the local objective.

[0144] In this embodiment, the consensus interaction power can be represented as a global consensus variable maintained by the park cloud center, representing the expected value of the interaction power of each side energy station. It can serve as a reference benchmark connecting global optimization and local optimization.

[0145] In one possible and specific implementation, the controller can use the latest penalty signal (updated Lagrange multipliers) and the obtained local information (received local optimization results from side stations) to calculate a new and better interactive power reference value according to the rules of the asynchronous alternating direction multiplier method.

[0146] In one possible and specific implementation, an average or weighted average can be used to perform the update. For example, the updated consensus interaction power value can be obtained by taking a weighted average or arithmetic average of the interaction power allocation value calculated by the cloud center based on the global model in the current iteration and the local expected interaction power value returned by the side station.

[0147] The step of generating the park supply guarantee strategy based on the global optimization model of the park cloud center, the local optimization models of each side energy station, and the final global coordination variables and local optimization results includes:

[0148] Step S1662: Based on the global optimization model of the park cloud center, the local optimization model of each side energy station, the final global coordination variables and local optimization results, determine at least one of the following control commands: the output plan command of the distributed resources within each traditional energy station; the scheduling time and output command of the mobile energy station; and the adjustment command of the adjustable load within the park.

[0149] In this embodiment, the output plan instructions for distributed resources within each traditional energy station can be determined based on the global optimization model of the park cloud center, the local optimization models of each side energy station, and the final global coordination variables and local optimization results. Specifically, for each traditional energy station, the controller can generate a detailed power output setpoint sequence for each controllable distributed resource within it during one or more subsequent scheduling periods. The output plan instructions for distributed resources within each traditional energy station can further include: photovoltaic output instructions, energy storage system instructions, and micro gas turbine instructions. More specifically, the photovoltaic output instruction can be represented as the active power output setpoint of the photovoltaic inverter, which is directly derived from the decision variables for the actual photovoltaic output in the local optimization results. The energy storage system instruction can be represented as the charging or discharging power setpoint of the energy storage in each time period and possible mode switching instructions. This value can be derived from the charging and discharging power decision variables of the energy storage in the local optimization results. The micro gas turbine instruction can be the start / stop state of the micro gas turbine and its active power output setpoint. This value can be derived from the output decision variables of the micro gas turbine in the local optimization results.

[0150] In this embodiment, the scheduling time and output command of the mobile energy station can be determined based on the global optimization model of the park cloud center, the local optimization models of each side energy station, and the final global coordination variables and local optimization results. Specifically, the scheduling time and output command of the mobile energy station can include a scheduling time command and an output command. The scheduling time command can represent the final time requirement for the mobile energy station to arrive at the designated power supply location, or the starting time for power supply. It can be derived from the effective power supply time determined by considering time lag in the optimization model. The output command can represent the power generation curve that the mobile generator needs to output after arrival, or the charging and discharging power plan of the mobile energy storage at the designated location. This value can correspond to the value of the decision variables related to the mobile energy station in the global optimization model after convergence.

[0151] In this embodiment, adjustment commands for adjustable loads within the park can be determined based on the global optimization model of the park cloud center, the local optimization models of each side energy station, and the final global coordination variables and local optimization results. Specifically, the adjustment commands for adjustable loads within the park can include: load interruption commands and load reduction commands. Load interruption commands can specify a list of load units that need to have their power completely cut off and the start and end times of the interruption. Load reduction commands can specify the load units that need to reduce their operating power, their target power value or load reduction ratio, and the duration.

[0152] Step S1664: Send the control command to the corresponding execution unit.

[0153] In this embodiment, the control system can encapsulate different types of control commands according to a predefined communication protocol and assign a unique address identifier to the target execution unit for each command or set of commands. For example, independent network addresses are assigned to the local controller of each traditional energy station, the on-board control system of each mobile generator vehicle, and the load management terminal.

[0154] In one specific implementation plan, a collaborative supply guarantee method based on asynchronous alternating direction multiplier method and energy station cloud-edge-end is provided.

[0155] It should be noted that:

[0156] (1) The amount of clean energy in the park is small and there are many points. Their decentralized and distributed access to the park makes it difficult to respond quickly to the park's dispatch instructions to ensure supply. The decentralized use of various distributed resources leads to low energy utilization.

