Cloud-edge collaborative heavy gas turbine health intelligent management system and management method

Abstract

The application discloses a cloud-edge cooperative heavy gas turbine health intelligent management system and a management method, an edge gateway, an edge computing micro cloud platform and a center cloud platform; the edge gateway is used for collecting operation data of the heavy gas turbine by using a protocol, and transmitting the operation data to the edge computing micro cloud platform in a target form; the edge computing micro cloud platform is used for performing data processing on the operation data through at least one first micro service, and transmitting obtained key parameters to the center cloud platform; and the center cloud platform is used for performing fault diagnosis through at least one second micro service based on the key parameters to obtain target early warning information, and transmitting the target early warning information to the edge computing micro cloud platform. According to actual application service requirements, the application can flexibly customize and dynamically configure resource configuration of the edge cloud, on the basis of realizing optimized configuration and efficient utilization of computing power and network resources, accurately adapts diversified health management requirements, and improves service quality of the system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent management, in particular to a cloud-edge collaborative heavy gas turbine health intelligent management system and management method. BACKGROUND

[0002] As a core equipment in the energy field, the heavy gas turbine has a complex structure, variable operating conditions and various failure modes. The traditional operation and maintenance mode highly depends on manual experience, has limited intelligence level, and is difficult to effectively identify the complex coupling relationship between failures, which easily leads to unplanned shutdown and significantly affects the safety and economy of energy supply. Therefore, to improve the reliability of equipment and reduce maintenance costs, a health management digital system has gradually become an important technical means.

[0003] In related technologies, the health management digital system generally relies on an industrial internet technology system, integrates cloud computing, big data, artificial intelligence and visualization tools, realizes real-time collection, storage and analysis of operation data, and supports equipment performance monitoring, degradation prediction and early warning to provide decision basis for predictive maintenance. However, the system mostly adopts an industrial internet architecture with relatively fixed function division, and the service boundary, data flow and business flow of the edge layer and the cloud platform lack flexibility, which makes it difficult for the system to dynamically adjust resource allocation according to actual needs, reduces flexibility, and reduces service quality. SUMMARY

[0004] The present application provides a cloud-edge collaborative heavy gas turbine health intelligent management system and management method to solve the technical problems of poor resource allocation flexibility and reduced service quality in the prior art.

[0005] Therefore, the present application provides a cloud-edge collaborative heavy gas turbine health intelligent management system, which can break the fixed division mode of edge cloud functions based on at least one first microservice in the edge computing micro cloud platform and at least one second microservice in the center cloud platform, so that the system can flexibly customize and dynamically configure the resource allocation of the edge cloud according to the actual application service demand, accurately adapt to the diversified equipment health management demand in different scenarios on the basis of realizing the optimized allocation and efficient utilization of computing power and network resources, and improve the service quality of the system.

[0006] Another object of the present application is to provide a cloud-edge collaborative heavy gas turbine health intelligent management method.

[0007] To achieve the above object, the present application provides a cloud-edge collaborative heavy gas turbine health intelligent management system, which comprises an edge gateway, an edge computing micro cloud platform and a center cloud platform.

[0008] The edge gateway is configured to collect operation data of the heavy-duty gas turbine by using a protocol and transmit the operation data in a target form to the edge computing micro cloud platform.

[0009] The edge computing micro cloud platform is configured to perform data processing on the operation data by using at least one first micro service and transmit the obtained key parameters to the central cloud platform.

[0010] The central cloud platform is configured to perform fault diagnosis by using at least one second micro service based on the key parameters to obtain target early warning information and transmit the target early warning information to the edge computing micro cloud platform.

[0011] The cloud-edge collaborative heavy-duty gas turbine health intelligent management system can further have the following additional technical features.

[0012] In the embodiment of the present application, the at least one first micro service includes a data preprocessing micro service, a feature extraction micro service and a data twin micro service.

[0013] The data preprocessing micro service is configured to perform preprocessing on the operation data to obtain preprocessed data.

[0014] The feature extraction micro service is configured to perform feature extraction on the preprocessed data to obtain a feature vector.

[0015] The data twin micro service is configured to perform calculation on the feature vector by using a digital twin model to obtain twin data and determine residual data based on the twin data as key parameters.

[0016] In the embodiment of the present application, the target early warning information includes first early warning information and / or second early warning information; and the at least one second micro service includes a trend prediction micro service, a monitoring and early warning micro service and a fault diagnosis micro service.

[0017] The trend prediction micro service is configured to obtain prediction data of the heavy-duty gas turbine by using a prediction algorithm based on historical operation data and the key parameters.

[0018] The monitoring and early warning micro service is configured to perform real-time monitoring on the prediction data to obtain first early warning information.

[0019] The fault diagnosis micro service is configured to perform comprehensive analysis on the prediction data of the heavy-duty gas turbine to obtain second early warning information.

[0020] In the embodiment of the present application, the system further includes a micro service migration platform,

[0021] The microservice migration platform is configured to, in response to a first target edge computing micro cloud platform triggering a first migration event, migrate at least one first microservice in the first target edge computing micro cloud platform to the central cloud platform, or, in response to the central cloud platform triggering a second migration event, migrate at least one second microservice in the central cloud platform to a second target edge computing micro cloud platform.

[0022] In the embodiment of the present application, the system further comprises a load balancing management platform;

[0023] The load balancing management platform is configured to dynamically manage container resources of microservices based on state space characteristics of a plurality of nodes corresponding to the same microservice through deep reinforcement learning.

[0024] In the embodiment of the present application, the system further comprises an application development platform;

[0025] The application development platform is configured to manage development of the edge computing micro cloud platform and the central cloud platform.

[0026] In the embodiment of the present application, the application development platform comprises a general configuration module, a deployment module, a computing management module and a differential configuration module.

[0027] The general configuration module is configured to obtain general configuration data.

[0028] The deployment module is configured to store images and / or files corresponding to the general configuration data to an image warehouse, an object storage, a configuration center and a model library respectively, and distribute the images and / or files in the image warehouse, the object storage, the configuration center and the model library to the central cloud platform and the edge computing micro cloud platform.

[0029] The computing management module is configured to obtain an algorithm module and generate new algorithm data based on the algorithm module.

[0030] The differential configuration module is configured to manage differential versions of the edge computing micro cloud platform.

