Cloud-edge collaborative microgrid power management system

Abstract

The application provides a cloud-edge collaborative micro-grid electric energy management system, and relates to the technical field of new energy and smart grid, comprising: a field sensing layer, an edge computing and control layer, a cloud collaborative layer and a user interaction layer; wherein the edge computing and control layer comprises: an intelligent edge energy control device deployed in an energy consumption field; the intelligent edge energy control device comprises: an AI computing localization and real-time strategy reasoning module, an off-network autonomous operation module and a multi-modal data access module; when the communication connection between the cloud collaborative layer and the edge computing and control layer is disconnected, the AI computing localization and real-time strategy reasoning module generates a first optimization decision based on local strategies and real-time data; the off-network autonomous operation module executes the first optimization decision. The edge can immediately take over and independently operate when the network is interrupted, thereby fundamentally solving the pain point that the traditional system is paralyzed when the network is interrupted.

Description

Technical Field

[0001] This invention relates to the fields of new energy and smart grid technology, and in particular to a cloud-edge collaborative microgrid power management system. Background Technology

[0002] As the global energy structure shifts towards cleaner and distributed energy, the large-scale deployment of photovoltaic power generation, energy storage systems, and electric vehicle charging facilities is driving the evolution of the power system from a traditional centralized model to a distributed model with multi-faceted interaction among power sources, grids, loads, and storage. Against this backdrop, new management and control architectures such as virtual power plants and integrated energy management systems have emerged. Through collaboration between the cloud and edge, they enable the aggregated scheduling and efficient utilization of distributed energy resources, becoming a key technological direction for solving the problem of renewable energy consumption and improving the flexibility of grid operation. Among these, the cloud-edge collaborative architecture, combining the global optimization capabilities of the cloud with the real-time response advantages of the edge, is widely used in the field of power management, aiming to balance the intermittency of distributed energy resources, the randomness of loads, and the stability requirements of grid operation.

[0003] Existing cloud-edge collaborative power management systems suffer from a fundamental architectural limitation: a tightly coupled, unidirectionally dependent master-slave control relationship exists between the edge and the cloud. Under this relationship, the system's real-time control and optimization decisions heavily rely on the continuous and stable issuance of commands from the cloud; it cannot cope with abnormal operating conditions such as network communication interruptions. Once disconnected from the cloud, the edge side, lacking complete local decision-making and control closed-loop capabilities, cannot maintain the system's autonomous and stable operation, posing a risk of instability. Summary of the Invention

[0004] One of the objectives of this invention is to provide a cloud-edge collaborative microgrid power management system. Through intelligent edge energy control devices deployed at energy consumption sites, the system enables on-site decision-making at energy consumption sites (stations) and can achieve cloud-edge collaborative control with the cloud collaborative layer. Furthermore, it enables independent operation when the network is interrupted, thereby fundamentally solving the pain point of traditional systems being paralyzed when the network is interrupted.

[0005] This invention provides a cloud-edge collaborative microgrid power management system, comprising: a field sensing layer, an edge computing and control layer, a cloud-cloud collaboration layer, and a user interaction layer; The edge computing and control layer includes: intelligent edge energy control devices deployed at energy consumption sites; the intelligent edge energy control devices include: an AI computing power localization and real-time strategy inference module, a multimodal data access module, and a SCADA server; The multimodal data access module accesses the field sensing layer to collect multimodal operation data of the microgrid and performs local analysis on the operation data to obtain real-time data; the AI ​​computing power localization and real-time policy reasoning module generates the first optimization decision based on the local policy and real-time data; the SCADA server executes the first optimization decision.

[0006] Preferably, the field perception layer collects multimodal operation data of the microgrid; the edge computing and control layer processes the operation data locally and sends the processed data to the cloud collaboration layer; the cloud collaboration layer generates a second optimization decision based on the received data and sends the second optimization decision to the edge computing and control layer for execution; the user interaction layer is used to display the operation status, charging information and the execution results of the optimization decision. Preferably, the cloud collaboration layer includes: the EMS cloud platform and the charging operation and maintenance platform; Among them, the EMS cloud platform is used for model training and the generation of the second optimization decision; The charging operation and maintenance platform is used for charging operation management, user interaction, system maintenance and alarm management.

[0007] Preferably, the user interaction layer includes: a mobile terminal, and / or a web terminal.

[0008] Preferably, the intelligent edge energy control device further includes: a firewall module, a data preprocessing module, a communication management module, and a network switching module; The communication management module is used to access the operational data collected by the field perception layer; the data preprocessing module preprocesses the accessed operational data; and the SCADA server sends the processed data to the cloud collaboration layer through the network exchange module.

[0009] Preferably, during and within a preset time after the network outage, the operational data accessed by the communication management module is directly transferred to the multimodal data access module for local network outage self-governance; after a preset time following the network outage, the multimodal data access module directly accesses the field sensing layer to collect multimodal operational data of the microgrid.

