Scalable microservice adaptive elastic architecture system for intelligent fault diagnosis

Abstract

The present invention discloses a scalable microservice adaptive elastic architecture system for intelligent fault diagnosis. The fault diagnosis system provided by the present invention performs reasonable architectural hierarchical classification on the software platform according to the application target paths of different services, utilizes the "end-edge-cloud" microservice architecture to perform scalable transformation of the software architecture, releases the coupling relationship between the various services of the fault diagnosis software platform, and conducts deep mining of fault diagnosis knowledge by constructing a self-organizing industrial knowledge graph, realizes the algorithm microservice instance structure prototype design drive and iteration capability, and at the same time, scenario-based customization and elastic scheduling of the execution status of each microservice virtual container node realizes comprehensive intelligent fault diagnosis for multiple scenarios and multiple devices.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a scalable microservice adaptive elastic architecture system for intelligent fault diagnosis. Background Art

[0002] A fault diagnosis platform leverages technologies like the Internet of Things and artificial intelligence to provide real-time monitoring of the operating status of industrial equipment, intelligent early warning, and fault analysis. This platform can help companies improve equipment reliability, safety, and efficiency, reduce repair costs and downtime losses, and enable predictive maintenance.

[0003] Currently, many manufacturers and research institutions at home and abroad have developed various types of fault diagnosis platforms to meet the needs of different industries and application scenarios. Currently, fault diagnosis platforms face the following technical problems:

[0004] 1. Lack of rapid and comprehensive fault diagnosis in multiple scenarios: Traditional fault diagnosis platforms can only be deployed, debugged, and diagnosed based on specific scenarios. They have long deployment cycles, poor adaptability to dynamic scenarios, and are unable to perform elastic scaling or switch fault diagnosis algorithms based on actual conditions.

[0005] 2. Inability to automatically build elastic instances: Traditional systems lack automated elastic instances to build a resilient architecture. The system cannot automatically adjust and flexibly scale, and lacks the ability to implement scenario-based customization.

[0006] 3. Insufficient scalability: The fault diagnosis platform needs to be able to handle changing business needs and system scale to ensure the platform's scalability and flexibility. However, traditional fault diagnosis methods may have difficulty coping with changing business needs and system scale, resulting in insufficient platform scalability.

[0007] 4. Insufficient monitoring capabilities for large-scale systems: Elastic architectures typically include hundreds or even thousands of services. However, a single request in a business scenario often requires complex chain processing across multiple services, multiple middleware, and multiple machines, making it difficult to troubleshoot these complex chains.

[0008] 5. Low fault diagnosis efficiency: Due to the complexity of the elastic architecture, fault diagnosis becomes more difficult, requiring analysis of large amounts of logs and metrics to identify the root cause of the fault. Traditional fault diagnosis methods may not meet these requirements, so new technologies and methods are needed to improve diagnostic efficiency.

[0009] 6. Low visualization: The fault diagnosis platform needs to provide an intuitive visualization interface so that users can quickly understand the status and trend of the fault. However, traditional fault diagnosis methods may lack a visualization interface, which makes it difficult for users to intuitively understand the fault status.

[0010] In summary, how to promptly diagnose faults in multiple scenarios and multiple devices is a technical problem that currently needs to be solved urgently by those skilled in the art. Summary of the Invention

[0011] In view of this, the present invention provides a scalable microservice adaptive elastic architecture system for intelligent fault diagnosis, which overcomes the defect of the existing technology that it cannot perform fault diagnosis on multiple scenarios and multiple devices in a timely manner.

[0012] In order to achieve the above object, the present invention provides the following technical solutions:

[0013] In a first aspect, an embodiment of the present invention provides a scalable microservice adaptive elastic architecture system for intelligent fault diagnosis. The architecture is divided into the following layers according to different application target paths: prototype layer, driver layer, edge layer and application layer, wherein:

[0014] The prototype layer is used to generate dynamic, real-time, and scalable multi-edge device topology network connection prototypes based on preset generation protocols, completing the communication and interaction functions between the edge and the cloud. At the same time, based on fault diagnosis knowledge mining and the weight gradient of each node connection in the self-organizing industrial knowledge graph, it updates and iterates the preset knowledge graph, generates a network structure prototype for the dynamic operation of the algorithm microservice instance, and dynamically sets the microservice startup process and optimization parameters.

