Workshop production line whole-process data processing and intelligent monitoring method and system

Abstract

The invention belongs to the field of industrial control, particularly relates to a workshop production line whole-process data processing and intelligent monitoring method and system, and aims to solve the problem that the whole production process is difficult to effectively monitor in related technologies. The method comprises the following steps: collecting asynchronous data streams from different sources in the whole production process; mapping each data point in the asynchronous data stream to a unified space-time reference framework; constructing a knowledge graph; inputting the knowledge graph into the space-time diagram convolutional network model to obtain an abnormal probability score of each node in the knowledge graph; and determining an abnormal probability node, carrying out reverse backtracking along a relation edge in the knowledge graph, positioning a source abnormal node, carrying out feature attribution analysis on state data of the source abnormal node, and generating a root cause interpretation report of an abnormal working condition. According to the method, the knowledge graph with the work order as the entity and the process flow as the relation chain is constructed, digital mapping and unified representation of the whole production flow are achieved, and the whole production flow can be effectively monitored.

Description

Technical Field

[0001] This invention belongs to the field of industrial control, and specifically relates to a method and system for data processing and intelligent monitoring of the entire process of a workshop production line. Background Technology

[0002] In the field of smart manufacturing, full-process monitoring of the workshop production line is crucial to ensuring production efficiency and product quality. Modern production lines typically contain numerous automated devices and complex processes. As the production process proceeds, massive amounts of multi-source, heterogeneous data streams are continuously generated. These data include, but are not limited to, time-series data from equipment sensors (such as vibration, temperature, and current), production work orders and material flow data from the Manufacturing Execution System (MES), and product quality data captured by vision inspection systems.

[0003] Currently, there are two main paradigms for monitoring production lines. One is the threshold-based single-point monitoring method, which sets static thresholds for key parameters of specific equipment or processes, triggering an alarm once the monitored value exceeds the limit. However, this method views each data point in isolation, lacks a holistic perspective on the entire production process, fails to understand the mutual influence and correlation between different process links, is prone to generating a large number of false alarms, and is difficult to pinpoint the root cause. The second approach is a monitoring method based on traditional machine learning models, which trains on historical data to detect anomalies. While this method improves the level of intelligence to some extent, its models typically rely on manually constructed features, making it difficult to effectively handle asynchronous, heterogeneous, multi-source streaming data, and thus difficult to accurately pinpoint the root cause.

[0004] In summary, existing methods are insufficient for effectively monitoring the entire production process and can only detect anomalies rather than diagnose root causes, resulting in limited effectiveness in real-world applications. Summary of the Invention

[0005] To address the aforementioned problems in existing technologies—namely, the inability of related technologies to effectively monitor the entire production process, their limited ability to detect anomalies rather than diagnose root causes, and their limited practical application—this invention, in its first aspect, proposes a method for full-process data processing and intelligent monitoring of a workshop production line, the method comprising: Collect asynchronous data streams from different sources throughout the entire production process; Each data point in the asynchronous data stream is mapped to a unified spatiotemporal reference framework, and each data point is bound to the production work order number and process node to obtain a standardized data unit. Based on the standardized data units, a knowledge graph is constructed with production work orders as entities and process flows as relational chains. The nodes of the knowledge graph are used to embed real-time status features containing process flows, and the edge weights are used to characterize the dependencies between process flows. The knowledge graph is input into a pre-built spatiotemporal graph convolutional network model to obtain the anomaly probability score of each node in the knowledge graph and identify the potential propagation path of abnormal conditions between process nodes; the spatiotemporal graph convolutional network model calculates the anomaly probability score and propagation path by capturing the spatial and temporal dependencies in the production process. Identify the abnormal probability nodes, and backtrack along the relation edges in the knowledge graph to locate the source abnormal node. Perform feature attribution analysis on the state data of the source abnormal node to generate a root cause explanation report for the abnormal operating conditions.

