Simulation method for automatic equipment control system

Abstract

The invention relates to the technical field of control system simulation, in particular to a simulation method for an automatic equipment control system. The method comprises the following steps: carrying out multi-source sensing data acquisition on a device group to obtain a device operation original data set; performing timestamp synchronization on the equipment operation original data set to obtain an equipment operation synchronization data set; performing heterogeneous information fusion on the equipment operation synchronous data set to obtain an equipment operation feature matrix; performing spatial relationship calculation on the equipment group based on the equipment operation feature matrix to obtain an initial topological graph of the equipment group; performing communication strength weight analysis on the initial topological graph of the equipment group to obtain a communication weight matrix between the equipment; and dynamically updating the device group initial topological graph according to the inter-device communication weight matrix to obtain a device group dynamic topological graph. According to the invention, the simulation precision of the automation equipment control system in a complex dynamic environment is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of control system simulation, and particularly to a simulation method for an automatic equipment control system. Background Art

[0002] In the field of logistics and warehousing, the efficient cooperation of automated equipment is the key to achieving rapid goods turnover and precise distribution. However, existing simulation technologies have significant defects in modeling equipment interactions in a dynamic environment, and this problem is particularly prominent in logistics and warehousing scenarios. Taking an Automatic Guided Vehicle (AGV) cluster as an example, when it executes real-time path planning in a warehouse, the interactions between devices are extremely complex. AGVs need to continuously exchange information such as position, speed, and task priority to achieve efficient path planning and task allocation. However, existing simulation methods cannot accurately represent the asymmetric information interaction characteristics between these devices. In actual operation, the communication delays between AGVs often vary due to differences in device performance, network layout, and task load, and this difference is often ignored in the simulation model, resulting in a large deviation between the simulation results and the actual operation situation. This spatio-temporal discretization modeling error of information interaction not only affects the cooperation efficiency of devices in normal operation, but also exposes serious reliability problems when dealing with sudden order peaks. In logistics and warehousing scenarios, fluctuations in order volume are normal, especially during e-commerce promotion periods, when the order volume will increase sharply within a short period of time. At this time, the AGV cluster needs to quickly respond and re-plan the task path to ensure that goods can be shipped out in a timely manner. However, due to the inability of the simulation system to accurately simulate the real interaction characteristics between devices, when encountering a sudden order peak in actual operation, the reliability of the system will drop by 83%. This means that in the case of a sharp increase in order volume, the cooperation efficiency of the AGV cluster cannot be effectively guaranteed, resulting in goods backlog and delivery delays. Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide a simulation method for an automatic equipment control system to solve at least one of the above technical problems.

[0004] To achieve the above object, a simulation method for an automatic equipment control system includes the following steps:

[0005] Step S1: Collect multi-source sensing data of the device group to obtain an original device operation data set; synchronize the timestamps of the original device operation data set to obtain a synchronized device operation data set; fuse heterogeneous information of the synchronized device operation data set to obtain a device operation feature matrix;

[0006] Step S2: Calculate the spatial relationship of the device group based on the device operation feature matrix to obtain the initial topology graph of the device group; conduct a communication intensity weight analysis on the initial topology graph of the device group to obtain the communication weight matrix between devices; dynamically update the initial topology graph of the device group according to the communication weight matrix between devices to obtain the dynamic topology graph of the device group;

[0007] Step S3: Construct the communication characteristic tensor of the dynamic topology graph of the device group to obtain the device communication characteristic data set; calculate the information flow routing based on the device communication characteristic data set to obtain the routing policy table between devices; conduct a communication delay distribution modeling according to the routing policy table between devices to obtain the interaction characteristic model of the device group;

[0008] Step S4: Conduct a hierarchical construction of the task buffer pool for the interaction characteristic model of the device group to obtain the device task buffer pool configuration data; perform task coding compression on the device group according to the device task buffer pool configuration data to obtain the device task allocation plan; conduct a load status evaluation on the device task allocation plan to obtain the system load warning index set;

[0009] Step S5: Based on the interaction characteristic model of the device group, conduct a simulation operation on the device group according to the system load warning index set to obtain the device simulation performance evaluation report; conduct an error analysis on the device simulation performance evaluation report to obtain the device simulation parameter optimization plan; obtain the spatio-temporal feature fusion model of the device group, and update the spatio-temporal feature fusion model of the device group according to the device simulation parameter optimization plan to obtain the device control simulation model.