[0031] To achieve the above purpose, another aspect of the present application provides a cloud-edge collaborative heavy gas turbine health intelligent management method, which comprises:

[0032] Collecting operation data of the heavy gas turbine by using a protocol;

[0033] Processing the operation data by at least one first microservice in the edge computing micro cloud platform to obtain key parameters;

[0034] Based on the key parameters, target early warning information is obtained by fault diagnosis through at least one second microservice in the central cloud platform.

[0035] In the embodiment of the application, the at least one first microservice includes a data preprocessing microservice, a feature extraction microservice and a data twin microservice; the key parameters are obtained by data processing on the running data through at least one first microservice in the edge computing micro cloud platform, including:

[0036] The preprocessed data is obtained by preprocessing the running data through the data preprocessing microservice;

[0037] The feature vector is obtained by feature extraction on the preprocessed data through the feature extraction microservice;

[0038] The twin data is obtained by calculation on the feature vector through the digital twin model in the data twin microservice, and the residual data obtained based on the twin data is determined as the key parameters.

[0039] In the embodiment of the application, the at least one second microservice includes a trend prediction microservice, a monitoring and early warning microservice and a fault diagnosis microservice; the target early warning information is obtained by fault diagnosis through at least one second microservice in the central cloud platform based on the key parameters, including:

[0040] The prediction data of the heavy gas turbine is obtained by a prediction algorithm based on the historical running data and the key parameters through the trend prediction microservice;

[0041] The first early warning information is obtained by real-time monitoring on the prediction data through the monitoring and early warning microservice;

[0042] The second early warning information is obtained by comprehensive analysis on the prediction data of multiple heavy gas turbines through the fault diagnosis microservice.

[0043] In the embodiment of the application, the method further includes:

[0044] In response to a first migration event triggered by a first target edge computing micro cloud platform, at least one first microservice in the first target edge computing micro cloud platform is migrated to the central cloud platform, or in response to a second migration event triggered by the central cloud platform, at least one second microservice in the central cloud platform is migrated to a second target edge computing micro cloud platform.

[0045] In the embodiment of the application, the method further includes:

[0046] Obtaining state space features corresponding to multiple nodes, wherein the nodes are container clusters corresponding to the same microservice;

[0047] Based on the state space features, the container resources of the microservice are dynamically managed through deep reinforcement learning.

[0048] In the embodiments of the present application, the method further comprises:

[0049] Obtaining general configuration data;

[0050] Storing the images and / or files corresponding to the general configuration data in the image repository, object storage, configuration center and model library, respectively;

[0051] In response to receiving the instruction issued by the center cloud platform, the images and / or files in the image repository, the object storage, the configuration center and the model library are issued to the center cloud platform and the edge computing micro cloud platform.

[0052] In the embodiments of the present application, the method further comprises:

[0053] Obtaining an algorithm module uploaded by a user, and generating new algorithm data based on the algorithm module;

[0054] Generating general configuration data based on the new algorithm data.

[0055] In the embodiments of the present application, the method further comprises:

[0056] Determining a target version file of the edge computing micro cloud platform;

[0057] In response to receiving an update instruction from the center cloud platform, issuing the target version file to the edge computing micro cloud platform.

[0058] The cloud-edge collaborative heavy gas turbine health intelligent management system and method of the embodiments of the present application comprise an edge gateway, an edge computing micro cloud platform and a center cloud platform. The edge gateway is used to collect operation data of the heavy gas turbine by using a protocol, and transmit the operation data to the edge computing micro cloud platform in a target form. The edge computing micro cloud platform is used to process the operation data by at least one first microservice, and transmit the obtained key parameters to the center cloud platform. The center cloud platform is used to perform fault diagnosis by at least one second microservice based on the key parameters to obtain target early warning information, and transmit the target early warning information to the edge computing micro cloud platform. Thus, based on at least one first microservice in the edge computing micro cloud platform and at least one second microservice in the center cloud platform, the present application breaks the fixed division mode of edge cloud functions, so that the system can flexibly customize and dynamically configure the resource configuration of the edge cloud according to the actual application service demand, on the basis of realizing the optimized configuration and efficient utilization of computing power and network resources, accurately adapts to the diversified equipment health management needs in different scenarios, and improves the service quality of the system.

[0059] Additional aspects and advantages of the present application will be better understood from the following descriptions, which presents a detailed description of some embodiments of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0060] The above and / or additional aspects and advantages of the present application will become more apparent by describing in detail some embodiments thereof with reference to the attached drawings in which:

[0061] Figure 1 is a structural schematic diagram of a cloud-edge collaborative heavy gas turbine health intelligent management system according to an embodiment of the present application;

[0062] Figure 2 is an interaction schematic diagram of a cloud-edge collaborative heavy gas turbine health intelligent management system according to an embodiment of the present application;

[0063] Figure 3 is a schematic diagram of microservice migration according to an embodiment of the present application;

[0064] Figure 4 is a flow schematic diagram of acquiring general configuration data according to an embodiment of the present application;

[0065] Figure 5 is a flow schematic diagram of health intelligent management system deployment according to an embodiment of the present application;

[0066] Figure 6 is a structural schematic diagram of health intelligent management system deployment according to an embodiment of the present application;

[0067] Figure 7 is a flow schematic diagram of a cloud-edge collaborative heavy gas turbine health intelligent management method according to an embodiment of the present application. DETAILED DESCRIPTION

[0068] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0069] In order to enable persons skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor should fall within the protection scope of the present application.

[0070] A heavy gas turbine health intelligent management system and a management method thereof are provided based on a cloud-edge collaborative heavy gas turbine control and protection system verification platform according to an embodiment of the present application.

[0071] Figure 1 is a structural schematic diagram of the cloud-edge collaborative heavy gas turbine health intelligent management system according to an embodiment of the present application, Figure 2 is an interaction schematic diagram of the cloud-edge collaborative heavy gas turbine health intelligent management system according to an embodiment of the present application.

[0072] As shown in Figure 1 and Figure 2 , the health intelligent management system comprises an edge gateway 1, an edge computing micro cloud platform 2 and a central cloud platform 3.

[0073] The edge gateway 1 is configured to collect the operation data of the heavy gas turbine by using a protocol and transmit the operation data to the edge computing micro cloud platform in a target form.

[0074] The edge computing micro cloud platform 2 is configured to perform data processing on the operation data by using at least one first micro service and transmit the obtained key parameters to the central cloud platform.

[0075] The central cloud platform 3 is configured to perform fault diagnosis by using at least one second micro service based on the key parameters to obtain target early warning information and transmit the target early warning information to the edge computing micro cloud platform.