[0157] (2) When extreme working conditions occur (such as fluctuations in clean energy output leading to supply and demand shortages), traditional distributed processing methods (such as the alternating direction multiplier method) rely mainly on synchronous calculation. In the case of communication delay, it is necessary to wait for the information of the remaining sub-regions to be transmitted before further processing can be carried out, which has problems such as slow calculation and poor adaptability.

[0158] (3) During the supply guarantee process in the park, the dispatch of mobile power generation vehicles takes a certain amount of time to arrive at the site. The impact of power supply delay is usually not taken into account during the supply guarantee process in the park, especially in the dispatch of mobile energy stations.

[0159] (4) Traditional park supply guarantee is based on distributed resource decentralized scheduling, which consumes a lot of manpower, material resources and economic costs. It lacks efficient centralized energy supply forms of energy stations of different types, and also lacks cloud-edge-end collaborative supply guarantee methods based on the collaborative processing of park cloud center, communication, side energy stations and end equipment.

[0160] To address the aforementioned shortcomings, this implementation proposes a collaborative power supply guarantee method for a park-wide cloud-management-edge-end based on the asynchronous alternating direction multiplier method and energy stations. This method falls under the field of power system optimization and operation technology, specifically involving a collaborative power supply guarantee method for a park-wide cloud-management-edge-end based on the asynchronous alternating direction multiplier method and energy stations. By integrating distributed resources within the park through energy stations and constructing a collaborative power supply guarantee framework based on this, under extreme operating conditions, traditional energy stations on the edge perform local optimization based on equipment-side information monitoring. Considering the impact of power supply delays from mobile energy stations, the park cloud center coordinates the traditional energy stations and mobile energy stations as a whole based on the asynchronous alternating direction multiplier method, ultimately obtaining the overall power supply guarantee strategy for the park. Cloud-edge communication is implemented via optical fiber, and edge-end communication is implemented based on the MQTT-SN lightweight communication protocol, effectively solving problems such as communication latency and slow computation speed during park power supply guarantee.

[0161] This embodiment provides a collaborative power supply method for park cloud-edge-device systems based on the asynchronous alternating direction multiplier method and energy stations. The implementation process includes the following steps:

[0162] 1. A collaborative power supply framework for cloud-management-edge-device systems in industrial parks based on energy stations.

[0163] The park comprises a park dispatch center, traditional energy stations, mobile energy stations, transformers, power lines, residential buildings, industrial plants, and commercial buildings. The park dispatch center centrally manages park resources, including traditional energy station dispatch, direct load regulation, and mobile energy station dispatch. There are three traditional energy stations, each supplying energy to one of the residential buildings, one of the industrial plants, and one of the commercial buildings, respectively. These three stations are interconnected via power lines and transformers, enabling energy supply under normal conditions and energy redundancy in emergencies. Each energy station consists of photovoltaic panels, energy storage, and a micro gas turbine. The mobile energy stations consist of mobile generators and mobile energy storage, primarily responsible for short-term power supply to the load under emergency or extreme conditions. In the park, the proportions of important loads and general loads are 20% and 80%, respectively. Important loads are non-adjustable rigid loads, referring to the continuous production loads in the industrial plant clusters. General loads refer to other loads, residential loads, and commercial loads in the industrial plant clusters. The proportions of adjustable loads in industrial, residential, and commercial general loads are 20%, 30%, and 40%, respectively. All of them are directly controlled loads, that is, they are directly connected to the park's load direct control system and can be directly dispatched and regulated by the park's dispatch center.

