A micro-grid cloud-edge collaborative autonomous control method and system, a storage medium and an electronic device

Abstract

The application provides a micro-grid cloud edge collaborative autonomous control method and system, a storage medium and an electronic device, which are applied to a three-level control architecture including a cloud decision layer, a back-end scheduling layer and an edge EMS control layer. The cloud decision layer generates photovoltaic output prediction values and load prediction values for a plurality of future prediction periods; the back-end scheduling layer generates a cloud scheduling strategy accordingly; the edge EMS control layer compares actual data with a collection time identifier with prediction values for corresponding prediction periods, selects a cloud optimal operation mode or an edge offline autonomous mode in combination with a deviation level and a communication state, actively switches to the edge offline autonomous mode when communication is normal but prediction is seriously mismatched, and generates local control instructions according to real-time field data; and when recovery conditions are met, the cloud scheduling is recovered in a manner of limiting energy storage power variation and recovering adjustable loads in batches. The scheme improves control continuity and operation adaptability under cloud strategy mismatch or communication abnormality.

Description

Technical Field

[0001] This application belongs to the field of microgrid control technology, and relates to a microgrid cloud-edge collaborative autonomous control method, system, storage medium and electronic device. Background Technology

[0002] With the development of new energy power generation technology, energy storage technology, and smart power distribution technology, distributed photovoltaics, energy storage devices, charging facilities, and various types of electrical loads are increasingly widely used in industrial parks, commercial buildings, and remote power supply areas. Microgrids can coordinate and manage distributed power sources, energy storage devices, electrical loads, and the external power grid within a certain area, and have become an important component of local utilization of new energy and flexible operation of the power system.

[0003] As the number of connected devices and the scale of operational data in microgrids continue to increase, microgrid operation and control are gradually exhibiting characteristics such as diverse equipment types, frequent changes in operating status, multiple control objectives, and complex operating conditions. During microgrid operation, it is necessary not only to ensure power supply continuity and system security, but also to consider electricity costs, renewable energy utilization rates, the operating status of energy storage devices, and load electricity demand.

[0004] Existing microgrid control technologies are still insufficient in adapting to changes in renewable energy output, electricity load, and energy storage status. When operating conditions change significantly, issues such as mismatch between control results and actual operating needs, untimely power regulation, power fluctuations at grid connection points, and unreasonable charging and discharging of energy storage may occur.

[0005] Furthermore, the continuity and reliability of microgrid control are easily affected by communication anomalies, data delays, control information failures, or equipment malfunctions. The load structure, renewable energy output characteristics, equipment operating status, and external environment of a microgrid change over time. Existing control parameters and operating strategies lack long-term adaptability, easily leading to problems such as decreased control accuracy, unstable operating performance, and poor adaptability to different application scenarios. Summary of the Invention

[0006] This application provides a cloud-edge collaborative autonomous control method, system, storage medium, and electronic device for microgrids, which addresses the problems of insufficient adaptability of existing microgrid control technologies to changes in operating conditions, poor control continuity and reliability under abnormal conditions, decreased accuracy, and unstable operating performance.

[0007] In a first aspect, this application provides a microgrid cloud-edge collaborative autonomous control method, comprising: a cloud decision layer generating photovoltaic power output forecasts and load forecasts corresponding to multiple future forecast periods, and sending the photovoltaic power output forecasts and load forecasts to a back-end scheduling layer; the back-end scheduling layer receiving microgrid operation data reported by an edge EMS control layer, generating a cloud scheduling strategy based on the photovoltaic power output forecasts, the load forecasts, and the microgrid operation data, and sending the photovoltaic power output forecasts, load forecasts, and the cloud scheduling strategy to the edge EMS control layer; the edge EMS control layer collecting actual photovoltaic power output, actual load power, energy storage state of charge, and grid connection point demand with a collection time identifier, determining the corresponding forecast period based on the collection time identifier, comparing the actual photovoltaic power output with the photovoltaic power output forecast corresponding to the forecast period, and comparing the actual load power with the load forecast corresponding to the forecast period. The predicted load values ​​are compared, and the deviation level is determined based on the comparison results. The edge EMS control layer detects the communication status between itself and the backend scheduling layer, and selects either cloud-optimized operation mode or edge offline autonomous mode based on the communication status and the deviation level. Specifically, when the communication status is abnormal, or when the communication status is normal and the deviation level is determined to be a preset severe abnormality level, the edge EMS control layer switches to the edge offline autonomous mode. In the cloud-optimized operation mode, the edge EMS control layer executes the cloud scheduling strategy. In the edge offline autonomous mode, the edge EMS control layer generates local control commands for controlling at least one of the energy storage system and adjustable load based on the grid connection point demand, the actual photovoltaic output, the actual load power, and the energy storage state of charge. When the communication status returns to normal and the deviation level returns to a preset normal level, the edge EMS control layer resumes executing the cloud scheduling strategy.

[0008] In one implementation of the first aspect, the cloud-based decision-making layer generates the photovoltaic output forecast value and the load forecast value based on historical microgrid operating data, environmental meteorological data, and real-time microgrid operating data; the historical operating data includes historical photovoltaic output and historical load power, and the environmental meteorological data includes at least one of irradiance and ambient temperature; the back-end scheduling layer generates the cloud-based scheduling strategy based on time-of-use pricing, grid connection point demand constraints, energy storage charging and discharging safety constraints, and equipment operation constraints, and the cloud-based scheduling strategy includes energy storage charging and discharging plans corresponding to the multiple forecast periods.

[0009] In one implementation of the first aspect, the plurality of prediction time periods correspond to a preset prediction time granularity; the edge EMS control layer collects the actual photovoltaic output and the actual load power according to a preset sampling period, and preprocesses the sampled data within an evaluation window to generate field evaluation data for judging the deviation level, the field evaluation data including photovoltaic output evaluation value and load power evaluation value; the edge EMS control layer determines the deviation level based on the deviation between the photovoltaic output evaluation value and the corresponding photovoltaic output prediction value and the deviation between the load power evaluation value and the corresponding load prediction value.

[0010] In one implementation of the first aspect, the deviation level includes at least the preset normal level, the preset general anomaly level, and the preset severe anomaly level; when the communication status is normal, the cloud scheduling strategy is executed under the preset normal level; under the preset general anomaly level, the cloud scheduling strategy continues to be executed and abnormal data is uploaded to the backend scheduling layer; under the preset severe anomaly level, the system switches to the edge offline autonomous mode; the operating mode switch is performed only when the corresponding switching condition is met for a consecutive preset number of evaluation windows of the deviation level, and the entry threshold for switching from the cloud-optimized operating mode to the edge offline autonomous mode is different from the exit threshold for reverting from the edge offline autonomous mode to the cloud-optimized operating mode.

[0011] In one implementation of the first aspect, under the edge offline autonomous mode, provided that the energy storage state of charge and energy storage charging / discharging power are within safe limits, the edge EMS control layer generates the local control command as follows: when the grid connection point demand meets the preset demand warning condition, the edge EMS control layer selects to execute one or more of the following in the order of stopping energy storage charging, controlling energy storage discharging when the energy storage state of charge is higher than the discharge lower limit, limiting the charging facility power, and cutting off non-critical loads, until the grid connection point demand meets the preset demand control condition; when the grid connection point demand does not meet the preset demand warning condition and When the actual output of the photovoltaic system exceeds the actual load power, the edge EMS control layer controls the energy storage system to absorb the surplus photovoltaic power that the on-site load has not absorbed. When the energy storage system reaches the charging limit condition, the energy storage charging is stopped, and the remaining surplus photovoltaic power is sent to the upstream grid or the photovoltaic output is restricted according to the grid connection conditions of the project. When the demand at the grid connection point does not meet the preset demand warning condition and there is no surplus photovoltaic power, the edge EMS control layer determines the charging state, discharging state, or standby state of the energy storage system according to the state of charge interval of the energy storage and the current time-of-use electricity price period, and restricts the operating power of the energy storage system.

[0012] In one implementation of the first aspect, after the communication state remains normal for a preset stable duration and the deviation level remains at the preset normal level for a preset number of consecutive evaluation windows, the edge EMS control layer compares the current local control command with the target control command corresponding to the cloud scheduling strategy, limits the change in energy storage power between adjacent control cycles, and restores the restricted adjustable load in batches according to load priority. After the current field operation state reaches the target state corresponding to the target control command, the cloud scheduling strategy is resumed.

[0013] In one implementation of the first aspect, the method further includes: the edge EMS control layer records the real-time operation data of the microgrid, the deviation level, the reason for the operation mode switching, the local control command and the execution result, and transmits the recorded data back to the cloud decision layer via the back-end scheduling layer; the cloud decision layer updates the prediction model based on the transmitted data, and the back-end scheduling layer updates at least one of the deviation judgment parameters and the cloud scheduling parameters based on the transmitted data.

[0014] Secondly, this application provides a microgrid cloud-edge collaborative autonomous control system, including a cloud decision-making layer, a back-end scheduling layer, and an edge EMS control layer. The cloud decision-making layer generates photovoltaic (PV) output forecasts and load forecasts for multiple future forecast periods, and sends these forecasts to the back-end scheduling layer. The back-end scheduling layer receives microgrid operation data reported by the edge EMS control layer, generates a cloud scheduling strategy based on the PV output forecasts, load forecasts, and microgrid operation data, and sends these strategies to the edge EMS control layer. The edge EMS control layer collects actual PV output, actual load power, energy storage state of charge, and grid connection point demand with a collection time identifier, determines the corresponding forecast period based on the collection time identifier, and sends the actual PV output... The power and actual load power are compared with the photovoltaic power output forecast and load forecast values ​​corresponding to the forecast period, respectively, and the deviation level is determined based on the comparison results. The communication status between the system and the back-end scheduling layer is detected, and either the cloud-optimized operation mode or the edge offline autonomous mode is selected based on the communication status and the deviation level. When the communication status is abnormal, or when the communication status is normal and the deviation level is determined to be a preset severe abnormality level, the system switches to the edge offline autonomous mode. The cloud scheduling strategy is executed in the cloud-optimized operation mode. In the edge offline autonomous mode, local control commands for controlling at least one of the energy storage system and adjustable load are generated based on the grid connection point demand, the actual photovoltaic power output, the actual load power, and the energy storage charge status. The cloud scheduling strategy is resumed when the communication status returns to normal and the deviation level returns to a preset normal level.

[0015] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0016] Fourthly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the method described in the first aspect.

[0017] As described above, the microgrid cloud-edge collaborative autonomous control method, system, storage medium, and electronic device described in this application have the following beneficial effects:

[0018] In this application, the photovoltaic output forecast and load forecast values ​​generated by the cloud decision-making layer are used not only by the back-end dispatch layer to generate cloud dispatch strategies, but also by the edge EMS control layer to determine the adaptability of the cloud dispatch strategies to the actual on-site operating conditions. Based on the data acquisition time marker, the edge EMS control layer compares the actual photovoltaic output and actual load power with the forecast values ​​for the corresponding forecast period, and selects the operating mode based on the deviation level and cloud-edge communication status. This allows the system to promptly switch to edge offline autonomous mode when communication is abnormal, and actively stops executing cloud dispatch strategies that are no longer applicable when communication is normal but the actual on-site operating conditions deviate significantly from the forecast conditions. After entering edge offline autonomous mode, the edge EMS control layer can generate local control commands in real time based on grid connection point demand, actual photovoltaic output, actual load power, and energy storage charge status to maintain microgrid power balance and on-site control continuity. The cloud dispatch strategy is then resumed after communication and on-site operating conditions are restored. Therefore, this application can improve the microgrid's adaptability to forecast deviations, communication anomalies and changes in field operating conditions, reduce control lag and operational risks caused by cloud scheduling strategy mismatch or communication interruption, and take into account both cloud global optimization capabilities and edge field autonomy capabilities. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the application scenario of the microgrid cloud-edge collaborative autonomous control system described in the embodiments of this application.