[0076] In the embodiment of the present application, the edge gateway 1 and the edge computing micro cloud platform 2 together form an edge computing system, which is matched with a single heavy gas turbine to realize the collection and monitoring of the operation data of the heavy gas turbine. The central cloud platform serves the application scenarios of multiple gas turbines (gas turbine cluster) across regions, and provides the functions of unified health state intelligent management of the gas turbine cluster, diversified gas turbine health management application customization and application expansion services.

[0077] Specifically, in the embodiment of the present application, the edge gateway can realize the docking of the heavy gas turbine control system and the special measurement system by using an industrial protocol to collect the operation data of the heavy gas turbine in real time. The industrial protocol can be any one of MODBUS, OPC-UA, Profibus, CAN, HART and Ethernet. In addition, in the embodiment of the present application, the edge gateway can also expand the industrial fieldbus protocol through plug-in protocol management.

[0078] In the embodiment of the present application, the edge gateway can convert the collected operation data into a unified standard communication and data protocol (such as Ethernet / MQTT) and transmit it to the edge computing micro cloud platform. In the embodiment of the present application, the operation data can include raw or calculated (such as filtered) state data such as power, speed, temperature, pressure, and valve position, as well as control data (such as control instructions, intermediate calculation values, and feedback data, etc.).

[0079] Further, in the embodiment of the present application, the at least one first microservice can include a data preprocessing microservice, a feature extraction microservice, and a data twin microservice.

[0080] In the embodiment of the present application, the data preprocessing microservice is used to preprocess the operation data to obtain preprocessed data; the feature extraction microservice is used to extract features from the preprocessed data to obtain a feature vector; and the data twin microservice is used to calculate the feature vector through a digital twin model to obtain twin data, and determine residual data based on the twin data as a key parameter.

[0081] In the embodiment of the present application, the data preprocessing microservice can clean, denoise, format standardize, and remove outliers from the operation data, and obtain preprocessed data adapted to different operating stages of the device through an adaptive algorithm.

[0082] In the embodiment of the present application, the feature extraction microservice can target extract feature parameters strongly related to key working condition identification and feature state point representation of the gas turbine from the preprocessed data, and construct a low-dimensional and high-discriminability feature vector, so as to ensure the accuracy of subsequent data twin microservice and trend prediction microservice.

[0083] In the embodiment of the present application, the data twin microservice can use a mechanism model, a data-driven model, or a mechanism and data-driven fusion model to construct a high-precision gas turbine digital twin, so as to output key parameters or performance indicators in real time. In the embodiment of the present application, the data twin microservice can also calculate the feature vector through a digital twin model to obtain twin data, and determine residual data based on the twin data as a key parameter.

[0084] In the embodiment of the present application, the corresponding residual data can be obtained by comparing the twin data with the operation data of the gas turbine, and the residual data is determined as a key parameter.

[0085] In the embodiment of the present application, the at least one second microservice can include a trend prediction microservice, a monitoring and early warning microservice, and a fault diagnosis microservice.

[0086] In the embodiment of the present application, the trend prediction microservice is used to obtain the prediction data of the heavy-duty gas turbine based on the historical operation data and the key parameters through a prediction algorithm; the monitoring and early warning microservice is used to monitor the prediction data in real time to obtain first early warning information; and the fault diagnosis microservice is used to comprehensively analyze the prediction data of multiple heavy-duty gas turbines to obtain second early warning information.

[0087] In the embodiment of the present application, the trend prediction microservice can obtain the prediction data of the heavy-duty gas turbine through Kalman filtering based on the historical operation data and the key parameters, so as to realize accurate prediction of the equipment trend. The method for obtaining the prediction data of the heavy-duty gas turbine is not limited in the embodiment of the present application. In the embodiment of the present application, the prediction data can be the prediction data corresponding to the key parameters.

[0088] In the embodiment of the present application, the monitoring and early warning microservice monitors the prediction data in real time, and matches the early warning conditions through the monitoring and early warning container to obtain the first early warning information of the health state degradation of the important equipment, component, system and whole machine of the heavy-duty gas turbine.

[0089] In the embodiment of the present application, the fault diagnosis microservice comprehensively analyzes the prediction data of the heavy-duty gas turbine through the fault diagnosis container to obtain the second early warning information. In one embodiment of the present application, the fault diagnosis microservice can comprehensively analyze the prediction data of a single heavy-duty gas turbine through the fault diagnosis container to obtain the second early warning information. In another embodiment of the present application, the fault diagnosis microservice can also comprehensively analyze the prediction data of multiple heavy-duty gas turbines of the same type through the fault diagnosis container to obtain the second early warning information.

[0090] In the embodiment of the present application, the fault diagnosis container can comprehensively analyze through fuzzy rules of fuzzy mathematics or pattern recognition methods under multiple working conditions. The method is not limited in the embodiment of the present application.

[0091] In the embodiment of the present application, after the first early warning information and the second early warning information are obtained through the monitoring and early warning microservice and the fault diagnosis microservice, the first early warning information, the second early warning information and the corresponding maintenance suggestions can be pushed to the corresponding edge computing micro cloud platform for reference for the maintenance of the corresponding heavy-duty gas turbine.

[0092] In the embodiment of the present application, the health intelligent management system can use a unified message middleware as a data flow transmission channel, and standardize the input and output formats to ensure reliable flow of data between different services, so as to ensure efficient cooperation between microservices.

[0093] In addition, in the embodiments of the present application, the first microservices and the second microservices can be independently deployed and flexibly expanded, and can be developed by using different technology stacks and programming languages, so that the technical flexibility advantage of the microservice architecture is fully released.

[0094] The cloud-edge collaborative heavy gas turbine health intelligent management system provided by the present application comprises an edge gateway, an edge computing micro cloud platform and a center cloud platform. The edge gateway is used for collecting operation data of the heavy gas turbine by using a protocol, and transmitting the operation data to the edge computing micro cloud platform in a target form. The edge computing micro cloud platform is used for performing data processing on the operation data by using at least one first microservice, and transmitting the obtained key parameters to the center cloud platform. The center cloud platform is used for performing fault diagnosis to obtain target early warning information by using at least one second microservice based on the key parameters, and transmitting the target early warning information to the edge computing micro cloud platform. Therefore, based on the at least one first microservice in the edge computing micro cloud platform and the at least one second microservice in the center cloud platform, the present application breaks the fixed division mode of the edge cloud function, so that the system can flexibly customize and dynamically configure the resource configuration of the edge cloud according to the actual application service demand, on the basis of realizing the optimized configuration and efficient utilization of the computing power and network resources, accurately adapts to the diversified equipment health management demand in different scenarios, and improves the service quality of the system.