[0010] Through multi-source sensing data acquisition, timestamp synchronization, and heterogeneous information fusion, the present invention can accurately characterize the complex interaction characteristics between devices, including the dynamic exchange of position, speed, and task priority information. This solves the problem in the prior art that the asymmetric information interaction between devices cannot be accurately simulated. Through the dynamic update of the topology graph, the communication state and the change of spatial relationship between devices can be reflected in real time. Through the construction of the communication characteristic tensor and the modeling of the communication delay distribution, the communication delay between devices can be accurately predicted. This enables the control system to fully consider the impact of communication delay on device collaboration during the simulation phase, so as to better handle complex situations of sudden order peaks during actual operation and avoid collaboration failures and task delays caused by communication delays. Through the hierarchical construction of the task buffer pool, task coding compression, and load status evaluation, the system load can be monitored in real time. Through the error analysis and parameter optimization of the device simulation performance evaluation report, combined with the incremental learning update of the spatio-temporal feature fusion model, the parameters and structure of the simulation model can be continuously optimized. This enables the simulation model to more accurately reflect the actual operation of the device, reduce the deviation between the simulation result and the actual operation, and improve the simulation accuracy. Through the multi-threaded parallel simulation of the initial configuration scheme of the device group simulation, the device group simulation operation trajectory set and performance evaluation report can be quickly generated. This enables the system to complete the simulation analysis of multiple operating conditions in a short time. Through coding compression, the storage and transmission volume of task data can be reduced, and the system's requirements for storage and communication resources can be lowered. In summary, the present invention improves the simulation accuracy of automated devices in complex dynamic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Other features, objects, and advantages of the present invention will become more apparent by reading the detailed description with reference to the following drawings:

[0012] Figure 1 The flowchart of the steps of a simulation method for an automatic device control system according to an embodiment is shown.

[0013] Figure 2 The detailed flowchart of step S41 according to an embodiment is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] The technical method of the present invention will be clearly and completely described below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0015] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0016] It should be understood that although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0017] To achieve the above object, please refer to Figures 1 to 2 , the present invention provides a simulation method for an automatic device control system, including the following steps:

[0018] Step S1: Collect multi-source sensing data for the device group to obtain the original device operation data set; synchronize the time stamps of the original device operation data set to obtain the synchronized device operation data set; perform heterogeneous information fusion on the synchronized device operation data set to obtain the device operation feature matrix;

[0019] Step S2: Calculate the spatial relationship of the device group based on the device operation feature matrix to obtain the initial topology map of the device group; perform communication intensity weight analysis on the initial topology map of the device group to obtain the communication weight matrix between devices; dynamically update the initial topology map of the device group according to the communication weight matrix between devices to obtain the dynamic topology map of the device group;

[0020] Step S3: Construct the communication characteristic tensor of the dynamic topology map of the device group to obtain the device communication characteristic data set; calculate the information flow routing based on the device communication characteristic data set to obtain the routing policy table between devices; perform communication delay distribution modeling according to the routing policy table between devices to obtain the interaction characteristic model of the device group;

[0021] Step S4: Hierarchically construct the task buffer pool of the interaction characteristic model of the device group to obtain the device task buffer pool configuration data; perform task coding compression on the device group according to the device task buffer pool configuration data to obtain the device task allocation scheme; evaluate the load status of the device task allocation scheme to obtain the system load warning index set;

[0022] Step S5: Based on the device group interaction characteristic model, simulate the operation of the device group according to the system load warning index set to obtain a device simulation performance evaluation report; conduct error analysis on the device simulation performance evaluation report to obtain a device simulation parameter optimization plan; obtain the device group spatio-temporal feature fusion model, and update the device group spatio-temporal feature fusion model according to the device simulation parameter optimization plan to obtain a device control simulation model.