[0020] Figure 2 This is a schematic diagram of the overall process of the microgrid cloud-edge collaborative autonomous control method described in the embodiments of this application.

[0021] Figure 3 This is a schematic diagram of the process for selecting the deviation level and operating mode as described in the embodiments of this application.

[0022] Figure 4 This is a schematic diagram of the general control flow of the edge offline autonomous mode described in the embodiments of this application.

[0023] Figure 5 This is a schematic diagram of the specific control flow of an example of the edge offline autonomous mode described in the embodiments of this application.

[0024] Figure 6 This is a schematic diagram of the smooth switching back operation mode in the embodiments of this application.

[0025] Figure 7 This is a schematic diagram of the functional modules of the microgrid cloud-edge collaborative autonomous control system in this application embodiment. Detailed Implementation

[0026] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0027] It should be noted that the accompanying drawings in the following embodiments are only schematic illustrations of the basic concept of this application. The drawings only show components or processes related to this application and do not represent the quantity, shape, size, or proportion of each component in actual implementation. In actual implementation, the system structure and component layout can be adjusted according to the microgrid capacity, the number of field devices, communication conditions, and control requirements.

[0028] Figure 1 This paper illustrates an application scenario and functional architecture of the microgrid cloud-edge collaborative autonomous control system 100 of this application. The system includes a cloud decision-making layer 101, a back-end scheduling layer 102, and an edge energy management system (EMS) control layer 103 (hereinafter referred to as the edge EMS control layer). The cloud decision-making layer 101 communicates with the back-end scheduling layer 102, and the back-end scheduling layer 102 communicates with the edge EMS control layer 103. The edge EMS control layer 103 communicates with metering or control equipment related to distributed photovoltaic 401, energy storage system 402, charging facilities 403, critical loads 404, general and non-critical loads 405, and the upstream power grid 406 in the field microgrid 104.

[0029] It should be understood that the cloud decision-making layer 101, the back-end scheduling layer 102, and the edge EMS control layer 103 are control layers divided according to function, and are not limited to three independent physical devices. Among them, the cloud decision-making layer 101 and the back-end scheduling layer 102 can be deployed on the same cloud server or on different servers that communicate with each other; the edge EMS control layer 103 is deployed in the microgrid field.

[0030] It should be noted that the cloud-based decision layer 101 is used to generate prediction results for future operating conditions. Specifically, the cloud-based decision layer 101 is equipped with a prediction model. Its input data includes historical microgrid operating data, environmental meteorological data, and real-time microgrid operating data. The output data includes photovoltaic power output prediction values ​​and load prediction values ​​corresponding to multiple future prediction periods. The prediction values ​​are arranged in chronological order to form a photovoltaic power output prediction sequence and a load prediction sequence.

[0031] The back-end scheduling layer 102 is used to aggregate edge-side data, generate cloud-based scheduling strategies, and complete strategy interaction with the edge side. Specifically, the back-end scheduling layer 102 acts as an optimization scheduling layer between the prediction results and field control. It receives the photovoltaic output prediction sequence and load prediction sequence output by the cloud decision layer 101, and receives microgrid operation data reported by the edge EMS control layer 103. The microgrid operation data is the operation data collected and reported by the field. The back-end scheduling layer 102 is used to integrate time-of-use pricing, grid demand, energy storage state of charge (SOC), and equipment operating boundaries to generate energy storage charging and discharging plans and cloud-based scheduling strategies for each predicted period in the future.

[0032] The edge EMS control layer 103 is used to collect field data, execute cloud scheduling strategies, determine deviation levels and communication status, and optionally determine the validity status of cloud data. It generates local control commands when cloud scheduling data becomes unusable. Specifically, the edge EMS control layer 103, as the main field control entity, continuously collects data on actual photovoltaic output, actual load power, charging facility power, energy storage SOC, energy storage charging and discharging status, and grid connection point-related data. When the communication status is normal and the deviation level is at a preset normal level, the edge EMS control layer 103 controls the field equipment according to the cloud scheduling strategy issued by the backend scheduling layer 102. When the communication status is abnormal, or when the communication status is normal but the deviation level is determined to be a preset severe abnormal level, the edge EMS control layer 103 switches to edge offline autonomous mode. Optionally, the edge EMS control layer 103 also switches to edge offline autonomous mode when the validity status of cloud data is abnormal. The edge EMS control layer 103 can be implemented by an industrial computer, field server, edge gateway, or computing control unit within a control cabinet.

[0033] It should be noted that, in addition to selecting the control mode according to the deviation level when communication is normal, when the communication link between the back-end scheduling layer and the edge EMS control layer is disconnected, the edge EMS control layer will determine the communication status as abnormal and switch to the edge offline autonomous mode.

[0034] Optionally, if the predicted value or the cloud scheduling strategy exceeds a preset valid time limit, or if the on-site operating status reaches a preset protection condition, the edge EMS control layer will determine the cloud data validity status as abnormal and switch to the edge offline autonomous mode. The cloud data validity status indicates whether the predicted value and cloud scheduling strategy issued by the cloud are still suitable as the basis for current on-site control; the preset valid time limit refers to the maximum time the predicted value or the cloud scheduling strategy is allowed to be used as the basis for current control after its generation or issuance; the preset protection condition refers to a preset safety condition where the equipment operating parameters or electrical parameters reach a point where protection control needs to be immediately executed by the edge side.

[0035] It should be noted that the edge EMS control layer can use a single field control device or be split into an edge gateway and a local controller. The edge gateway is responsible for field data acquisition, communication forwarding, and operating mode determination, while the local controller is responsible for executing energy storage charging and discharging, charging facility power limiting, and load shelving actions.

[0036] The field microgrid 104 in this application is the control object of the aforementioned three-level architecture. The field microgrid 104 includes a power supply side, an energy storage side, and a power consumption side. The power supply side includes at least distributed photovoltaic 401 and an upstream grid 406; the energy storage side includes an energy storage system 402; and the power consumption side includes charging facilities 403, critical loads 404, and general and non-critical loads 405.

[0037] It should be noted that this application uses the aforementioned distributed photovoltaic power generation as an example for illustration, but the scope of protection of this application is not limited thereto. In practical applications, the distributed photovoltaic power generation unit can be completely replaced by a wind turbine generator or other types of new energy power generation units. Those skilled in the art can flexibly select according to the actual application scenario, and all of these fall within the scope of the technical concept of this application.

[0038] It should be understood that a data uplink and a policy downlink are formed between the cloud decision-making layer 101, the backend scheduling layer 102, and the edge EMS control layer 103. The edge EMS control layer 103 uploads the field operation data to the backend scheduling layer 102; the backend scheduling layer 102 summarizes the field operation data, provides the data for model updates to the cloud decision-making layer 101, and uses the field operation data to optimize scheduling; the cloud decision-making layer 101 sends the predicted values ​​to the backend scheduling layer 102; the backend scheduling layer 102 generates a cloud scheduling policy based on the predicted values ​​and the field operation data, and sends the predicted values ​​and the cloud scheduling policy to the edge EMS control layer 103.

[0039] It should be understood that the adjustable load referred to in this application refers to a load whose power consumption can be changed according to the power limiting command or status command issued by the edge EMS control layer 103, including at least one of charging facility 403 and general and non-critical load 405.

[0040] like Figure 2 As shown, the microgrid cloud-edge collaborative autonomous control method of this application includes steps S101 to S108.

[0041] In step S101, the cloud decision layer generates photovoltaic output forecast values ​​and load forecast values ​​corresponding to multiple future forecast periods (that is, photovoltaic output forecast values ​​and load forecast values ​​for future preset time sequences), and sends the forecast values ​​to the backend scheduling layer.

[0042] It should be understood that the multiple prediction periods correspond to a preset prediction time granularity. The prediction time granularity characterizes the time interval between adjacent prediction periods and also determines the duration of each prediction period. In other words, the cloud-based decision layer uses the prediction time granularity as the basic unit to divide the prediction period into multiple consecutive and equally long prediction periods, and generates corresponding photovoltaic output prediction values ​​and load prediction values ​​for each prediction period. For example, when the prediction time granularity is 15 minutes, each prediction period in the prediction sequence is 15 minutes, and adjacent prediction periods are arranged consecutively.

[0043] In step S102, the backend scheduling layer receives the microgrid operation data reported by the edge EMS control layer, generates an energy storage charging and discharging plan and a cloud scheduling strategy based on the predicted value, the microgrid operation data, time-of-use electricity price and operation constraints, and sends the predicted value and the cloud scheduling strategy to the edge EMS control layer.

[0044] In step S103, the edge EMS control layer collects or reads the actual photovoltaic output, actual load power, energy storage SOC, and grid connection point demand with the collection time identifier according to the preset sampling period.

[0045] In step S104, the edge EMS control layer determines the corresponding prediction period based on the acquisition time identifier, compares the actual photovoltaic output and actual load power with the photovoltaic output prediction value and load prediction value of the corresponding prediction period, and determines the deviation level based on the comparison results.

[0046] In step S105, the edge EMS control layer detects the communication status between itself and the backend scheduling layer, and selects either the cloud-optimized operation mode or the edge offline autonomous mode based on the communication status and deviation level.

[0047] Optionally, the edge EMS control layer also detects the validity status of cloud data and switches to edge offline autonomous mode when the validity status of cloud data is abnormal.

[0048] Step S106: In the edge offline autonomous mode, the edge EMS control layer generates local control commands based on the grid connection point demand, actual photovoltaic output, actual load power, energy storage charge status, current time-of-use electricity price period, and load priority.

[0049] Step S107: After the communication status remains normal and the deviation level continuously recovers to the normal level, the edge EMS control layer smoothly switches the current local control command back to the target control command corresponding to the cloud scheduling strategy and resumes the execution of the cloud scheduling strategy.

[0050] In step S108, the edge EMS control layer transmits the running data, mode switching records, local control commands and execution results back to the backend scheduling layer. The cloud decision layer updates the prediction model, and the backend scheduling layer updates the deviation judgment parameters or cloud scheduling parameters.

[0051] It should be understood that the numbering of the above steps is used to illustrate the data and control relationship, and does not limit the steps to be performed in exactly the same time order in all implementations; among them, field data acquisition, communication status detection and deviation judgment can be performed in parallel periodically.

[0052] The following further explains the specific implementation method of generating cloud prediction data by the cloud decision layer 101 in step S101. The cloud prediction data includes photovoltaic power output prediction values ​​and load prediction values ​​corresponding to multiple future prediction periods.

[0053] In some specific embodiments, step S101 can be further refined into steps S101a to S101c.