[0095] In the embodiments of the present application, the device hardware resources of the edge computing micro cloud platform are relatively fixed, and after the health management application continuously grows, the computing resources of the edge computing micro cloud platform may be limited. Based on this, when the computing resources of the edge computing micro cloud platform are insufficient, the computing task with low real-time requirement can be migrated to the center cloud platform to utilize the powerful and elastic computing power thereof; when the network bandwidth is insufficient, the task can be calculated nearby on the edge computing micro cloud platform.

[0096] Based on the above description, the health intelligent management system can further comprise a microservice migration platform 4.

[0097] In the embodiment of the present application, when the computing resource (such as CPU) of the edge computing micro cloud platform is less than a preset threshold, the edge computing micro cloud platform can trigger a first migration event, at this time, the edge computing micro cloud platform is a first target edge computing micro cloud. In the embodiment of the present application, when performing container load balancing, the center cloud platform can trigger a second migration event, and at least one second microservice in the center cloud platform is migrated to a second target edge computing micro cloud platform.

[0098] In the embodiment of the present application, in response to the first target edge computing micro cloud platform triggering a migration event, at least one first microservice in the first target edge computing micro cloud platform is migrated to the center cloud platform, that is, at least one of the data preprocessing container, the feature extraction container and the digital twin container in the center cloud platform is configured, so that at least one first microservice can be migrated to the center cloud platform, and the computing load of the edge computing micro cloud platform is reduced. In the embodiment of the present application, the migrated first microservice can be set according to the required computing resource of each first microservice.

[0099] In the embodiment of the present application, in response to the center cloud platform triggering a second migration event, at least one second microservice in the center cloud platform is migrated to a second target edge computing micro cloud platform, that is, at least one second microservice is executed on the second target edge computing micro cloud platform. In the embodiment of the present application, the second target edge computing micro cloud platform can be a computing-capable edge computing micro cloud platform determined in load balancing. In the embodiment of the present application, the migrated second microservice can be set according to the required computing resource of each second microservice.

[0100] Figure 3 A schematic diagram of microservice migration is proposed for the embodiment of the present application. As shown in Figure 3 The data preprocessing microservice, feature extraction microservice and data twin microservice in the edge computing micro cloud platform can be migrated to the center cloud platform, so that the data preprocessing microservice can obtain preprocessed data through the center cloud platform; the feature extraction microservice obtains the feature vector by performing feature extraction on the preprocessed data; the data twin microservice calculates the twin data by using the digital twin model on the feature vector, and determines the residual data based on the twin data as the key parameter.

[0101] In the embodiment of the present application, in order to realize the automatic and intelligent resource adjustment of the first microservice and the second microservice corresponding containers, the health intelligent management system can flexibly customize and configure the software and services of the edge computing micro cloud platform and the center cloud platform according to actual needs, and then realize the optimal configuration of computing power and network resources, so as to effectively cope with the complex and changeable needs in the field of heavy gas turbine health intelligent management, meet the diversified configuration of health management applications in different scenarios, and improve the adaptation ability of the system to various service modes, data processing methods and business processes.

[0102] In the embodiment of the present application, each first microservice and each second microservice can be instantiated and deployed in an independent container for running. Multiple microservice instances can also be deployed to form a container cluster for running according to the response requirements of the application service. That is, due to the sharing of container cluster resources by various health management applications, in the scenarios of application quantity fluctuation, service mode switching, data processing method change, etc., the computing power resources need to be dynamically allocated through load balancing, so as to ensure that the application meets the timeliness and throughput requirements. In the embodiment of the present application, the health management application can be the health management application corresponding to each component of the heavy gas turbine, that is, each heavy gas turbine can correspond to multiple health management applications.

[0103] Based on the above description, the health intelligent management system can further include a load balancing management platform 5, wherein the load balancing management platform 5 is configured to dynamically manage the container resources of the microservice based on the state space characteristics of multiple nodes, wherein the nodes can be the container cluster corresponding to the same microservice.

[0104] In the embodiment of the present application, the container is the running carrier of the microservice, and the resource (CPU / memory / network IO) usage indicators of the container can be collected by the container runtime. Then, the resource usage indicators of all nodes are summarized, and the container resources are dynamically managed through deep reinforcement learning, so as to realize the expansion and contraction operation and traffic scheduling of the container cluster (i.e. microservice instance), realize the dynamic allocation and scheduling optimization of computing resources, and realize high utilization of resources while ensuring system performance.

[0105] In the embodiment of the present application, the method of dynamically managing the container resources of the microservice based on the state space characteristics of multiple nodes through deep reinforcement learning can include: dynamically managing the container resources of the microservice through a target Actor-Critic (actor-critic) network based on the state space characteristics of multiple nodes.

[0106] In the embodiments of the present application, the target Actor-Critic network comprises a policy network (Actor) and a value network (Critic). μ,σ The policy network outputs a discrete action probability distribution π

[0107] Specifically, in the embodiments of the present application, the state space feature may be:

[0108]

[0109] The node-level resource indicator can be a multi-dimensional state vector. The multi-dimensional state vector corresponding to the node-level resource indicator can be obtained by the resource usage of each node. For example, assuming that each node has N vectors (i.e., N micro-service instances), the multi-dimensional state vector corresponding to the node-level resource indicator can be:

[0110]

[0111] The node-level resource indicator can be a multi-dimensional state vector. The multi-dimensional state vector corresponding to the node-level resource indicator can be obtained by the resource usage of each node. For example, assuming that each node has N vectors (i.e., N micro-service instances), the multi-dimensional state vector corresponding to the node-level resource indicator can be:

[0112]

[0113] Further, the cluster-level load statistical indicator can be composed of a global load variance and a task queue average length. Specifically, the cluster-level load statistical indicator can be:

[0114]

[0115] The global load variance is used to measure the resource usage balance among nodes. The global load variance can be:

[0116]

[0117] The task queue average length is the length of the pending queue. The task queue average length can be:

[0118]

[0119] Further, the task characteristic setting state value can be:

[0120] ​​​​​​

[0121] where task type is encoded as a one-hot vector (e.g. compute-intensive = [1, 0], I / O-intensive = [0, 1]); task priority is normalized to [0, 1] with high-priority tasks close to 1.