[0015] The driver layer is used to complete the distributed cluster management of microservice virtual container nodes and edge device nodes through the microservice driver and automated node scheduling launcher with a unified interface;

[0016] The edge layer is used to collect and pre-process abnormal data from edge device nodes and communicate data with the cloud;

[0017] The application layer is used to complete the remote communication of diagnostic results and the human-machine interface interaction function of edge devices through the development and compilation of preset fault algorithm programs.

[0018] Optionally, the intelligent fault diagnosis system also includes: connecting to the edge microservice instance through the EMQX distributed IoT message platform, the edge layer publishes the pre-processed data to the EMQX message platform and forwards it to the cloud time series database, and at the same time the edge layer monitors and executes the command information issued by the cloud through the EMQX message platform.

[0019] Optionally, the step of generating a dynamic, real-time, and scalable multi-edge device topology network connection prototype according to a preset generation protocol to complete the communication interaction function between the edge and the cloud includes:

[0020] Build scalable topic and client ID generation protocols for the cloud and edge respectively;

[0021] Decoupling the interaction between the edge and the cloud through the generation protocol;

[0022] According to the generated protocol, a dynamic, real-time, and scalable multi-edge device topology network connection prototype is generated to complete the communication interaction function between the edge and the cloud.

[0023] Optionally, the steps of updating and iterating a preset knowledge graph based on fault diagnosis knowledge mining and weight gradients of connections between nodes in the self-organizing industrial knowledge graph, generating a network structure prototype for dynamic operation of the algorithm microservice instance, and dynamically setting the microservice startup process and optimization parameters include:

[0024] Constructing a fault diagnosis knowledge triple, wherein the triple includes: a head entity node in the triple, a tail entity node in the triple, and a relationship vector between the head entity node and the tail entity node;

[0025] Through distributed representation learning optimization method, knowledge mining of implicit relationships between entity nodes is carried out;

[0026] After evaluation by staff, the implicit relationships are updated and iterated into the preset knowledge graph;

[0027] Based on the fault diagnosis knowledge triples, we define an evaluation indicator vector. We use the fuzzy analytic hierarchy process to quantify the relationship between the evaluation indicators, and use this to initialize the weights for each evaluation indicator. We then continuously iterate the indicator weight gradients for nodes and connections during actual user applications in different scenarios to generate a network structure prototype for the dynamic operation of the algorithm microservice instance.

[0028] Generate microservice instances based on the network structure prototype that dynamically runs the algorithm microservice instance, and dynamically set the microservice startup process and optimization parameters.

[0029] Optionally, the distributed cluster management of microservice virtual container nodes and edge device nodes is completed through the microservice driver and automated node scheduling launcher of the unified interface, including:

[0030] By parsing the unique hash identifier of the incoming driver microservice, the type and number of the microservice are obtained and assigned to the pre-built launcher corresponding to the microservice. The launcher performs the image packaging operation of the corresponding microservice and allocates the corresponding network port number for the microservice. Based on the microservice's attached information, a multi-container orchestration file is generated or added to it.

[0031] Dynamically optimize its virtual machine startup parameters based on its network port number and resource usage, call the automated startup command script, use the preset virtual containerization technology to deploy and start microservices, and use cluster management tools to manage the distributed cluster of microservice virtual container nodes and edge device nodes.

[0032] Optionally, the driver layer further includes: a node scheduling scoring rule, which is used to complete the flexible scheduling of the execution status of each virtual container node according to the calculation result of the scheduling scoring function.

[0033] Optionally, the node scheduling scoring rule is calculated using the following formula:

[0034]

[0035] Among them, f(cpu,mem,pro,dist) represents the node scheduling scoring rule function, cpu represents the CPU load function under the current request, mem represents the memory usage function under the current request, pro represents the process prototype function of the current request, dist represents the distance between the current request node and the device, C is the control scaling degree, and req is the number of HTTP requests received by the current system.