[0054] Step S101a: Periodic acquisition of multi-source data.

[0055] Specifically, the cloud-based decision-making layer periodically acquires historical operating data, environmental meteorological data, and real-time operating data of the microgrid.

[0056] The historical operating data is used to characterize the historical operating patterns of the microgrid and may include historical photovoltaic output, historical load power, and historical energy storage SOC; the environmental meteorological data is used to characterize the external environmental conditions affecting photovoltaic power generation and may include at least one of ambient temperature and irradiance; the real-time operating data of the microgrid is the field feedback data reflecting the current or recent operating status of the microgrid operating data, provided by the back-end scheduling layer, and may include at least one of current actual photovoltaic output, actual load power, energy storage SOC, and equipment operating status.

[0057] Step S101b: Model inference generates time-series prediction sequences.

[0058] Specifically, the cloud-based decision-making layer inputs the aforementioned multi-source data into the prediction model and outputs the photovoltaic power output prediction value and load prediction value corresponding to multiple prediction periods in the future, thereby forming a continuous photovoltaic power output prediction sequence and load prediction sequence.

[0059] It should be noted that the prediction time granularity (i.e., the unit prediction period) corresponding to the prediction sequence is not fixed and can be flexibly set by those skilled in the art according to actual scheduling needs and system response capabilities.

[0060] Preferably, the prediction time granularity is set to 15 minutes. In other words, the time-series prediction sequence is arranged in consecutive 15-minute intervals, and each interval corresponds to at least one set of matching photovoltaic output prediction values ​​and load prediction values. The time-series prediction sequence may include n sets of corresponding photovoltaic output prediction values ​​and load prediction values ​​for interval 1, interval 2, up to interval n.

[0061] It should be noted that the prediction model can be a general time-series prediction model trained based on a large amount of historical data, or a customized prediction model adapted and trained using specific operating data of the target microgrid. It should be understood that the core inventive concept of this application lies in the synergistic logic between the application of prediction results, deviation hierarchical control, operating mode switching, and edge offline autonomous mechanism; therefore, no necessary limitations are placed on the specific network structure or training algorithm of the prediction model.

[0062] Step S101c: Rolling update and distribution of the predicted sequence.

[0063] Specifically, the cloud-based decision-making layer repeatedly executes the above data acquisition and prediction inference process according to a preset cycle, providing the back-end scheduling layer with prediction sequences (including photovoltaic output prediction sequences and load prediction sequences) that are updated over time. After each new prediction sequence is generated, the cloud-based decision-making layer synchronously sends the prediction sequence to the back-end scheduling layer.

[0064] It should be noted that the prediction sequences generated by the cloud-based decision-making layer perform two core functions. On the one hand, they are used by the back-end scheduling layer to generate energy storage charging and discharging plans for future time periods. On the other hand, they are sent by the back-end scheduling layer to the edge EMS control layer along with the cloud-based scheduling strategy. For any sampling time, the edge EMS control layer determines the prediction time period to which the sampling time belongs based on the acquisition time identifier, and obtains the photovoltaic output prediction value and load prediction value corresponding to that prediction time period from the two prediction sequences, respectively.

[0065] It should be noted that the prediction sequence is formed by arranging the predicted values ​​corresponding to multiple future prediction periods in chronological order; the predicted value is the specific data corresponding to a certain prediction period in the prediction sequence; the prediction benchmark referred to in this application refers to the photovoltaic power output prediction value and load prediction value corresponding to the current prediction period obtained from the corresponding prediction sequence according to the collection time identifier, which is used to compare with the actual photovoltaic power output and actual load power of the same prediction period.

[0066] Therefore, the same prediction result generated by the cloud decision layer has two consecutive related uses in this application: one is to participate in the global optimization on the cloud side (backend scheduling layer); the other is to be used on the edge side (edge ​​EMS control layer) to determine whether the future working conditions on which the global optimization is based still match the actual working conditions on site.

[0067] The following is a further explanation of step S102.

[0068] In some specific embodiments, step S102 can be further refined into steps S102a to S102d.

[0069] In step S102a, the back-end scheduling layer receives the photovoltaic power output prediction sequence and load prediction sequence output by the cloud decision layer, and receives the microgrid operation data reported by the edge EMS control layer.

[0070] The prediction sequence is formed by arranging the predicted values ​​corresponding to multiple prediction periods in chronological order. The predicted values ​​include the photovoltaic output prediction value and the load prediction value corresponding to each prediction period.

[0071] The microgrid operation data includes energy storage state of charge, equipment operating status, and actual power data. The equipment operating status characterizes the start-up / shutdown or adjustable status of the energy storage system, charging facilities, and adjustable loads. The actual power data includes at least one of the following: actual photovoltaic output, actual load power, energy storage charging / discharging power, and grid connection point power. The grid connection point refers to the location where the microgrid connects to the upstream power grid and exchanges power. Grid connection point power characterizes the power the microgrid obtains from or transmits to the upstream power grid.

[0072] Furthermore, the backend scheduling layer also obtains time-of-use pricing and grid demand limits corresponding to the project. The time-of-use pricing includes pricing information for peak, flat, and valley periods; the grid demand limits are used to restrict the upper limit of power that the grid connection point can obtain from the upstream grid.

[0073] In step S102b, the back-end scheduling layer performs optimized scheduling based on the predicted value, microgrid operation data, time-of-use electricity price, and operation constraints.

[0074] The operational constraints include grid demand constraints, energy storage charging and discharging safety constraints, and equipment operation constraints. Grid demand constraints are used to ensure that the power at the grid connection point does not exceed the grid demand limit; energy storage charging and discharging safety constraints are used to limit the energy storage state of charge, charging power, and discharging power to within permissible ranges; and equipment operation constraints are used to limit the start-up and shutdown status, power adjustment range, or adjustable time periods of charging facilities and adjustable loads.

[0075] It should be noted that the optimized scheduling of the back-end scheduling layer aims at system operational stability, operational economy, and renewable energy utilization rate. System operational stability refers to ensuring that the grid connection point power, energy storage state of charge, and equipment operating status are within acceptable limits while meeting the load's electricity demand. Operational economy refers to arranging energy storage charging and discharging and adjustable load operation in conjunction with time-of-use pricing and power purchase agreements. Renewable energy utilization rate refers to prioritizing the use of on-site photovoltaic power output to reduce situations where photovoltaic power is not consumed on-site.

[0076] It should also be noted that the weights of each scheduling objective can be configured according to the project's operational needs. For example, in scenarios with high requirements for power supply continuity, the weight of the stability objective can be increased; in industrial and commercial parks, the weight of the economic objective can be increased; and in green and low-carbon projects, the weight of the renewable energy utilization rate objective can be increased.

[0077] Step S102c: Based on the optimized scheduling results, the backend scheduling layer generates an energy storage charging and discharging plan and a cloud scheduling strategy.

[0078] The energy storage charging and discharging plan refers to the energy storage operation arrangement corresponding to each predicted period, including the target charging power, target discharging power, or standby state; the cloud scheduling strategy refers to the control information used to guide the edge EMS control layer to execute microgrid control, which includes at least the energy storage charging and discharging plan, and may also include charging facility power limits and adjustable load status.

[0079] In step S102d, the backend scheduling layer sends the photovoltaic output prediction sequence, load prediction sequence, and cloud scheduling strategy to the edge EMS control layer. When the communication status is normal and the deviation level is within the normal range, the edge EMS control layer controls the energy storage system and adjustable load according to the cloud scheduling strategy.

[0080] It should be noted that this application does not limit the specific optimization algorithm used by the backend scheduling layer, as long as the optimization method can obtain an executable energy storage charging and discharging plan and cloud scheduling strategy under the stated scheduling objectives and operational constraints. "Executable" means that the generated control information does not violate grid demand constraints, energy storage charging and discharging safety constraints, and equipment operation constraints, and can be issued by the edge EMS control layer to the corresponding controlled equipment for execution.

[0081] The following is a further explanation of step S103.

[0082] In some specific embodiments, step S103 can be further refined into steps S103a to S103c.

[0083] Step S103a: The edge EMS control layer collects actual on-site operating data according to the preset sampling period.

[0084] Specifically, the edge EMS control layer communicates with field devices or metering devices to collect or read actual photovoltaic output, actual load power, energy storage SOC, and grid connection point demand, and adds or associates collection time identifiers to the actual field operation data. The collection time identifiers for actual photovoltaic output, actual load power, and energy storage SOC represent the time of data collection; the collection time identifier for grid connection point demand includes time information associated with the corresponding demand statistical period.

[0085] In one specific implementation, the cloud-based decision-making layer generates photovoltaic output prediction sequences and load prediction sequences with a prediction time granularity of 15 minutes, while the edge EMS control layer collects or reads actual on-site operating data with a sampling period of 50ms. The prediction time granularity is used to divide the prediction period corresponding to cloud-based prediction and scheduling, and the sampling period is used to monitor changes in the on-site operating status within the prediction period. Therefore, a prediction period can include multiple sampling times and their corresponding actual on-site operating data.

[0086] It should be noted that the on-site sampling period of this application is not limited to 50ms. It can be set according to the data update period of the on-site metering equipment, the processing capability of the edge device and the control response requirements, but the sampling period should be shorter than the prediction time granularity.

[0087] In step S103b, the edge EMS control layer matches the actual on-site operating data with the predicted values ​​for the corresponding prediction period based on the acquisition time identifier.

[0088] Specifically, the edge EMS control layer determines the prediction period to which the sampling time belongs based on the acquisition time identifier, and obtains the photovoltaic output prediction value corresponding to the prediction period from the photovoltaic output prediction sequence, and obtains the load prediction value corresponding to the prediction period from the load prediction sequence.

[0089] The actual photovoltaic (PV) output and the predicted PV output use the same metering boundary, and the actual load power and the predicted load power use the same metering boundary, so that the actual data and the predicted values ​​correspond to each other in terms of time, data type, and statistical range. The metering boundary refers to the statistical object and statistical range corresponding to the power data. The actual PV output and the predicted PV output correspond to the same PV system, and the actual load power and the predicted load power correspond to the same load statistical range.

[0090] In step S103c, the edge EMS control layer preprocesses the actual on-site operating data collected during the corresponding prediction period to generate on-site evaluation data for deviation judgment. The on-site evaluation data includes photovoltaic output evaluation values ​​and load power evaluation values.

[0091] Specifically, because the sampling period of the edge EMS control layer is shorter than the prediction time granularity, multiple sampling data points for actual photovoltaic output and actual load power can be obtained within the same prediction period. The edge EMS control layer filters or statistically aggregates the sampling data within a preset evaluation window to obtain photovoltaic output evaluation values ​​and load power evaluation values, and acquires the energy storage SOC and grid connection point demand within the corresponding evaluation window.

[0092] The edge EMS control layer combines the photovoltaic output evaluation value, load power evaluation value, energy storage SOC, grid connection point demand, and the photovoltaic output forecast value and load forecast value for the corresponding forecast period to form a current operating status data group, and uses the current operating status data group for subsequent deviation calculation and deviation level determination.

[0093] Through the above processing, the edge EMS control layer can match the high-frequency collected on-site actual operation data with the time-segmented predicted values ​​generated in the cloud in terms of time and measurement boundaries, avoiding direct comparison of data at different times or in different statistical ranges due to differences in sampling period and prediction time granularity.

