Edge computing resource scheduling and self-healing method and device based on containerized micro-service

Abstract

The invention discloses an edge computing resource scheduling and self-healing method and device based on containerized micro-service. The method comprises the following steps: acquiring operation data in a multi-energy scene through a heterogeneous terminal; when a certain micro-service in the edge computing gateway breaks down, the edge computing gateway based on the containerized micro-service architecture dynamically dispatches operation of the micro-service to realize fault self-healing of the micro-service; when a certain edge computing gateway in the edge computing gateway cluster formed through the negotiation of the local area network breaks down, another healthy gateway is selected to bear the responsibility of the broken-down edge computing gateway; when the edge computing gateway detects that the operation data reach a trigger condition, executing control by adopting a scheduling strategy based on event driving, computing a new output distribution scheme of each energy unit through edge-cloud cooperation, and issuing an adjustment instruction to the corresponding energy equipment by the edge computing gateway according to the output distribution scheme; the system reliability is improved, the real-time response capability is enhanced, and the energy utilization rate is optimized.

Description

Technical Field

[0001] This application belongs to the fields of industrial automation control and power information technology, and specifically relates to an edge computing resource scheduling and self-healing method and device based on containerized microservices. Background Technology

[0002] With the explosive growth in the number of IoT devices and the expansion of new energy access, traditional centralized cloud computing processing models face numerous challenges. In industrial control and smart grid scenarios, a large number of sensors and actuators generate real-time data. If all of this data is uploaded to the cloud for processing, it not only puts high pressure on network bandwidth but also makes it difficult to respond promptly to rapid changes on-site due to communication latency (typical cloud response latency is tens of milliseconds or even higher). Edge computing, by deploying computing gateways closer to the data source, offloads some computation and decision-making locally, reducing response latency to below 10 milliseconds and significantly improving real-time performance and reliability.

[0003] Existing edge computing gateways typically employ embedded hardware, supporting various industrial and IoT protocols to access heterogeneous terminal devices and achieve on-site data acquisition and preliminary processing. However, in traditional solutions, each edge gateway often operates independently, lacking a collaborative mechanism between gateways. If a gateway fails, monitoring and control tasks in its connected area may be interrupted, creating a single point of failure risk. While some systems use cloud redundancy to take over tasks, cloud intervention involves significant communication latency, making it difficult to restore local services promptly. Furthermore, the fault tolerance of edge nodes themselves is mostly limited to software self-monitoring and automatic restart, which is insufficient for distributed self-healing at the entire edge network level.

[0004] On the other hand, in new energy microgrids or regional energy internet, multiple energy forms such as hydropower, wind power, and photovoltaics are often deployed simultaneously (i.e., "hydro-wind-solar" multi-energy complementary systems). The energy dispatch of such systems is highly dynamic and uncertain: for example, instantaneous changes in wind speed can cause fluctuations in wind turbine output, changes in solar intensity can affect photovoltaic power output, and sudden increases in load can lead to supply-demand imbalances. Traditionally, adjustments to the output of each power source are made through pre-set dispatch plans or manual instructions from the dispatch center. However, this approach cannot quickly respond to sudden events on-site, potentially missing opportunities to utilize surplus wind and solar energy or causing frequency and voltage fluctuations. How to leverage the real-time capabilities of edge computing to automatically sense events on-site and trigger coordinated control, dynamically optimizing multi-energy output, is currently a research hotspot in the field of smart grids.

[0005] Edge computing platforms based on containerized microservice architecture offer excellent elasticity and manageability. By encapsulating functional modules as containerized microservices, services can be deployed, started, or stopped on demand, enabling dynamic resource scheduling and fault isolation. When a microservice experiences an anomaly, the container orchestration system can automatically restart or migrate the service, achieving self-healing capabilities. However, in current technologies, edge-side container orchestration is often limited to a single gateway device, lacking effective mechanisms for collaborative fault tolerance in multi-gateway distributed environments. Furthermore, in energy management scenarios, integrating event-driven control strategies with containerized platforms remains an unresolved issue.

[0006] In summary, the existing technology has the following shortcomings: 1) It lacks a fault-tolerant mechanism for collaborative autonomy among multiple edge gateways, and it is difficult for a neighboring gateway to take over its function in a timely manner when a single gateway fails; 2) It lacks an event-driven scheduling method for multi-energy collaborative scenarios such as water, wind and solar power, and cannot make full use of the real-time performance of edge computing to optimize energy allocation. Summary of the Invention

[0007] To address the aforementioned issues, this application provides a method for edge computing resource scheduling and self-healing based on containerized microservices, the improvement of which includes: Obtain operational data in multi-energy scenarios through heterogeneous terminals; When a microservice inside the edge computing gateway fails, the edge computing gateway based on the containerized microservice architecture can achieve self-healing of the microservice by dynamically scheduling the operation of the microservice. When one of the edge computing gateways in an edge computing gateway cluster formed through local area network negotiation fails, another healthy gateway is selected to take over the responsibilities of the failed edge computing gateway. When the edge computing gateway detects that the running data has reached the trigger condition, it uses an event-driven scheduling strategy to execute control. Through edge-cloud collaboration, it calculates a new power allocation scheme for each energy unit and issues adjustment commands to the corresponding energy devices according to the power allocation scheme. Among them, the edge computing gateway based on the containerized microservice architecture encapsulates the microservices of data acquisition, protocol conversion, edge analysis, device control and communication management into independent containers within the edge computing gateway.

[0008] Optionally, the edge computing gateway based on a containerized microservice architecture achieves microservice fault self-healing by dynamically scheduling the execution of microservices through a container orchestration system, including: Edge computing gateways based on containerized microservice architectures deploy and schedule microservices through a container orchestration system. They dynamically adjust the operation of each container based on resource utilization and policies, and set health monitoring and automatic restart policies for critical microservices to achieve automatic fault recovery.