[0036] Optionally, after the container is started, the launcher executes the corresponding daemon thread to collect and monitor logs of the microservice virtual container, which is used by development and operation personnel to locate functional failures and observe the operation status in real time.

[0037] Optionally, after the container is started, the method further includes: automatically capturing and alarming an exception that occurs in the microservice virtual container, and then restarting the container and isolating and suspending it to wait for development and operation personnel to troubleshoot the exception.

[0038] Optionally, the interactive functions include: user interaction interface, gateway routing forwarding, multi-tenant permission management, workflow editing and driving, digital twin three-dimensional engine multi-source heterogeneous database synchronization management, and data protocol bridging conversion.

[0039] The technical solution of the present invention has the following advantages:

[0040] 1. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis provided by the present invention divides the network architecture into application layer, prototype layer, driver layer and edge layer, and connects the prototypes through a dynamic real-time scalable multi-edge device-end topology network, thereby eliminating the high coupling relationship between the edge device nodes providing data collection and management and the cloud nodes providing fault diagnosis applications, as well as between providing application function resource services and providing fault diagnosis algorithm support instances, thereby achieving highly flexible and scalable elastic expansion capabilities.

[0041] 2. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis provided by the present invention deeply mines fault diagnosis knowledge through the construction of a knowledge graph, builds a network structure prototype for the dynamic operation of algorithm microservice instances, thereby realizing the generation capability of the fault diagnosis algorithm microservice instance structure, and gradually iteratively optimizes the knowledge graph by updating and iterating the industrial knowledge graph according to user usage.

[0042] 3. The scalable microservices adaptive elastic architecture system for intelligent fault diagnosis provided by this invention effectively manages resources and loads in distributed systems through node scheduling and scoring rules. It also achieves load balancing, assigning tasks to the nodes best suited to handle them. This ensures that every node in the system can operate efficiently, avoiding resource overloads on some nodes and resource waste on others. It also implements scenario-based, customized, elastic scheduling of the execution status of each virtual container node, improving the system's flexibility, reliability, and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1 A module composition diagram of a specific example of a scalable microservice adaptive elastic architecture system for intelligent fault diagnosis provided by an embodiment of the present invention;

[0045] Figure 2 A block diagram of the prototype generation of the dynamic operation network structure of the algorithm microservice instance provided in an embodiment of the present invention.

[0046] Figure 3 A flowchart of distributed cluster management of microservice virtual container nodes and edge device nodes provided by an embodiment of the present invention.

[0047] Figure 4 A schematic diagram of a microservice call provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0049] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0050] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components; wireless connections or wired connections. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0051] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0052] Example

[0053] The embodiment of the present invention provides a scalable microservice adaptive elastic architecture system for intelligent fault diagnosis, including: a prototype layer, a driver layer, an edge layer, and an application layer. Among them, the prototype layer is mainly a construction service set for dynamically configuring various edge device terminals and fault diagnosis algorithm abstract structure prototypes, the driver layer is mainly a microservice driver and an automated node scheduling starter set that implements a unified interface, the edge layer is mainly a set of edge device nodes and a data preprocessing service set, and the application layer is mainly a set of various upper-layer applications based on the architecture. The fault diagnosis system provided by the present invention performs reasonable architectural layering and classification on the software platform according to the application target path of different services, utilizes the "end-edge-cloud" microservice architecture to perform scalable transformation of the software architecture, removes the coupling relationship between the various services of the fault diagnosis software platform, conducts deep mining of fault diagnosis knowledge by constructing a self-organizing industrial knowledge graph, realizes algorithm microservice instance structure prototype design drive and iterative hot update iteration capability, and at the same time, scenario-based customization and elastic scheduling of the execution status of each microservice virtual container node realizes comprehensive intelligent fault diagnosis for multiple scenarios and multiple devices.