[0010] Preferably, the AI ​​computing power localization and real-time policy inference module generates a first optimization decision based on local policies and real-time data, including: Extract on-site video data of photovoltaic modules from real-time data; Identify photovoltaic panel shading or foreign object coverage in on-site video data of photovoltaic modules; When photovoltaic panel obstruction or foreign object coverage is detected, the maximum power point tracking strategy of the corresponding inverter is changed from the incremental conductance method to the intelligent hybrid algorithm. The maximum power point tracking strategy of the inverter in the intelligent hybrid algorithm is as follows: The illuminance, ambient temperature, output voltage, and output current of the photovoltaic module are collected and converted into a four-dimensional input vector by A / D conversion. When constructing the four-dimensional input vector, a sliding window filter is used to remove high-frequency noise, and the parameters are mapped to the [0,1] interval through normalization to provide standardized input data for subsequent algorithms. Then, a fuzzy logic algorithm is used to infer the step size adjustment coefficient. After coarse adjustment using the step size adjustment coefficient, an improved particle swarm optimization algorithm is used for precise adjustment to achieve the optimal operating point. Finally, a neural network is used for dynamic correction.

[0011] Preferably, the microgrid includes: smart meters, photovoltaic modules and inverters, energy storage units, charging piles, environmental sensors, output links and input links.

[0012] Preferably, the AI ​​computing power localization and real-time policy inference module generates a first optimization decision based on local policies and real-time data, including: Analyze real-time data to determine power generation, remaining storable electricity in energy storage units, and electricity consumption of charging piles; The electrical energy status is determined by comprehensively analyzing the power generation, the remaining storable electricity of the energy storage unit, and the electricity consumption of the charging pile; When the power status is pending external transmission, the local policy is retrieved to determine the connection order of the output links corresponding to the pending external transmission status. The output links are connected based on the order of their connection. Among them, the state of waiting to be exported represents the power state when the total power consumption of the charging pile is less than the total power generation and the remaining storable power is less than or equal to the preset first threshold. The construction of the output link connection order in the local strategy is determined based on the analysis of monitoring data of each output link before the network outage. The specific analysis process is as follows: Based on the type of power demand target connected to the output link, a first priority value is determined using a preset first priority value quantification table; based on the equity value fed back after the power demand target response, a second priority value is determined using a preset second priority value quantification table; based on the loss parameters of each output link, a third priority value is determined using a preset third priority value quantification table; based on the capacity of each output link, a fourth priority value is determined using a preset fourth priority value table; and the sums of the first, second, third, and fourth priority values ​​are sorted in descending order to form the output link connection order.

[0013] The present invention has the following beneficial effects: By deploying intelligent edge energy control devices at energy consumption sites, on-site decision-making at the energy consumption sites (stations) is achieved, enabling cloud-edge collaborative control with the cloud layer. Furthermore, it can operate independently during network outages, fundamentally solving the pain point of traditional systems paralyzing upon network failure. The combination of local autonomous operation and cloud-based global optimization enhances the safety, real-time performance, economy, and scalability of the power system. It supports localized processing of multimodal data, further enriching decision-making dimensions while improving the capacity for renewable energy absorption and the economic efficiency and stability of system operation.

[0014] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of a cloud-edge collaborative microgrid power management system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the multi-dimensional operational data categories collected by the field perception layer in an embodiment of the present invention; Figure 3 This is a topology diagram of a cloud-edge collaborative microgrid power management system according to an embodiment of the present invention; Figure 4 This is a topology diagram of another cloud-edge collaborative microgrid power management system in an embodiment of the present invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0018] This invention provides a cloud-edge collaborative microgrid power management system, such as... Figure 1 As shown, it includes: a field perception layer, an edge computing and control layer, a cloud collaboration layer, and a user interaction layer; The edge computing and control layer includes: intelligent edge energy control devices deployed at energy consumption sites; the intelligent edge energy control devices include: an AI computing power localization and real-time strategy inference module, a multimodal data access module, and a SCADA server; The multimodal data access module connects to the field perception layer to collect multimodal operational data of the microgrid and performs local analysis to obtain real-time data. The AI ​​computing power localization and real-time strategy inference module generates a first optimization decision based on the local strategy and real-time data. The SCADA server executes the first optimization decision. The local strategy is the control strategy stored in the intelligent edge energy control device, specifically including: the inverter's maximum power point tracking strategy, the scheduling strategy for the electricity generated by the photovoltaic modules, the control strategy for the output link, and the control strategy for the input link. The control strategy for the output link includes: the control strategy that specifies the connection order of the output link. The control strategy for the input link includes: the control strategy that specifies the access rules. The local strategy is a rule table configured by the cloud collaboration layer when communicating with the cloud collaboration layer and is updated periodically by the cloud collaboration layer.