[0164] The park's critical load supply is ensured through a cloud-edge-device collaborative supply guarantee system. The cloud center serves as the park's dispatch center, the edge refers to three traditional energy stations and mobile energy stations, the device refers to the energy equipment within the energy stations, and the network refers to the communication between the cloud, edge, and device. Cloud-edge communication is achieved via fiber optic cable, while edge-device communication is based on the MQTT-SN lightweight communication protocol. In extreme cases, sensors at the energy equipment end collect real-time data on equipment operating voltage, current, power, and energy storage, transmitting this data to the energy station edge servers for preprocessing, fusion, and simple analysis (such as threshold analysis and anomaly analysis). The energy station edge servers then transmit the analysis results to the park's cloud center. The park's cloud center further optimizes the overall system based on the power output provided by each energy station, considering factors such as economy and reliability. Dispatch instructions are then issued to each traditional and mobile energy station. Traditional energy stations formulate their own output plans, considering internal equipment operating constraints, power constraints from interactions with other energy stations (including mobile energy stations), and load reliability constraints. The optimized results are fed back to the cloud center, which further optimizes the instructions based on the feedback from the energy stations. Ultimately, the optimal collaborative supply guarantee solution is obtained through iterative optimization. In this process, the iterative optimization between the cloud center and the energy station is based on the asynchronous alternating direction multiplier method. Compared with the traditional alternating direction multiplier method, the asynchronous alternating direction multiplier method enables the cloud center to make scheduling decisions by updating only some of the energy station output information. It can achieve faster convergence speed while ensuring scheduling effect and meeting the requirements of rapid supply guarantee.

[0165] 2. Supply guarantee and dispatch model of the park's cloud center

[0166] 2.1 Objective Function

[0167] The cloud center in the park aims to minimize both economic and reliability costs, where economic costs include energy purchase costs. Mobile energy station dispatch costs Reliability cost is the cost of load failure. Reliability cost can also be understood as supply guarantee cost. The lower the reliability cost, the better the supply guarantee effect, as detailed below:

[0168] (1);

[0169] (2);

[0170] (3);

[0171] (4);

[0172] In the formula, This can be expressed as the unit price of electricity purchased from the external power grid at time t; This represents the active power purchased by the park from the external power grid at time t; The unit cost of generating electricity for a mobile generator vehicle (which is a type of mobile energy station); This represents the power generation capacity of the mobile generator in the mobile energy station at time t. This represents the unit cost of operating a mobile energy storage device (belonging to a mobile energy station) at time t; and These are the charging power and discharging power of the mobile energy storage device at time t, respectively. , , These represent the unit power loss cost caused by power outages for industrial, residential, and commercial loads at time t, respectively. , , These represent the unmet power (i.e., unloaded power) for industrial, residential, and commercial loads at time t. t represents time, and T represents the dispatch period, with a time interval of 15 minutes and a value of 96 for T.

[0173] 2.2 Constraints

[0174] 2.2.1 Energy Purchase Cost Constraints

[0175] (5);

[0176] In the formula, , These represent the upper and lower limits of the power purchase capacity at time t, respectively. This can be expressed as the technical or contractual upper limit of the power that can be purchased from the external power grid at time t; It can be expressed as the technical or contractual lower limit of the power that can be purchased from the external power grid at time t. It can be expressed as the active power value that the park plans to purchase from the external power grid at time t.

[0177] 2.2.2 Constraints of Mobile Energy Stations

[0178] 2.2.2.1 Constraints of Mobile Generator Vehicle

[0179] (6);

[0180] In the formula, As a decision variable, it can represent the planned power generation capacity of the mobile generator vehicle numbered m at time t; This is a preset constant or upper bound, representing the maximum allowable power (power limit) that the mobile generator vehicle numbered m can generate at time t. is a preset constant or lower bound, representing the minimum output (power lower limit) required for the mobile generator car numbered m to maintain stable operation at time t.

[0181] 2.2.2.2 Constraints of Mobile Energy Storage

[0182] (7);

[0183] (8);

[0184] In the formula, and Both are decision variables, representing the charging power and discharging power of the mobile energy storage device numbered m at time t, respectively. and Both are preset constants, representing the maximum allowable charging power and the maximum allowable discharging power of the mobile energy storage device at time t, respectively. Let m be the decision variable, representing the energy stored in the mobile energy storage device numbered m at time t. and These are preset constants, representing the maximum upper limit (capacity) and minimum lower limit of energy storage for the mobile energy storage device, respectively. Let m be the decision variable, representing the energy value stored in the mobile energy storage device numbered m at the next time t+1. and The preset efficiency coefficients (ranging from 0 to 1) represent the charging efficiency and discharging efficiency of the mobile energy storage device, respectively. and These are preset boundary condition values. This represents the initial energy of the moving energy storage at the beginning of the scheduling period (t=0); This represents the energy value of the mobile energy storage at the end of the scheduling period (t=T). t is the time index in the optimization model, and m is the index of the mobile energy station.