[0009] Optionally, the selection of another health gateway by the collaborative management service to assume the responsibilities of the fault edge computing gateway includes: The collaborative management service selects a standby edge computing gateway node in the edge computing gateway cluster to automatically load the containerized microservice instance of the failed node, establish a communication link with the heterogeneous terminal connected to the failed edge computing gateway, and continue execution from the most recently synchronized state data.

[0010] Optionally, the execution control using an event-driven scheduling strategy, which calculates a new power allocation scheme for each energy unit through edge-cloud collaboration based on a multi-objective optimization strategy, and issues adjustment commands from the edge computing gateway to the corresponding energy devices according to the power allocation scheme, includes: The edge computing gateway cluster invokes pre-deployed optimization algorithms or control strategies in the cloud to perform execution control in response to detected events; By collecting operational data, energy output status, and event trigger records uploaded from the edge via the cloud, big data analysis and model training are performed to obtain artificial intelligence models for resource scheduling or equipment control. These models can quickly calculate new output allocation schemes for each energy unit and then distribute effective artificial intelligence models for resource scheduling or equipment control to the edge deployment. According to the power distribution plan, the edge computing gateway issues adjustment commands to the corresponding energy devices.

[0011] Optionally, the step of the edge computing gateway detecting that the running data has reached the trigger condition includes: The wind speed rises above the set threshold within a unit of time and the wind turbine is not currently operating at full capacity; the local load increases above the set standard value in a short period of time and the voltage shows a downward trend; the photovoltaic output power drops by more than a certain amount in a short period of time. The trigger condition is met when the edge computing gateway detects that at least one of the above conditions is met.

[0012] Based on the same inventive concept, this application also provides an edge computing resource scheduling and self-healing device based on containerized microservices, the improvement of which includes: The data acquisition unit is used to acquire operational data in multi-energy scenarios through heterogeneous terminals; The microservice fault unit is used to enable the edge computing gateway, based on a containerized microservice architecture, to achieve self-healing of microservices by dynamically scheduling the operation of microservices when a microservice inside the edge computing gateway fails. The gateway failure unit is used to select another healthy gateway to assume the responsibilities of the failed edge computing gateway when one of the edge computing gateways in the edge computing gateway cluster formed through local area network negotiation fails. The event fault unit is used to execute control using an event-driven scheduling strategy when the edge computing gateway detects that the running data has reached the trigger condition. It calculates a new power allocation scheme for each energy unit through edge-cloud collaboration and issues adjustment instructions to the corresponding energy equipment according to the power allocation scheme. Among them, the edge computing gateway based on the containerized microservice architecture encapsulates the microservices of data acquisition, protocol conversion, edge analysis, device control and communication management into independent containers within the edge computing gateway.

[0013] Optionally, the edge computing gateway based on a containerized microservice architecture achieves microservice fault self-healing by dynamically scheduling the execution of microservices through a container orchestration system, including: Edge computing gateways based on containerized microservice architectures deploy and schedule microservices through a container orchestration system. They dynamically adjust the operation of each container based on resource utilization and policies, and set health monitoring and automatic restart policies for critical microservices to achieve automatic fault recovery.

[0014] Optionally, the selection of another health gateway by the collaborative management service to assume the responsibilities of the fault edge computing gateway includes: The collaborative management service selects a standby edge computing gateway node in the edge computing gateway cluster to automatically load the containerized microservice instance of the failed node, establish a communication link with the heterogeneous terminal connected to the failed edge computing gateway, and continue execution from the most recently synchronized state data.

[0015] Optionally, the execution control using an event-driven scheduling strategy, which calculates a new power allocation scheme for each energy unit through edge-cloud collaboration based on a multi-objective optimization strategy, and issues adjustment commands from the edge computing gateway to the corresponding energy devices according to the power allocation scheme, includes: The edge computing gateway cluster invokes pre-deployed optimization algorithms or control strategies in the cloud to perform execution control in response to detected events; By collecting operational data, energy output status, and event trigger records uploaded from the edge via the cloud, big data analysis and model training are performed to obtain artificial intelligence models for resource scheduling or equipment control. These models can quickly calculate new output allocation schemes for each energy unit and then distribute effective artificial intelligence models for resource scheduling or equipment control to the edge deployment. According to the power distribution plan, the edge computing gateway issues adjustment commands to the corresponding energy devices.

[0016] Optionally, the step of the edge computing gateway detecting that the running data has reached the trigger condition includes: The wind speed rises above the set threshold within a unit of time and the wind turbine is not currently operating at full capacity; the local load increases above the set standard value in a short period of time and the voltage shows a downward trend; the photovoltaic output power drops by more than a certain amount in a short period of time. The trigger condition is met when the edge computing gateway detects that at least one of the above conditions is met.

[0017] This application provides a method and apparatus for edge computing resource scheduling and self-healing based on containerized microservices, comprising: acquiring operational data in multi-energy scenarios through heterogeneous terminals; when a microservice within an edge computing gateway fails, the edge computing gateway based on a containerized microservice architecture achieves self-healing of the microservice by dynamically scheduling the operation of the microservice; when an edge computing gateway in an edge computing gateway cluster formed through local area network negotiation fails, another healthy gateway is selected to assume the responsibilities of the failed edge computing gateway; when the edge computing gateway detects that the operational data reaches the triggering condition, it executes control using an event-driven scheduling strategy, calculates a new power allocation scheme for each energy unit through edge-cloud collaboration, and issues adjustment commands to the corresponding energy devices according to the power allocation scheme; Through multi-gateway autonomous collaboration, when a single edge gateway fails, it can quickly and automatically switch over, with other gateways seamlessly taking over the tasks, avoiding single points of failure and significantly improving the overall availability of the system.