[0019] This invention achieves local processing and decision-making of multimodal data from power plants through a first optimization decision determined by a local strategy, enabling autonomous operation even in the event of a network outage. Since the intelligent edge energy control device is deployed at the energy consumption site, such as a photovoltaic-storage-charging power plant, on-site data is collected and analyzed in real time, eliminating the need for data transmission and decision reception stages, achieving millisecond-level response. This means that the intelligent edge energy control device achieves local data autonomy even in the event of a network outage, including autonomous scheduling strategies and autonomous equipment control. For power plants equipped with intelligent edge energy control devices, regardless of whether they are connected to or disconnected from the network, rapid local data acquisition and strategy response are possible, with strategy response times reaching milliseconds, maximizing economic benefits such as peak-valley arbitrage and energy storage utilization. Furthermore, when the intelligent edge energy control device is connected to the network, the cloud collaboration layer can update the local strategy in real time. In the event of a network outage, inertial decisions are made based on the local strategy to ensure the continuous and stable operation of the power plant.

[0020] In addition, the determination of whether the communication connection between the cloud collaboration layer, edge computing and control layer is broken can be made by monitoring the heartbeat packets used in the communication.

[0021] In addition, the field perception layer collects multimodal operational data of the microgrid; the edge computing and control layer processes the operational data locally and sends the processed data to the cloud collaboration layer; the cloud collaboration layer generates a second optimization decision based on the received data and sends the second optimization decision to the edge computing and control layer for execution; the user interaction layer displays the operating status, charging information, and execution results of the optimization decision; when communication is restored after a network outage, the intelligent edge energy control device first uploads the latest operational data during the interruption to the cloud to complete status synchronization, ensuring that the cloud has a comprehensive understanding of the current system operating conditions, and then automatically switches back to the collaborative control mode to receive the cloud optimization strategy again. Furthermore, under normal communication conditions, using the cloud-edge collaborative control mode, the intelligent edge energy control device receives and executes the global optimization strategy issued by the cloud collaboration layer while performing local control based on local strategies, fully leveraging the cloud's global coordination and optimization capabilities to ensure the overall operational efficiency of the system. The cloud collaboration layer trains deep learning models based on historical data from multiple sites and coordinates global optimization and cross-site energy scheduling. The global optimization strategy includes: the second optimization strategy.

[0022] The microgrid comprises: smart meters, photovoltaic modules and inverters, energy storage units, charging piles, environmental sensors, output links, and input links. Photovoltaic modules and inverters convert solar energy into electrical energy; charging piles charge electric vehicles; environmental sensors detect environmental parameters within the microgrid; energy storage units store the electrical energy generated by the photovoltaic modules and inverters; output links are used to output electrical energy and can be configured in multiple ways to connect to other nearby electrical devices, the power grid, or the microgrid itself; input links are used to receive external electrical energy and can also be configured in multiple ways to connect to other nearby power generation equipment, the power grid, or the microgrid itself; the configuration of input and output links enables energy scheduling between microgrids and the output of electrical energy from microgrids to other electrical devices; smart meters are configured at various nodes of the power grid to monitor electrical parameters; and the core function of the field sensing layer is to comprehensively collect operational data from all devices on the microgrid. Smart meters primarily capture core power parameters such as voltage, current, and power. Photovoltaic modules and inverters mainly collect power generation and equipment operating status. Energy storage units collect data such as state of charge, charging and discharging efficiency, and operational faults. Charging piles collect real-time charging power, charging time, and equipment operating conditions. Environmental sensors collect environmental information such as on-site temperature, humidity, light, and smoke, providing complete data support for subsequent scheduling decisions.

[0023] The cloud-based collaboration layer comprises the EMS cloud platform and the charging operation and maintenance platform. The EMS cloud platform is used for model training and generating secondary optimization decisions. The charging operation and maintenance platform is used for charging operation management, user interaction, system operation and maintenance, and alarm management. User interaction refers to interactions related to operation management. Based on historical operating data from multiple sites, the EMS cloud platform trains and optimizes models for load forecasting and energy storage scheduling, generating globally optimal control strategies. The EMS cloud platform focuses on operational functions such as charging order management, payment settlement, and user account maintenance. The two layers collaborate to achieve the dual goals of technical optimization and commercial operation. For example, primary optimization decisions include: output link connection, input link access, and inverter maximum power point tracking (MPPT) strategy; secondary optimization decisions include: output link connection, input link access, inverter MPPT strategy, equipment shutdown control, and energy scheduling strategies between sites.

[0024] The user interaction layer includes mobile terminals and / or web terminals. Through this layer, users are displayed the overall operating status, real-time data of each device, charging progress, cost details, and fault alarm information. It supports remote viewing and operation, improving the energy user experience and ease of maintenance.

[0025] Furthermore, to achieve adaptability of the local strategy, when the network is reconnected after an outage, the cloud-based collaboration layer updates the local strategy and continuously updates it at the configured update interval during reconnection. Outages are usually low-probability events and the outage time is generally not too long. During this period, the probability of fluctuations in external electricity prices and link availability is even smaller. Therefore, the local strategy is to maintain the stable operation of the microgrid under outage autonomy, rather than prioritizing optimal operation. Suboptimal operation is also acceptable.