[0185] 2.2.4 Power Balance Constraints

[0186] Since the dispatching of mobile generators requires a certain amount of time, this model incorporates the power supply delay effect of mobile energy stations. Furthermore, it innovatively simplifies the power supply delay effect of mobile energy stations by rounding up using a rounding function. Therefore, the power balance constraints are as follows:

[0187] (9);

[0188] In the formula, Let be the decision variable, representing the active power purchased by the park from the external power grid at time t; Let be the decision variable, representing the time t+ The planned power generation capacity of the mobile generator vehicle numbered m; specifically, K is a known or estimated quantity, representing the travel time required for the mobile generator vehicle to travel from its standby location to the target power supply location. This represents the floor function, which can convert continuous movement time K into integer multiples of 15-minute basic scheduling intervals, thereby determining the number of delay periods before power supply takes effect; The physical meaning could be that a dispatch command is issued at time t, but the actual power supply contribution of the mobile generator will be... It only becomes apparent after a certain period of time. and Let m represent the discharge power and charging power of the mobile energy storage device numbered m at time t, respectively. These represent the power supplied by the park's cloud center to traditional energy stations that provide power to industrial, residential, and commercial loads, respectively.

[0189] 3. Optimization model for ensuring power supply to traditional energy stations on the edge of the industrial park

[0190] 3.1 Objective Function

[0191] Traditional energy stations must balance supply and demand with economic efficiency, reliability, and clean energy integration, including the cost of clean energy integration. Energy storage operating costs Cost of generating electricity from micro gas turbines Reliability cost Energy interaction costs The details are as follows:

[0192] (10);

[0193] (11);

[0194] (12);

[0195] (13);

[0196] (14);

[0197] (15);

[0198] In the formula, The total operating cost of the energy station supplying energy to type n load, where n is an index parameter with values ​​{ind,res,com}, representing the energy station supplying energy to industrial load / residential load / commercial load respectively; The penalty cost related to the consumption of photovoltaic clean energy at this site is the economic penalty incurred when the actual output of photovoltaic power is lower than the predicted value, which is used to incentivize the maximization of clean energy consumption. Costs related to the operation of the energy storage equipment at this station; Fuel costs associated with the micro gas turbine power generation at this station; Costs related to the reliability of power supply to this station, namely the cost of losses due to power outages caused by insufficient power supply; Costs associated with power exchange between this station and the park's cloud center or other energy stations; The price of photovoltaic curtailment penalty at time t. Let t be the predicted value of the photovoltaic output of this station. The actual output of the photovoltaic system at this station at time t. Let t be the unit operating cost of the energy storage equipment at this station. Let t be the charging power of the energy storage at this station. Let t be the discharge power of the energy stored at this station. Let t be the unit load loss cost of this station. Let t be the power of the station when it loses load. Let t be the unit power generation cost of the micro gas turbine at this station. To determine the actual output power of the micro gas turbine at this station, Let t be the unit energy interaction cost of this station. This represents the interaction power between energy stations. A positive value indicates that the power flows into the energy station, a negative value indicates that the power flows out of the energy station, and a value of 0 indicates that the energy station does not interact with other energy stations.

[0199] 3.2 Constraints

[0200] 3.2.1 Photovoltaic power output constraints

[0201] (16);

[0202] 3.2.2 Energy Storage Constraints

[0203] (17);

[0204] (18);

[0205] In the formula, Let be the decision variable, representing the planned active power output of the photovoltaic equipment of an energy station serving type n loads at time t; The upper bound is set as a preset value, representing the predicted maximum photovoltaic output of the energy station at time t. and Let be the decision variables, representing the charging power and discharging power of the energy station's energy storage at time t, respectively. and are preset constants, representing the maximum allowable charging power and the maximum allowable discharging power of the energy storage at time t, respectively. and These are preset constants, representing the maximum upper limit (rated capacity) and minimum lower limit of energy storage capacity of the energy storage device, respectively. , , These represent the energy stored in the traditional energy station at times t, t+1, the initial state, and T, respectively. , These represent the energy storage charging and discharging efficiency in the energy station.

