Industrial service network anomaly reconstruction method, device and system, and storage medium

Abstract

The invention discloses an industrial service network anomaly reconstruction method, device and system, and a storage medium. The method comprises the following steps: S1, constructing an industrial service cooperation network; dividing the sub-graphs into different sub-graphs according to plant area positions and service cluster categories; s2, in different sub-graphs, propagation embedding is carried out through interaction and cooperation information of the industrial main body and the service cluster in the sub-graphs, and complex high-order information in the different sub-graphs is obtained; s3, aggregating the high-order information of the different sub-graphs to obtain feature information after fusion of the different sub-graphs; and S4, performing prediction by using an inner product according to the embedded representation of the fused industrial subject and service. By adopting the technical scheme of the invention, the problem that the precision of network reconstruction is restricted due to the fact that a traditional model cannot fully learn position attributes and cluster characteristics of industrial services is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial service networks, and in particular relates to an industrial service network abnormal reconstruction method and device, system, and storage medium. Background Art

[0002] With the development of technologies such as the Industrial Internet of Things (IIoT), edge computing, and digital twins, equipment, processes, and procedures in industrial production environments can be digitally modeled as industrial services, enabling ubiquitous connectivity and intelligent collaboration through the Industrial Internet. In complex industrial scenarios, the industrial service networks formed by enterprises around production needs exhibit distinct domain characteristics. These supply and demand networks, composed of equipment collaboration relationships, process matching characteristics, and production process preferences, can be defined as industrial service collaboration networks. However, due to dynamic factors such as production line adjustments, equipment migration, or supply chain restructuring, existing industrial service collaboration relationships may break down, leading to local or global failures in the industrial service network. This phenomenon is referred to as industrial service network decoupling. How to achieve self-healing reconstruction of industrial service networks in dynamic industrial scenarios, proactively adapt to new production environments without human intervention, and provide continuously optimized service support for intelligent manufacturing systems has become a key issue in the development of the Industrial Internet.

[0003] To restructure the industrial service network, intelligent service replacement is required for decoupled network nodes. To ensure accurate network repair, a recommendation system is used to accurately recommend each missing node in the network. Precise service matching is achieved by mining production process characteristics and historical collaboration records. Dynamic reconstruction technology based on the industrial service collaboration network can effectively maintain the continuity of production systems and enhance the adaptability of industrial services.

[0004] Existing industrial service recommendation systems based on graph neural networks mostly use global embedding learning, which struggles to effectively capture the dynamic characteristics of industrial service clusters. The heterogeneity of industrial equipment, the sensitivity of process parameters, and the differences in standardization of service interfaces prevent traditional models from fully learning the location attributes and cluster characteristics of industrial services, limiting the accuracy of network reconstruction. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an industrial service network abnormal reconstruction method and device, system and storage medium.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] An industrial service network abnormal reconstruction method, comprising:

[0008] Step S1: Build an industrial service collaboration network and divide it into different subgraphs according to factory locations and service cluster categories;

[0009] Step S2: In different subgraphs, the interaction and collaboration information between the industrial entities and service clusters in the subgraphs is propagated and embedded to obtain complex high-level information in different subgraphs;

[0010] Step S3: Aggregate high-order information of different sub-graphs to obtain fused feature information of different sub-graphs;

[0011] Step S4: Use the inner product to make predictions based on the embedded representation of the fused industrial entity and service.

[0012] Preferably, in step S1, based on the service collaboration records of industrial entities and the location area information of industrial plants, industrial services are classified into different types of service clusters and an industrial service collaboration network is constructed.

[0013] Preferably, in step S3, high-order information of different sub-graphs is aggregated, and features of layers 0-k are merged using a feature fusion layer to obtain fused feature information of different sub-graphs.

[0014] The present invention also provides an industrial service network abnormal reconstruction device, comprising:

[0015] The first processing module is used to build an industrial service collaboration network and divide it into different subgraphs according to factory location and service cluster category;

[0016] The second processing module is used to propagate and embed the interaction and collaboration information between the industrial entities and service clusters in different subgraphs, thereby obtaining complex high-level information in different subgraphs;

[0017] The third processing module is used to aggregate the high-order information of different sub-graphs to obtain the feature information after the different sub-graphs are fused;

[0018] The fourth processing module is used to make predictions using the inner product based on the embedded representation of the fused industrial subject and service.

[0019] Preferably, the first processing module classifies industrial services into different types of service clusters and constructs an industrial service collaboration network based on the service collaboration records of industrial entities and the location area information of industrial plants.

[0020] Preferably, the third processing module aggregates high-order information of different sub-graphs, and uses a feature fusion layer to merge features of 0-k layers to obtain fused feature information of different sub-graphs.

[0021] The present invention also provides an industrial service network abnormal reconstruction system, comprising: a memory and a processor, wherein the memory stores a computer program executed by the processor, and the computer program executes the industrial service network abnormal reconstruction method when executed by the processor.

[0022] The present invention also provides a storage medium, on which a computer program is stored, and the computer program executes the industrial service network abnormal reconstruction method when running.