[0023] In this embodiment, first, an industrial-grade sensor network (such as temperature, vibration, current, and pressure sensors) is used to collect the original device operation data set, and timestamp synchronization processing is performed through MATLAB to ensure data time consistency. Then, the principal component analysis (PCA) method is used to fuse heterogeneous information in the synchronized data to generate a device operation feature matrix. Based on this matrix, the initial topology graph of the device group is constructed using the MATLAB graph theory toolbox and three-dimensional space scanning technology, and the communication weight matrix is calculated by analyzing communication traffic and bandwidth utilization, and then the topology graph is dynamically updated. Subsequently, a communication characteristic tensor is constructed, the information flow routing is calculated, and an interaction characteristic model is established. On this basis, a hierarchical construction of the task buffer pool and task coding compression are performed to form a task allocation plan, and the load status is evaluated to generate a warning index set. MATLAB Simulink is used for simulation operation to generate a performance evaluation report, and the simulation parameters are optimized through error analysis. Finally, combined with the device simulation parameter optimization plan, incremental learning update is performed on the spatio-temporal feature fusion model to obtain a device control simulation model.

[0024] Preferably, step S1 includes the following steps:

[0025] Step S11: Monitor the thermal field distribution during the operation of the device group to obtain a device temperature field data set;

[0026] Specifically, a suitable infrared thermal imager can be selected, such as the FLIR T640 model. The infrared thermal imager is installed at a fixed position in the warehousing system so that it can cover the main operation area of the device group. During the operation of the device, the infrared thermal imager collects the temperature data on the surface of the device at a frequency of 30 frames per second and transmits the data to the data acquisition server through a wireless network. The data processing software on the server performs real-time analysis on the collected temperature data to generate a two-dimensional thermal map of the device temperature field. By analyzing the thermal map, the temperature distribution of different parts of the device during operation can be clearly seen, thereby obtaining a device temperature field data set.

[0027] Step S12: Collect the vibration characteristics of the device group to obtain a device vibration signal time series data set;

[0028] Specifically, a suitable acceleration sensor can be selected, such as the ADXL355 model. Multiple acceleration sensors are respectively installed on the key components of the device, such as the drive wheels and motors of the AGV. During the operation of the device, the acceleration sensors collect vibration signals at a sampling frequency of 1000 times per second and transmit the data to the data acquisition module through a wired connection. The data acquisition module performs preliminary processing on the collected vibration signals, including filtering and noise reduction operations. The processed vibration signals are converted into time series data and stored in the database to form a time series dataset of device vibration signals.

[0029] Step S13: Perform current sampling on each device execution component in the device group to obtain a device current feature dataset;

[0030] Specifically, a suitable current transformer can be selected, such as the current transformer with the model CT-500, whose measurement range is 0 - 500A and the accuracy is 0.2 level, which can meet the current measurement requirements of most industrial devices. The current transformer is installed on the power supply line of the device execution component to ensure that it can accurately measure the current passing through the device. At the same time, use the NI USB-6341 data acquisition card, which has a high sampling rate and multi-channel input function and can collect the analog signals output by the current transformer in real time. During the operation of the device, the current transformer converts the current signal of the device execution component into a voltage signal, and the data acquisition card samples the voltage signal at a sampling frequency of 1000 times per second and transmits the data to the computer through the USB interface. The data processing software on the computer filters and normalizes the collected current data to eliminate noise and standardize the data format. The processed current data is stored in the database to form a device current feature dataset.