[0094] It should be noted that the prediction time granularity mentioned in this application refers to the time interval between adjacent prediction time periods; the sampling period refers to the time interval between two adjacent collections or readings of actual on-site operating data by the edge EMS control layer, and the sampling period is shorter than the prediction time granularity.

[0095] The data collection time identifier mentioned in this application refers to the time information used to characterize the actual time of data collection during on-site operation; the prediction period refers to the time interval formed by dividing the future prediction cycle according to the prediction time granularity, with each prediction period corresponding to the photovoltaic power output prediction value and the load prediction value.

[0096] The evaluation window mentioned in this application refers to the time range selected within the corresponding prediction period for collecting multiple sampled data; the photovoltaic output evaluation value and load power evaluation value refer to the power characterization values ​​obtained after filtering or statistically aggregating the corresponding sampled data within the evaluation window.

[0097] The actual load power mentioned in this application refers to the load power collected within the same metering boundary as the load forecast value; the energy storage SOC refers to the state of charge parameter used to characterize the current remaining power state of the energy storage system.

[0098] The grid connection point demand mentioned in this application refers to a parameter determined by the grid connection point metering device according to preset demand statistics rules, used to characterize the power level obtained by the microgrid from the upstream grid during the current demand statistics period. The grid connection point demand is used to subsequently determine whether the microgrid is approaching or has reached the grid demand limit.

[0099] The following combination Figure 3The deviation level determination in step S104 and the operation mode selection in step S105 are further explained. The edge EMS control layer determines the deviation level based on the preset deviation range of the photovoltaic output deviation and the load power deviation. The deviation level can be determined by the higher level of the photovoltaic output deviation and the load power deviation, or by a weighted combination of the two. The corresponding deviation range is preset based on the microgrid capacity, photovoltaic installed capacity, load fluctuation characteristics, and the equipment's allowable adjustment range.

[0100] In one embodiment, the deviation levels are divided into Level 1 deviation, Level 2 deviation, and Level 3 deviation, corresponding to the preset normal level, preset general abnormal level, and preset severe abnormal level, respectively. The communication status and the validity status of cloud data are detected separately by the edge EMS control layer and are not used as the deviation level determined by the comparison between actual data and predicted values. Figure 3 This demonstrates a basic process for selecting the operating mode based on communication status and deviation level; abnormal cloud data validity status is an optional switching condition and is not included in the process. Figure 3 It is shown repeatedly in the middle.

[0101] (1) Level 1 deviation corresponds to the normal level, indicating that the deviation between the actual evaluation value and the corresponding predicted value is within the allowable range. Under Level 1 deviation, the edge EMS control layer determines that the cloud prediction result and cloud scheduling strategy are still suitable for the current field conditions and continues to execute the cloud scheduling strategy.

[0102] It should be understood that under Level 1 deviation, the edge EMS control layer operates in a cloud-optimized mode. The edge EMS control layer receives energy storage charging and discharging plans and cloud-based scheduling strategies from the backend scheduling layer and executes corresponding control commands based on the current time period.

[0103] It should be noted that the edge EMS control layer continues to collect actual photovoltaic output, actual load power, and energy storage SOC while executing cloud-based scheduling strategies, and continuously compares the actual photovoltaic output and actual load power with the predicted photovoltaic output and load values ​​for the current forecast period. In other words, the cloud-optimized operation mode does not stop edge judgment, but rather the edge EMS control layer monitors the adaptation status between the plan and the on-site operating conditions while executing the cloud plan.

[0104] (2) Level 2 deviation corresponds to a general anomaly level, indicating that the deviation exceeds the allowable range but does not reach the severe anomaly threshold. Under Level 2 deviation, the edge EMS control layer continues to execute the cloud scheduling strategy and uploads the abnormal data and corresponding time information to the backend scheduling layer.

[0105] It should be noted that when the deviation progresses from Level 1 to Level 2, the edge EMS control layer maintains the cloud-optimized operation mode but uploads abnormal data to the backend scheduling layer. The backend scheduling layer can use the abnormal data for subsequent scheduling analysis, while the cloud decision layer can use the abnormal data for predictive model updates.

[0106] Optionally, under the general abnormal state corresponding to the secondary deviation, audible and visual alarms and remote message push can be added to send abnormal information to the site or remote operation and maintenance terminal.

[0107] In addition to selecting the control mode based on the aforementioned deviation level, when the communication link between the backend scheduling layer and the edge EMS control layer is disconnected, the edge EMS control layer will determine the communication status as abnormal and switch to the edge offline autonomous mode.

[0108] Optionally, if the predicted value or cloud scheduling strategy exceeds a preset effective time limit, or if the on-site operating status reaches a preset protection condition, the edge EMS control layer will determine the cloud data validity status as abnormal and switch to edge offline autonomous mode. The preset protection condition refers to a preset safety condition where the device's operating parameters or electrical parameters reach a point where protection control needs to be immediately executed by the edge side.

[0109] It should be noted that when the deviation level reaches level three, the communication status is abnormal, or the cloud data validity status is abnormal as described in the implementation method for detecting the validity status of cloud data, the edge EMS control layer will no longer use the cloud scheduling strategy as the current control basis, but will read the current field data and start local closed-loop control.

[0110] It should be understood that the cloud-optimized operation mode can utilize cloud computing resources and global data for comprehensive scheduling across multiple time periods; the edge offline autonomous mode can maintain the continuity of on-site control when cloud scheduling data fails or cannot be transmitted. The two modes are connected by deviation levels and communication status; in the implementation method for detecting the validity status of cloud data, they are also connected by the validity status of the cloud data.

[0111] Furthermore, to suppress the frequent switching between the two operating modes (hereinafter referred to as mode jitter) caused by random fluctuations in instantaneous sampling data and to ensure the stability of the microgrid's operating status, the edge EMS control layer adopts an anti-jitter mechanism that combines continuous evaluation window judgment and hysteresis dual thresholds to perform mode switching, as follows:

[0112] Firstly, the continuous evaluation window determination rule. The edge EMS control layer does not directly trigger mode switching based on the deviation results of a single sampling. Instead, it only executes a formal mode switching operation when the current deviation level meets the corresponding mode switching conditions within a consecutive preset number of evaluation windows. This filters out false triggers caused by instantaneous data spikes and random fluctuations. The duration of the evaluation window can be set to be greater than the sampling period but less than the prediction time step. Within each evaluation window, the deviation level can be determined based on the statistical values ​​of multiple sampled data.

[0113] Secondly, a hysteresis-based dual-threshold design. Different entry and exit thresholds are set for the entry and exit processes of the operating mode. For example, when switching from cloud-optimized operating mode to edge offline autonomous mode, a higher deviation threshold is used as the entry condition; when returning from edge offline autonomous mode to cloud-optimized operating mode, a lower deviation threshold is used as the exit condition. This threshold hysteresis prevents repeated switching of operating modes when the deviation fluctuates slightly near the critical value.

[0114] Based on the above mechanism, the switching of operating modes in this application is not solely determined by whether communication between the cloud and the edge is connected. Even if the communication link remains normal, if the predicted values ​​issued by the cloud and the actual operating conditions on site continue to be severely mismatched, and the entry conditions for the edge offline autonomous mode have been met, the edge EMS control layer can still actively switch to the edge offline autonomous mode, stop executing the mismatched cloud scheduling strategy, and ensure the operation safety and power balance of the microgrid through local control.

[0115] The following combination Figure 4 The edge offline autonomous control process in step S106 will be further explained.

[0116] like Figure 4 As shown, after determining the entry into the edge offline autonomous mode in step S105, the edge EMS control layer stops using the cloud scheduling strategy as the current control basis, and acquires real-time field data, determines the field operating status, and generates and issues local control commands according to the preset control cycle.

[0117] In a factory microgrid application scenario, the real-time data includes at least the factory's base load power, charging facility power, actual photovoltaic output, energy storage SOC, grid connection point demand, current time-of-use pricing period, and load priority. The factory's base load power refers to the power consumed to maintain factory production and daily operation, excluding charging facilities and other individually adjustable loads. The actual load power can be determined based on the factory's base load power, charging facility power, and the power of other electrical equipment, but does not include the charging power of the energy storage system when calculating surplus photovoltaic power. The energy storage SOC characterizes the ratio of the current remaining electricity of the energy storage system to its available capacity. The grid connection point demand refers to a parameter determined by the grid connection point metering device according to preset demand statistics rules, characterizing the power level obtained by the microgrid from the upper-level grid during the current demand statistics period. The grid connection point refers to the location where the microgrid connects to the upper-level grid and exchanges electricity. The current time-of-use pricing period refers to the peak, normal, or valley period to which the current moment belongs.

[0118] In some specific embodiments, edge offline autonomous control includes load demand management, photovoltaic surplus power consumption, energy storage SOC hierarchical management and time-of-use pricing-linked power control. The above controls are not executed independently of each other, but are determined by the edge EMS control layer within the same control cycle based on the on-site operating status and control priority to obtain the final local control command.

[0119] In this application, the safety constraints on energy storage SOC and charging / discharging power are safety boundaries that must be met when generating local control commands. Under the premise of meeting these safety boundaries, the edge offline autonomous control determines the control priority in the order of demand control, photovoltaic (PV) integration, and economic regulation. Specifically, demand control is used to prevent the grid connection point's demand from exceeding the grid's demand limit; PV integration is used to improve the utilization rate of on-site PV power; and economic regulation is used to reduce the microgrid's electricity costs based on time-of-use pricing. The specific judgment process is as follows:

[0120] (a) Demand assessment at grid connection points

[0121] The edge EMS control layer first compares the grid-connected point demand with a preset demand warning threshold. The demand warning threshold is a pre-set judgment value based on the grid demand limit corresponding to the microgrid, used to initiate demand control in advance before the grid-connected point demand reaches the grid demand limit. The preset demand warning condition refers to the grid-connected point demand reaching or exceeding the demand warning threshold; the preset demand control condition refers to the grid-connected point demand being reduced to below a preset demand recovery target after adjustment, where the demand recovery target is not higher than the demand warning threshold.

[0122] When the demand at the grid connection point reaches or exceeds the aforementioned demand warning threshold, the edge EMS control layer performs load demand management to reduce the power obtained by the microgrid from the upper-level grid. Specifically, the control logic is as follows: when the energy storage system is charging, the charging power of the energy storage system is preferentially reduced or the energy storage system is stopped from charging; when the energy storage SOC is higher than the allowable discharge lower limit, the energy storage system is controlled to discharge; when the available discharge power of the energy storage system is still insufficient to reduce the demand at the grid connection point, the power consumption of the charging facilities is limited; when limiting the power of the charging facilities still cannot meet the demand control requirements, non-critical loads are limited or cut off according to load priority.

[0123] The discharge power of the energy storage system, the power reduction of the charging facilities, and the power reduction of non-critical loads are determined based on the demand gap, which refers to the power purchase required to reduce the demand at the grid connection point to below the demand warning threshold.