[0018] Event-driven scheduling strategies enable edge systems to sense and respond to changes in field conditions within milliseconds, automatically executing optimized control and significantly reducing response latency compared to traditional cloud-based scheduling. This is particularly beneficial for scenarios with frequent power output fluctuations, such as wind and solar power, allowing for timely adjustments to control strategies to maintain stable grid operation.

[0019] Through intelligent scheduling with edge-cloud collaboration, clean energy is maximized while ensuring load demand. When wind and solar power output is sufficient, the edge control strategy prioritizes increasing wind turbine and solar power output and reducing hydropower output for water storage and backup. Conversely, when wind and solar power output is insufficient, hydropower output is rapidly increased to fill the gap, achieving a dynamic optimal combination of wind, hydro, and solar energy, significantly improving the utilization rate of renewable energy and reducing wind and solar curtailment.

[0020] This application is compatible with heterogeneous industrial equipment access and power system safety specifications, supports the integration of local closed-loop control and cloud intelligence, and is suitable for various application scenarios such as smart power distribution rooms, microgrid energy management, and industrial process control. It has good versatility and practical value.

[0021] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 The diagram illustrates the implementation steps of an edge computing resource scheduling and self-healing method based on containerized microservices provided in this application. Figure 2 The flowchart of an edge computing resource scheduling and self-healing method based on containerized microservices provided in this application is shown. Figure 3 This paper presents an organizational diagram of an edge computing resource scheduling and self-healing device based on containerized microservices, as provided in this application. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] Example 1 This application provides a method for edge computing resource scheduling and self-healing based on containerized microservices, such as... Figure 1 ,include: Obtain operational data in multi-energy scenarios through heterogeneous terminals; When a microservice inside the edge computing gateway fails, the edge computing gateway based on the containerized microservice architecture can achieve self-healing of the microservice by dynamically scheduling the operation of the microservice. When one of the edge computing gateways in an edge computing gateway cluster formed through local area network negotiation fails, another healthy gateway is selected to take over the responsibilities of the failed edge computing gateway. When the edge computing gateway detects that the running data has reached the trigger condition, it uses an event-driven scheduling strategy to execute control. Through edge-cloud collaboration, it calculates a new power allocation scheme for each energy unit and issues adjustment commands to the corresponding energy devices according to the power allocation scheme. Among them, the edge computing gateway based on the containerized microservice architecture encapsulates the microservices of data acquisition, protocol conversion, edge analysis, device control and communication management into independent containers within the edge computing gateway.

[0026] Optionally, the edge computing gateway based on a containerized microservice architecture achieves microservice fault self-healing by dynamically scheduling the execution of microservices through a container orchestration system, including: Edge computing gateways based on containerized microservice architectures deploy and schedule microservices through a container orchestration system. They dynamically adjust the operation of each container based on resource utilization and policies, and set health monitoring and automatic restart policies for critical microservices to achieve automatic fault recovery.

[0027] Optionally, the selection of another health gateway by the collaborative management service to assume the responsibilities of the fault edge computing gateway includes: The collaborative management service selects a standby edge computing gateway node in the edge computing gateway cluster to automatically load the containerized microservice instance of the failed node, establish a communication link with the heterogeneous terminal connected to the failed edge computing gateway, and continue execution from the most recently synchronized state data.

[0028] Optionally, the execution control using an event-driven scheduling strategy, which calculates a new power allocation scheme for each energy unit through edge-cloud collaboration based on a multi-objective optimization strategy, and issues adjustment commands from the edge computing gateway to the corresponding energy devices according to the power allocation scheme, includes: The edge computing gateway cluster invokes pre-deployed optimization algorithms or control strategies in the cloud to perform execution control in response to detected events; By collecting operational data, energy output status, and event trigger records uploaded from the edge via the cloud, big data analysis and model training are performed to obtain artificial intelligence models for resource scheduling or equipment control. These models can quickly calculate new output allocation schemes for each energy unit and then distribute effective artificial intelligence models for resource scheduling or equipment control to the edge deployment. According to the power distribution plan, the edge computing gateway issues adjustment commands to the corresponding energy devices.

[0029] Optionally, the step of the edge computing gateway detecting that the running data has reached the trigger condition includes: The wind speed rises above the set threshold within a unit of time and the wind turbine is not currently operating at full capacity; the local load increases above the set standard value in a short period of time and the voltage shows a downward trend; the photovoltaic output power drops by more than a certain amount in a short period of time. The trigger condition is met when the edge computing gateway detects that at least one of the above conditions is met.