[0054] One of the main benefits of prototyping is getting feedback early in the design process from the people who will use the product. Prototyping allows designers to investigate various aspects of the design and user experience before committing to a final solution. Prototyping is a form of exploration in an interactive environment, often using software, to prototype interfaces and simulate their functionality.

[0055] In the embodiment of the present invention, Figure 1 As shown, the intelligent fault diagnosis system with an adaptive network architecture includes a prototype layer, a driver layer, an edge layer, and an application layer. The prototype layer generates a dynamic, real-time, and scalable multi-edge device topology network connection prototype based on a preset generation protocol, enabling communication and interaction between the edge and the cloud. Simultaneously, based on fault diagnosis knowledge mining and the weight gradients of each node connection in the self-organizing industrial knowledge graph, it iterates the preset knowledge graph, generates a network structure prototype for the dynamic operation of algorithm microservice instances, and dynamically sets the microservice startup process and optimization parameters. The driver layer manages the distributed cluster of microservice virtual container nodes and edge device nodes through a unified interface microservice driver and automated node scheduling initiator. The edge layer collects and pre-processes abnormal data from edge device nodes and communicates with the cloud. The application layer uses developed and compiled preset fault algorithm programs to remotely communicate diagnostic results and implement human-machine interface interaction on edge devices. This decouples the edge device nodes that provide data collection and management from the cloud nodes that provide fault diagnosis algorithm knowledge, achieving highly flexible and scalable elastic expansion capabilities and solving the problem of comprehensive fault diagnosis in multiple scenarios and devices.

[0056] In the embodiment of the present invention, Figure 1 As shown, the scalable microservice adaptive elastic architecture system for intelligent fault diagnosis also includes: connecting to the edge microservice instance through the EMQX distributed IoT messaging platform, the edge layer publishes pre-processed data to the EMQX messaging platform and forwards it to the cloud time series database, and the edge layer monitors and executes the command information issued by the cloud through the EMQX messaging platform.

[0057] EMQX natively supports distributed cluster architecture, which can handle a large number of clients and messages while ensuring high availability, fault tolerance and scalability.

[0058] In an embodiment of the present invention, a dynamic, real-time, and scalable multi-edge device-side topology network connection prototype is generated according to a preset generation protocol to complete the steps of the communication interaction function between the edge and the cloud, including: constructing scalable Topic and client ID generation protocols for the cloud and edge respectively; decoupling the interaction between the edge and the cloud through the generation protocol; generating a dynamic, real-time, and scalable multi-edge device-side topology network connection prototype according to the generation protocol to complete the communication interaction function between the edge and the cloud. By privately deploying the EMQX distributed IoT messaging platform and building a highly flexible and scalable topic generation protocol, the interaction between the edge and the cloud is decoupled, and a dynamic, real-time, and scalable multi-edge device-side topology network connection prototype is generated, thereby realizing the customization of the edge device-side connection interaction process, as well as the data two-way binding structure and protocol design based on the Message Queuing Telemetry Transport (MQTT) protocol.

[0059] In practice, a Topic acts as a transmission medium between message publishers (Pub) and subscribers (Sub). Devices can send and receive messages through Topics, thus enabling communication between the server and the device.

[0060] In the embodiments of the present invention, all edge devices involved exchange data and instructions through the MQTT IoT data transmission protocol. MQTT can provide real-time and reliable messaging services for connecting remote devices with minimal code and limited bandwidth. First, it is necessary to build highly flexible and scalable topic and client ID generation protocols for the cloud and edge, respectively. The construction formula is as follows:

[0061] Topic={PL,PR,ED,"cloud" / "edge"}

[0062] ID client ={PL,PR,ED,N,"cloud" / "edge"}

[0063] Among them, PL represents the platform name, PR represents the project name, ED represents the service name, N represents the edge or cloud node number, and "cloud" / "edge" represents the cloud or edge. This protocol is used to decouple the interaction between the edge and the cloud. That is, both the cloud and the edge can realize the interaction of data and instructions by broadcasting or listening to the topics agreed upon in the protocol, and the sender and receiver of the message are uniquely identified by the client ID. Finally, a dynamic, real-time, and scalable topological network connection prototype for multiple edge devices is generated according to the protocol, which enables the customization of the connection and interaction process between multiple edge devices and the cloud, and opens up a cloud-edge two-way binding (data uplink - instruction downlink) channel based on the MQTT protocol.