[0124] (II) Judgment of surplus photovoltaic power

[0125] When the demand at the grid connection point does not reach the aforementioned demand warning threshold, the edge EMS control layer further determines whether there is surplus photovoltaic power based on the actual photovoltaic output and the actual load power. The surplus photovoltaic power refers to the portion of photovoltaic power exceeding the current load consumption when the actual photovoltaic output is greater than the current actual load power of the microgrid.

[0126] When surplus photovoltaic (PV) power exists, the edge EMS control layer controls the energy storage system to absorb the surplus PV power, provided that the energy storage SOC has not reached the allowable charging limit and the energy storage system has charging capabilities. The target charging power of the energy storage system does not exceed the smaller of the surplus PV power and the allowable charging power of the energy storage. The charging limitation condition refers to the energy storage SOC reaching the allowable charging limit, or the energy storage system being prohibited from continuing charging due to equipment status or safety protection requirements. When the energy storage system cannot continue to absorb all the surplus PV power, the remaining PV power is transmitted to the upper-level grid, provided the project's grid connection conditions allow for power transmission to the upper-level grid. The project's grid connection conditions refer to the operating conditions applicable when the microgrid exchanges power with the upper-level grid, including at least whether reverse power transmission is allowed and the permitted power range for transmission to the upper-level grid.

[0127] (III) Time-of-use electricity price linkage control

[0128] When the demand at the grid connection point does not reach the demand warning threshold and there is no surplus photovoltaic power, the edge EMS control layer determines the charging, discharging, or standby status of the energy storage system based on the energy storage SOC and the current time-of-use electricity price. When the energy storage SOC is within the allowable charging and discharging range, the system is controlled to charge during off-peak hours, discharged during peak hours, and its operating status is determined based on the actual load power and the grid connection point demand during normal periods.

[0129] It should be noted that the safety boundary of energy storage SOC takes precedence over the economic control corresponding to time-of-use pricing. When the energy storage SOC reaches the lower limit of permissible discharge, energy storage discharge will stop even if it is currently during peak hours; when the energy storage SOC reaches the upper limit of permissible charging, energy storage charging will stop even if it is currently during off-peak hours or there is surplus photovoltaic power. The upper limit of permissible charging and the lower limit of permissible discharge are preset according to the safety operation requirements of the energy storage system.

[0130] (iv) Control command generation and closed-loop iteration

[0131] The edge EMS control layer generates local control commands based on the above judgment results. The local control commands include charging commands, discharging commands or standby commands issued to the energy storage system, power limiting commands issued to the charging facilities, and power limiting commands or disconnection commands issued to the control equipment of non-critical loads.

[0132] It should be understood that when conflicting control requirements arise from different control logics within the same control cycle, the edge EMS control layer determines the final control command in the order of energy storage safety, demand control, photovoltaic power consumption, and economic adjustment. In other words, under the premise of meeting the energy storage SOC and charging / discharging power safety constraints, the edge EMS control layer prioritizes grid-connected point demand control; when the grid-connected point demand does not reach the demand warning threshold, it then executes photovoltaic surplus power consumption and time-of-use pricing linkage control.

[0133] It should be noted that during edge offline autonomy, critical loads are given priority in power supply. Critical loads refer to loads that should not be restricted or disconnected due to production continuity, safety, or basic operational needs; non-critical loads refer to loads whose power can be temporarily reduced or stopped when necessary; adjustable loads include charging facilities and other electrical equipment whose operating power, start / stop status, or operating time can be adjusted by the edge EMS control layer. The load priority is used to indicate the power supply sequence of different loads during power limiting or load shedding, and can be pre-set according to production processes, equipment importance, and power demand before the microgrid is put into operation. When performing demand control, the edge EMS control layer prioritizes restricting lower-priority loads and retains power supply to higher-priority loads after demand control requirements are met.

[0134] After generating and issuing local control commands for the current control cycle, the edge EMS control layer enters the next control cycle, re-collects real-time data from the field, and adjusts local control commands based on the updated grid connection demand, actual photovoltaic output, actual load power, and energy storage SOC, thereby forming a closed-loop control in the edge offline autonomous state.

[0135] It should be noted that the sampling period mentioned above refers to the interval between the edge EMS control layer reading real-time data from the field device, which determines the granularity of the system's acquired operating condition information; the control period refers to the interval between the edge EMS control layer executing a complete decision-making cycle and issuing commands, which determines the system's response speed. In this application, the sampling period and the control period can use the same time scale or be configured independently. Typically, the sampling period is shorter than or equal to the control period, enabling the edge EMS control layer to generate local control commands based on the latest field data for each decision, avoiding control deviations due to outdated data, thereby ensuring the timeliness and accuracy of offline autonomous control.

[0136] The following combination Figure 4 Further explanation of load demand control is provided.

[0137] The grid connection point demand refers to the current demand value output by the grid-connected metering equipment according to preset demand statistics rules, used for demand control; the grid connection point power purchase refers to the power obtained by the microgrid from the upstream grid at the current sampling time. The edge EMS control layer can update the grid connection point demand based on the grid connection point power purchase and the preset demand statistics rules. The demand red line is a demand limit value determined based on the project contract demand or the upper limit of grid connection point power control set by the project; the demand warning threshold is lower than the demand red line, used to trigger demand control before the grid connection point demand reaches the demand red line; the demand recovery target is not higher than the demand warning threshold, used to determine whether the demand control has met the preset demand control conditions. For example, the demand warning threshold can be 95% of the demand red line, and the demand recovery target can be 90% of the demand red line. When the grid connection point demand reaches or exceeds the demand warning threshold, the edge EMS control layer performs load demand management, and the control sequence is as follows:

[0138] (a) Stop charging energy storage

[0139] Energy storage charging increases grid connection demand, so the edge EMS control layer first adjusts the energy storage charging power to zero, causing the energy storage system to stop charging.

[0140] (ii) Controlling the discharge of stored energy

[0141] After energy storage charging stops, the edge EMS control layer determines whether the energy storage SOC is higher than the discharge safety lower limit. If it is higher, the energy storage system has the capability to participate in demand regulation, and the edge EMS control layer controls the energy storage system to discharge. The energy storage discharge power is determined based on the power gap between the current demand at the grid connection point and the demand recovery target, and is limited by the allowable discharge power of the energy storage device. By discharging energy storage to provide partial power to the field load, the power obtained by the grid connection point from the upstream grid can be reduced.

[0142] If the energy storage SOC has reached or fallen below the discharge safety limit, the edge EMS control layer will prohibit the energy storage from continuing to discharge in order to avoid deep energy storage depletion.

[0143] (iii) Limiting the power of adjustable loads

[0144] While controlling the discharge of energy storage, the edge EMS control layer can limit the power of adjustable loads such as charging facilities. Reducing the power of adjustable loads can directly reduce the demand at the grid connection point. If the energy storage SOC has reached the lower limit of safe discharge and cannot continue to discharge, the edge EMS control layer will first limit the power of charging facilities and other adjustable loads.

[0145] (iv) Cut off non-critical loads

[0146] When the above measures are still insufficient to reduce the demand at the grid connection point to the demand recovery target, the edge EMS control layer will cut off non-critical loads according to the preset load priority.

[0147] The above control sequence forms a continuous control chain of "stopping energy storage charging → energy storage discharging → limiting adjustable loads → cutting off non-critical loads" to prevent the demand at the grid connection point from continuing to rise and to ensure the power supply to critical loads.

[0148] The following combination Figure 4 Further explanation is provided regarding the utilization of surplus photovoltaic power.

[0149] When the demand at the grid connection point does not reach the demand warning threshold, the edge EMS control layer compares the actual photovoltaic output with the actual load power to determine whether there is surplus photovoltaic power. The surplus photovoltaic power refers to the power in the actual photovoltaic output that exceeds the on-site load power demand when the actual photovoltaic output is greater than the actual load power.

[0150] Furthermore, when the actual output of photovoltaic power exceeds the actual load power, the on-site load prioritizes photovoltaic power supply, and the edge EMS control layer controls the charging of the energy storage system based on the unabsorbed surplus photovoltaic power of the on-site load. The charging power of the energy storage system is jointly limited by the allowable charging power and the energy storage SOC. The energy storage SOC characterizes the ratio of the current remaining electricity of the energy storage system to its available capacity; the allowable charging power refers to the current maximum charging power determined based on the rated power of the energy storage converter, the allowable charging rate of the battery, and the energy storage SOC.

[0151] Furthermore, when the surplus photovoltaic power is less than or equal to the allowable charging power of the energy storage, the edge EMS control layer controls the energy storage system to charge at a power adapted to the surplus photovoltaic power, so that the surplus photovoltaic power is preferentially stored in the energy storage system.

[0152] When the surplus photovoltaic power exceeds the allowable charging power of the energy storage, the edge EMS control layer controls the energy storage system to charge according to the allowable charging power. If the project allows reverse power transmission, the remaining surplus photovoltaic power that the energy storage system cannot absorb will be transmitted to the upper-level grid through the grid connection point.

[0153] It should be noted that when the energy storage SOC reaches the preset charging limit, the edge EMS control layer stops energy storage charging and handles the remaining surplus photovoltaic power according to the project's grid connection conditions. For grid-connected microgrids that allow surplus power to be fed into the grid, the surplus photovoltaic power can be fed into the upstream grid; for islanded microgrids that do not have the conditions for surplus power to be fed into the grid or projects that restrict reverse power transmission, photovoltaic output can be restricted, or adjustable loads with power regulation capabilities can be activated to absorb the surplus photovoltaic power.

[0154] Therefore, the edge EMS control layer determines the distribution method of surplus photovoltaic power among the energy storage system, the upstream grid and adjustable load based on the actual output of photovoltaic power, the actual load power, the energy storage SOC and the grid connection conditions of the project.

[0155] The following combination Figure 4 Further explanation of the graded management and control of energy storage SOC.

[0156] It should be noted that the energy storage SOC is used to reflect the current state of charge of the energy storage system. To balance energy storage safety, on-site power regulation requirements, and operational economy, the edge EMS control layer performs different charging and discharging controls on the energy storage system according to the range of the energy storage SOC.

[0157] In one specific embodiment, 15%, 50%, 80%, and 90% are used as SOC grading thresholds to divide the energy storage SOC into five intervals.

[0158] (1) When the energy storage SOC is below 15%, the energy storage system is judged to be in the low power range. The edge EMS control layer prohibits the energy storage system from continuing to discharge in order to avoid the energy storage system being adversely affected by deep discharge.

[0159] (2) When the energy storage SOC is not less than 15% and less than 50%, the energy storage system can participate in the charging and discharging regulation according to the preset control rules. However, when performing discharge control, the energy storage SOC should be ensured to be not less than the allowable discharge lower limit.

[0160] (3) When the energy storage SOC is not less than 50% and not more than 80%, the energy storage system has relatively sufficient available power and remaining charging capacity. The edge EMS control layer can combine the grid connection point demand, actual load power and current time-of-use electricity price period to control the energy storage system to charge, discharge or standby.

[0161] (4) When the energy storage SOC is higher than 80% and lower than 90%, the energy storage system is close to the preset charging limit. The edge EMS control layer reduces the energy storage charging power to slow down the rate at which the energy storage SOC continues to rise.

[0162] (5) When the energy storage SOC is not lower than 90%, the edge EMS control layer stops charging the energy storage to avoid overcharging of the energy storage system. If there is still surplus photovoltaic power at this time, the surplus power will be fed into the grid, the photovoltaic output will be limited, or the adjustable load power will be increased according to the grid connection conditions of the project.