[0026] The intelligent edge energy control device also includes: a firewall module, a data preprocessing module, a communication management module, a SCADA server, and a network switching module; The communication management module is used to access operational data collected by the field perception layer; the data preprocessing module preprocesses the accessed operational data; and the SCADA server sends the processed data to the cloud collaboration layer through the network exchange module. Although both the communication management module and the multimodal data access module are used to access operational data collected by the field perception layer, their specific functions differ. They can be configured as two parallel modules, accessing data simultaneously, but with different data flows: the communication management module's data flows to the data preprocessing module, while the multimodal data access module's data flows to the AI ​​computing power localization and real-time policy inference module. Alternatively, they can be configured as two serial modules, connecting the multimodal data access module to the communication management module. In this case, the communication management module's data is copied twice, one copy flowing to the data preprocessing module and the other to the multimodal data access module.

[0027] As the core control unit on the edge side, the intelligent edge energy control device ensures low-latency data transmission between the firewall module, data preprocessing module, communication management module, SCADA server, AI computing power localization and real-time policy inference module, network-off autonomous operation module, multimodal data access module, and network switching module via internal bus connections. The firewall module provides physical isolation between internal and external networks, eliminating the risk of external network attacks. The communication management module is compatible with multiple communication protocols such as Modbus and TCP / IP, allowing access to various types of terminal devices. The SCADA server monitors the operating status of each device in real time and executes control commands. The AI ​​computing power localization and real-time policy inference module is equipped with an AI acceleration chip, providing local model training and inference capabilities. The network switching module enables data interaction and command distribution with the cloud collaboration layer.

[0028] To ensure stable system operation in complex environments and enhance device compatibility, the intelligent edge energy control device supports multi-protocol adaptation, enabling connection to photovoltaic inverters, energy storage converters, charging piles, wind turbines, monitoring equipment, gate barriers, electricity meters, integrated protection devices, and meteorological sensors. Multi-protocol adaptation is a key design feature for improving device compatibility. The intelligent edge energy control device is compatible with multiple mainstream communication protocols such as Modbus, TCP / IP, and 104 protocol, allowing direct communication with different types and manufacturers of devices, including photovoltaic inverters, energy storage converters, and charging piles. This broad device access capability allows the system to integrate various core devices such as photovoltaic, energy storage, charging, and meteorological monitoring equipment, achieving centralized management and control of various energy consumption and monitoring resources. It also adapts to different device configurations in various scenarios, further enhancing the system's environmental adaptability and device compatibility.

[0029] like Figure 3 and Figure 4 The figures shown represent specific system architectures of this application. Figure 3The AI ​​computing power node is equipped with an AI computing power localization and real-time policy inference module, the network switch is equipped with a network switching module, and the communication management machine is equipped with a communication management module. Figure 3 The architecture primarily describes the data flow when working with the cloud-edge collaboration layer system; Figure 4 The EMS localization unit is equipped with AI computing power localization and real-time strategy inference modules, multimodal data access modules, and SCADA servers, etc. like Figure 2 As shown, the operational data includes: structured power data and multimodal unstructured data; The structured power data includes voltage, current, power, and energy storage state of charge data; the multimodal unstructured data includes on-site video data of photovoltaic modules and environmental audio data.

[0030] The on-site sensing layer collects structured power data. Voltage, current, and power data directly reflect the energy transmission and consumption status of the power system, while energy storage state-of-charge data accurately reflects the available capacity of the energy storage system, providing core quantitative basis for power balance scheduling and charge / discharge control. Multimodal unstructured data supplements system operation information from multiple perspectives. On-site video data of photovoltaic modules can intuitively present the physical state of the photovoltaic panels, and environmental audio data can help determine whether there is abnormal noise in the equipment operation, corroborating the structured power data. The combined collection of structured and unstructured data breaks the traditional single sensing mode that relies solely on power parameters. It covers the core operating indicators of the power system while also taking into account the physical state of the equipment, instantaneous dynamics, and environmental correlation information, achieving a comprehensive and in-depth perception of the system's operating status and providing more comprehensive data support for subsequent intelligent decision-making.

[0031] In one embodiment, during and within a preset time period following a network outage, the operational data accessed by the communication management module is directly transferred to the multimodal data access module for local network outage autonomous processing. After the preset time following the network outage, the multimodal data access module directly accesses the field sensing layer to collect multimodal operational data of the microgrid. The system provided in this embodiment differs from other embodiments in that it is controlled by the cloud when connected to the network, and by the edge side after a network outage. The preset time ranges from 1 minute to 20 minutes. In one embodiment, the AI ​​computing power localization and real-time policy inference module generates a first optimization decision based on local policies and real-time data, including: Extract on-site video data of photovoltaic modules from real-time data; Identify photovoltaic panel shading or foreign object coverage in on-site video data of photovoltaic modules; When photovoltaic panel obstruction or foreign object coverage is detected, the maximum power point tracking strategy of the corresponding inverter is changed from the incremental conductance method to the intelligent hybrid algorithm.