[0006] In some preferred embodiments, mapping individual data points in the asynchronous data stream to a unified spatiotemporal reference framework includes: Define time offset and coordinate transformation rules for each type of data stream. The coordinate transformation rules are used to map the physical or logical location of the data source to a unified linear spatial coordinate system with the production line entrance as the origin. The timestamps of all data points are corrected using the time offset. The coordinate transformation rules are used to map data points to specific process nodes in the spatial coordinate system, so that each data point in the asynchronous data stream obtains a globally unique spatiotemporal coordinate (T, S), where T is a correction timestamp accurate to milliseconds and S is the spatial coordinate of the corresponding process node.

[0007] In some preferred embodiments, the step of constructing a knowledge graph based on the standardized data units, with production work orders as entities and process flows as relational chains, includes: An incremental graph update algorithm is used to continuously monitor the input standardized data units; For the new standardized data unit, its production order number and process node information are located to the corresponding nodes and edges in the knowledge graph; By using the node embedding update function, the new state features are fused with the original embedding vector of the node to update the real-time state representation of the node. At the same time, the edge weights between nodes are dynamically adjusted according to the flow time interval between the current batch and the previous batch in adjacent processes to represent the changes in material flow efficiency and dependence intensity between processes in real time.

[0008] In some preferred embodiments, the spatiotemporal graph convolutional network model captures the spatial and temporal dependencies in the production process, including: By employing graph convolution operations, the feature representation of the target node is updated by aggregating the state feature information of the target node's neighboring nodes, thereby obtaining the spatial dependency of the target node, where the target node is any node in the knowledge graph. At each node, a gated recurrent unit or a long short-term memory network is introduced to model the feature representation of the node itself arranged in time series, so as to capture the temporal dependence of each node.

[0009] In some preferred embodiments, the spatiotemporal graph convolutional network model is trained using an integrated adversarial training strategy, the training process of which includes: A training set was constructed using knowledge graph sample data under historical normal and abnormal operating conditions to perform preliminary training on the spatiotemporal graph convolutional network model. The projective gradient descent method is used to perturb the node feature vectors of the knowledge graph samples in the training set to generate adversarial samples for the current model, and these adversarial samples are added to the training set to form an enhanced training dataset. Using the enhanced training dataset, the spatiotemporal graph convolutional network model is retrained with the goal of minimizing the joint loss of the model on normal and adversarial samples until the joint loss reaches the expected target.

[0010] In some preferred embodiments, the feature attribution analysis of the state data of the source anomaly node includes: The multi-dimensional state data of the source anomaly node is input into the trained prediction model, and the contribution value of each input feature to the model output is calculated. All features are sorted according to their contribution values, and the top K features with the largest contribution values ​​are selected as key root cause features, where K > 0; Based on the key root cause characteristics, a root cause explanation report is determined, which is used to characterize the degree of influence of multi-dimensional state data on the abnormal operating conditions.

[0011] In some preferred embodiments, the method further includes: Perform similarity matching between the current anomaly pattern characteristics and the historical case library; Retrieve a set of successful handling strategies for similar anomalies from the past, and conduct a comprehensive evaluation based on the effectiveness, cost, and suitability of the strategies for the current scenario; The top-scoring disposal strategies will be uploaded to the workshop production line management platform.

[0012] In some preferred embodiments, the method further includes: Based on the current state of the knowledge graph and the identified abnormal source nodes and propagation paths, a discrete event simulation model is constructed. The discrete event simulation model uses the abnormal probability score output by the spatiotemporal graph convolutional network model as the initial infection intensity and the edge weights in the knowledge graph as the intensity coefficient and probability of abnormal propagation. Run the discrete event simulation model to simulate the dynamic process of abnormal operating conditions propagating to subsequent process nodes over time steps, and predict the abnormal state changes of each node at each time step. Based on the results of the abnormal state changes, an abnormal impact assessment report is output. The abnormal impact assessment report is used to characterize the expected impact range and severity of the abnormal operating conditions on the workshop production line.