[0122] In embodiments of the present application, the output action can be where target node is the target node number of the resource to be adjusted, e.g. , and is the adjustment amount for the target node resource quota, e.g. CPU share , memory limit represents the increase / decrease ratio of the target node CPU / memory quota, e.g. ΔCPU = 0.1 represents an increase of 10% CPU quota.

[0123] Further, in embodiments of the present application, the value network estimates the state value .

[0124] Further, in embodiments of the present application, the value network can determine the state value through a reward function where the reward function can be:

[0125]

[0126] where task delay is the time consumption from task submission to completion, α, is a hyperparameter identifying the weight, which needs to be tuned according to the scene.

[0127] where in embodiments of the present application, the resource balance degree is:

[0128]

[0129] In embodiments of the present application, the overload penalty is:

[0130]

[0131] where , is an overload threshold, e.g. 0.9.

[0132] And, in the embodiment of the application, the target Actor-Critic network is obtained by training the Actor-Critic network. In the embodiment of the application, the network parameters in the Actor-Critic network can be updated by a target function TD Error (time difference error) during the training of the Actor-Critic network, until the Actor-Critic network converges, and the target Actor-Critic network is obtained.

[0133] In the embodiment of the application, the target function TD Error is:

[0134]

[0135] Wherein, is a discount factor, is .

[0136] In the embodiment of the application, the target function TD Error is the difference between the predicted value of the Critic network and (the actual reward + the predicted value of the next state), and this error is used to update the Critic (to make its prediction more accurate) and the Actor (to optimize its strategy) at the same time.

[0137] And, in the embodiment of the application, after obtaining the target Actor-Critic network through the above steps, the intelligent allocation and dynamic scaling of container cluster resources can be obtained through the target Actor-Critic network, and high-response and high-reliability load balancing support can be provided for the heavy gas turbine health intelligent management system.

[0138] Further, in the embodiment of the application, the load balancing management platform can also dynamically manage the container resources of the microservice through the scaling strategy. Specifically, in the embodiment of the application, the scaling condition is that if the CPU usage of node j exceeds the threshold value θ for T seconds high , the following scaling strategy is used, wherein the scaling strategy is:

[0139]

[0140] Wherein, in the embodiment of the application, the scaling condition is that if the CPU usage of node j is lower than the threshold value θ for T seconds low , the following way is used to scale down.

[0141]

[0142] And, in the embodiment of the application, based on the above capacity expansion and contraction strategy, the following scenarios can be realized: in the gas turbine health management application, tasks with high real-time requirements are defined as high priority to ensure their allocation to low-delay nodes; when the health management application computing demand at the edge increases, resources are automatically expanded to meet the computing power demand; when a node fails, tasks are quickly migrated to a backup node based on a deep reinforcement learning strategy to ensure service continuity.

[0143] In the embodiment of the application, the above-mentioned health intelligent management system can further include an application development platform 6, wherein the application development platform is used to manage the development of the edge computing micro cloud platform and the central cloud platform.

[0144] In the embodiment of the application, the above-mentioned application development platform can include a general configuration module, a deployment module, a computing management module, and a differentiated configuration module.

[0145] In the embodiment of the application, the health management platform in the related art only supports customized application development in terms of application service expansion functions, and the expansion application development process is not standardized and the standards are not unified, resulting in low overall development efficiency. The health intelligent management system does not reserve a general application expansion interface and framework, and developers need to rebuild the underlying logic for each specific device health management service requirement, resulting in low software reuse rate and a lot of repetitive development work. In addition, there is no uniform expansion application development specification in the industry, and different developers have different development habits and technology selection, resulting in uneven code quality. In addition, due to inconsistent data formats and calling interfaces, expansion applications are difficult to work together, integration is difficult, and the functionality of the health intelligent management system is hindered. Therefore, the above-mentioned general configuration module can be used to obtain general configuration data to complete application development in a determined process and configuration, thereby improving the application development efficiency.

[0146] Specifically, in the embodiment of the application, the above-mentioned general configuration module can be used to obtain general configuration data. In the embodiment of the application, Figure 4 A flowchart for obtaining general configuration data is provided in the embodiment of the application. As Figure 4 shown, the method for obtaining general configuration data can include the following steps:

[0147] Step 401, obtaining data input of the health management application.

[0148] In the embodiment of the application, the data input of the health management application can be defined. In the embodiment of the application, the health management application can be connected with the data that needs to be actually collected when the health management application is issued.

[0149] Step 402, obtaining the first configuration of the data preprocessing container corresponding to the data preprocessing microservice.

[0150] In the embodiments of the present application, the first configuration can include a data preprocessing algorithm and a first rule. In the embodiments of the present application, the data preprocessing algorithm and the first rule can be defined.

[0151] In step 403, a second configuration of a feature extraction container corresponding to a feature extraction microservice is obtained.

[0152] In the embodiments of the present application, the second configuration can include a second rule of feature extraction. In the embodiments of the present application, the second rule of feature extraction can be defined.

[0153] In step 404, a third configuration of a data twin container corresponding to a data twin microservice is obtained.

[0154] In the embodiments of the present application, the third configuration can include a configuration of a target data twin model.

[0155] In the embodiments of the present application, the method for determining the configuration of the target digital twin model can include the following steps:

[0156] In step 4041, a preset digital twin model is selected.

[0157] In step 4042, the preset digital twin model is trained based on a running data set of the object to obtain a trained digital twin model.

[0158] In step 4043, the trained digital twin model is tested by using a test data set to obtain model configuration parameters of the target data twin model.

[0159] In the embodiments of the present application, after the model configuration parameters of the target data twin model are determined through the above steps, the target data twin model can be deployed to a health management application, and the display of the integrated front-end page configuration data and charts can be combined.

[0160] In step 405, a fourth configuration of a trend prediction container corresponding to a trend prediction microservice is obtained.

[0161] In the embodiments of the present application, the fourth configuration can include trend prediction parameters, trend prediction output and alarm conditions. In the embodiments of the present application, the fourth configuration can be defined.

[0162] In step 406, output data of the health management application is obtained, and the application is interfaced with an edge computing micro cloud platform when the application is issued.

[0163] In the embodiment of the present application, based on the obtained general configuration data, the application development process can be specified, and the problem of inconsistent application development standards can be solved, so that the application development can be completed with the specified process and configuration, and the application development efficiency is improved.