[0163] It should be noted that the energy storage SOC tiered management and grid connection point demand management work together. When the grid connection point demand reaches the demand warning threshold, but the energy storage SOC has reached the allowable discharge lower limit, the edge EMS control layer no longer reduces the grid connection point demand by discharging the energy storage. Instead, it reduces the grid connection point demand by reducing the power of charging facilities, limiting non-critical loads, or cutting off non-critical loads.

[0164] It should be understood that the above SOC grading thresholds are only used to illustrate one specific implementation method and do not constitute a limitation on other energy storage systems. In practical applications, corresponding SOC grading thresholds can be set according to battery type, battery health status, allowable depth of charge and discharge, and project operation strategy.

[0165] It should be noted that the energy storage SOC safety constraint is a safety boundary that must be met when generating local control commands. The edge EMS control layer must not allow the energy storage SOC to exceed the allowable range through charging or discharging. Under the premise of meeting the aforementioned energy storage SOC safety constraint, grid connection point demand regulation takes precedence over the consumption of surplus photovoltaic power and the economic regulation corresponding to time-of-use pricing.

[0166] The following combination Figure 4 Further explanation is provided regarding the power control linked to time-of-use electricity pricing for energy storage.

[0167] It should be understood that the edge EMS control layer can obtain the time-of-use electricity price period corresponding to the current moment. The time-of-use electricity price period includes peak period, normal period and valley period, and each electricity price period corresponds to a different purchase price.

[0168] The edge EMS control layer can set corresponding charging power limits and discharging power limits for the energy storage system based on the energy storage rated power, energy storage SOC, the grid demand limit corresponding to the project, and the time-of-use electricity price period.

[0169] For example, during off-peak hours, when the electricity purchase price is relatively low, the edge EMS control layer can control the energy storage system to charge when the demand at the grid connection point does not reach the demand warning threshold and the energy storage SOC is lower than the preset charging upper limit, so as to reserve electricity for subsequent peak load periods.

[0170] During normal periods, the edge EMS control layer can control the energy storage system to participate in charge and discharge regulation with relatively small power based on the actual load power, actual photovoltaic output and grid connection point demand, so as to keep the grid connection point power within the target range.

[0171] During peak hours, due to the relatively high electricity purchase price, the edge EMS control layer can prioritize controlling the discharge of the energy storage system when the energy storage SOC is higher than the allowable discharge lower limit, in order to reduce the high-priced electricity that the microgrid obtains from the upper-level grid.

[0172] In one implementation, different time-of-use pricing periods correspond to different upper limits for energy storage operation power. For example, the upper limit for energy storage discharge power during peak hours can be higher than the upper limit for energy storage discharge power during normal hours. The specific power upper limit can be preset based on the rated power of the energy storage system, the allowable charge / discharge rate of the energy storage battery, and the project load level.

[0173] It should be understood that time-of-use pricing-linked power control should comply with the energy storage SOC safety constraints and the grid connection point demand constraints. When it is during peak hours but the energy storage SOC has reached the allowable discharge lower limit, the edge EMS control layer stops energy storage discharge; when it is during off-peak hours but the grid connection point demand has reached the demand warning threshold, the edge EMS control layer no longer controls the energy storage system to charge in order to utilize low-priced electricity.

[0174] Therefore, in the edge offline autonomous mode, the edge EMS control layer can adjust the energy storage charging and discharging power based on real-time field data, while meeting the requirements of energy storage safety and grid connection point demand control, and using time-of-use pricing.

[0175] See Figure 5 The following describes a specific implementation method based on the control logic under the communication disconnection state.

[0176] In this embodiment, a positive energy storage power indicates that the energy storage system is discharging, a negative energy storage power indicates that the energy storage system is charging, and a zero energy storage power indicates that the energy storage system is in standby mode.

[0177] Let the factory's base load power be P. F The charging facility has a power rating of P. EV The actual output of photovoltaic power is P PV The energy storage capacity is P B Without considering the power of other branches and system losses, the power purchased by the grid connection point can be expressed as (P F +P EV -P PV -P BIt should be understood that the grid connection point demand refers to the demand value determined by the grid-connected metering device based on the power purchased by the grid connection point within the demand statistics period, according to preset demand statistics rules. The edge EMS control layer determines the current grid connection point demand based on the power purchased by the grid connection point and the preset demand statistics rules.

[0178] According to the above notation conventions, when the energy storage system is charging, P B When P is negative, charging the energy storage system increases the power drawn from the upstream grid at the grid connection point; when the energy storage system discharges, P... B If the value is positive, the discharge of energy storage will reduce the power obtained by the grid connection point from the upstream grid.

[0179] In this embodiment, the demand limit refers to the demand limit value determined based on the project contract demand or the upper limit of the grid connection point demand control set by the project.

[0180] When the demand at the grid connection point reaches 95% of the aforementioned demand threshold, the edge EMS control layer initiates grid connection point demand management.

[0181] When P B A value less than 0 indicates that the energy storage system is currently in a charging state. The edge EMS control layer first sets P... B The value is adjusted to 0 to stop the energy storage system from charging, thus preventing the energy storage charging from continuing to increase the power purchased by the grid connection point. After the energy storage charging stops, the edge EMS control layer re-determines the grid connection point demand. If the grid connection point demand still does not meet the preset demand control conditions, the subsequent control in energy storage discharge, limiting the power of charging facilities, and limiting or cutting off non-critical loads will continue to be executed.

[0182] When P B When the energy storage SOC is not less than 0 and is higher than 15%, the edge EMS control layer determines the energy storage discharge demand based on the difference between the current demand at the grid connection point and 90% of the demand threshold, in order to reduce the grid connection point demand to around 90% of the demand threshold. The energy storage discharge power does not exceed 750 kW. If the energy storage discharge is still insufficient to meet the demand control requirements, the edge EMS control layer can also reduce the charging facility power and, if necessary, stop the operation of the charging facility.

[0183] When the grid connection point demand reaches 95% of the aforementioned demand threshold and the energy storage SOC is not higher than 15%, the edge EMS control layer will adjust the energy storage power to zero, prohibit the energy storage system from continuing to discharge, and reduce the grid connection point demand by stopping the operation of charging facilities and limiting or cutting off non-critical loads.

[0184] When the demand at the grid connection point is less than 95% of the aforementioned demand threshold, and the sum of the factory's basic load power and the charging facility power is less than the actual output of the photovoltaic system, it indicates that the actual output of the photovoltaic system is greater than the on-site load power demand, and there is surplus photovoltaic power.

[0185] When there is surplus photovoltaic power and the energy storage SOC is below 90%, the edge EMS control layer controls the energy storage system to charge. The target charging power of the energy storage does not exceed the smaller of the surplus photovoltaic power, 750 kW, and the allowable charging power of the energy storage. When the energy storage SOC is above 80% but below 90%, the allowable charging power of the energy storage is reduced according to a preset derating rule. For the remaining surplus photovoltaic power that the energy storage system cannot absorb, it is fed into the upstream grid if the project allows reverse power transmission; if the project does not allow reverse power transmission, the photovoltaic output is limited or the power absorbed by the adjustable load is increased.

[0186] When there is surplus photovoltaic power and the energy storage SOC is not less than 90%, the edge EMS control layer will adjust the energy storage power to 0. If the project allows reverse power transmission, the surplus photovoltaic power will be fed into the upstream grid; if the project does not allow reverse power transmission, the photovoltaic output will be limited or the power absorbed by the adjustable load will be increased.

[0187] When the grid connection point's demand is lower than 95% of the aforementioned demand threshold, and the sum of the factory's base load power and the charging facility power is not less than the actual photovoltaic output, it indicates that there is a net electricity demand on site. This net electricity demand can be determined based on the difference between the factory's base load power, the charging facility power, and the actual photovoltaic output.

[0188] When there is net electricity demand on site and the energy storage SOC is not less than 50%, during peak hours, the energy storage discharge power is the smaller of the net electricity demand, 750 kW, and the allowable energy storage discharge power; during normal hours, the energy storage discharge power is the smaller of the net electricity demand, 375 kW, and the allowable energy storage discharge power.

[0189] When there is net electricity demand on site, the energy storage SOC is not less than 50%, and the current period is during off-peak hours, if the energy storage SOC is less than 80%, the edge EMS control layer will determine the smaller of 250 kW, the allowable charging power of the energy storage, and the remaining demand margin at the grid connection point as the target charging power for the energy storage; if the energy storage SOC is not less than 80%, the edge EMS control layer will control the energy storage system to remain in standby mode. The remaining demand margin at the grid connection point refers to the available adjustment margin between the demand warning threshold and the current demand at the grid connection point.

[0190] When there is net electricity demand on site and the energy storage SOC is below 50%, the energy storage system remains in standby during peak and flat periods; during valley periods, the edge EMS control layer determines the smaller of 750 kW, the allowable charging power of the energy storage, and the remaining demand margin at the grid connection point as the target charging power of the energy storage.

[0191] The 750kW, 375kW, and 250kW figures mentioned above are example power values ​​from a specific project and do not constitute a limitation on the rated power or charging / discharging power of the energy storage system in this application. In other embodiments, corresponding power values ​​can be set according to the rated power of the energy storage converter, the allowable charge / discharge rate of the energy storage battery, the grid-connected capacity of the project, and the on-site load level. When actually implementing off-peak charging control, the energy storage charging power is also limited by the remaining demand margin at the grid connection point; the remaining demand margin at the grid connection point refers to the difference between the demand warning threshold and the current demand at the grid connection point. If charging at the example power would cause the grid connection point demand to reach the demand warning threshold, then the energy storage charging power will be reduced or energy storage charging will be stopped.

[0192] The aforementioned specific control process incorporates grid connection point demand management, photovoltaic surplus power consumption, energy storage SOC hierarchical management and time-of-use pricing-linked power control into the same judgment process, enabling the edge EMS control layer to generate a determined energy storage target status, energy storage target power and load control status based on real-time field data in each local control cycle.

[0193] The following combination Figure 6 Further explanation is provided regarding the recovery and smooth back-switch of cloud-edge control.

[0194] In edge offline autonomous mode, the edge EMS control layer continuously monitors its communication status with the backend scheduling layer. Once the communication link is restored, the edge EMS control layer re-receives the photovoltaic output prediction sequence and load prediction sequence generated by the cloud decision layer, as well as the cloud scheduling strategy generated by the backend scheduling layer, and determines whether the communication status has remained normal for a preset stable duration. The preset stable duration refers to the preset time during which the communication link needs to remain normal before initiating a smooth backslip. Optionally, the edge EMS control layer also determines whether the cloud data validity status has returned to normal based on whether the re-received prediction values ​​and the cloud scheduling strategy are within a preset valid time limit.

[0195] After receiving the recovered prediction sequence and cloud scheduling strategy, the edge EMS control layer does not immediately stop all edge offline autonomous control. Instead, it first redetermines the deviation level based on the current actual on-site operating data.

[0196] Specifically, the edge EMS control layer compares the actual photovoltaic output with the photovoltaic output forecast value corresponding to the current forecast period, and compares the actual load power with the load forecast value corresponding to the current forecast period, and redetermines the deviation level based on the comparison results.