[0013] In a second aspect, the present invention proposes a full-process data processing and intelligent monitoring system for a workshop production line, comprising: The data acquisition module is used to collect asynchronous data streams from different sources throughout the entire production process; The preprocessing module is used to map each data point in the asynchronous data stream to a unified spatiotemporal reference framework, and bind each data point with the production work order number and process node to obtain standardized data units. The feature extraction module is used to construct a knowledge graph based on the standardized data unit, with production work orders as entities and process flow as relational chains. The nodes of the knowledge graph are used to embed real-time status features containing process flow, and the edge weights are used to represent the dependencies between processes. The model output module is used to input the knowledge graph into a pre-built spatiotemporal graph convolutional network model to obtain the anomaly probability score of each node in the knowledge graph and identify the potential propagation path of abnormal conditions between process nodes; the spatiotemporal graph convolutional network model calculates the anomaly probability score and propagation path by capturing the spatial and temporal dependencies in the production process. The report generation module is used to determine the abnormal probability nodes, and to backtrack along the relation edges in the knowledge graph to locate the source abnormal node. The module then performs feature attribution analysis on the state data of the source abnormal node to generate a root cause explanation report for the abnormal operating conditions.

[0014] The beneficial effects of this invention are: (1) Traditional methods often monitor individual equipment or processes in isolation, failing to grasp the interconnected effects of the entire process. This invention effectively integrates multi-source heterogeneous data from the equipment, operation, quality, and material layers through a unified spatiotemporal reference framework. Simultaneously, by constructing a knowledge graph with work orders as entities and process flows as relational chains, it achieves digital mapping and unified representation of the entire production process. This enables the monitoring system to gain a global perspective on the production status, understand the dependencies and constraints between processes, and achieve a leap from single-point monitoring to system-level monitoring, thereby enabling effective monitoring of the entire production process.

[0015] (2) This invention can automatically and quickly locate the source abnormal node by backtracking along the relation edges in the knowledge graph, which completely changes the inefficient mode of manual source tracing. Combined with the characteristic attribution analysis of the source node status data, it automatically generates a root cause explanation report with data support (e.g., "The vibration amplitude of the spindle at station 3 increased abnormally by 0.5μm at time XX, resulting in the hole diameter machining error at station 5"), freeing analysts from complicated data mining and allowing them to directly focus on the essence of the problem, effectively saving troubleshooting time on the production line. Attached Figure Description

[0016] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart illustrating a method for full-process data processing and intelligent monitoring of a workshop production line proposed in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of a workshop production line full-process data processing and intelligent monitoring system proposed in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a computer system used to implement the methods and system embodiments of this application. Detailed Implementation

[0017] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] like Figure 1 As shown, the first embodiment of the present invention provides a method for full-process data processing and intelligent monitoring of a workshop production line. The method includes steps S10-S50, each of which is described in detail below: Step S10: Collect asynchronous data streams from different sources throughout the entire production process; Step S20: Map each data point in the asynchronous data stream to a unified spatiotemporal reference framework, and bind each data point with the production work order number and process node to obtain a standardized data unit. Step S30: Based on the standardized data unit, construct a knowledge graph with production work orders as entities and process flow as relational chains. The nodes of the knowledge graph are used to embed real-time status features containing process flow, and the edge weights are used to characterize the dependencies between processes. Step S40: Input the knowledge graph into the pre-built spatiotemporal graph convolutional network model to obtain the anomaly probability score of each node in the knowledge graph and identify the potential propagation path of abnormal conditions between process nodes; the spatiotemporal graph convolutional network model calculates the anomaly probability score and propagation path by capturing the spatial and temporal dependencies in the production process. Step S50: Determine the abnormal probability node, and backtrack along the relation edges in the knowledge graph to locate the source abnormal node. Perform feature attribution analysis on the state data of the source abnormal node to generate a root cause explanation report of the abnormal working condition.