[0164] Further, in the embodiment of the present application, the deployment module can be used to store the images and / or files corresponding to the general configuration data in the image warehouse, object storage, configuration center and model library respectively, and download the images and / or files in the image warehouse, object storage, configuration center and model library to the central cloud platform and edge computing micro cloud platform. In the embodiment of the present application, Figure 5 A flowchart of deploying a health intelligent management system is provided in the embodiment of the present application, Figure 6 A structural diagram of deploying a health intelligent management system is provided in the embodiment of the present application. As shown in Figure 5 and Figure 6 The method for deploying a health intelligent management system can include the following steps:

[0165] Step 501, store the images and / or files corresponding to the general configuration data in the image warehouse, object storage, configuration center and model library respectively.

[0166] In the embodiment of the present application, the general configuration data and the computing process are split, and the container image obtained after splitting is stored in the image warehouse.

[0167] In addition, in the embodiment of the present application, the configuration file of each image is uploaded to the configuration center.

[0168] Further, in the embodiment of the present application, the model file formed in the general configuration data is stored in the model library.

[0169] In the embodiment of the present application, the dependent environment and installation configuration script required by the above image are stored in the object storage.

[0170] Step 502, in response to receiving the downloading instruction of the central cloud platform, download the images and / or files in the image warehouse, object storage, configuration center and model library to the central cloud platform and edge computing micro cloud platform.

[0171] In the embodiment of the present application, in response to receiving the downloading instruction of the central cloud platform, the central cloud platform and the edge computing micro cloud platform can respectively download the required deployment images and / or files to the central cloud platform and the edge computing micro cloud platform through the application downloading deployment service.

[0172] In addition, in the embodiment of the present application, when the central cloud platform and the edge computing micro cloud platform update the existing application model, the model library can be downloaded to the target central cloud platform and edge computing micro cloud platform.

[0173] Further, in the embodiments of the present application, the tasks of image pulling, configuration file management, dependent environment preparation and model file acquisition can be completed by functions and resources in third-party tools such as Kubernetes.

[0174] In step 503, the central cloud platform and the edge computing micro cloud platform start running based on the images and / or files.

[0175] In the embodiments of the present application, when the images and / or files obtained by the central cloud platform and the edge computing micro cloud platform correspond to new applications, the central cloud platform and the edge computing micro cloud platform can start running; when the images and / or files obtained by the central cloud platform and the edge computing micro cloud platform correspond to updated models, the screenshots are updated according to the information of the new model files, the configurations are updated and the running is restarted.

[0176] In the embodiments of the present application, through the above steps, the whole process integration of development, deployment, running and optimization of the health management application can be realized, and the system adaptability and operation and maintenance efficiency in complex industrial scenarios are significantly improved.

[0177] In addition, in the embodiments of the present application, in order to ensure the automatic configuration and data consistency of the health intelligent management system, the present application can determine a unified data acquisition interface protocol, a data bus topic naming rule and a field name rule, and through the above standardized protocol and rule, the health intelligent management system can automatically configure the data acquisition interface, avoiding manual configuration, thereby improving the automation degree and operation efficiency of the system.

[0178] In the embodiments of the present application, the data acquisition interface protocol is used to ensure the data communication consistency between the edge computing micro cloud platform, the central cloud platform and the health management application, defines the data transmission format, the message structure and the data verification rule, adopts the standardized data format, ensures the scalability and flexibility of the data, and the data field agrees on the field type, data precision and unit specification. Among them, the data transmission can uniformly adopt the message queue, through the agreement of the standardized topic naming rule, the readability of the data management and the maintainability of the system can be improved, the hierarchical and naming specification are followed, the clear structure is ensured, and the management is facilitated.

[0179] Specifically, in the embodiments of the present application, the above naming rule can be organized in the following format: <device type> / <device ID> / <subsystem> / <application name> / <computing flow>; the field name rule defines the naming specification of all fields in the data acquisition system, ensures the consistency of all acquired fields in the system, the field name should be concise, clear and descriptive, so as to facilitate subsequent data processing and management, including naming format, field type prefix, field description and the like.

[0180] Further, in the embodiments of the present application, when the health intelligent management system is deployed, the system automatically generates field names according to the collection requirements of the device. Through the automatically generated field names, the health management application can directly read, parse and process the data, avoiding the trouble of manual configuration.

[0181] In addition, in the embodiments of the present application, during the deployment process, the system automatically generates message bus topic instances according to predefined naming rules without manual configuration. By defining the data collection interface protocol, topic naming rules and field name rules, the platform can automatically complete the following without manual configuration: automatically generating interfaces according to the predefined protocol; automatically generating topics according to the device and data type without manually specifying the topic; automatically generating field names to ensure the consistency of data structure and name; seamless collaboration between the edge computing micro cloud platform and the central cloud platform to realize real-time configuration, data collection and operation of the health management application.

[0182] Further, in the embodiments of the present application, through the collaborative work of the central cloud platform and the edge computing micro cloud platform, the health intelligent management system realizes the integration of development and use. Among them, the application developers can complete the application configuration and script distribution on the central cloud platform, and the edge computing micro cloud platform is responsible for execution and real-time feedback. The whole process runs through the whole process of construction, testing and deployment.

[0183] In addition, in the embodiments of the present application, through cloud service support, the health intelligent management system can realize automatic application deployment, data matching and real-time operation without human intervention, improving management efficiency and system flexibility; by defining data collection interface protocol, topic naming rules and field name rules, an automatic data configuration and management scheme is provided, avoiding the complexity of manual configuration and ensuring the consistency and efficiency of data.

[0184] In the embodiments of the present application, the above-mentioned computing management module can be used to obtain an algorithm module and generate new algorithm data based on the algorithm module.

[0185] In the embodiments of the present application, the above-mentioned algorithm module can be uploaded by a user. In addition, in the embodiments of the present application, the obtained algorithm module and the existing algorithm module can be concatenated and combined to obtain new algorithm data, so as to realize a new computing flow task. Therefore, by configuring the algorithm module, the algorithm flow of the application can be flexibly configured to meet the special health management application and algorithm development requirements.

[0186] In addition, in the embodiment of the present application, the developer can upload the algorithm modules one by one, and in each algorithm module, the corresponding configuration is performed according to the configuration items required by the module, so that the complex and differentiated health management application development can be realized by combining multiple algorithm modules through the addition of algorithm modules.

[0187] Further, in the embodiment of the present application, after the new algorithm data is generated, the general configuration data can also be generated based on the new algorithm data, so that the algorithm can be updated by using the configuration data.