[0206] 3.2.3 Load Supply Constraints

[0207] (19);

[0208] In the formula, Let be the decision variable, representing the unmet load power, i.e., the unloaded power, within the power supply range of an energy station serving type n loads (industrial ind, residential res, or commercial com) at time t; t is a preset constant, representing the lower limit of the allowable power loss at time t for this station. t is a preset constant, representing the maximum allowable power loss at time t for this station.

[0209] 3.2.4 Constraints of Micro Gas Turbines

[0210] (20);

[0211] In the formula, t is a preset constant, representing the minimum stable output power allowed by this micro gas turbine technology at time t. Let be the decision variable, representing the planned power generation of the micro gas turbine at time t. t is a preset constant, representing the rated maximum output power of the micro gas turbine at time t.

[0212] 3.2.5 Interactive Power Constraints

[0213] (twenty one)

[0214] In the formula, For decision variables, it represents the net exchange power between the station and the park cloud center (or other stations through the cloud center) at time t; This is a preset constant, representing the lower limit of the allowed interactive power at time t; t is a preset constant, representing the upper limit of the allowed interactive power at time t.

[0215] 3.2.6 Power Balance Constraints

[0216] (twenty two)

[0217] In the formula, The power generation capacity of photovoltaic equipment. This refers to the discharge power of the energy storage device. The charging power for energy storage devices, The power generation capacity of the micro gas turbine. This refers to the interaction power between the station and external networks. This represents the actual load power satisfied by the station.

[0218] 4. A collaborative supply guarantee solution for cloud-edge-device systems in industrial parks based on the asynchronous alternating direction multiplier method.

[0219] Traditional alternating direction multiplier methods require all edge energy stations to complete their solutions before the cloud center can coordinate overall. However, due to potential communication delays between the cloud center, edge traditional energy stations, and end-devices, the optimization strategies of the edge energy stations may not reach the cloud center simultaneously, increasing the cloud center's waiting time. This results in slow computation speed for the park's supply guarantee strategy, making it unsuitable for the rapid response requirements under extreme conditions. Therefore, unlike the traditional synchronous alternating direction multiplier method, an asynchronous alternating direction multiplier method is designed for collaborative supply guarantee between the park's cloud management system and its edge devices. The steps are as follows:

[0220] 4.1 Sensors at the energy equipment end collect data such as equipment operating voltage, current, power, and energy storage status in real time, and transmit them to the side-side traditional energy station server based on the MQTT-SN lightweight communication protocol. The side-side traditional energy station server determines whether extreme operating conditions have occurred based on threshold analysis (such as whether there is a significant power shortage in the next 15 minutes). If no extreme operating conditions occur, the side-side energy station operates according to the established normal strategy. If an extreme operating condition is determined to have occurred, the side-side energy station transmits the determination result to the park cloud center through fiber optic communication, starts the park supply guarantee model, and sets the park extreme operating condition as the initial operating condition of the park cloud center supply guarantee scheduling model and the traditional energy station supply guarantee optimization model.

[0221] 4.2 Introducing Lagrange multipliers and punishment factors The following penalty costs will be added to the optimization models for the park cloud center and traditional energy stations, respectively:

[0222] (twenty three);

[0223] (twenty four);

[0224] In the formula, This refers to the additional penalty cost incurred by the park cloud center model in the current iteration due to the incomplete satisfaction of coupling constraints; These are the Lagrange multipliers (dual variables) corresponding to the park cloud center model. This is the penalty factor (penalty parameter) for the park cloud center model. , as well as Both are decision variables, representing the interactive power values ​​calculated by the park's cloud center at time t, which are planned to be allocated to traditional energy stations that supply energy to industrial loads (ind), residential loads (res), and commercial loads (com). , as well as These represent the energy station interaction power supplying energy to industrial, residential, and commercial loads, respectively. This represents the interactive power of the energy station supplying energy to a load of type n. This is the additional penalty cost incurred by the nth side energy station (n can represent an industrial, residential, or commercial station) in the current iteration due to the failure to satisfy coupling constraints; The Lagrange multiplier corresponding to the nth side energy station; The penalty factor for the nth side energy station. This represents the expected interaction power between the nth side energy station and the cloud center (or other stations) after its own optimization calculation at time t. The input value from the cloud center (which is a known quantity in the current iteration for the side station model) represents the reference value of the interaction power calculated by the cloud center at time t and planned to be allocated to the nth side station.