[0020] Compared to existing technologies that view individual data points or devices in isolation, this embodiment addresses the entire production process from a systemic and interconnected perspective.

[0021] Based on this embodiment, discrete and asynchronous production events are transformed into a structured, computable, and dynamically evolving knowledge graph in the digital world. On this basis, graph neural network technology is used to diagnose, monitor, and trace the entire production process.

[0022] In this embodiment, asynchronous data streams from different sources are collected throughout the entire production process. Specifically, this includes: At the equipment level, real-time operating parameters and status logs generated by PLCs, sensors (temperature, vibration, pressure), CNC numerical control systems, robot controllers, etc. are collected. For the operational layer, work order information, process start / end time, and personnel operation records are collected from the MES (Manufacturing Execution System). For the quality layer, the results of QC (quality control) inspection equipment and image data from the vision inspection system are collected. For the material layer, material flow information is collected from RFID and barcode scanners.

[0023] Furthermore, mapping each data point in the asynchronous data stream to a unified spatiotemporal reference framework includes: Define time offset and coordinate transformation rules for each type of data stream. The coordinate transformation rules are used to map the physical or logical location of the data source to a unified linear spatial coordinate system with the production line entrance as the origin. The timestamps of all data points are corrected using the time offset. The coordinate transformation rules are used to map data points to specific process nodes in the spatial coordinate system, so that each data point in the asynchronous data stream obtains a globally unique spatiotemporal coordinate (T, S), where T is a correction timestamp accurate to milliseconds and S is the spatial coordinate of the corresponding process node.

[0024] Furthermore, the construction of a knowledge graph based on the standardized data units, with production work orders as entities and process flows as relational chains, includes: An incremental graph update algorithm is used to continuously monitor the input standardized data units; For the new standardized data unit, its production order number and process node information are located to the corresponding nodes and edges in the knowledge graph; By using the node embedding update function, the new state features are fused with the original embedding vector of the node to update the real-time state representation of the node. At the same time, the edge weights between nodes are dynamically adjusted according to the flow time interval between the current batch and the previous batch in adjacent processes to represent the changes in material flow efficiency and dependence intensity between processes in real time.

[0025] Furthermore, the spatiotemporal graph convolutional network model captures the spatial and temporal dependencies in the production process, including: By employing graph convolution operations, the feature representation of the target node is updated by aggregating the state feature information of the target node's neighboring nodes, thereby obtaining the spatial dependency of the target node, where the target node is any node in the knowledge graph. At each node, a gated recurrent unit or a long short-term memory network is introduced to model the feature representation of the node itself arranged in time series, so as to capture the temporal dependence of each node.

[0026] Furthermore, the spatiotemporal graph convolutional network model is trained using an integrated adversarial training strategy, and its training process includes: A training set was constructed using knowledge graph sample data under historical normal and abnormal operating conditions to perform preliminary training on the spatiotemporal graph convolutional network model. The projective gradient descent method is used to perturb the node feature vectors of the knowledge graph samples in the training set to generate adversarial samples for the current model, and these adversarial samples are added to the training set to form an enhanced training dataset. Using the enhanced training dataset, the spatiotemporal graph convolutional network model is retrained with the goal of minimizing the joint loss of the model on normal and adversarial samples until the joint loss reaches the expected target.

[0027] Furthermore, the feature attribution analysis of the state data of the source abnormal node includes: The multi-dimensional state data of the source anomaly node is input into the trained prediction model, and the contribution value of each input feature to the model output is calculated. All features are sorted according to their contribution values, and the top K features with the largest contribution values ​​are selected as key root cause features, where K > 0; Based on the key root cause characteristics, a root cause explanation report is determined, which is used to characterize the degree of influence of multi-dimensional state data on the abnormal operating conditions.