[0032] In this embodiment, the AI ​​computing power localization and real-time strategy reasoning module first performs image recognition processing on the on-site video data of the photovoltaic module to accurately capture whether there are obstructions or foreign objects covering the surface of the photovoltaic panel that affect the power generation efficiency. When the photovoltaic panel is found to be obstructed or covered by foreign objects, the maximum power point tracking strategy of the inverter is adjusted to optimize the working parameters of the photovoltaic module in real time, effectively offsetting the impact of external factors on the power generation efficiency, and ultimately achieving the improvement of photovoltaic utilization efficiency and precise control of the photovoltaic module status. Among them, the incremental conductance method is based on the mathematical derivative principle to determine... Under certain conditions, the maximum power point can be accurately located; the intelligent hybrid algorithm integrates fuzzy logic, neural networks or particle swarm optimization to locate the maximum power point, can identify multiple peaks in the PV curve, and quickly jump out of local optima, which is suitable for component occlusion scenarios.

[0033] The maximum power point tracking strategy for inverters using the intelligent hybrid algorithm consists of three main stages: fuzzy logic pre-judgment, improved particle swarm optimization for precise tracking, and neural network dynamic correction. First, the illuminance, ambient temperature, output voltage, and output current of the photovoltaic module are collected and converted into a four-dimensional input vector using an A / D converter. During the construction of the four-dimensional input vector, a sliding window filter is used to remove high-frequency noise, and the parameters are normalized to the [0,1] interval to provide standardized input data for subsequent algorithms. Then, the step size adjustment coefficient is inferred by the fuzzy logic algorithm. After coarse adjustment using the step size adjustment coefficient, the improved particle swarm optimization algorithm performs precise adjustment to achieve the optimal operating point. Finally, a neural network is used for dynamic correction. Among them, the fuzzy logic algorithm is a dual-input single-output fuzzy inference system based on expert experience. The input is the voltage change ΔV and the power change ΔP, and the output is the disturbance step size adjustment coefficient K. The fuzzy inference process is as follows: ΔV and ΔP are divided into 5 fuzzy subsets: "negative large (NB), negative small (NS), zero (Z), positive small (PS), and positive large (PB)". The step size adjustment command is generated by the Mamdani inference method to realize the rapid location of the maximum power point region. The improved particle swarm optimization algorithm uses the following particle encoding method: the operating voltage is used as the particle position vector, each particle represents a candidate operating point, and the population size is set to 20-30 particles; an adaptive weight adjustment is adopted: a nonlinear inertial weight strategy is introduced, with a larger weight (0.8-1.0) used in the initial stage to ensure global search capability, and then linearly reduced to 0.4-0.6 in the later stage of iteration to enhance local search accuracy; the absolute value of the error between photovoltaic output power and theoretical maximum power is used as the fitness index; when the fitness change is less than 0.1% for 5 consecutive iterations or the maximum number of iterations (50 times) is reached, the search is terminated and the optimal operating point is output. The neural network structure employs a 3-layer feedforward neural network. The input layer has three nodes representing light intensity, temperature, and voltage; the hidden layer has 16 neurons (using the ReLU activation function); and the output layer has one node representing the optimal voltage prediction value. Offline training mechanism: A training dataset is constructed based on historical meteorological data and photovoltaic characteristic curves. The backpropagation algorithm is used to optimize the network weights, enabling dynamic modeling of environmental changes. Online correction strategy: When the deviation between the optimal operating point output by the PSO algorithm and the neural network prediction value exceeds 5%, a weight update mechanism is triggered to correct the model parameters in real time to adapt to environmental changes. Hybrid Algorithm Switching Mechanism: In the fuzzy logic pre-judgment stage, fuzzy logic control is used to achieve rapid coarse localization, guiding the operating point to the vicinity of the maximum power point. In the improved particle swarm optimization (PSO) precise tracking stage, an improved PSO algorithm is activated for local fine-grained search, ensuring convergence to the maximum power point within a short time. In the neural network dynamic correction stage, the optimal operating point is predicted in real time using a neural network. When the rate of change of environmental parameters exceeds 20% / min, the PSO algorithm is triggered to re-search, achieving dynamic tracking.

[0034] In one embodiment, the AI ​​computing power localization and real-time policy inference module generates a first optimization decision based on local policies and real-time data, including: Analyze real-time data to determine power generation, remaining storable electricity in energy storage units, and electricity consumption of charging piles; The electrical energy status is determined by comprehensively analyzing the power generation, the remaining storable electricity of the energy storage unit, and the electricity consumption of the charging pile; When the power status is pending external transmission, the local policy is retrieved to determine the connection order of the output links corresponding to the pending external transmission status. The output links are connected based on the order of their connection. The state of pending external transmission refers to the energy state when the total electricity consumption of the charging pile is less than the total power generation and the remaining storable electricity is less than or equal to a preset first threshold.