[0225] The updated formula is:

[0226] (25);

[0227] (26);

[0228] in, and Let be the Lagrange multipliers of the park's cloud center (main station), and let represent the values ​​at the k-th and k+1-th iterations, respectively. and Let be the Lagrange multipliers of the side energy station n (industrial, residential, or commercial station), and let represent the values ​​at the k-th and k+1-th iterations, respectively. and The penalty factor (step size parameter) is used for the park cloud center and the side energy station n, respectively. , and Let be the decision variable, representing the interactive power allocated to each type of edge energy station calculated by the park cloud center at time t in the k-th iteration. Specifically, represent the power allocation values ​​for industrial load (ind), residential load (res), and commercial load (com), respectively. , and Let be the reference value of the expected interactive power of the side energy stations, representing the expected interactive power of each side station (industrial, residential, commercial) after local optimization at time t in the k-th iteration. is the reference value of the interactive power of the edge energy station n in the (k+1)th iteration. It is a consensus variable in the algorithm and is used to coordinate the power exchange target between the cloud center and the edge station. This represents the interaction power value allocated by the park cloud center to the side station n at time t during the k-th iteration. This represents the expected interaction power value calculated locally by the side station n at time t in the k-th iteration. This represents the number of iterations. For convergence accuracy.

[0229] 4.3 Load initial values =0, = =0, = =100, k=0, =0.01.

[0230] 4.4 The park's cloud center broadcasts interactive power to all adjacent traditional energy stations via fiber optic communication. and The traditional energy station on the side solves its own supply guarantee optimization model based on this and the edge MQTT-SN lightweight communication protocol.

[0231] 4.5 The traditional energy station on the side transmits the supply guarantee optimization results to the park cloud center through optical fiber communication. In order to avoid the slow calculation speed of the park supply guarantee strategy and the difficulty in adapting to the response requirements under extreme working conditions caused by the communication delay between the traditional energy station cloud and the side, when the park cloud center receives at least one supply guarantee interaction strategy from the traditional energy station on the side, that is, when it updates the interaction power according to formula (25). , .

[0232] 4.6 k=k+1, determine whether the convergence condition (26) is satisfied. If it is satisfied, stop the iteration and output the supply guarantee strategy of the cloud center and energy station; otherwise, return to 4.4.

[0233] This implementation method has significant advantages: (1) Energy stations naturally have the advantage of efficient energy utilization. By integrating distributed resources through various energy stations in the park, the energy utilization rate can be improved, and the efficiency of distributed resource scheduling can be enhanced. (2) The asynchronous alternating direction multiplier method is used to handle the cloud-edge-end collaborative supply guarantee problem in the park, which can avoid the waiting time caused by communication delay and improve the calculation speed and environmental adaptability of the supply guarantee method. (3) The supply guarantee scheduling of the park cloud center takes into account the power supply delay of mobile energy stations and simplifies the process by using an up-rounding function, resulting in a supply guarantee strategy that is more realistic and reasonable. (4) The park supply guarantee based on the cloud-edge-end collaborative method can make full use of the existing resources in the park and improve the supply guarantee response speed and supply guarantee effect.

[0234] This implementation provides a cloud-edge-device collaborative power supply framework for the park based on energy stations, including the park cloud center, edge energy stations (mainly referring to traditional energy station scheduling), internal equipment of the device-side energy stations, and cloud-edge-device communication methods.

[0235] This implementation establishes a park cloud center power supply scheduling model, especially a power supply time delay effect model for mobile energy stations during the scheduling process; this implementation establishes a park-side traditional energy station power supply optimization model; the park cloud-edge-end collaborative power supply method based on the asynchronous alternating direction multiplier method and energy stations provided in this implementation includes four parts: a park cloud-edge-end collaborative power supply framework based on energy stations, a park cloud center power supply scheduling model, a park-side traditional energy station power supply optimization model, and a park cloud-edge-end collaborative power supply solution method based on the asynchronous alternating direction multiplier method; the park cloud-edge-end collaborative power supply solution method based on the asynchronous alternating direction multiplier method provided in this implementation includes six steps: extreme condition determination, power supply model loading, initial value loading, local power supply optimization of side-side traditional energy stations, global power supply scheduling optimization of the park cloud center, and convergence judgment.