[0188] In the embodiment of the present application, the above-mentioned differentiated configuration module can be used to manage the differentiated versions of the edge computing micro cloud platform. Specifically, in the embodiment of the present application, the method for managing the differentiated versions of the edge computing micro cloud platform can include: determining a target version file of the edge computing micro cloud platform, and in response to receiving an update instruction of the center cloud platform, issuing the target version file to the edge computing micro cloud platform.

[0189] In the embodiment of the present application, the user can store the configuration of each edge computing micro cloud in the health state monitoring service configuration of the center cloud platform.

[0190] In the embodiment of the present application, the target version file of each edge computing micro cloud platform can be determined through the health state monitoring service configuration in the center cloud platform, and the target version file of each edge computing micro cloud platform is configured and issued to the edge computing micro cloud platform, so that the edge computing micro cloud platform completes the deployment and operation of the health management application, realizes the differentiated services of the edge computing micro cloud platform configuration, and does not depend on the local deployment, thereby realizing the flexible configuration of the edge cloud software and service.

[0191] In addition, in the embodiment of the present application, the health state monitoring service configuration can include the configuration of each health management application (such as data collection and storage, and input and output docking of the application), and the data synchronization method of the edge computing micro cloud platform and the center cloud platform. In the center cloud platform, the health state monitoring service configuration corresponding to each edge computing micro cloud platform and the historical version are stored, and a new version can be obtained by modifying one of the historical versions and saving it.

[0192] For example, in the embodiment of the present application, after the health state monitoring service configuration is performed for a gas turbine in the center cloud platform, the interface can be used to control the issuance and online of the health management application on the gas turbine, and the edge computing micro cloud platform and the center cloud platform can pull the corresponding image and resources through object storage, image warehouse, and execute the corresponding script to realize the deployment and effectiveness of the health management application. Based on this, the center cloud platform can quickly, flexibly and conveniently update and configure the services and software of the edge computing micro cloud platform, thereby greatly improving the flexibility of the services.

[0193] Further, in the embodiment of the present application, the edge computing micro cloud platform and the center cloud platform may be in a network inaccessible state, at which time the health management application can be issued offline.

[0194] In the embodiment of the present application, the center cloud platform packs the micro service into an image file, and exports the database storage, message bus and other configuration information together, packs the image file and the configuration information into a compressed file, exports the file in the form of a file, and transmits it to the edge computing micro cloud platform through a physical medium such as a CD or a disk. The edge computing micro cloud platform imports the compressed package, and completes the import of the image file, the configuration and the writing of the environment variable through a shell script, and the start of the container, to complete the issuance of the health management application in the edge computing micro cloud platform. Based on this, the application management function of the edge computing micro cloud platform and the center cloud platform in the network inaccessible state can be effectively solved.

[0195] To achieve the above-mentioned embodiments, Figure 7 A cloud-edge collaborative heavy gas turbine health intelligent management method is proposed for the embodiment of the present application, as shown in the figure, which can include the following steps: Figure 7

[0196] Step 701, collecting the operation data of the heavy gas turbine by using a protocol;

[0197] Step 702, obtaining the key parameters by processing the operation data through at least one first micro service in the edge computing micro cloud platform;

[0198] Step 703, obtaining the target early warning information by diagnosing the fault through at least one second micro service in the center cloud platform based on the key parameters.

[0199] In the embodiment of the present application, the above-mentioned at least one first micro service can include a data preprocessing micro service, a feature extraction micro service and a data twin micro service. In the embodiment of the present application, the method of obtaining the key parameters by processing the operation data through at least one first micro service in the edge computing micro cloud platform can include: preprocessing the operation data through the data preprocessing micro service to obtain preprocessed data; extracting features from the preprocessed data through the feature extraction micro service to obtain a feature vector; calculating the feature vector through a digital twin model in the data twin micro service to obtain twin data, and determining the residual data based on the twin data as the key parameters.

[0200] ​Further, in the embodiments of the present application, the at least one second microservice can include a trend prediction microservice, a monitoring and early warning microservice, and a fault diagnosis microservice. In addition, the method of obtaining target early warning information based on the key parameters through the at least one second microservice in the central cloud platform for fault diagnosis can include: obtaining predicted data of the heavy gas turbine through a trend prediction microservice based on historical operation data and key parameters, and through a prediction algorithm; obtaining first early warning information through real-time monitoring of the predicted data by a monitoring and early warning microservice; and obtaining second early warning information through comprehensive analysis of the predicted data of the plurality of heavy gas turbines by a fault diagnosis microservice.

[0201] Further, in the embodiments of the present application, the method can further include: in response to the first target edge computing micro cloud platform triggering a first migration event, migrating at least one first microservice in the first target edge computing micro cloud platform to the central cloud platform, or in response to the central cloud platform triggering a second migration event, migrating at least one second microservice in the central cloud platform to the second target edge computing micro cloud platform.

[0202] Further, in the embodiments of the present application, the method can further include the following steps:

[0203] Obtaining state space features corresponding to a plurality of nodes, wherein the nodes are container clusters corresponding to the same microservice; and based on the state space features, dynamically managing the container resources of the microservice through deep reinforcement learning.

[0204] Further, in the embodiments of the present application, the method can further include the following steps:

[0205] Obtaining general configuration data;

[0206] Storing images and / or files corresponding to the general configuration data to an image repository, an object storage, a configuration center, and a model library, respectively;

[0207] In response to receiving a downlink instruction of the central cloud platform, downlinking the images and / or files in the image repository, the object storage, the configuration center, and the model library to the central cloud platform and the edge computing micro cloud platform.

[0208] Further, in the embodiments of the present application, the method can further include the following steps:

[0209] Obtaining an algorithm module uploaded by a user, and generating new algorithm data based on the algorithm module;

[0210] Generating general configuration data based on the new algorithm data.

[0211] Further, in the embodiments of the present application, the method can further include the following steps:

[0212] Determine a target version file of the edge computing micro cloud platform;

[0213] In response to receiving the update instruction of the center cloud platform, the target version file is issued to the edge computing micro cloud platform.

[0214] The cloud-edge collaborative heavy gas turbine health intelligent management method provided by the application comprises the following steps: collecting operation data of a heavy gas turbine by using a protocol; performing data processing on the operation data by at least one first micro service in an edge computing micro cloud platform to obtain key parameters; and performing fault diagnosis based on the key parameters by at least one second micro service in a center cloud platform to obtain target early warning information. Thus, based on at least one first micro service in an edge computing micro cloud platform and at least one second micro service in a center cloud platform, the application breaks the fixed division mode of edge cloud functions, so that the resource configuration of the edge cloud can be flexibly customized and dynamically configured according to actual application service requirements, on the basis of realizing the optimized configuration and efficient utilization of computing power and network resources, the diversified equipment health management requirements in different scenarios are accurately adapted, and the service quality is improved.