[0028] Furthermore, the method also includes: Perform similarity matching between the current anomaly pattern characteristics and the historical case library; Retrieve a set of successful handling strategies for similar anomalies from the past, and conduct a comprehensive evaluation based on the effectiveness, cost, and suitability of the strategies for the current scenario; The top-scoring disposal strategies will be uploaded to the workshop production line management platform.

[0029] Furthermore, the method also includes: Based on the current state of the knowledge graph and the identified abnormal source nodes and propagation paths, a discrete event simulation model is constructed. The discrete event simulation model uses the abnormal probability score output by the spatiotemporal graph convolutional network model as the initial infection intensity and the edge weights in the knowledge graph as the intensity coefficient and probability of abnormal propagation. Run the discrete event simulation model to simulate the dynamic process of abnormal operating conditions propagating to subsequent process nodes over time steps, and predict the abnormal state changes of each node at each time step. Based on the results of the abnormal state changes, an abnormal impact assessment report is output. The abnormal impact assessment report is used to characterize the expected impact range and severity of the abnormal operating conditions on the workshop production line.

[0030] Although the steps in the above embodiments are described in the above order, those skilled in the art will understand that in order to achieve the effect of this embodiment, different steps do not need to be executed in such an order. They can be executed simultaneously (in parallel) or in a reverse order. These simple variations are all within the protection scope of this invention.

[0031] The second embodiment of the present invention proposes a full-process data processing and intelligent monitoring system for a workshop production line, which is used to implement the method involved in the first embodiment above.

[0032] The aforementioned system can be a software module, comprising several instructions stored in memory. The processor can access this memory, invoke the instructions, and execute them to complete the full-process data processing and intelligent monitoring method for the workshop production line described in the various embodiments above. In some embodiments, the full-process data processing and intelligent monitoring system for the workshop production line can also be built from hardware devices. For example, the system can be built from one or more chips, which can coordinate with each other to complete the full-process data processing and intelligent monitoring method for the workshop production line described in the various embodiments above. Furthermore, the system can also be built from various logic devices, such as general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), microcontrollers, ARM (AcornRISC) or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination of these components.

[0033] Please refer to Figure 2 , Figure 2 The diagram shows the structure of a full-process data processing and intelligent monitoring system for the workshop production line, including: The data acquisition module 210 is used to collect asynchronous data streams from different sources throughout the entire production process; The preprocessing module 220 is used to map each data point in the asynchronous data stream to a unified spatiotemporal reference framework, and bind each data point with the production work order number and process node to obtain standardized data units. The feature extraction module 230 is used to construct a knowledge graph based on the standardized data unit, with production work orders as entities and process flow as relational chains. The nodes of the knowledge graph are used to embed real-time status features containing process flow, and the edge weights are used to characterize the dependency relationship between process flow. The model output module 240 is used to input the knowledge graph into a pre-built spatiotemporal graph convolutional network model to obtain the anomaly probability score of each node in the knowledge graph and identify the potential propagation path of abnormal conditions between process nodes; the spatiotemporal graph convolutional network model calculates the anomaly probability score and propagation path by capturing the spatial and temporal dependencies in the production process. The report generation module 250 is used to determine the abnormal probability node, and to backtrack along the relation edges in the knowledge graph to locate the source abnormal node, perform feature attribution analysis on the state data of the source abnormal node, and generate a root cause explanation report of the abnormal working condition.

[0034] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the system described above can be found in the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0035] It should be noted that the workshop production line full-process data processing and intelligent monitoring system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiments can be merged into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing the various modules or steps and are not considered as an improper limitation of the present invention.

[0036] The third embodiment of the present invention provides a computer-readable storage medium storing computer instructions, which are executed by the computer to realize the above-described method for full-process data processing and intelligent monitoring of the workshop production line.

[0037] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the electronic device and computer-readable storage medium described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0038] Those skilled in the art will recognize that the modules and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. The programs corresponding to the software modules and method steps can be placed in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art. To clearly illustrate the interchangeability of electronic hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the invention.