[0031] Step S14: Monitor the pressure distribution during the operation of the device group to obtain a device pressure field dataset;

[0032] Specifically, a suitable pressure sensor can be selected, such as the pressure sensor with the model MPX5700AP, whose measurement range is 0 - 50 psi and the accuracy is 0.5%. Multiple pressure sensors are installed on the key parts of the device, such as the tires of the AGV and the pipes of the hydraulic system. During the operation of the device, the pressure sensors collect pressure data at a sampling frequency of 50 times per second and transmit the data to the central control unit through a wireless communication module (such as the ZigBee module). The central control unit performs real-time analysis on the received pressure data, including data calibration and anomaly detection. The processed pressure data is stored in the database of the industrial-grade server to form a device pressure field dataset.

[0033] Step S15: Denote the device temperature field dataset, the device vibration signal time series dataset, the device current feature dataset, and the device pressure field dataset as the original device operation dataset, and synchronize the timestamps of the original device operation dataset to obtain the synchronized device operation dataset;

[0034] Specifically, a Dell PowerEdge R740 server can be selected as the data processing center. At the same time, the SQLServer database management system is used to store and manage various types of data. At the data acquisition end, data acquisition devices such as infrared thermal imagers, acceleration sensors, current transformers, and pressure sensors are connected to the server through a network switch. During the acquisition process, a unified timestamp generator, such as a GPS clock synchronization module with an accuracy of up to 1 millisecond, is configured for each device's data acquisition module. When the device operation data is acquired, the data acquisition module attaches the timestamp to each data point and transmits the data to the server. The data synchronization software (such as Apache Kafka) on the server receives and processes this timestamped data, aligns and integrates the data from different sources according to the timestamp, and forms the original device operation dataset.

[0035] Step S16: Perform heterogeneous information fusion on the synchronized device operation dataset to obtain the device operation feature matrix.

[0036] Specifically, MATLAB can be selected for heterogeneous information fusion. During the data fusion process, first, normalize the temperature, vibration, current, and pressure data in the synchronized device operation dataset. For example, use the Z-Score normalization method to normalize the current and pressure data. The preprocessed data is imported into the MATLAB environment, and by writing a custom fusion algorithm, different types of feature data are fused. The fusion algorithm adopts the principal component analysis (PCA) method to extract the main features from each dataset and map them into a unified feature space. For example, the principal components of the temperature data, the principal components of the vibration signal, the principal components of the current feature, and the principal components of the pressure data are weighted and combined to form a comprehensive feature vector. Through the above steps, the heterogeneous information is fused into a device operation feature matrix, which can comprehensively reflect the multi-dimensional state information of the device during operation.

[0037] Through multi-dimensional monitoring of the operating status of the device group (including key features of the thermal field, vibration, current, and pressure), timestamp synchronization, and heterogeneous information fusion, the present invention can comprehensively and accurately capture various physical states and behavioral characteristics of the device during operation. This helps to more realistically reflect the operating conditions of the device under actual complex working conditions.

[0038] Preferably, step S2 includes the following steps:

[0039] Step S21: Perform a three-dimensional space scan on the device group to obtain device operation scenario point cloud data;

[0040] Specifically, a three-dimensional laser scanner FARO Focus3D X 330 can be installed at a fixed position in the warehousing system to ensure that it can cover the main operation area of the device group. During the scanning process, the scanner performs an omnidirectional scan of the device and its surrounding environment at a scanning rate of 1 million points per second, generating high-density point cloud data. After the scanning is completed, the FAROSCENE software is used to preprocess the point cloud data, including operations such as denoising, filtering, and point cloud stitching. For example, by removing abnormal points and noise points, the quality of the point cloud data is improved; by stitching the point cloud data from multiple perspectives, the complete device operation scenario point cloud data is generated.