[0197] If the communication status has not been maintained normally for a preset stable duration, or the deviation level is still at the preset severe anomaly level, or the cloud data validity status described in the implementation method for detecting the validity status of cloud data is still abnormal, it indicates that the conditions for resuming the execution of the cloud scheduling strategy have not been met. At this time, the edge EMS control layer continues to execute edge offline autonomous control.

[0198] When the communication status remains normal for a preset stable duration, and the deviation level remains at the preset normal level for a preset number of consecutive evaluation windows, it indicates that the actual on-site operating status has returned to the normal deviation range. In the implementation method for detecting the effective status of cloud data, it should also be required that the effective status of cloud data has returned to normal. The edge EMS control layer compares the differences between the local current energy storage power, charging facility status, and adjustable load status with the target energy storage power, target charging facility status, and target adjustable load status corresponding to the cloud scheduling strategy.

[0199] Based on the aforementioned differences, the edge EMS control layer limits the change in energy storage power between adjacent control cycles and restores the restricted or disconnected adjustable loads in batches according to load priority, so that the current operating state of the edge side gradually approaches the target operating state corresponding to the cloud scheduling strategy.

[0200] The energy storage power variation refers to the change in the target energy storage power between two adjacent control cycles. The allowable range of the energy storage power variation can be preset based on the allowable power variation range of the energy storage equipment and the demand control requirements of the grid connection point. The number of adjustable loads or load power restored in each batch can also be preset based on the remaining available capacity of the grid connection point.

[0201] During the smooth shedding process, the edge EMS control layer continues to monitor the actual photovoltaic output, actual load power, energy storage SOC, and grid connection point demand. When the on-site operating conditions change again, the edge EMS control layer adjusts the shedding process based on the updated on-site operating data to avoid directly executing cloud-based scheduling strategies that are incompatible with the current on-site operating conditions.

[0202] When the current operating state of the edge reaches the target operating state corresponding to the cloud scheduling strategy, or when the difference between the current local control command and the target control command corresponding to the cloud scheduling strategy enters the preset allowable range, the edge EMS control layer stops executing the edge offline autonomous control and resumes executing the cloud scheduling strategy issued by the backend scheduling layer.

[0203] The smooth back-cut is used to avoid the charging and discharging power of the energy storage system from jumping directly from the current power to a target power with a large difference, and at the same time to avoid the simultaneous resumption of operation of multiple adjustable loads that have been restricted or cut off, which would cause a new demand impact on the grid connection point.

[0204] If the communication status becomes abnormal again during the smooth rollback process, the deviation level reaches the preset severe abnormality level again, or the cloud data validity status becomes abnormal again in the implementation of detecting the validity status of cloud data, the edge EMS control layer stops this rollback and continues to execute edge offline autonomous control.

[0205] The following provides further explanation of the closed-loop iteration of the running data in this application.

[0206] The edge EMS control layer records actual microgrid operating data in both cloud-optimized operation mode and edge offline autonomous mode. The recorded actual operating data includes actual photovoltaic output, actual load power, energy storage SOC, energy storage charging and discharging power, grid connection point power, and on-site control results.

[0207] The edge EMS control layer also records deviation levels, mode switching times, mode switching reasons, and local control commands generated and executed during edge offline autonomy.

[0208] The reasons for the mode switching include at least one of the following: communication link disconnection, abnormal cloud data validity status, and serious deviation between actual on-site operating data and corresponding predicted values; wherein, the abnormal cloud data validity status may be caused by the predicted value or cloud scheduling strategy exceeding the preset valid time limit, or the on-site operating status reaching the preset protection condition.

[0209] During the communication link disconnection, the edge EMS control layer saves the actual operating data, mode switching records, and local control commands in the edge-side storage device; after the communication link is restored, the edge EMS control layer sends the data saved during the disconnection back to the back-end scheduling layer.

[0210] The backend scheduling layer aggregates the data transmitted back from the edge EMS control layer and provides the actual photovoltaic output and actual load power to the cloud decision layer. The cloud decision layer compares the photovoltaic output forecast with the actual photovoltaic output for the corresponding time period and the load forecast with the actual load power for the corresponding time period, and updates the corresponding prediction model based on the subsequently accumulated on-site actual operation data.

[0211] The backend scheduling layer adjusts at least one of the deviation judgment parameters and cloud scheduling parameters based on the actual execution results of the cloud scheduling strategy and the field control results during edge offline autonomy. The deviation judgment parameters include at least one of a preset deviation range, a number of continuous evaluation windows, an entry threshold, an exit threshold, and a preset stability duration. The cloud scheduling parameters include weights corresponding to stability, economy, and renewable energy utilization rate, as well as at least one of grid demand constraints, energy storage charging and discharging safety constraints, and equipment operation constraints. The backend scheduling layer sends the updated deviation judgment parameters to the edge EMS control layer for subsequent deviation level judgment and operating mode switching.

[0212] The aforementioned closed-loop update is based on actual on-site operational data and does not require the edge EMS control layer to maintain real-time communication with the backend scheduling layer during edge offline autonomy. After the communication link is restored, by supplementing and transmitting the operational records during the disconnection period, the cloud decision-making layer and the backend scheduling layer can obtain the actual on-site operational status and control execution results during the disconnection period.

[0213] Thus, through a closed-loop processing procedure of "on-site data acquisition - cloud prediction - backend scheduling - edge execution or autonomy - operation data feedback - prediction model and scheduling parameter update", the system can continuously accumulate operation data under different meteorological conditions, load states and abnormal operating conditions during long-term operation, and use the accumulated data to improve subsequent prediction and scheduling results.

[0214] Optionally, for multi-microgrid clusters, a cluster collaborative scheduling function can be added to the back-end scheduling layer to perform power mutual assistance and load transfer between microgrids, based on the fact that each microgrid has its own edge EMS control layer.

[0215] The scope of protection of the microgrid cloud-edge collaborative autonomous control method described in this application is not limited to the execution order of the steps listed in this embodiment. Any solution implemented by adding, subtracting, or replacing steps in the prior art based on the principles of this application is included within the scope of protection of this application.

[0216] This application also provides a microgrid cloud-edge collaborative autonomous control system, which can implement the microgrid cloud-edge collaborative autonomous control method described in this application. However, the implementation device of the microgrid cloud-edge collaborative autonomous control method described in this application includes, but is not limited to, the structure of the microgrid cloud-edge collaborative autonomous control system listed in this embodiment. All structural modifications and substitutions of the prior art made based on the principles of this application are included within the protection scope of this application.

[0217] like Figure 7 As shown, the cloud-based decision layer 101 includes a photovoltaic forecasting module 111, a load forecasting module 112, and a model update module 113. The photovoltaic forecasting module 111 is used to generate a photovoltaic output forecasting sequence, the load forecasting module 112 is used to generate a load forecasting sequence, and the model update module 113 is used to update the forecasting model based on the actual data transmitted from the field.

[0218] The backend scheduling layer 102 includes a data aggregation module 121, an optimized scheduling module 122, and a strategy interaction module 123. The data aggregation module 121 is used to aggregate the field data reported by the edge EMS control layer 103; the optimized scheduling module 122 is used to generate energy storage charging and discharging plans based on the prediction sequence and operating constraints; and the strategy interaction module 123 is used to send the photovoltaic output prediction sequence, load prediction sequence, and cloud scheduling strategy to the edge EMS control layer 103.

[0219] The edge EMS control layer 103 includes a data acquisition module 131, a deviation determination module 132, a mode switching module 133, an offline autonomous module 134, and a back-switch control module 135.

[0220] The data acquisition module 131 is used to acquire data on actual photovoltaic output, actual load power, energy storage SOC, current energy storage power, and grid connection point, all with an acquisition time identifier. The deviation determination module 132 is used to determine the corresponding forecast period based on the acquisition time identifier, compare the actual photovoltaic output and actual load power with the predicted photovoltaic output and load values ​​corresponding to that forecast period, and determine the deviation level.

[0221] The mode switching module 133 is used to select either the cloud-optimized operation mode or the edge offline autonomous mode according to the deviation level and communication status; optionally, the mode switching module 133 is also used to detect the validity status of cloud data and switch to the edge offline autonomous mode when the validity status of cloud data is abnormal.

[0222] The offline autonomous module 134 is used to perform load demand management, photovoltaic surplus power consumption, energy storage SOC hierarchical management and time-of-use pricing-linked power control.

[0223] The back-off control module 135 is used to compare the current local control command with the target control command corresponding to the cloud scheduling strategy after the communication status has been continuously normal for a preset stable time and the deviation level has been maintained at the preset normal level for a preset number of consecutive evaluation windows. It then limits the change in energy storage power between adjacent control cycles and restores the adjustable load in batches until the cloud scheduling strategy is resumed.

[0224] The above functional modules are logically divided for ease of explanation. In actual implementation, multiple functional modules can be executed by the same processor or by multiple communicating processing devices.

[0225] This application embodiment also provides an electronic device, which can be a server, industrial computer or control device for implementing the cloud decision layer 101, the back-end scheduling layer 102 or the edge EMS control layer 103.

[0226] The electronic device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it implements the corresponding steps in the above embodiments, such as predictive data generation, optimized scheduling, actual data acquisition, deviation level determination, operation mode switching, edge offline autonomy, smooth back-switching, and operation data backhaul.

[0227] This application also provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the above-described microgrid cloud-edge collaborative autonomous control method.

[0228] For example, a computer-readable storage medium may be a random access memory, a read-only memory, a flash memory, a hard disk, a solid-state drive, an optical disk, or other non-transitory storage medium capable of storing program code.

[0229] In summary, this application provides a cloud-edge collaborative autonomous control method, system, storage medium, and electronic device for microgrids. Through a hierarchical collaborative architecture of cloud-based predictive decision-making and edge-based autonomous execution, combined with deviation classification judgment, multi-condition mode switching, anti-jitter mechanism, local multi-objective autonomous control, and smooth back-switch mechanism, it achieves at least the following beneficial technical effects:

[0230] 1. Layered collaboration between cloud and edge, balancing global optimization with local real-time reliability.

[0231] The system adopts a three-tiered functional architecture consisting of a cloud-based decision-making layer, a back-end scheduling layer, and an edge EMS control layer. The cloud leverages computing resources to achieve multi-period prediction and multi-objective optimized scheduling, balancing operational economy with renewable energy utilization. The edge layer, located close to field equipment, performs high-frequency data acquisition and local real-time control, reducing the impact of communication latency or interruptions on the continuity of field control. These two layers work together to balance the global optimization capabilities of the cloud with the on-site response capabilities of the edge.

[0232] 2. Dual-dimensional mode switching judgment improves adaptability to extreme working conditions.

[0233] The switching of operating modes is not solely based on cloud-edge communication connectivity; the deviation level between predicted values ​​and actual operating conditions is also incorporated as a criterion. Even if the communication link remains normal, when there is a persistent and severe mismatch between cloud-based predictions and actual on-site conditions, the edge can proactively switch to edge offline autonomous mode and cease executing mismatched cloud-based scheduling strategies. This improves operational adaptability under conditions such as sudden weather changes and load surges, and reduces operational risks caused by scheduling mismatches.