[0039] The following is for reference. Figure 3 It shows a schematic diagram of the structure of a computer system used to implement the methods and system embodiments of this application. Figure 3The server shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0040] like Figure 3 As shown, the computer system includes a Central Processing Unit (CPU) 301, which can perform various appropriate actions and processes based on programs stored in Read Only Memory (ROM) 302 or programs loaded from storage section 308 into Random Access Memory (RAM) 303. The RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An Input / Output (I / O) interface 305 is also connected to the bus 304.

[0041] The following components are connected to the input / output interface 305: an input section 306 including a keyboard, mouse, etc.; an output section 307 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on the drive 310 as needed so that computer programs read from them can be installed into the storage section 308 as needed.

[0042] Specifically, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by central processing unit 301, it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, system, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, system, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, system, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0043] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0044] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0045] The terms “first”, “second”, etc., are used to distinguish similar objects, not to describe or indicate a specific order or sequence.

[0046] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to those processes, methods, articles, or apparatus / systems.

[0047] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for full-process data processing and intelligent monitoring of a workshop production line, characterized in that, include: Collect asynchronous data streams from different sources throughout the entire production process; Each data point in the asynchronous data stream is mapped to a unified spatiotemporal reference framework, and each data point is bound to the production work order number and process node to obtain a standardized data unit. Based on the standardized data units, a knowledge graph is constructed with production work orders as entities and process flows as relational chains. The nodes of the knowledge graph are used to embed real-time status features containing process flows, and the edge weights are used to characterize the dependencies between process flows. The knowledge graph is input into a pre-built spatiotemporal graph convolutional network model to obtain the anomaly probability score of each node in the knowledge graph and identify the potential propagation path of abnormal conditions between process nodes; the spatiotemporal graph convolutional network model calculates the anomaly probability score and propagation path by capturing the spatial and temporal dependencies in the production process. Identify the abnormal probability nodes, and backtrack along the relation edges in the knowledge graph to locate the source abnormal node. Perform feature attribution analysis on the state data of the source abnormal node to generate a root cause explanation report for the abnormal operating conditions.

2. The method for full-process data processing and intelligent monitoring of a workshop production line according to claim 1, characterized in that, The step of mapping each data point in the asynchronous data stream to a unified spatiotemporal reference framework includes: Define time offset and coordinate transformation rules for each type of data stream. The coordinate transformation rules are used to map the physical or logical location of the data source to a unified linear spatial coordinate system with the production line entrance as the origin. The timestamps of all data points are corrected using the time offset. The coordinate transformation rules are used to map data points to specific process nodes in the spatial coordinate system, so that each data point in the asynchronous data stream obtains a globally unique spatiotemporal coordinate (T, S), where T is a correction timestamp accurate to milliseconds and S is the spatial coordinate of the corresponding process node.

3. The method for full-process data processing and intelligent monitoring of a workshop production line according to claim 1, characterized in that, The construction of a knowledge graph based on the standardized data units, with production work orders as entities and process flows as relational chains, includes: An incremental graph update algorithm is used to continuously monitor the input standardized data units; For the new standardized data unit, its production order number and process node information are located to the corresponding nodes and edges in the knowledge graph; By using the node embedding update function, the new state features are fused with the original embedding vector of the node to update the real-time state representation of the node. At the same time, the edge weights between nodes are dynamically adjusted according to the flow time interval between the current batch and the previous batch in adjacent processes to represent the changes in material flow efficiency and dependence intensity between processes in real time.

4. The method for full-process data processing and intelligent monitoring of a workshop production line according to claim 1, characterized in that, The spatiotemporal graph convolutional network model captures the spatial and temporal dependencies in the production process, including: By employing graph convolution operations, the feature representation of the target node is updated by aggregating the state feature information of the target node's neighboring nodes, thereby obtaining the spatial dependency of the target node, where the target node is any node in the knowledge graph. At each node, a gated recurrent unit or a long short-term memory network is introduced to model the feature representation of the node itself arranged in time series, so as to capture the temporal dependence of each node.