[0064] In an embodiment of the present invention, according to the weight gradient of each node connection of the fault diagnosis knowledge mining and the self-organizing industrial knowledge graph, the preset knowledge graph is updated and iterated, and a network structure prototype of the dynamic operation of the algorithm microservice instance is generated, and the microservice startup process and optimization parameters are dynamically set. The steps include: constructing a fault diagnosis knowledge triple, the triple includes: the head entity node in the triple, the tail entity node in the triple, and the relationship vector between the head entity node and the tail entity node; using a distributed representation learning optimization method to perform knowledge mining of the implicit relationship between the entity nodes; after the implicit relationship is evaluated by the staff, it is updated and iterated into the preset knowledge graph; defining an evaluation index vector according to the fault diagnosis knowledge triple, using the fuzzy hierarchical analysis method to quantify the relationship between the evaluation indicators, as the initialization weight of each evaluation indicator, and continuously iteratively updating the index weight gradient of the nodes and connections in the actual application process of users in different scenarios, generating a network structure prototype of the dynamic operation of the algorithm microservice instance; generating a microservice instance according to the network structure prototype of the dynamic operation of the algorithm microservice instance, and dynamically setting the microservice startup process and optimization parameters.

[0065] In a specific embodiment, if Figure 2 As shown, starting from data acquisition and collection, the data here includes not only device data, but also status data of controllers such as PLC, alarm data and historical data of SCADA system, manually filled fault data, etc. This is only an example and is not limited to this. In actual application, the corresponding data is selected according to the actual situation.

[0066] Data structuring processing is to format the collected multi-source heterogeneous data according to pre-defined standards and process it into the required structured data; entity linking is to generate a graph structure based on structured data and link the entity nodes in the graph structure according to the structured data.

[0067] Ontology modeling is the process of modeling all nodes and links in a graph structure and describing them in the form of a model. The specific implementation of distributed representation learning has been clearly stated in the article, and its goal is to identify implicit relational information related to equipment fault diagnosis. "RDBMS, Graph DBMS" represent relational databases and graph-structured databases, respectively.

[0068] Knowledge reasoning is the process of inferring unknown knowledge based on existing knowledge. The specific process has been reflected in the knowledge mining reasoning on the right. The low-code fault diagnosis platform obtains fault diagnosis knowledge mining in the knowledge graph to realize the registration algorithm microservice instance, set the startup process and optimization parameters, and finally generate the dynamic operation network structure of the microservice instance. The low-code fault diagnosis platform is also responsible for equipment data collection and knowledge representation extraction, and feeds it back to real-time resources.

[0069] In a specific embodiment, the prototype design and knowledge mining of the fault diagnosis algorithm process are first implemented through the graph structure. First, the fault diagnosis knowledge triple G is defined as follows:

[0070] G={(h,l,t)}∈{MI×FI×RI×AI}

[0071] Among them, MI represents the device information set, FI represents the fault information set, RI represents the relationship set between nodes, AI represents the algorithm information set, h represents the head entity node in the triplet G, t represents the tail entity node in the triplet G, and l represents the relationship vector between the head entity node h and the tail entity node t.