[0035] This embodiment provides a first optimization decision generation method when external power transmission is required. Firstly, real-time data analysis determines whether external power transmission is needed. Specifically, this analysis can be determined by power generation, the remaining storable capacity of energy storage units, and the power consumption of charging piles. Power generation is the total power generation of photovoltaic power generation per unit time; the remaining storable capacity of energy storage units is the amount of electricity that the energy storage units can still store, i.e., the difference between the total capacity and the existing capacity; the power consumption of charging piles is calculated based on the charging rate of the charging piles in use; the specific indication of needing external power transmission is that the total power consumption of charging piles is less than the total power generation and the remaining storable capacity is less than or equal to a preset first threshold. Specifically, the total power consumption of charging piles is the sum of the power consumption of all charging piles; the total power generation is the sum of the power generation of all photovoltaic modules; and any value between 1% and 5% of the total capacity can be used as the first threshold. Connections are requested sequentially according to the output link connection order specified in the local strategy; once connected, the photovoltaic power is transmitted externally. The output links include lines connecting to nearby power grids, electrical equipment, power plants, and other power demand targets. The construction of the output link connection order in the local strategy is determined based on the analysis of monitoring data of each output link before the network outage. The specific analysis process is as follows: Based on the type of power demand target connected to the output link, a first priority value is determined using a preset first priority value quantification table; based on the equity value fed back after the power demand target response, a second priority value is determined using a preset second priority value quantification table; based on the loss parameters of each output link, a third priority value is determined using a preset third priority value quantification table; based on the capacity of each output link, a fourth priority value is determined using a preset fourth priority value table; and the sums of the first, second, third, and fourth priority values ​​are sorted in descending order to form the output link connection order. The first, second, third, and fourth priority value quantification tables are pre-configured by professionals. In the first priority value quantification table, the first priority value is associated with the type of electricity demand target. In the second priority value quantification table, the second priority value is associated with the equity value, which can be specifically represented as the price per unit of electricity. In the third priority value quantification table, the third priority value is associated with the loss parameter, which is specifically represented as the loss ratio. In the fourth priority value quantification table, the capacity to be accommodated is associated with the fourth priority value. The loss ratio is the difference between the total output electricity and the actual received electricity, and the ratio of the two.

[0036] After a network outage, the AI ​​computing power localization and real-time strategy inference module can optimize the output link connection order based on historical output link connectivity. Specific optimizations include: counting the number of output link connection requests rejected after a preset time period from the current point in time; querying a preset correction table to determine the correction value to be deducted; and then reordering the output link connection order by deducting the sum of the first, second, third, and fourth priority values ​​based on the correction value. The preset time period can be set to within one month.

[0037] In one embodiment, the AI ​​computing power localization and real-time policy inference module generates a first optimization decision based on local policies and real-time data, and further includes: Analyze real-time data to determine the status of each input link; When a single input link is in the state of pending access, based on the local policy, combined with the power generation, the remaining storable power of the energy storage unit and the power consumption of the charging pile, it is determined whether to access it. When it is determined to access it, at least one device that can be used as an electrical load is simultaneously identified from the battery cluster of the energy storage unit and the charging pile as the target object. Based on the target object, execute the access action; When multiple input links are in the state of pending access, the local policy is used to determine the accessible input links and the target objects corresponding to each input link, based on the power generation, the remaining storable power of the energy storage unit, and the power consumption of the charging pile. Based on the accessible input links and target objects, perform access actions.

[0038] This embodiment monitors the data on the input link to determine whether there is power access. When there is, the input link is connected to the charging pile or energy storage unit that needs power, and the connection is determined according to the actual situation to meet the external power access request and realize the rational scheduling of power.