5. A method for full-process data processing and intelligent monitoring of a workshop production line according to claim 1 or 4, characterized in that, The spatiotemporal graph convolutional network model is trained using an integrated adversarial training strategy, and its training process includes: A training set was constructed using knowledge graph sample data under historical normal and abnormal operating conditions to perform preliminary training on the spatiotemporal graph convolutional network model. The projective gradient descent method is used to perturb the node feature vectors of the knowledge graph samples in the training set to generate adversarial samples for the current model, and these adversarial samples are added to the training set to form an enhanced training dataset. Using the enhanced training dataset, the spatiotemporal graph convolutional network model is retrained with the goal of minimizing the joint loss of the model on normal and adversarial samples until the joint loss reaches the expected target.

6. The method for full-process data processing and intelligent monitoring of a workshop production line according to claim 1, characterized in that, The feature attribution analysis of the state data of the source abnormal node includes: The multi-dimensional state data of the source anomaly node is input into the trained prediction model, and the contribution value of each input feature to the model output is calculated. All features are sorted according to their contribution values, and the top K features with the largest contribution values ​​are selected as key root cause features, where K > 0; Based on the key root cause characteristics, a root cause explanation report is determined, which is used to characterize the degree of influence of multi-dimensional state data on the abnormal operating conditions.

7. The method for full-process data processing and intelligent monitoring of a workshop production line according to claim 1, characterized in that, The method further includes: Perform similarity matching between the current anomaly pattern characteristics and the historical case library; Retrieve a set of successful handling strategies for the current anomaly in the past, and conduct a comprehensive evaluation based on the effectiveness, cost and adaptability of the strategy to the current scenario; The top-scoring disposal strategies will be uploaded to the workshop production line management platform.

8. The method for full-process data processing and intelligent monitoring of a workshop production line according to claim 7, characterized in that, The method further includes: Based on the current state of the knowledge graph and the identified abnormal source nodes and propagation paths, a discrete event simulation model is constructed. The discrete event simulation model uses the abnormal probability score output by the spatiotemporal graph convolutional network model as the initial infection intensity and the edge weights in the knowledge graph as the intensity coefficient and probability of abnormal propagation. Run the discrete event simulation model to simulate the dynamic process of abnormal operating conditions propagating to subsequent process nodes over time steps, and predict the abnormal state changes of each node at each time step. Based on the results of the abnormal state changes, an abnormal impact assessment report is output. The abnormal impact assessment report is used to characterize the expected impact range and severity of the abnormal operating conditions on the workshop production line.

9. A full-process data processing and intelligent monitoring system for a workshop production line, characterized in that, include: The data acquisition module is used to collect asynchronous data streams from different sources throughout the entire production process; The preprocessing module is used to map each data point in the asynchronous data stream to a unified spatiotemporal reference framework, and bind each data point with the production work order number and process node to obtain standardized data units. The feature extraction module is used to construct a knowledge graph based on the standardized data unit, with production work orders as entities and process flow as relational chains. The nodes of the knowledge graph are used to embed real-time status features containing process flow, and the edge weights are used to represent the dependencies between processes. The model output module is used to input the knowledge graph into a pre-built spatiotemporal graph convolutional network model to obtain the anomaly probability score of each node in the knowledge graph and identify the potential propagation path of abnormal conditions between process nodes; the spatiotemporal graph convolutional network model calculates the anomaly probability score and propagation path by capturing the spatial and temporal dependencies in the production process. The report generation module is used to determine the abnormal probability nodes, and to backtrack along the relation edges in the knowledge graph to locate the source abnormal node. The module then performs feature attribution analysis on the state data of the source abnormal node to generate a root cause explanation report for the abnormal operating conditions.

Images