[0030] Example 2 Based on the same inventive concept, this application also provides a preferred embodiment of an edge computing resource scheduling and self-healing method based on containerized microservices, such as... Figure 2 ,include: Step 1: Heterogeneous terminal devices access at least one edge computing gateway through multiple protocols, and the edge computing gateway collects data from the terminal devices and performs preliminary control. Step 2: Deploy edge applications on the edge computing gateway using a containerized microservice architecture. Encapsulate functional modules such as data preprocessing, protocol conversion, local analysis, and device control into independent containers. The container orchestration system dynamically schedules the operation of each microservice to adapt to changes in resource load, and achieves local self-healing of microservice failures through health monitoring and automatic restart. Step 3: Implement local control closed loop at the edge computing gateway. When the network is disconnected from the cloud or an anomaly occurs, the edge computing gateway autonomously completes the control operation of the terminal device based on the preset control logic and local real-time data to maintain the basic stability of the field system. Step 4: The cloud platform performs global model training and optimization analysis based on the data uploaded by the edge computing gateway to obtain an artificial intelligence model for resource scheduling or device control. The model is then deployed and run on the edge computing gateway, enabling the edge to use local inference results to guide real-time decision-making and achieve synergy between cloud training and edge inference. Step 5: Construct an autonomous collaborative network of multiple edge computing gateways: Through a local area network, multiple edge computing gateways are automatically discovered and connected to each other, exchange status information to negotiate and form an edge gateway cluster, and realize unified collaborative management of edge resources. Step 6: Configure a distributed fault tolerance mechanism within the edge gateway cluster: Each edge computing gateway monitors the running status of each other. When a faulty gateway in the cluster is detected to be offline, another normally operating gateway takes over the containerized microservice tasks carried by the faulty gateway and resumes operation from the pre-synchronized status data, so as to realize the fault self-healing takeover of edge computing tasks. Step 7: Set event monitoring rules and monitor the on-site environment and load status in real time, including at least parameters such as wind speed, light intensity, and local load power. When a predefined event triggering condition is detected (including rapid changes in wind speed, sudden increases in load, or sudden drops in photovoltaic output), a corresponding event triggering signal is generated at the edge computing gateway. Step 8: Execute event-driven resource scheduling and energy optimization strategies based on event trigger signals: The edge computing gateway cluster calls pre-deployed optimization algorithms or control strategies for the detected events, calculates the output power adjustment of each energy unit (hydropower unit, wind turbine and photovoltaic power generation device) under edge-cloud collaboration, and generates a multi-energy coordinated control scheme; Step 9: According to the control scheme, the edge computing gateway issues adjustment commands to the corresponding energy devices to dynamically adjust the output distribution of each energy unit in order to optimize the power supply and demand balance: reduce hydropower output for energy storage when wind and solar energy are in surplus, increase hydropower output to make up for the gap when renewable energy is insufficient, and balance the frequency and voltage of the power grid in real time. Step 10: Ensure high availability of data and control command transmission through multi-channel redundant communication. The edge computing gateway simultaneously accesses wired and wireless network links and evaluates link quality in real time. When the main link fails or its performance degrades, it automatically switches to the backup link. It also adopts multi-path parallel transmission or retransmission mechanisms to improve the reliability of critical data. Step 11: Take security protection measures to ensure that the system complies with the power network security specifications, including encrypted data transmission between the edge computing gateway and the terminal device and the cloud, two-way identity authentication between the terminal and the gateway, and secure isolation of the microservice container runtime environment, thereby preventing unauthorized access and data tampering.

[0031] Optionally, in step 5, multiple edge computing gateways automatically discover each other and exchange identification information through the zero-configuration discovery protocol of local area networking, use a distributed leader election or consensus algorithm to determine the cluster management method, and maintain a unified cluster member list and a shared status data table for collaborative scheduling.

[0032] Optionally, the task takeover of the faulty gateway in step 6 includes: instantiating the critical containerized microservices running on the faulty gateway by a pre-specified backup gateway or a gateway dynamically selected according to the load, loading the running status or cached data of the faulty gateway in its most recent synchronization, and after the backup gateway takes over, establishing a communication link with the terminal device connected to the faulty gateway and continuing to execute the unfinished tasks.

[0033] Optionally, the predefined event triggering conditions in step 7 include one or more of the following: the change in wind speed within a preset time window exceeds a threshold, the increase in load power within a preset time exceeds a threshold, the grid frequency deviation exceeds a threshold, and the rate of decline in photovoltaic output exceeds a threshold. When the above conditions are met, a trigger signal for the corresponding event is generated at the edge computing gateway.

[0034] Optionally, the optimization algorithm called by the edge computing gateway cluster in step 8 is based on a multi-objective optimization strategy, taking into account real-time power balance, economy and equipment safety constraints. The algorithm uses a cloud-trained intelligent model or a pre-set rule curve. When an event is triggered, it quickly calculates the target output value of each energy source with parameters such as the current availability margin of wind power and photovoltaic power and the reservoir energy storage status as input.

[0035] Optionally, the adjustment commands issued by the edge computing gateway to the energy devices in step 9 include: adjusting the opening of the guide vanes of the water turbine or the number of units put into operation to change the hydropower output, sending the power set value to the power controller of the wind turbine to change the wind turbine output, and adjusting the output power percentage of the photovoltaic inverter, thereby completing the dynamic adjustment of the output of each energy source on a time scale of milliseconds to seconds.

[0036] Optionally, the multi-channel redundant communication in step 10 includes an edge computing gateway equipped with a cellular mobile communication module and a wired Ethernet interface. When the main communication link is interrupted or the delay exceeds a preset threshold, the gateway switches to a backup link within 50 milliseconds to continue transmitting data. Real-time control commands are sent simultaneously via at least two different types of links and require confirmation from the other end to improve the reliability of critical control information transmission.

[0037] Optionally, in step 11, the edge computing gateway adopts protection measures that comply with the safety specifications of secondary equipment in the power system, including a device authentication mechanism based on a unique hardware identifier and digital certificate, 256-bit encryption of communication data, and integrity verification and security log auditing of edge-side control commands, to ensure that the edge scheduling and control process meets the safety requirements of the power industry.

[0038] Optionally, the event triggering logic includes: within a preset time window t, when a change in wind speed Δv = |v(t) - v(t - Δt)| or a change in load power ΔP = |P(t) is detected... P(t When Δv(t) exceeds its respective preset threshold, i.e., if (Δv>Vth)∨(ΔP>Pth), the event is triggered. v(t) is the real-time wind speed measurement at time t; P(t) is the real-time load power (or system power) at time t; Δt is the preset time window length; Δv is the wind speed change within the window Δt; ΔP is the power change within the window Δt; Vth is the wind speed change trigger threshold; Pth is the power change trigger threshold; ∨: logical "OR" operator; the event is triggered if either condition is met.

[0039] Optionally, it also includes an energy dispatch optimization algorithm; the optimization objective function of the algorithm is expressed in mathematical form as follows: ; The constraints include that the total output power of the system at each time step equals the load demand at that time step, i.e.