[0215] It should be understood that the various forms of flow shown above can be used to reorder, add or delete steps. For example, the steps described in the present disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions of the present disclosure can be achieved, which is not limited herein.

[0216] The above specific embodiments do not constitute a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements within the spirit and principles of the present disclosure should be included in the protection scope of the present disclosure.

Claims

1. A cloud-edge collaborative intelligent health management system for heavy-duty gas turbines, characterized in that, This includes edge gateways, edge computing micro-cloud platforms, and central cloud platforms; The edge gateway is used to collect the operating data of the heavy-duty gas turbine using a protocol, and transmit the operating data to the edge computing micro-cloud platform in target form; The edge computing micro-cloud platform is used to process the running data through at least one first microservice and transmit the obtained key parameters to the central cloud platform. The central cloud platform is used to obtain target early warning information by performing fault diagnosis through at least one second microservice based on the key parameters, and to transmit the target early warning information to the edge computing micro cloud platform. The at least one first microservice includes a data preprocessing microservice, a feature extraction microservice, and a data twin microservice; The data preprocessing microservice is used to preprocess the running data to obtain preprocessed data; The feature extraction microservice is used to extract features from the preprocessed data to obtain feature vectors. The data twin microservice is used to calculate twin data from the feature vector using a digital twin model, and to determine the residual data obtained based on the twin data as key parameters. The target early warning information includes a first early warning information and / or a second early warning information; The at least one second microservice includes a trend prediction microservice, a monitoring and early warning microservice, and a fault diagnosis microservice; The trend prediction microservice is used to obtain the prediction data of the heavy-duty gas turbine based on historical operating data and the key parameters through a prediction algorithm. The monitoring and early warning microservice is used to monitor the predicted data in real time and obtain the first early warning information; The fault diagnosis microservice performs comprehensive analysis on the prediction data of the heavy-duty gas turbine to obtain the second early warning information.

2. The management system according to claim 1, characterized in that, The system also includes a microservice migration platform. The microservice migration platform is configured to migrate at least one first microservice in the first target edge computing microcloud platform to the central cloud platform in response to a first migration event triggered by the first target edge computing microcloud platform, or to migrate at least one second microservice in the central cloud platform to the second target edge computing microcloud platform in response to a second migration event triggered by the central cloud platform.

3. The management system according to claim 1, characterized in that, The system also includes a load balancing management platform; The load balancing management platform is used to dynamically manage the container resources of microservices based on the state space characteristics of multiple nodes through deep reinforcement learning, wherein the nodes are container clusters corresponding to the same microservice.

4. The management system according to claim 1, characterized in that, The system also includes an application development platform; The application development platform is used to manage the development of the edge computing micro-cloud platform and the central cloud platform.

5. The management system according to claim 4, characterized in that, The application development platform includes a general configuration module, a deployment module, a computing management module, and a differentiated configuration module; The general configuration module is used to obtain general configuration data; The deployment module is used to store the images and / or files corresponding to the general configuration data to the image repository, object storage, configuration center and model library respectively, and to distribute the images and / or files in the image repository, object storage, configuration center and model library to the central cloud platform and the edge computing micro cloud platform; The computing management module is used to acquire the algorithm module and generate new algorithm data based on the algorithm module; The differentiated configuration module is used to manage differentiated versions of the edge computing micro-cloud platform.

6. A cloud-edge collaborative intelligent health management method for heavy-duty gas turbines, characterized in that, include: The protocol is used to collect operating data from heavy-duty gas turbines; Key parameters are obtained by processing the running data through at least one first microservice in the edge computing micro-cloud platform. Based on the key parameters, target early warning information is obtained by fault diagnosis through at least one second microservice in the central cloud platform. The at least one first microservice includes a data preprocessing microservice, a feature extraction microservice, and a data twin microservice; Key parameters are obtained by processing the runtime data through at least one first microservice in the edge computing micro-cloud platform, including: The data preprocessing microservice is used to preprocess the running data to obtain preprocessed data. The feature extraction microservice extracts features from the preprocessed data to obtain a feature vector. The feature vector is calculated using the digital twin model in the data twin microservice to obtain twin data, and the residual data obtained based on the twin data is determined as the key parameter. The at least one second microservice includes a trend prediction microservice, a monitoring and early warning microservice, and a fault diagnosis microservice; the step of obtaining target early warning information by fault diagnosis through at least one second microservice in the central cloud platform based on the key parameters includes: Based on historical operating data and the key parameters, the trend prediction microservice obtains the predicted data for the heavy-duty gas turbine through a prediction algorithm. The predicted data is monitored in real time through the monitoring and early warning microservice to obtain the first early warning information; The fault diagnosis microservice comprehensively analyzes the prediction data of multiple heavy-duty gas turbines to obtain the second early warning information.

7. The method according to claim 6, characterized in that, The method further includes: In response to a first migration event triggered by a first target edge computing micro-cloud platform, at least one first microservice in the first target edge computing micro-cloud platform is migrated to the central cloud platform; or, in response to a second migration event triggered by the central cloud platform, at least one second microservice in the central cloud platform is migrated to the second target edge computing micro-cloud platform.

8. The method according to claim 6, characterized in that, The method further includes: Obtain the state space features corresponding to multiple nodes, wherein the nodes are container clusters corresponding to the same microservice; Based on the state space characteristics, deep reinforcement learning is used to dynamically manage the container resources of microservices.

9. The method according to claim 6, characterized in that, The method further includes: Obtain general configuration data; The images and / or files corresponding to the general configuration data are stored in the image repository, object storage, configuration center, and model library, respectively. In response to receiving instructions from the central cloud platform, the images and / or files in the image repository, the object storage, the configuration center, and the model library are distributed to the central cloud platform and the edge computing micro cloud platform.

10. The method according to claim 9, characterized in that, The method further includes: Obtain the algorithm module uploaded by the user, and generate new algorithm data based on the algorithm module; Generate general configuration data based on the data from the new algorithm.

11. The method according to claim 9, characterized in that, The method further includes: Determine the target version file of the edge computing micro-cloud platform; In response to receiving the update instruction from the central cloud platform, the target version file is sent to the edge computing micro cloud platform.

Images