[0236] like Figure 2 As shown, according to an embodiment of the present invention, a campus cloud-edge-end collaborative power supply device is provided, characterized in that the device is configured in the controller of the campus cloud center, the controller of the campus cloud center is communicatively connected to multiple edge-side energy stations, and the device includes:

[0237] The acquisition module is used to acquire park operation data; wherein, the park operation data includes at least the operation status information of the plurality of side energy stations;

[0238] The determination module is used to determine whether to activate the collaborative supply guarantee mode based on the park's operational data.

[0239] An optimization module, in response to the activation of the collaborative supply guarantee mode, uses an asynchronous alternating direction multiplier method to iteratively solve a solution based on a preset global optimization model of the park cloud center and local optimization models of each side energy station to obtain the park supply guarantee strategy. The global optimization model of the park cloud center is a model representing the overall economic efficiency and reliability of the park in a coordinated manner; the local optimization models of each side energy station are models representing the internal distributed resource operation optimization; and the asynchronous alternating direction multiplier method is a distributed optimization algorithm used for asynchronous collaborative computation between the park cloud center and the multiple side energy stations.

[0240] According to an embodiment of the present invention, an electronic device is provided; please refer to... Figure 3 The electronic device in this embodiment may include one or more of the following components: a processor, a network interface, memory, non-volatile memory, and one or more application programs, wherein the one or more application programs may be stored in non-volatile memory and configured to be executed by one or more processors, and the one or more programs are configured to perform the methods as described in the foregoing method embodiments.

[0241] According to embodiments of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a computer, causes the computer to perform the method described in any of the above embodiments.

[0242] According to embodiments of the present invention, a computer program product comprising instructions is also provided, which, when executed by a computer, cause the computer to perform a method in any of the above embodiments. The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for ensuring supply through cloud-edge-device collaboration in a park, characterized in that, The method is executed by the controller of the park cloud center, which is connected to multiple side-side energy stations via communication. The method includes: Acquire park operation data; wherein, the park operation data includes at least the operation status information of the plurality of side energy stations; Based on the park's operational data, determine whether to activate the collaborative supply guarantee mode; In response to the activation of the collaborative supply guarantee mode, based on the preset global optimization model of the park cloud center and the local optimization model of each side energy station, the asynchronous alternating direction multiplier method is used for iterative solution to obtain the park supply guarantee strategy; wherein, the global optimization model of the park cloud center is a model used to characterize the overall economic efficiency and reliability collaborative optimization of the park, the local optimization model of each side energy station is a model used to characterize the internal distributed resource operation optimization, and the asynchronous alternating direction multiplier method is a distributed optimization algorithm used to perform asynchronous collaborative calculation between the park cloud center and the multiple side energy stations.

2. The method according to claim 1, characterized in that, The steps for obtaining park operation data include: The operating status information of the multiple edge energy stations is obtained through the cloud-edge communication network. The operating status information is uploaded by the corresponding edge energy station to the controller in the park cloud center. The operating status information uploaded by any edge energy station is obtained by processing the raw operating data collected by the corresponding end-side device of the energy station through the edge server of the energy station. The cloud-edge communication network is a fiber optic communication network. The edge server and the end-side device communicate via the MQTT-SN protocol.

3. The method according to claim 1, characterized in that, The step of determining whether to activate the collaborative supply guarantee mode based on the park's operational data includes: Based on the operating status information of the multiple side energy stations, determine whether there is a power supply shortage; If a power supply shortage is detected, the coordinated power supply mode is activated.

4. The method according to claim 1, characterized in that, The steps for responding to the activation of the collaborative supply guarantee mode, based on the preset global optimization model of the park cloud center and the local optimization model of each side energy station, and using the asynchronous alternating direction multiplier method for iterative solution to obtain the park supply guarantee strategy, include: The initialization step is configured to set the initial global coordination variables and convergence conditions required by the asynchronous alternating direction multiplier method; The step of executing the iterative loop is configured to execute multiple iterative loops until the convergence condition is met; wherein, a single iterative loop includes: sending the global coordination variable in the current iteration to each side energy station; receiving the local optimization result returned by at least one side energy station; wherein, each side energy station solves its local optimization model in parallel based on the received global coordination variable to obtain its own local optimization result, and the controller executes the subsequent steps immediately after receiving at least one local optimization result without waiting for all side energy stations; updating the global coordination variable based on the received local optimization result, and determining whether the updated result meets the convergence condition; The execution strategy generation step is configured to generate the park supply guarantee strategy based on the global optimization model of the park cloud center, the local optimization models of each side energy station, and the final global coordination variables and local optimization results when the convergence condition is met.