[0039] When a single input link is in the "pending access" state, access is determined by combining the power generation, the remaining storable capacity of the energy storage unit, and the power consumption of the charging pile. This includes: determining the power increase / decrease parameter based on the power generation, the remaining storable capacity of the energy storage unit, and the power consumption of the charging pile; querying a pre-configured access threshold table based on the power increase / decrease parameter and the remaining storable capacity to obtain the access threshold; sampling the power parameters of the input link; comparing the sampled power parameters with the access threshold; allowing access if the sampled power parameters are less than or equal to the access threshold, otherwise not allowing access. Specifically, the power increase / decrease parameter is the ratio of the total power generation minus the total power consumption to the remaining storable capacity. The access threshold table is pre-built, and the power increase / decrease parameter, remaining storable capacity, and access threshold correspond to each other. The rules for building the access threshold table include: the larger the power increase / decrease parameter, the smaller the access threshold; the smaller the remaining storable capacity, the smaller the access threshold; and the access threshold in the access threshold table represents the maximum amount of power allowed to be accessed per unit time. Specifically, when multiple input links are in the "pending access" state, access is determined by combining the power generation, the remaining storable capacity of the energy storage unit, and the power consumption of the charging pile. This includes: determining power level adjustment parameters based on power generation, the remaining storable capacity of the energy storage unit, and the power consumption of the charging pile; querying a pre-configured access threshold table based on the power level adjustment parameters and the remaining storable capacity to obtain the access threshold; sampling the power parameters of the input links; comparing the sampled power parameter data with the access threshold; allowing access when the sum of the sampled power parameter data of each input link is less than or equal to the access threshold; and allowing access when any input link... Access is not permitted if the sampled power parameters of a road exceed the access threshold. When neither of these two conditions applies, input links are combined to determine multiple access scenarios. The benefits of each access scenario are evaluated, and the input link with the highest evaluated benefit value is selected. The evaluation process involves: determining the evaluation value based on the difference between a pre-set benchmark value and the access benefit value; assigning weights based on the ratio of the sampled power parameters of each input link in each access scenario; and obtaining the benefit value through a weighted sum of the evaluation value and the weights. The benchmark value must be greater than the access benefit value. When access is confirmed, at least one device that can serve as an electrical load is simultaneously identified from the battery clusters of the energy storage unit and the charging piles as the target object, including: The system iterates through the battery clusters and charging stations of the energy storage unit to filter out potential targets. The selection criteria for battery clusters are whether the current power level is less than or equal to 80% of the total capacity. The selection criteria for charging stations are whether they use the power of the energy storage unit as the access point. Determine the allowed access parameters for each candidate target; the allowed access parameters include the amount of electricity allowed to be accessed per unit time. Based on each allowed access parameter, determine the minimum access unit; calculate the greatest common divisor of each allowed access parameter, and use the greatest common divisor as the minimum access unit; Based on the minimum access unit, each candidate target and input link is unitized to obtain the first sequence unit and the second sequence unit; the first sequence unit is obtained by fuzzy unitization of each candidate target based on the minimum access unit; the second sequence unit is obtained by fuzzy unitization of each input link based on the minimum access unit; fuzzy unitization is the operation of splitting each candidate target or each output link into an integer number of minimum access units; the minimum access unit is a certain power value within a unit of time as the minimum unit of analysis; Number the first sequence unit and the second sequence unit; The first sequence unit and the second sequence unit are simulated and connected separately, and the simulated connection is evaluated. The first sequence unit and the second sequence unit are paired and connected one by one according to the evaluation value from largest to smallest. The battery cluster and charging pile of the energy storage unit corresponding to the second sequence unit paired with the first sequence unit are taken as the target objects.

[0040] Simulated connections were performed on the first and second sequence units respectively, and the simulated connections were evaluated, specifically including: The simulated connection includes the loss parameters of the lines and equipment that need to be traversed after connecting the first sequence unit and the second sequence unit. The evaluation value is obtained by querying a pre-configured evaluation value table through the loss parameters; the larger the loss parameter in the evaluation value table, the smaller the evaluation value.

[0041] Although the duration of network outages is generally not too long, there are still strategies to deal with prolonged outages. Local policy updates can be performed by analyzing data manually entered by staff. To ensure the security of the update process through analysis of manually entered data, in one embodiment, the intelligent edge energy control device further includes a risk monitoring module for monitoring the update of local policies after a network outage. The risk monitoring module performs the following operations: Configure the initial time threshold (any time between 2 minutes and 30 minutes), the update license time (any time between 2 minutes and 30 minutes), and the timer; After the network is disconnected, the timer starts counting. When the timer reaches the first time threshold, staff are allowed to update the local policy within the permitted update time. When the license renewal period ends, the timer is reset to zero and restarted. In addition, the risk monitoring module can perform risk analysis on specific updates. To achieve this, a policy library must first be configured, which stores relevant policies for microgrid power management. The policy library categorizes the policies; local policies are constructed by selecting one policy from each category. Furthermore, when constructing the policy library, a risk value is configured for switching from one policy to another within the same category. The relevant policies can be configured by professionals, and the risk value can be configured based on their experience. The specific monitoring actions of the risk monitoring module include: Based on the overall strategy database, determine the risk value of switching from the current strategy in the local strategy to the strategy updated by the staff; If the risk value is greater than or equal to the risk threshold, the update will be rejected. The formula for determining the risk threshold is as follows: ; In the formula, Indicates the risk threshold. The first constant is pre-configured (any value between 10 and 100). Reset the timer by a certain number of times; The initial correction value configured for the corresponding staff member; The weights corresponding to the staff members; The upper limit of the pre-configured weights; The lower limit of the pre-configured weights; To count the number of times the staff member's update operation evaluation result was positive in the historical network outage update operation records; Count the number of times the staff member's update operation evaluation result was reversed in the historical network outage update operation records; , For the pre-configured adjustment value, ; The steps for determining whether the evaluation result of the staff member's update operation in the historical network outage update operation record is positive or negative are as follows: When staff first connect to the network after updating, the switching status of the updated strategy is determined; if no strategy is switched after connecting to the network, the evaluation result is positive. When switching policies after connecting to the network, the risk value is determined based on the overall policy database as the first reference value; The risk value of directly switching from the policy before the network disconnection to the policy after the network connection is determined based on the policy database and used as the second reference value; When the first reference value is greater than or equal to the second reference value, the evaluation result is reversed; When the first reference value is less than the second reference value, the evaluation result is positive.