[0072] Then, the knowledge mining of implicit relationships between entity nodes is carried out through the distributed representation learning optimization method. First, the semantic weight vector θ is added i ∈R k , R k To be added to the head and tail entities h and t corresponding to the relation l, the knowledge representation model of the fault diagnosis entity node is as follows:

[0073]

[0074] in, is the Hadamma product operator, h i and t i By the relationship l i The semantic representation of the head and tail entity vectors. In order to improve the semantic association accuracy of the fault diagnosis knowledge mining entity nodes, the similarity of the entity nodes is calculated based on the Mahalanobis distance, and the implicit relationship between the fault diagnosis knowledge mining entity nodes is mined. The score function constructed based on this is as follows:

[0075]

[0076] Among them, f l (h, t) is the score function of the workshop resource data triple G, and W l is a relation-specific symmetric non-negative weight matrix corresponding to the adaptive metric. Finally, the optimization objective function is given as follows:

[0077]

[0078] in,[·] + is its upper bound value greater than 0, γ is the maximum semantic boundary interval of positive and negative triples, ||W l || Fis the F-norm of the matrix, C controls the scaling, and λ controls the regularization of the adaptive weight matrix. The above equation shows that when the result of this equation is minimized, there may be an implicit relationship between the corresponding head entity node h and the tail entity node t. This implicit relationship is evaluated by engineers and then hot-updated and iterated into the industrial knowledge graph.

[0079] Finally, an evaluation metric vector (algorithm performance, usage frequency, sampling type, and device adaptability) is defined for the fault diagnosis knowledge triple G. The fuzzy analytic hierarchy process (FAHP) is then used to quantify the relationships between the metrics. Weights for each evaluation metric are initialized, forming a distinct hierarchical structure. The metric weight gradients for nodes and connections are then iteratively updated during actual user applications in different scenarios. Finally, based on the fault diagnosis knowledge, the weight gradients for nodes and connections are mined to generate a network structure prototype for the dynamic operation of the scenario-based algorithm microservice instance. Based on this, the algorithm microservice instance is registered, and the startup process and optimization parameters are set. Through prototyping and knowledge mining of the fault diagnosis algorithm process, the metric weight gradients for nodes and connections of the fault diagnosis algorithm knowledge are continuously iteratively updated, ultimately generating a network structure prototype for the dynamic operation of the scenario-based algorithm microservice instance and constructing the fault diagnosis algorithm instance process.

[0080] In an embodiment of the present invention, the distributed cluster management of microservice virtual container nodes and edge device nodes is completed through the microservice driver and automated node scheduling launcher of a unified interface, including: obtaining the type and number information of the microservice by parsing the unique hash identifier of the incoming required driving microservice, and assigning it to the pre-constructed launcher corresponding to the microservice. The launcher executes the image packaging operation of the corresponding microservice, and assigns the corresponding network port number to the microservice, and generates or adds it to the multi-container orchestration file based on the microservice attachment information; dynamically optimizes its virtual machine startup parameters based on its network port number and resource occupancy, and calls the automated startup command script, uses the preset virtual containerization technology to deploy and start the microservice, and uses the cluster management tool to manage the distributed cluster of microservice virtual container nodes and edge device nodes.

[0081] In a specific embodiment, if Figure 3 As shown, first, the unique hash identifier of the required driver microservice is parsed to obtain the type and number information of the microservice, and then assigned to the pre-constructed starter corresponding to the microservice. The starter performs the image packaging operation of the corresponding microservice, generates or adds it to the multi-container orchestration file according to the microservice attachment information, and assigns the corresponding network port number to it.

[0082] Dynamically optimize its JVM virtual machine startup parameters based on its I / O and resource usage, call the automated startup command script, use Docker virtual containerization technology to deploy and start microservices, and use cluster management tools to complete the distributed cluster management of microservice virtual container nodes and edge device nodes.

[0083] In an embodiment of the present invention, the driver layer also includes: a node scheduling scoring rule, which is used to elastically schedule the execution status of each virtual container node based on the calculation results of the scheduling scoring function. The node scheduling scoring rule is mainly used to solve resource management and load balancing problems in distributed systems. It is mainly based on the real-time traffic load and resource usage of the node to score, and then sorts and selects the best nodes based on the score. However, it fails to take into account the execution process and prototype structure of the microservice instance itself, which may cause mismatches between service nodes.