[0234] 3. Dual anti-shake mechanism to suppress frequent shaking in the operating mode.

[0235] By combining continuous evaluation window judgment with hysteresis dual threshold design, the system can filter out mode false triggering caused by instantaneous sampling spikes and random power fluctuations, reduce repeated switching of operating modes near the critical value, ensure stable microgrid operation, and reduce the operational impact caused by frequent equipment start-ups or power jumps.

[0236] 4. Hierarchical priority and local autonomy, with orderly consideration of multiple objectives under disconnection conditions.

[0237] The edge offline autonomous mode uses the safety constraints of energy storage SOC and charging / discharging power as the necessary safety boundaries. Within these boundaries, control priorities are determined according to the order of demand management, photovoltaic power consumption, and economic adjustment. This integrates grid-connected point demand management, photovoltaic surplus power consumption, energy storage SOC hierarchical protection, and time-of-use pricing-linked control into a unified judgment logic. In scenarios where cloud-based dispatch is not required, this mode can ensure the safe operation of energy storage batteries and reduce the risk of grid-connected point demand exceeding limits, while also improving the local consumption of renewable energy and taking into account electricity costs under time-of-use pricing.

[0238] 5. Gradual and smooth back-switch eliminates power surges during mode switching.

[0239] After communication is restored and the operating conditions return to normal, a gradual and smooth switching mechanism is adopted. By limiting the change in energy storage power between adjacent control cycles and restoring adjustable loads in batches according to load priority, the local operating status gradually approaches the target state of cloud scheduling, reducing the power impact and demand fluctuations at the grid connection point caused by large jumps in energy storage power or concentrated load restoration, and achieving a smooth switch between the two operating modes.

[0240] 6. Full-scenario data closed-loop iteration, continuous self-optimization of system performance.

[0241] It supports local caching of operational data during network outages and supplementary data transmission after communication is restored, enabling the cloud to obtain operational records during the outage period. Updating the prediction model, deviation judgment parameters, and cloud scheduling parameters based on actual field operational data improves the adaptability of subsequent prediction results and scheduling strategies to field conditions.

[0242] 7. The architecture is flexible and versatile, adaptable to various project implementation scenarios.

[0243] The solution is adaptable to microgrids with different operating modes, such as grid-connected and islanded types, and supports various grid connection conditions, such as surplus power feeding into the grid and restricted reverse power transmission. At the same time, it can flexibly configure SOC classification thresholds, power limits, deviation thresholds and scheduling target weights according to project needs. It is suitable for various microgrid application scenarios such as industrial and commercial parks and factory areas, and has strong versatility and engineering implementation value. It can also be smoothly extended to multi-microgrid cluster collaborative scheduling scenarios.

[0244] It should be understood that the above embodiments are used to illustrate the technical solutions of this application, and are not intended to limit this application. Equivalent substitutions or modifications made by those skilled in the art without departing from the technical concept of this application should be included within the protection scope of this application.

[0245] In the embodiments provided in this application, it should be understood that the disclosed systems or methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of modules / units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of devices or modules or units may be electrical, mechanical, or other forms.

[0246] The modules / units described as separate components may or may not be physically separate. The components shown as modules / units may or may not be physical modules; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules / units can be selected to achieve the objectives of the embodiments of this application, depending on actual needs. For example, the functional modules / units in the various embodiments of this application may be integrated into one processing module, or each module / unit may exist physically separately, or two or more modules / units may be integrated into one module / unit.

[0247] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0248] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0249] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A cloud-edge collaborative autonomous control method for microgrids, characterized in that, include: The cloud-based decision-making layer generates photovoltaic power output forecasts and load forecasts for multiple future forecast periods, and sends the photovoltaic power output forecasts and load forecasts to the back-end scheduling layer. The back-end scheduling layer receives microgrid operation data reported by the edge EMS control layer, generates a cloud scheduling strategy based on the photovoltaic output forecast, the load forecast, and the microgrid operation data, and sends the photovoltaic output forecast, the load forecast, and the cloud scheduling strategy to the edge EMS control layer. The edge EMS control layer collects the actual photovoltaic output, actual load power, energy storage state of charge, and grid connection point demand with the collection time identifier. It determines the corresponding prediction period based on the collection time identifier, compares the actual photovoltaic output with the predicted photovoltaic output value corresponding to the prediction period, and compares the actual load power with the predicted load value corresponding to the prediction period. It determines the deviation level based on the comparison results. The edge EMS control layer detects the communication status between itself and the backend scheduling layer, and selects either cloud-optimized operation mode or edge offline autonomous mode based on the communication status and the deviation level. When the communication status is abnormal, or when the communication status is normal and the deviation level is determined to be a preset severe abnormal level, the edge offline autonomous mode is switched. In the cloud-optimized operation mode, the edge EMS control layer executes the cloud scheduling strategy; in the edge offline autonomous mode, the edge EMS control layer generates local control commands for controlling at least one of the energy storage system and adjustable load based on the grid connection point demand, the actual photovoltaic output, the actual load power, and the energy storage state of charge; when the communication status returns to normal and the deviation level returns to the preset normal level, the edge EMS control layer resumes executing the cloud scheduling strategy.

2. The microgrid cloud-edge collaborative autonomous control method according to claim 1, characterized in that, The cloud-based decision-making layer generates the photovoltaic output forecast and the load forecast based on the microgrid's historical operating data, environmental meteorological data, and real-time operating data; the historical operating data includes historical photovoltaic output and historical load power, and the environmental meteorological data includes at least one of irradiance and ambient temperature. The back-end scheduling layer generates the cloud scheduling strategy based on time-of-use pricing, grid connection point demand constraints, energy storage charging and discharging safety constraints, and equipment operation constraints. The cloud scheduling strategy includes energy storage charging and discharging plans corresponding to the multiple predicted time periods.

3. The microgrid cloud-edge collaborative autonomous control method according to claim 1, characterized in that, The multiple prediction time periods correspond to the preset prediction time granularity; The edge EMS control layer collects the actual photovoltaic output and the actual load power according to a preset sampling period, and preprocesses the sampled data within an evaluation window to generate on-site evaluation data for judging the deviation level. The on-site evaluation data includes photovoltaic output evaluation value and load power evaluation value. The edge EMS control layer determines the deviation level based on the deviation between the photovoltaic output evaluation value and the corresponding photovoltaic output prediction value, and the deviation between the load power evaluation value and the corresponding load prediction value.

4. The microgrid cloud-edge collaborative autonomous control method according to claim 3, characterized in that, The deviation levels include at least the preset normal level, the preset general abnormal level, and the preset severe abnormal level; When the communication status is normal, the cloud scheduling strategy is executed under the preset normal level; under the preset general abnormal level, the cloud scheduling strategy continues to be executed and abnormal data is uploaded to the backend scheduling layer; under the preset severe abnormal level, the edge offline autonomous mode is switched. The operation mode switch is performed only when the corresponding switching conditions are met for a consecutive preset number of evaluation windows of the deviation level, and the entry threshold for switching from the cloud-optimized operation mode to the edge offline autonomous mode is different from the exit threshold for reverting from the edge offline autonomous mode to the cloud-optimized operation mode.

5. The microgrid cloud-edge collaborative autonomous control method according to claim 1, characterized in that, In the edge offline autonomous mode, provided that the energy storage state of charge and energy storage charging and discharging power are within safe limits, the edge EMS control layer generates the local control commands in the following manner: When the demand at the grid connection point meets the preset demand warning conditions, the edge EMS control layer selects to execute one or more of these in the following order: stop energy storage charging, control energy storage discharge when the energy storage state of charge is higher than the discharge lower limit, limit the power of charging facilities, and cut off non-critical loads, until the demand at the grid connection point meets the preset demand control conditions. When the demand at the grid connection point does not meet the preset demand warning condition and the actual output of the photovoltaic system is greater than the actual load power, the edge EMS control layer controls the energy storage system to absorb the unabsorbed photovoltaic surplus power of the on-site load; when the energy storage system reaches the charging limit condition, the energy storage charging is stopped, and the remaining photovoltaic surplus power is sent to the upper-level grid or the photovoltaic output is limited according to the grid connection conditions of the project. When the demand at the grid connection point does not meet the preset demand warning condition and there is no surplus photovoltaic power, the edge EMS control layer determines the charging state, discharging state, or standby state of the energy storage system based on the charging state interval of the energy storage and the current time-of-use electricity price period, and limits the operating power of the energy storage system.

6. The microgrid cloud-edge collaborative autonomous control method according to claim 1, characterized in that, After the communication status remains normal for a preset stable duration and the deviation level remains at the preset normal level for a preset number of consecutive evaluation windows, the edge EMS control layer compares the current local control command with the target control command corresponding to the cloud scheduling strategy, limits the change in energy storage power between adjacent control cycles, and restores the restricted adjustable load in batches according to load priority. After the current field operation status reaches the target status corresponding to the target control command, the cloud scheduling strategy is resumed.

7. The microgrid cloud-edge collaborative autonomous control method according to claim 1, characterized in that, Also includes: The edge EMS control layer records the microgrid's real-time operating data, the deviation level, the reason for the operating mode switching, local control commands and execution results, and transmits the recorded data back to the cloud decision layer via the back-end scheduling layer. The cloud-based decision-making layer updates the prediction model based on the returned data, and the back-end scheduling layer updates at least one of the deviation judgment parameters and cloud scheduling parameters based on the returned data.

8. A microgrid cloud-edge collaborative autonomous control system, characterized in that, This includes a cloud-based decision-making layer, a backend scheduling layer, and an edge EMS control layer; The cloud-based decision-making layer is used to generate photovoltaic power output forecasts and load forecasts for multiple future forecast periods, and to send the photovoltaic power output forecasts and load forecasts to the back-end scheduling layer. The back-end scheduling layer is used to receive microgrid operation data reported by the edge EMS control layer, generate a cloud scheduling strategy based on the photovoltaic output forecast, the load forecast and the microgrid operation data, and send the photovoltaic output forecast, the load forecast and the cloud scheduling strategy to the edge EMS control layer. The edge EMS control layer is used to collect actual photovoltaic output, actual load power, energy storage state of charge, and grid connection point demand with a collection time identifier. It determines the corresponding prediction period based on the collection time identifier, compares the actual photovoltaic output and actual load power with the predicted photovoltaic output and load values ​​corresponding to the prediction period, and determines the deviation level based on the comparison results. It detects the communication status between itself and the backend scheduling layer, and selects either a cloud-optimized operation mode or an edge offline autonomous mode based on the communication status and the deviation level. When the communication status is abnormal, or when the communication status is normal and the deviation level is determined to be a preset severe abnormality level, it switches to the edge offline autonomous mode. In the cloud-optimized operation mode, it executes the cloud scheduling strategy. In the edge offline autonomous mode, it generates local control commands for controlling at least one of the energy storage system and adjustable load based on the grid connection point demand, the actual photovoltaic output, the actual load power, and the energy storage state of charge. It resumes execution of the cloud scheduling strategy when the communication status returns to normal and the deviation level returns to a preset normal level.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the microgrid cloud-edge collaborative autonomous control method as described in any one of claims 1 to 7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, When the processor executes the computer program, it implements the microgrid cloud-edge collaborative autonomous control method as described in any one of claims 1 to 7.

Images