[0040] Pi(t) is the output power of the i-th energy unit (i=1,…,N) at time t; Pout(t) is the total output power of the system at time t; L(t) is the load demand power at time t; T is the total number of discrete time steps in the scheduling cycle; N is the number of grid-connected energy units; J is the optimization objective function value, representing the quadratic sum of power fluctuations; min is the solution to Pi(t) that minimizes the objective function J. The expression for the total output power of the system is:

[0041] The overall structure of this application is as follows: Autonomous collaboration and fault-tolerant cooperation mechanism among multiple edge computing gateways: Edge gateways automatically discover each other through the local area network and negotiate to form an autonomous alliance. When a gateway fails, other gateways in the alliance can detect it and take over its work through task takeover, state synchronization, etc., to achieve distributed fault self-healing. Specifically, each gateway runs a lightweight collaborative management service and monitors each other's heartbeats. When a node is detected to be offline or fail, the backup node automatically loads the containerized microservice instance of the failed node and continues to run from the most recently synchronized state data, thereby minimizing the impact of the failure on the system.

[0042] Event-driven scheduling and energy optimization strategies for hydropower, wind power, and solar power co-generation scenarios: In complex scenarios involving multiple energy sources such as hydropower, wind power, and solar power, an event-driven edge scheduling strategy is introduced. The edge gateway monitors key events in real time (such as sudden changes in wind speed, sudden changes in sunlight, or load surges). When a trigger condition is detected, resource scheduling and multi-energy coordinated control strategies are automatically activated. Under edge-cloud collaboration, the optimal output power allocation scheme for each energy unit is calculated, and the output of hydro turbines, wind turbines, and solar inverters is dynamically adjusted to maximize the utilization of renewable energy and maintain grid stability while ensuring load demand.

[0043] Heterogeneous terminal access: Various heterogeneous terminals such as field sensors, measurement and control devices and actuators are connected to the edge computing gateway through multiple protocols, such as industrial fieldbus (Modbus, CAN), IoT protocols (MQTT, CoAP) or power-specific protocols (DL / T 645, IEC 61850), to ensure compatible access for different types of devices.

[0044] Containerized microservice architecture: The edge gateway adopts a containerized microservice architecture, encapsulating functional modules such as data acquisition, protocol conversion, edge analysis, device control, and communication management into independent containers. A container orchestration system is used to deploy and schedule microservices, dynamically adjusting the operation of each container based on resource utilization and policies (such as scaling up or down as needed or migrating to other nodes). Health monitoring and automatic restart policies are set for critical services to achieve automatic fault recovery.

[0045] Local control closed loop: Local control logic is pre-deployed on the edge gateway, such as a PID controller or local rule engine based on edge-side measurements. In the event of network failure or emergency, the edge gateway can independently complete the predetermined control tasks (such as power switching, local load shedding, etc.) on its own, forming a control closed loop from the device end to the edge end, ensuring the safe and stable operation of field devices.

[0046] Cloud-edge collaborative AI: The cloud platform performs deep learning modeling and analysis on historical data to train AI models for prediction and optimized control (such as new energy output prediction models, load prediction models, or optimized scheduling decision models). After training, the models are deployed to edge gateways and run in real time locally through inference engines, achieving a combination of cloud training and edge inference. On the one hand, the powerful computing capabilities of the cloud are used for global optimization and long-term model training; on the other hand, the edge utilizes the obtained models to quickly make local decisions, improving response speed and reducing the computing and communication burden on the cloud.

[0047] Multi-channel adaptive communication: The communication links between the edge gateway and the cloud, as well as between the gateway itself, are designed in a multi-channel redundancy mode. Preferably, the gateway simultaneously possesses multiple network interfaces such as wired Ethernet, 5G / 4G cellular networks, Wi-Fi, or private networks. The system monitors the network quality of each channel in real time (such as latency, packet loss rate, and signal strength). When the primary channel experiences an interruption or performance degradation, it automatically switches to the backup channel, completing link switching within milliseconds to ensure communication continuity. Important control commands can be sent in parallel across multiple channels or employ an acknowledgment and retransmission mechanism to further improve communication reliability.

[0048] Security Protection: The entire edge computing system meets the requirements of power system network security specifications. Data transmission between the edge gateway, terminal devices, and the cloud uses encryption protocols (such as TLS / SSL) to ensure confidentiality, and device access requires two-way authentication to prevent unauthorized access. Simultaneously, the gateway itself employs secure boot and sandbox isolation technologies to protect against malicious code, and critical control containers run in a trusted execution environment to ensure that edge-side control decisions are not tampered with or interfered with.

[0049] The edge computing system of this application comprises several edge computing gateways interconnected via a local area network to form an edge gateway cluster. The cluster connects to a cloud management platform via a high-bandwidth, low-latency network. Each edge gateway connects to a group of field terminal devices through multiple protocol interfaces. These devices may include sensors, actuators, and distributed energy devices (such as hydroelectric generators, wind turbines, and photovoltaic inverters). The edge gateways adopt a containerized microservice architecture, running several containerized service modules within them, such as data acquisition and preprocessing services, protocol conversion services, local analysis and AI inference services, device control services, and communication management services. The container orchestration module is responsible for managing the deployment, startup, and shutdown of these microservices and monitoring their operational status.

[0050] In this embodiment, heterogeneous terminal devices in the industrial field access the edge gateway via different protocols. For example, several measurement sensors access the gateway via Modbus, smart meters via DL / T 645 protocol, video surveillance cameras via RTSP stream, and environmental sensors via MQTT protocol and WiFi. The same gateway has multiple protocol adaptation modules (which can be implemented as independent containers) to uniformly convert data from different sources into an internally common data model. For example, the protocol conversion service parses Modbus register data into semantically meaningful measurement values ​​and MQTT messages into internal objects, which are then passed to a data preprocessing service for filtering, aggregation, and anomaly detection. The processed key data is stored in the gateway's local cache database and simultaneously sent to other service modules and the cloud platform that have subscribed to the corresponding topics via a publish / subscribe mechanism.