[0084] The node scheduling scoring rule is calculated using the following formula:

[0085]

[0086] Here, f(cpu, mem, pro, dist) represents the node scheduling scoring rule function, cpu represents the CPU load function under the current request, mem represents the memory usage function under the current request, pro represents the process prototype function of the current request, dist represents the distance between the current request node and the device, C controls the scaling factor, and req represents the number of HTTP requests currently received by the system. Based on the calculation results of the scheduling scoring function, scenario-based, customized, and elastic scheduling of the execution status of each virtual container node is implemented. By incorporating the architectural prototype dimension and the device node distance dimension, an adaptive node scheduling scoring function is constructed, enabling scenario-based, customized, and elastic scheduling of the execution status of each virtual container node, avoiding mismatches of edge device nodes.

[0087] In an embodiment of the present invention, after the container is started, the launcher executes the corresponding daemon thread to collect and monitor logs for the microservice container, making it easier for development and operation personnel to locate functional failures and observe the operation status in real time. It mainly uses a distributed globally unique ID to connect the same request distributed on each service node in series, restore the call relationship, track system problems, analyze call data, and count system indicators, thereby realizing a session-level monitoring log tracking capability.

[0088] In a specific embodiment, if Figure 4As shown in the figure, "Request X" represents the Xth http request initiated by the user; "Return X" represents the Xth http response returned by the front-end to the user; A represents the front-end interface service; "B, C, D, E" represent the various back-end and middleware microservices called in the system after the user initiates "Request X". The specific microservices called are determined by the actual user request, and this is only for reference; RESTful is an architectural style of an application programming interface (API) that uses HTTP requests to access and use data, and this represents an http request that conforms to this style.

[0089] In an embodiment of the present invention, after the container is started, the following steps are further included: automatically capturing and alarming the exceptions that occur in the microservice container, and then restarting the container and isolating and suspending it to wait for development and operation personnel to troubleshoot the exceptions.

[0090] In this embodiment of the present invention, interactive functions include: a user interface, gateway routing and forwarding, multi-tenant permission management, workflow editing and driving, synchronization management of multi-source heterogeneous databases within the digital twin 3D engine, and data protocol bridging and conversion. Based on the application layer of the architecture described in this invention, a low-code, configurable, intelligent fault diagnosis software platform can be formed, enabling users to perform highly customized interactive operations for fault diagnosis.

[0091] The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis provided in the embodiments of the present invention divides the software architecture into an application layer, a prototype layer, a driver layer, and an edge layer. This decouples the edge device nodes that provide data collection and management from the cloud nodes that provide fault diagnosis algorithm knowledge, achieving highly flexible and scalable elastic expansion capabilities. By implementing a self-organizing knowledge graph, fault diagnosis knowledge is deeply mined, and a knowledge network of fault diagnosis algorithms is constructed, thereby achieving the ability to generate a microservice instance structure for the fault diagnosis algorithm. The knowledge graph is gradually iteratively optimized as users use it, allowing for timely fault diagnosis in multiple scenarios and devices.

[0092] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications derived therefrom remain within the scope of protection of the present invention.