[0041] Step S22: Perform multi-angle image acquisition of the operation scenario of the device group to obtain a multi-perspective image set of the operation scenario, and register and fuse the device operation scenario point cloud data and the multi-perspective image set of the operation scenario to obtain a three-dimensional mapping model of the operation scenario;

[0042] Specifically, a Basler ace series camera can be selected for multi-angle image acquisition of the operation scenario, with a resolution of 2048×2048 pixels and a frame rate of up to 50 frames per second. In the device operation scenario, multiple cameras are arranged to take pictures of the device and its operation environment from different angles respectively. For example, 4 cameras are arranged to take pictures from the front, side, top, and back respectively to obtain multi-perspective images of the device. During the shooting process, the cameras synchronously acquire images through a trigger signal. The acquired image data is transmitted to the data processing server. MATLAB and OpenCV libraries are used for image processing and registration fusion. First, the multi-perspective images acquired are corrected and de-distorted. Feature points in the images are extracted through a feature point matching algorithm (such as the SIFT or ORB algorithm), and the transformation matrix between the images is calculated. The multi-perspective images are registered using the transformation matrix and aligned to the same coordinate system. Finally, a three-dimensional mapping model of the operation scenario is generated.

[0043] Step S23: Perform feature space mapping projection on the device operation feature matrix and the three-dimensional mapping model of the operation scenario to obtain a spatio-temporal feature fusion model of the device group;

[0044] Specifically, Autodesk Maya can be selected to process the stereoscopic mapping model of the operating scenario. At the same time, MATLAB is used as the main data processing tool. During the data processing, the device operation feature matrix (including temperature, vibration, current, and pressure features) is imported into the MATLAB environment. These feature data are mapped to the corresponding positions of the stereoscopic mapping model of the operating scenario. For example, according to the spatial position information of the devices, the operation feature data of each device are matched with the point cloud data in the stereoscopic mapping model, and the feature data are projected to the corresponding positions of the model. Finally, a spatio-temporal feature fusion model containing device operation features and spatial position information is generated.

[0045] Step S24: Calculate the spatial distances between devices in the device group based on the spatio-temporal feature fusion model of the device group to obtain the inter-device communication distance matrix, and perform spatial clustering on the device group according to the inter-device communication distance matrix to obtain the device group spatial distribution data;

[0046] Specifically, MATLAB can be selected to process the spatial data in the spatio-temporal feature fusion model. In MATLAB, the Euclidean distance formula is used to calculate the spatial distances between devices. For example, for each pair of devices in the device group, according to their three-dimensional coordinates in the spatio-temporal feature fusion model, the straight-line distance between them is calculated, and these distances are stored as the inter-device communication distance matrix. The K-means clustering algorithm is used to perform spatial clustering on the device group. During the clustering process, an appropriate number of clustering centers is selected (for example, according to the scale and distribution of the device group, 3 to 5 clustering centers are selected). By iteratively optimizing the positions of the clustering centers, the device group is divided into multiple spatial distribution regions. Finally, the clustering results are stored as the device group spatial distribution data, including the central positions, device members, and boundary range information of each cluster.

[0047] Step S25: Construct a minimum spanning tree for the device group spatial distribution data to obtain the basic connection graph of the device group, and supplement redundant links to the device group based on the basic connection graph of the device group to obtain the network connectivity graph of the device group;

[0048] Specifically, the spatial distribution data of the device group can be imported into Gephi, including the location of each device and the clustering information it belongs to. Based on this data, the Prim algorithm is used to construct a minimum spanning tree (MST). The Prim algorithm starts from a starting node and gradually selects the nearest node to join the spanning tree until all nodes are connected. For example, an AGV located in the center of the warehouse is selected as the starting node. By calculating the communication distance between it and other devices, the device with the nearest distance is gradually added to the spanning tree, and finally a basic connection graph covering all devices is formed. By analyzing the connectivity and reliability of the network, appropriate links are selected for redundant supplementation. For example, for devices that are close but not directly connected, additional communication links are added. Finally, the network after supplementing redundant links is saved as the network connectivity graph of the device group.