[0051] The container orchestration module on the edge gateway is responsible for dynamically scheduling the operation of microservice containers based on preset policies and resource conditions. For example, when the number of connected devices and data traffic increases over a period of time, the orchestration module can automatically start additional analytics service containers to share the load based on CPU / memory utilization; during idle periods, it reduces redundant containers to release resources. Each service running within a container has a heartbeat report or monitoring probe, and the orchestration module periodically checks its status. If a service experiences an abnormal shutdown or becomes unresponsive, the corresponding container is immediately restarted; if multiple restarts fail consecutively, an alarm is triggered, and the service container can be migrated to another gateway in the cluster (if the service supports cross-node operation). For instance, if the data analytics service fails on the gateway, the system can start a container instance with the same image on the gateway to continue the analytics tasks.

[0052] This application introduces a multi-edge gateway collaboration mechanism. Each edge gateway automatically discovers each other upon startup using a local area network discovery protocol. For example, it can broadcast its existence using mDNS / UDP multicast. Upon receiving the broadcast, neighboring gateways exchange identity information via handshake and establish a peer-to-peer connection. The collaborative management services within the gateway cluster elect a master coordinating node (or a decentralized consensus algorithm can be used instead of a master). Each gateway maintains a cluster member list and a shared state table, periodically monitoring the online status of other members via heartbeat messages. Once a member gateway's heartbeat times out or it explicitly reports a failure, the cluster determines that the gateway has failed and gone offline.

[0053] When a gateway (e.g., a gateway) fails, the autonomous cluster mechanism immediately initiates a self-healing process. Assume the gateway was originally running several critical containerized services (e.g., device control services). First, the collaborative management service selects another healthy gateway (e.g., a gateway) to take over the responsibilities of the failed gateway. The selection strategy can be based on the real-time load and task matching of each gateway, choosing a gateway with sufficient resources and corresponding service images as the takeover node. The takeover node retrieves the most recently synchronized critical status data from the shared status table of the failed node.

[0054] For example, the gateway periodically copies the status of its device control services (current device operating mode, control parameters, etc.) to the cluster's shared storage or sends it directly to partner nodes for backup; therefore, the gateway can obtain this status when taking over. Subsequently, the gateway starts the container instance corresponding to the failed service (if it was not previously running locally) through the container orchestration module, loads the aforementioned status data, and establishes communication with the terminal device originally connected to the gateway to take over control. (Where the physical connection of the terminal device allows, the device may be connected to multiple gateways simultaneously or be accessible through another gateway via the network; for devices that can only be physically connected to a single gateway, compensation control can be synchronously compensated after the failed gateway recovers.) The entire process, from the occurrence of a failure to the completion of the new service takeover, can be completed within seconds, ensuring uninterrupted business operation. After the failed gateway recovers and comes back online, it can automatically obtain the data logs accumulated during operation from the currently taking over node, update its status, and enable it to seamlessly rejoin the cluster and resume operation.

[0055] This paper describes event-driven scheduling in a multi-energy collaborative scenario involving hydropower, wind power, and solar power. An edge gateway cluster is deployed within a regional energy system comprising a hydropower station, a wind power plant, and a solar power station. Each gateway acquires key environmental and operational parameters through sensors, such as wind speed sensor data, solar intensity meter data, reservoir water level and available water volume, real-time load demand, and grid frequency and voltage. This embodiment sets several trigger event rules, such as: "Wind speed surge event"—triggered when wind speed increases beyond a set threshold within a unit of time and the wind turbine is not currently operating at full capacity; "Load surge event"—triggered when local load increases by more than X% in a short period and voltage shows a downward trend; "Solar intensity drop event"—triggered when solar power output decreases by a certain amount within a short period, and so on. The edge gateway continuously runs an event detection module to analyze the sensor data stream, and immediately generates an event message locally once any event condition is met.

[0056] When an event occurs, the edge gateway cluster executes scheduling and control actions based on predefined strategies and intelligent algorithms. For example, for a "sudden increase in wind speed" event, the scheduling strategy is to prioritize the use of wind power's additional output and reduce hydropower output to store hydropower for backup, while ensuring load demand is met. The specific process is as follows: After the event is triggered, the edge gateway invokes its locally deployed energy optimization model (trained and distributed from the cloud, for example, based on reinforcement learning or optimization algorithms), inputting data such as the current wind turbine power margin, the current photovoltaic output level, the remaining storage capacity of the reservoir, and the load gap. The model quickly calculates and provides a new output allocation scheme for each energy unit, such as increasing wind turbine output from 80% to 100% of its rated capacity, reducing the number of hydropower units in operation by one to lower output, and maintaining the current output of photovoltaics. If the cloud is online when the event is triggered, the edge can also report the event to the cloud, which then optimizes the scheduling scheme based on broader information and then issues control commands. However, in its design, edge nodes can independently perform local optimizations to avoid cloud latency.

[0057] Based on the optimization results, the edge gateway sends control commands to field actuators via the device control service. These commands include adjusting the turbine guide vane opening, sending regulation instructions to the wind turbine converter, and modifying the output settings of the photovoltaic inverter, enabling dynamic adjustment of power distribution. The entire event response process, from detection to completion, is completed within hundreds of milliseconds to one second, significantly faster than traditional centralized dispatching. Similarly, for "load surge" events, the edge coordinates with various power sources to increase output: if wind turbines and photovoltaics have capacity, their output is immediately increased; otherwise, hydropower output is rapidly increased, or even standby turbine units are activated to respond to the load. Simultaneously, edge nodes can automatically request the activation of local energy storage devices to provide discharge support when necessary. Through this event-driven collaborative control strategy, the system can maintain supply and demand balance and grid parameter stability under various operating conditions.