[0042] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A cloud-edge collaborative microgrid power management system, characterized in that, include: The layers consist of: on-site perception layer, edge computing and control layer, cloud collaboration layer, and user interaction layer. The edge computing and control layer includes: intelligent edge energy control devices deployed at energy consumption sites; the intelligent edge energy control devices include: an AI computing power localization and real-time strategy inference module, a multimodal data access module, and a SCADA server; The multimodal data access module accesses the field sensing layer to collect multimodal operation data of the microgrid and performs local analysis on the operation data to obtain real-time data; the AI ​​computing power localization and real-time policy reasoning module generates the first optimization decision based on the local policy and real-time data; the SCADA server executes the first optimization decision.

2. The cloud-edge collaborative microgrid power management system as described in claim 1, characterized in that, The field sensing layer collects multimodal operation data of the microgrid; the edge computing and control layer processes the operation data locally and sends the processed data to the cloud collaboration layer. The cloud collaboration layer generates a second optimization decision based on the received data and sends the second optimization decision to the edge computing and control layer for execution; the user interaction layer is used to display the running status, charging information and the execution results of the optimization decision.

3. The cloud-edge collaborative microgrid power management system as described in claim 1, characterized in that, The cloud-based collaboration layer includes: the EMS cloud platform and the charging operation and maintenance platform; Among them, the EMS cloud platform is used for model training and the generation of the second optimization decision; The charging operation and maintenance platform is used for charging operation management, user interaction, system maintenance and alarm management.

4. The cloud-edge collaborative microgrid power management system as described in claim 1, characterized in that, The user interaction layer includes: mobile terminals, and / or, web terminals.

5. The cloud-edge collaborative microgrid power management system as described in claim 1, characterized in that, The intelligent edge energy control device also includes: a firewall module, a data preprocessing module, a communication management module, and a network switching module; The communication management module is used to access the operational data collected by the field perception layer; the data preprocessing module preprocesses the accessed operational data; and the SCADA server sends the processed data to the cloud collaboration layer through the network exchange module.

6. The cloud-edge collaborative microgrid power management system as described in claim 1, characterized in that, During the moment of network outage and within a preset time thereafter, the operational data accessed by the communication management module is directly transferred to the multimodal data access module for local network outage self-governance process; after a preset time following the network outage, the multimodal data access module directly accesses the field sensing layer to collect multimodal operational data of the microgrid.

7. The cloud-edge collaborative microgrid power management system as described in claim 1, characterized in that, The AI ​​computing power localization and real-time policy inference module generates the first optimization decision based on local policies and real-time data, including: Extract on-site video data of photovoltaic modules from real-time data; Identify photovoltaic panel shading or foreign object coverage in on-site video data of photovoltaic modules; When photovoltaic panel obstruction or foreign object coverage is detected, the maximum power point tracking strategy of the corresponding inverter is changed from the incremental conductance method to the intelligent hybrid algorithm. The maximum power point tracking strategy of the inverter in the intelligent hybrid algorithm is as follows: The illuminance, ambient temperature, output voltage, and output current of the photovoltaic module are collected and converted into a four-dimensional input vector by A / D conversion. When constructing the four-dimensional input vector, a sliding window filter is used to remove high-frequency noise, and the parameters are mapped to the [0,1] interval through normalization to provide standardized input data for subsequent algorithms. Then, a fuzzy logic algorithm is used to infer the step size adjustment coefficient. After coarse adjustment using the step size adjustment coefficient, an improved particle swarm optimization algorithm is used for precise adjustment to achieve the optimal operating point. Finally, a neural network is used for dynamic correction.

8. The cloud-edge collaborative microgrid power management system as described in claim 1, characterized in that, Microgrids include: Smart meters, photovoltaic modules and inverters, energy storage units, charging piles, environmental sensors, output links and input links.

9. The cloud-edge collaborative microgrid power management system as described in claim 8, characterized in that, The AI ​​computing power localization and real-time policy inference module generates the first optimization decision based on local policies and real-time data, including: Analyze real-time data to determine power generation, remaining storable electricity in energy storage units, and electricity consumption of charging piles; The electrical energy status is determined by comprehensively analyzing the power generation, the remaining storable electricity of the energy storage unit, and the electricity consumption of the charging pile; When the power status is pending external transmission, the local policy is retrieved to determine the connection order of the output links corresponding to the pending external transmission status. The output links are connected based on the order of their connection. Among them, the state of waiting to be exported represents the power state when the total power consumption of the charging pile is less than the total power generation and the remaining storable power is less than or equal to the preset first threshold. The construction of the output link connection order in the local strategy is determined based on the analysis of monitoring data of each output link before the network outage. The specific analysis process is as follows: Based on the type of power demand target connected to the output link, a first priority value is determined using a preset first priority value quantification table; based on the equity value fed back after the power demand target response, a second priority value is determined using a preset second priority value quantification table; based on the loss parameters of each output link, a third priority value is determined using a preset third priority value quantification table; based on the capacity of each output link, a fourth priority value is determined using a preset fourth priority value table; and the sums of the first, second, third, and fourth priority values ​​are sorted in descending order to form the output link connection order.

Images