5. The method according to claim 4, characterized in that, The global optimization model of the park cloud center includes a first objective function and a first constraint condition; wherein, the first objective function is configured to represent the total operating cost of the park, and includes an energy purchase cost item, a mobile energy station scheduling cost item, and a load shedding cost item; wherein, the energy purchase cost item represents the cost of purchasing electricity from the external power grid, the mobile energy station scheduling cost item represents the cost of calling mobile power generation equipment and mobile energy storage equipment, and the load shedding cost item represents the load outage loss caused by insufficient power supply; The constraints include at least a power balance constraint; wherein the power balance constraint is configured to consider the power supply delay effect caused by the dispatching and moving process of mobile energy stations when balancing the total power supply and total load power of the park; wherein the power supply delay effect is incorporated into the power balance constraint through a time delay model; wherein the time delay model is configured to quantify the moving time of mobile energy stations using an up-rounding function to determine their effective power supply time in the power balance.

6. The method according to claim 5, characterized in that, The local optimization model for each side energy station includes a second objective function and a second constraint; wherein the second objective function is configured to characterize the local operating cost of the corresponding energy station and includes at least one of the following: The clean energy consumption cost item is used to characterize the penalty cost incurred due to the incomplete utilization of the photovoltaic forecast output; The energy storage operating cost item is used to characterize the cost generated by the charging and discharging operations of the energy storage equipment in the energy station. The power generation cost item is used to characterize the cost of power generation by the micro gas turbines within the energy station. The local offload cost item is used to characterize the cost caused by the failure to meet load demand within the power supply range of the energy station; The energy interaction cost item is used to characterize the cost incurred by the energy station in interacting with the park's cloud center for power. The second constraint includes at least a local power balance constraint to ensure the instantaneous balance between power supply and power consumption within the energy station.

7. The method according to claim 4, characterized in that, The step of updating the global coordination variable based on the received local optimization results includes: Based on the local optimization results of at least one side energy station that have been received so far, update the Lagrange multipliers in the global coordination variables; Based on the updated Lagrange multipliers and the local optimization results of the at least one side energy station, the consensus interaction power in the global coordination variable is updated; The step of generating the park supply guarantee strategy based on the global optimization model of the park cloud center, the local optimization models of each side energy station, and the final global coordination variables and local optimization results includes: Based on the global optimization model of the park cloud center and the local optimization model of each side energy station, the final global coordination variables and local optimization results determine at least one of the following control commands: the output plan command for distributed resources within each traditional energy station; the scheduling time and output command for mobile energy stations; and the adjustment command for adjustable loads within the park. The control command is sent to the corresponding execution unit.

8. A cloud-edge-end collaborative supply guarantee device for industrial parks, characterized in that, The device is configured in the controller of the park cloud center, which is communicatively connected to multiple side-side energy stations. The device includes: The acquisition module is used to acquire park operation data; wherein, the park operation data includes at least the operation status information of the plurality of side energy stations; The determination module is used to determine whether to activate the collaborative supply guarantee mode based on the park's operational data. An optimization module, in response to the activation of the collaborative supply guarantee mode, uses an asynchronous alternating direction multiplier method to iteratively solve a solution based on a preset global optimization model of the park cloud center and local optimization models of each side energy station to obtain the park supply guarantee strategy. The global optimization model of the park cloud center is a model representing the overall economic efficiency and reliability of the park in a coordinated manner; the local optimization models of each side energy station are models representing the internal distributed resource operation optimization; and the asynchronous alternating direction multiplier method is a distributed optimization algorithm used for asynchronous collaborative computation between the park cloud center and the multiple side energy stations.

9. An electronic device, characterized in that, include: A memory, and one or more processors communicatively connected to the memory; The memory stores instructions that can be executed by the one or more processors to cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

Images