[0023] The present invention constructs an industrial service collaboration network based on different locations; uses subgraph construction technology to construct corresponding subgraphs for each service cluster in the network to depict the characteristic information of the service cluster, and learns the dynamic change characteristics of the service cluster characteristics and service network reconstruction at different locations; designs a graph neural network recommendation model that integrates location and collaboration information to deeply capture the comprehensive interaction between each service cluster and service characteristics at different locations, and fully learn the characteristic information of dynamic service clusters; and finally realizes cross-domain service collaborative recommendation to solve the problem that traditional models cannot fully learn the location attributes and cluster characteristics of industrial services, which restricts the accuracy of network reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0025] Figure 1 This is a flow chart of the abnormal reconstruction method of the industrial service network according to an embodiment of the present invention;

[0026] Figure 2 Schematic diagram of the principle of the abnormal reconstruction method of the industrial service network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0028] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0029] Example 1:

[0030] like Figure 1 、 2 As shown, an embodiment of the present invention provides an industrial service network abnormal reconstruction method, including:

[0031] Step S1: Based on the service collaboration records of industrial entities and the location and area information of industrial plants, industrial services are classified into different types of service clusters and an industrial service collaboration network is constructed; the network is divided into different subgraphs according to the plant location and service cluster category;

[0032] Step S2: In different subgraphs, the interaction and collaboration information between the industrial entities and service clusters in the subgraphs is propagated and embedded to obtain complex high-level information in different subgraphs;

[0033] Step S3: Aggregate the high-order information of different sub-graphs, and use the feature fusion layer to merge the features of the 0-k layers to obtain the fused feature information of different sub-graphs;

[0034] Step S4: Use the inner product to make predictions based on the embedded representations of the fused industrial entities and services. Use the recommendation results obtained for industrial entities at different factory locations to restore the service clusters of each category and repair the industrial service collaboration network to achieve proactive recommendations.

[0035] As an implementation method of an embodiment of the present invention, in step S1, in order to fully characterize the characteristic information of industrial entities in different factory locations and to be able to make active recommendations when the user arrives at a new location. The embodiment of the present invention proposes the concept of an industrial service collaboration network. Due to the different factory locations, the service clusters will change, and the original industrial service collaboration network will collapse. Therefore, the embodiment of the present invention is based on constructing an industrial service collaboration network based on different factory locations and different categories of service clusters, using a subgraph generation layer to complete subgraph extraction, and fully learning the characteristics of the industrial service collaboration network, different service clusters and factory locations.

[0036] In order to generate service clusters at new plant locations and fully learn similar features in categories, the present invention divides services into different service clusters. Specifically, each service consists of an initial feature vector e s express.

[0037] In order to accurately classify services, the embodiment of the present invention uses a neural network to perform classification prediction and classify services into different service clusters, namely,

[0038]

[0039] in, Represents the layer 0 embedding of the service, Represents the layer 1 embedding of a service.

[0040] After categorizing services into service clusters, the industrial service collaboration network can be divided into different subgraphs based on factory location. Furthermore, relationships within supply chain collaboration information are also important features. When navigating a new factory location, services from previous supply chain collaboration information may be referenced. Similarly, expert domain knowledge may also be referenced. Therefore, embodiments of the present invention integrate supply chain collaboration information and expert domain knowledge into each subgraph for embedded learning. Finally, the definition of a subgraph is shown in the following equation.

[0041]

[0042] in, represents the subgraph set obtained by dividing the original network according to the plant location and service cluster; u pq is a user in location p and cluster q; E pq represents the expert of position p and cluster q; F pq represents a social user at location p and cluster q; S pq Represents a service at location p and cluster q.

[0043] As an implementation method of the embodiment of the present invention, in step S2, the initial embedding of the industrial subject u and the service s is represented as and The calculation formula is shown in Eq.

[0044]

[0045] Since the direct interaction between industrial entities and service clusters, supply chain collaboration relationships and experts provides the most direct and important information, in the first-order propagation, all nodes participate in the graph convolution operation, as shown in the formula:

[0046]

[0047] For higher-order graph convolutions, in order to avoid introducing noise and learn the characteristic information of the interaction between the location of each plant and the industrial entities and services in each service cluster, all convolutional propagation operations are performed within the subgraph. In this embodiment of the present invention, the location area and category of the service cluster are unique. Therefore, each service will only appear in one subgraph, and the industrial entities that directly interact with it will appear in this subgraph. However, due to the interaction with other services, these industrial entities will also appear in other subgraphs. In order to learn the embedding of industrial entities in different subgraphs, this embodiment of the present invention defines high-order propagation as follows:

[0048]

[0049] Through this calculation method, the node embedding learned in each subgraph is only propagated within the subgraph, fully learning the embedding information under different service clusters in different factory locations.

[0050] As an implementation method of the present invention, in step S3, since all industrial entities in the subgraph interact with the services in the service cluster, they can also be regarded as potentially having similar preferences. At the same time, the introduction of supply chain collaboration relationships and experts also provides indirect interaction between industrial entities and the services in the subgraph. uS It represents the industrial subject features learned under this category at this location, and also contains potential similar interest features. After k layers of graph convolution, since industrial subjects can appear in multiple subgraphs, the final representation of industrial subjects is a combination of embeddings learned in different subgraphs, as shown in Eq.