Claims

1. A scalable microservice adaptive elastic architecture system for intelligent fault diagnosis, characterized by: The architecture is divided into the following layers according to different application target paths: prototype layer, driver layer, edge layer and application layer, among which, The prototype layer is used to generate dynamic, real-time, and scalable multi-edge device topology network connection prototypes based on preset generation protocols, completing the communication and interaction functions between the edge and the cloud. At the same time, based on fault diagnosis knowledge mining and the weight gradient of each node connection in the self-organizing industrial knowledge graph, it updates and iterates the preset knowledge graph, generates a network structure prototype for the dynamic operation of the algorithm microservice instance, and dynamically sets the microservice startup process and optimization parameters. The driver layer is used to complete the distributed cluster management of microservice virtual container nodes and edge device nodes through the microservice driver and automated node scheduling launcher with a unified interface; The edge layer is used to collect and pre-process abnormal data from edge device nodes and communicate data with the cloud; The application layer is used to complete the remote communication of diagnostic results and the human-machine interface interaction function of edge devices through the development and compilation of preset fault algorithm programs; The step of generating a dynamic, real-time, and scalable multi-edge device topology network connection prototype according to a preset generation protocol to complete the communication interaction function between the edge and the cloud includes: Build scalable topic and client ID generation protocols for the cloud and edge respectively; Decoupling the interaction between the edge and the cloud through the generation protocol; Generate a dynamic, real-time, and scalable multi-edge device topology network connection prototype based on the generated protocol to complete the communication interaction function between the edge and the cloud; The steps of mining fault diagnosis knowledge and the weight gradient of each node connection of the self-organizing industrial knowledge graph, updating and iterating the preset knowledge graph, generating a network structure prototype for the dynamic operation of the algorithm microservice instance, and dynamically setting the microservice startup process and optimization parameters include: Constructing a fault diagnosis knowledge triple, wherein the triple includes: a head entity node in the triple, a tail entity node in the triple, and a relationship vector between the head entity node and the tail entity node; Through distributed representation learning optimization method, knowledge mining of implicit relationships between entity nodes is carried out; After evaluation by staff, the implicit relationships are updated and iterated into the preset knowledge graph; Based on the fault diagnosis knowledge triples, we define an evaluation indicator vector. We use the fuzzy analytic hierarchy process to quantify the relationship between the evaluation indicators, and use this to initialize the weights for each evaluation indicator. We then continuously iterate the indicator weight gradients for nodes and connections during actual user applications in different scenarios to generate a network structure prototype for the dynamic operation of the algorithm microservice instance. Generate microservice instances based on the network structure prototype that dynamically runs the algorithm microservice instance, and dynamically set the microservice startup process and optimization parameters.

2. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis according to claim 1 is characterized in that: Also includes: By connecting to the edge microservice instance through the EMQX distributed IoT messaging platform, the edge layer publishes pre-processed data to the EMQX messaging platform and forwards it to the cloud-based time series database. At the same time, the edge layer listens to and executes the command information issued by the cloud through the EMQX messaging platform.

3. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis according to claim 1 is characterized in that: The distributed cluster management of microservice virtual container nodes and edge device nodes is completed through the microservice driver and automated node scheduling starter of the unified interface, including: By parsing the unique hash identifier of the incoming driver microservice, the type and number of the microservice are obtained and assigned to the pre-built launcher corresponding to the microservice. The launcher performs the image packaging operation of the corresponding microservice and allocates the corresponding network port number for the microservice. Based on the microservice's attached information, a multi-container orchestration file is generated or added to it. Dynamically optimize its virtual machine startup parameters based on its network port number and resource usage, call the automated startup command script, use the preset virtual containerization technology to deploy and start microservices, and use cluster management tools to manage the distributed cluster of microservice virtual container nodes and edge device nodes.

4. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis according to claim 1 is characterized in that: The driver layer further includes: a node scheduling scoring rule, which is used to complete the flexible scheduling of the execution status of each virtual container node according to the calculation result of the scheduling scoring function.

5. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis according to claim 4 is characterized in that: The node scheduling scoring rule is calculated using the following formula: Among them, f(cpu,mem,pro,dist) represents the node scheduling scoring rule function, cpu represents the CPU load function under the current request, mem represents the memory usage function under the current request, pro represents the process prototype function of the current request, dist represents the distance between the current request node and the device, C is the control scaling degree, and req is the number of HTTP requests received by the current system.

6. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis according to claim 3 is characterized in that: After the container is started, the launcher executes the corresponding daemon thread to collect and monitor logs of the microservice virtual container, which is used by development and operation personnel to locate functional failures and observe the operation status in real time.

7. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis according to claim 6 is characterized in that: After the container is started, it also includes: automatically capturing and alarming the exceptions that occur in the microservice virtual container, and then restarting the container and isolating it and suspending it to wait for development and operation personnel to troubleshoot the exceptions.

8. The scalable microservice adaptive elastic architecture system for intelligent fault diagnosis according to claim 1, characterized in that: The interactive functions include: user interaction interface, gateway routing forwarding, multi-tenant permission management, workflow editing and driving, digital twin 3D engine, multi-source heterogeneous database synchronization management and data protocol bridging conversion.

Images