[0058] While the edge can autonomously perform real-time control, this application still fully leverages cloud computing capabilities for long-term optimization and global coordination. The cloud platform continuously collects operational data uploaded from the edge, including energy output, load status, and event trigger records, and performs big data analysis and model training. For example, the cloud uses this data to train and update predictive models, improving the accuracy of predicting short-term fluctuations in wind power and load; or it uses reinforcement learning to train more sophisticated scheduling strategies offline in a digital simulation environment. Validated new models and strategies are deployed to the edge, enabling the edge control strategy to continuously evolve. Furthermore, the cloud platform can serve as a central dispatch and command hub, playing a role in coordinating larger areas or multiple microgrids: for example, when multiple regional energy systems interact, the cloud coordinates the power support and surplus / deficit adjustments of each edge, thereby achieving global optimization.

[0059] In this embodiment, each edge gateway is equipped with both fiber optic Ethernet and 5G uplinks, as well as a wireless mesh network for lateral communication between gateways. By default, the Ethernet uplink serves as the primary link, automatically switching to the 5G link as the primary link when latency or packet loss exceeds a threshold. Critical data is simultaneously transmitted in parallel through both links to prevent loss. Between gateways, a self-organizing wireless link ensures uninterrupted communication within the cluster even if connection to the cloud is lost. For security, both edge-to-cloud and edge-to-edge communications are transmitted via VPN encrypted tunnels. Gateways and devices have unique digital certificates for two-way authentication, preventing unauthorized devices from accessing the control network. Each control command includes an integrity checksum, which the receiver verifies before execution to prevent tampering. These security measures ensure the reliability and anti-interference capabilities of the method in power applications.

[0060] In summary, this application demonstrates its feasibility and superiority through specific implementation methods. The dynamic scheduling and self-healing method for edge computing resources based on containerized microservices achieves high availability, autonomous collaboration, and intelligent optimization control of edge computing systems in multi-node, multi-energy environments, and can be widely applied in fields such as smart power distribution networks, distributed energy management, and industrial automation.

[0061] Example 3 Based on the same inventive concept, this application also provides an edge computing resource scheduling and self-healing device based on containerized microservices, such as... Figure 3 ,include: The data acquisition unit is used to acquire operational data in multi-energy scenarios through heterogeneous terminals; The microservice fault unit is used to enable the edge computing gateway, based on a containerized microservice architecture, to achieve self-healing of microservices by dynamically scheduling the operation of microservices when a microservice inside the edge computing gateway fails. The gateway failure unit is used to select another healthy gateway to assume the responsibilities of the failed edge computing gateway when one of the edge computing gateways in the edge computing gateway cluster formed through local area network negotiation fails. The event fault unit is used to execute control using an event-driven scheduling strategy when the edge computing gateway detects that the running data has reached the trigger condition. It calculates a new power allocation scheme for each energy unit through edge-cloud collaboration and issues adjustment instructions to the corresponding energy equipment according to the power allocation scheme. Among them, the edge computing gateway based on the containerized microservice architecture encapsulates the microservices of data acquisition, protocol conversion, edge analysis, device control and communication management into independent containers within the edge computing gateway.

[0062] Optionally, the edge computing gateway based on a containerized microservice architecture achieves microservice fault self-healing by dynamically scheduling the execution of microservices through a container orchestration system, including: Edge computing gateways based on containerized microservice architectures deploy and schedule microservices through a container orchestration system. They dynamically adjust the operation of each container based on resource utilization and policies, and set health monitoring and automatic restart policies for critical microservices to achieve automatic fault recovery.

[0063] Optionally, the selection of another health gateway by the collaborative management service to assume the responsibilities of the fault edge computing gateway includes: The collaborative management service selects a standby edge computing gateway node in the edge computing gateway cluster to automatically load the containerized microservice instance of the failed node, establish a communication link with the heterogeneous terminal connected to the failed edge computing gateway, and continue execution from the most recently synchronized state data.

[0064] Optionally, the execution control using an event-driven scheduling strategy, which calculates a new power allocation scheme for each energy unit through edge-cloud collaboration based on a multi-objective optimization strategy, and issues adjustment commands from the edge computing gateway to the corresponding energy devices according to the power allocation scheme, includes: The edge computing gateway cluster invokes pre-deployed optimization algorithms or control strategies in the cloud to perform execution control in response to detected events; By collecting operational data, energy output status, and event trigger records uploaded from the edge via the cloud, big data analysis and model training are performed to obtain artificial intelligence models for resource scheduling or equipment control. These models can quickly calculate new output allocation schemes for each energy unit and then distribute effective artificial intelligence models for resource scheduling or equipment control to the edge deployment. According to the power distribution plan, the edge computing gateway issues adjustment commands to the corresponding energy devices.

[0065] Optionally, the step of the edge computing gateway detecting that the running data has reached the trigger condition includes: The wind speed rises above the set threshold within a unit of time and the wind turbine is not currently operating at full capacity; the local load increases above the set standard value in a short period of time and the voltage shows a downward trend; the photovoltaic output power drops by more than a certain amount in a short period of time. The trigger condition is met when the edge computing gateway detects that at least one of the above conditions is met.