[0051]

[0052] The embodiment of the present invention derives the embedding of aggregated industrial subject services across different subgraphs. arrive and to Subsequently, the embeddings obtained at each layer are further integrated to form the final representation as shown in Eq.

[0053]

[0054] As an implementation of an embodiment of the present invention, in step S4, through multi-layer propagation and aggregation within the subgraph, the embodiment of the present invention only utilizes the feature information learned within the subgraph, avoiding the influence of noise such as factory location and service clusters. Through subgraph aggregation and feature fusion, the final representation of industrial entities and services is obtained. Finally, the embodiment of the present invention predicts the industrial entity's preference for services by calculating the inner product between the industrial entity and the service, which is calculated as follows:

[0055]

[0056] By using inner product calculations, we predict the industrial entity's location within the current factory and obtain specific services that match that location. After recommending the industrial entity within that location, we use the recommended results to populate the industrial service network, filling in missing items and updating incorrect items.

[0057] Example 2:

[0058] An embodiment of the present invention further provides an industrial service network abnormal reconstruction device, comprising:

[0059] The first processing module is used to build an industrial service collaboration network and divide it into different subgraphs according to factory location and service cluster category;

[0060] The second processing module is used to propagate and embed the interaction and collaboration information between the industrial entities and service clusters in different subgraphs, thereby obtaining complex high-level information in different subgraphs;

[0061] The third processing module is used to aggregate the high-order information of different sub-graphs to obtain the feature information after the different sub-graphs are fused;

[0062] The fourth processing module is used to make predictions using the inner product based on the embedded representation of the fused industrial subject and service.

[0063] As an implementation of an embodiment of the present invention, the first processing module classifies industrial services into different types of service clusters and constructs an industrial service collaboration network based on the service collaboration records of industrial entities and the location area information of industrial plants.

[0064] Preferably, the third processing module aggregates high-order information of different sub-graphs, and uses a feature fusion layer to merge features of 0-k layers to obtain fused feature information of different sub-graphs.

[0065] Example 3:

[0066] An embodiment of the present invention further provides an industrial service network abnormal reconstruction system, comprising: a memory and a processor, wherein the memory stores a computer program executed by the processor, and the computer program executes an industrial service network abnormal reconstruction method when executed by the processor.

[0067] Example 4:

[0068] An embodiment of the present invention further provides a storage medium having a computer program stored thereon, and the computer program executes the industrial service network abnormal reconstruction method when running.

[0069] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A method for abnormal reconstruction of an industrial service network, characterized in that: include: Step S1: Build an industrial service collaboration network and divide it into different subgraphs according to factory locations and service cluster categories; Step S2: In different subgraphs, the interaction and collaboration information between the industrial entities and service clusters in the subgraphs is propagated and embedded to obtain complex high-level information in different subgraphs; Step S3: Aggregate high-order information of different sub-graphs to obtain fused feature information of different sub-graphs; Step S4: Use the inner product to make predictions based on the embedded representation of the fused industrial entity and service.

2. The abnormal reconstruction method of the industrial service network according to claim 1, characterized in that: In step S1, based on the service collaboration records of industrial entities and the location area information of industrial plants, industrial services are classified into different types of service clusters and an industrial service collaboration network is constructed.

3. The abnormal reconstruction method of the industrial service network according to claim 2, characterized in that: In step S3, the high-order information of different sub-graphs is aggregated, and the features of the 0-k layers are merged using the feature fusion layer to obtain the fused feature information of different sub-graphs.

4. An industrial service network abnormal reconstruction device, characterized in that: include: The first processing module is used to build an industrial service collaboration network and divide it into different subgraphs according to factory location and service cluster category; The second processing module is used to propagate and embed the interaction and collaboration information between the industrial entities and service clusters in different subgraphs, thereby obtaining complex high-level information in different subgraphs; The third processing module is used to aggregate the high-order information of different sub-graphs to obtain the feature information after the different sub-graphs are fused; The fourth processing module is used to make predictions using the inner product based on the embedded representation of the fused industrial subject and service.

5. The abnormal reconstruction device for an industrial service network according to claim 4, characterized in that: The first processing module classifies industrial services into different types of service clusters and constructs an industrial service collaboration network based on the service collaboration records of industrial entities and the location area information of industrial plants.

6. The abnormal reconstruction device for an industrial service network according to claim 5, characterized in that: The third processing module aggregates the high-order information of different sub-graphs and uses the feature fusion layer to merge the features of the 0-k layers to obtain the fused feature information of different sub-graphs.

7. An industrial service network abnormal reconstruction system, characterized in that: include: A memory and a processor, wherein the memory stores a computer program executed by the processor, and when the computer program is executed by the processor, the method for abnormal reconstruction of an industrial service network according to any one of claims 1 to 3 is executed.

8. A storage medium, characterized in that: The storage medium stores a computer program, which, when running, executes the abnormal reconstruction method for an industrial service network according to any one of claims 1 to 3.

Images