[0066] Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for edge computing resource scheduling and self-healing based on containerized microservices, characterized in that, include: Obtain operational data in multi-energy scenarios through heterogeneous terminals; When a microservice inside the edge computing gateway fails, the edge computing gateway based on the containerized microservice architecture can achieve self-healing of the microservice by dynamically scheduling the operation of the microservice. When one of the edge computing gateways in an edge computing gateway cluster formed through local area network negotiation fails, another healthy gateway is selected to take over the responsibilities of the failed edge computing gateway. When the edge computing gateway detects that the running data has reached the trigger condition, it uses an event-driven scheduling strategy to execute control. Through edge-cloud collaboration, it calculates a new power allocation scheme for each energy unit and issues adjustment commands to the corresponding energy devices according to the power allocation scheme. Among them, the edge computing gateway based on the containerized microservice architecture encapsulates the microservices of data acquisition, protocol conversion, edge analysis, device control and communication management into independent containers within the edge computing gateway.

2. A method as claimed in claim 1, characterized in that The edge computing gateway based on a containerized microservice architecture achieves self-healing of microservices by dynamically scheduling the execution of microservices through a container orchestration system, including: Edge computing gateways based on containerized microservice architectures deploy and schedule microservices through a container orchestration system. They dynamically adjust the operation of each container based on resource utilization and policies, and set health monitoring and automatic restart policies for critical microservices to achieve automatic fault recovery.

3. A method as claimed in claim 1, characterized in that The selection of another health gateway by the collaborative management service to assume the responsibilities of the fault edge computing gateway includes: The collaborative management service selects a standby edge computing gateway node in the edge computing gateway cluster to automatically load the containerized microservice instance of the failed node, establish a communication link with the heterogeneous terminal connected to the failed edge computing gateway, and continue execution from the most recently synchronized state data.

4. A method as claimed in claim 1, characterized in that The control is executed using an event-driven scheduling strategy. Through edge-cloud collaboration and a multi-objective optimization strategy, a new power allocation scheme for each energy unit is calculated. According to this power allocation scheme, the edge computing gateway issues adjustment commands to the corresponding energy devices, including: The edge computing gateway cluster invokes pre-deployed optimization algorithms or control strategies in the cloud to perform execution control in response to detected events; By collecting operational data, energy output status, and event trigger records uploaded from the edge via the cloud, big data analysis and model training are performed to obtain artificial intelligence models for resource scheduling or equipment control. These models can quickly calculate new output allocation schemes for each energy unit and then distribute effective artificial intelligence models for resource scheduling or equipment control to the edge deployment. According to the power distribution plan, the edge computing gateway issues adjustment commands to the corresponding energy devices.

5. A method as claimed in claim 1, characterized in that The step of the edge computing gateway detecting that the running data has reached the trigger condition includes: The wind speed rises above the set threshold within a unit of time and the wind turbine is not currently operating at full capacity; the local load increases above the set standard value in a short period of time and the voltage shows a downward trend; the photovoltaic output power drops by more than a certain amount in a short period of time. The trigger condition is met when the edge computing gateway detects that at least one of the above conditions is met.

6. An edge computing resource scheduling and self-healing device based on containerized microservices, characterized in that, include: The data acquisition unit is used to acquire operational data in multi-energy scenarios through heterogeneous terminals; The microservice fault unit is used to enable the edge computing gateway, based on a containerized microservice architecture, to achieve self-healing of microservices by dynamically scheduling the operation of microservices when a microservice inside the edge computing gateway fails. The gateway failure unit is used to select another healthy gateway to assume the responsibilities of the failed edge computing gateway when one of the edge computing gateways in the edge computing gateway cluster formed through local area network negotiation fails. The event fault unit is used to execute control using an event-driven scheduling strategy when the edge computing gateway detects that the running data has reached the trigger condition. It calculates a new power allocation scheme for each energy unit through edge-cloud collaboration and issues adjustment instructions to the corresponding energy equipment according to the power allocation scheme. Among them, the edge computing gateway based on the containerized microservice architecture encapsulates the microservices of data acquisition, protocol conversion, edge analysis, device control and communication management into independent containers within the edge computing gateway.

7. A device as claimed in claim 6, characterised in that The edge computing gateway based on a containerized microservice architecture achieves self-healing of microservices by dynamically scheduling the execution of microservices through a container orchestration system, including: Edge computing gateways based on containerized microservice architectures deploy and schedule microservices through a container orchestration system. They dynamically adjust the operation of each container based on resource utilization and policies, and set health monitoring and automatic restart policies for critical microservices to achieve automatic fault recovery.

8. A device as claimed in claim 6, characterised in that The selection of another health gateway by the collaborative management service to assume the responsibilities of the fault edge computing gateway includes: The collaborative management service selects a standby edge computing gateway node in the edge computing gateway cluster to automatically load the containerized microservice instance of the failed node, establish a communication link with the heterogeneous terminal connected to the failed edge computing gateway, and continue execution from the most recently synchronized state data.

9. The apparatus of claim 6, wherein, The control is executed using an event-driven scheduling strategy. Through edge-cloud collaboration and a multi-objective optimization strategy, a new power allocation scheme for each energy unit is calculated. According to this power allocation scheme, the edge computing gateway issues adjustment commands to the corresponding energy devices, including: The edge computing gateway cluster invokes pre-deployed optimization algorithms or control strategies in the cloud to perform execution control in response to detected events; By collecting operational data, energy output status, and event trigger records uploaded from the edge via the cloud, big data analysis and model training are performed to obtain artificial intelligence models for resource scheduling or equipment control. These models can quickly calculate new output allocation schemes for each energy unit and then distribute effective artificial intelligence models for resource scheduling or equipment control to the edge deployment. According to the power distribution plan, the edge computing gateway issues adjustment commands to the corresponding energy devices.

10. The apparatus as described in claim 6, characterized in that, The step of the edge computing gateway detecting that the running data has reached the trigger condition includes: The wind speed rises above the set threshold within a unit of time and the wind turbine is not currently operating at full capacity; the local load increases above the set standard value in a short period of time and the voltage shows a downward trend; the photovoltaic output power drops by more than a certain amount in a short period of time. The trigger condition is met when the edge computing gateway detects that at least one of the above conditions is met.

Images