[0049] Step S26: Perform topological reconstruction on the network connectivity graph of the device group to obtain the initial topological graph of the device group;

[0050] Specifically, the data of the network connectivity graph of the device group can be imported into the MATLAB environment, including node positions, link information, and the weights of redundant links. Graph theory algorithms are used to perform topological reconstruction on the network connectivity graph. For example, the Kruskal algorithm is used to optimize the network. This algorithm gradually constructs a spanning tree by selecting the edge with the smallest weight, while avoiding forming loops. During the topological reconstruction process, the link weights are adjusted according to the communication intensity and task requirements between devices. For example, for device pairs with frequent communication and high task priorities, higher weights are assigned to preferentially retain or strengthen the links between them. Through multiple iterative optimizations, the initial topological graph of the device group is finally obtained.

[0051] Step S27: Perform an analysis of the communication intensity weights on the initial topological graph of the device group to obtain the communication weight matrix between devices, and dynamically update the initial topological graph of the device group according to the communication weight matrix between devices to obtain the dynamic topological graph of the device group.

[0052] Specifically, for the detailed implementation process of this embodiment, please refer to the sub-steps of Step S27.

[0053] Through three-dimensional space scanning, multi-angle image acquisition, and registration and fusion, the present invention constructs a three-dimensional mapping model of the operating scenario, which ensures that the positional relationships of devices in complex scenarios can be accurately captured. Through feature space mapping projection, the deep fusion of the device operating state and spatial position information is realized. Through spatial distance calculation and spatial clustering, the spatial distribution of the device group can be scientifically divided, providing a reasonable basis for the construction of network connections. Through the construction of a minimum spanning tree and redundant link supplementation, the communication connectivity between devices is ensured, while enhancing the robustness of the network. Finally, through the analysis of communication intensity weights and dynamic topology update, the topology structure can be adjusted in real time according to the actual communication capabilities and task requirements between devices.

[0054] Preferably, step S27 includes the following steps:

[0055] Step S271: Collect the inter-device communication traffic of the device group to obtain an inter-device communication traffic data set, and perform packet statistics on the inter-device communication traffic data set to obtain an inter-device communication frequency matrix;

[0056] Specifically, in the network environment of the device group, Wireshark can be deployed on key network nodes, such as the mirror port of a switch. Set the filtering rules of Wireshark to focus on capturing the communication packets between devices, such as filtering based on the IP address or MAC address of the device. During the data collection process, Wireshark captures the communication traffic at a speed of thousands of packets per second and saves it as a file in the.pcap format. Subsequently, use a Python script in combination with the scapy library to process the captured packets. By writing a Python script, read the.pcap file, extract the source address, destination address, and timestamp information of each packet. Based on this information, count the communication frequency between devices, such as calculating the number of packets exchanged between each pair of devices per unit time (e.g., per minute). Finally, organize the statistical results into an inter-device communication frequency matrix, where each element in the matrix represents the communication frequency between a pair of devices.

[0057] Step S272: Perform bandwidth utilization statistics on the inter-device communication frequency matrix to obtain a device communication bandwidth occupancy matrix, and allocate communication priorities to the device group according to the device communication bandwidth occupancy matrix to obtain a device communication priority weight set;

[0058] Specifically, the inter-device communication frequency matrix can be imported into the MATLAB environment, and at the same time, obtain the communication bandwidth information between devices, which can be obtained through the management interface of network devices (such as switches). For example, query the bandwidth usage of the switch port through SNMP (Simple Network Management Protocol) to obtain the bandwidth utilization rate of each device port. In MATLAB, write a script to calculate the bandwidth occupancy between each pair of devices. According to the communication frequency matrix and bandwidth information, calculate the bandwidth occupancy of each pair of devices per unit time, such as through the formula: bandwidth occupancy = communication frequency × average packet size. Organize the calculation results into a device communication bandwidth occupancy matrix, where each element in the matrix represents the bandwidth occupancy between a pair of devices. Next, allocate communication priorities to the device group according to the bandwidth occupancy matrix. For example, set the priority allocation rule: device pairs with a bandwidth occupancy exceeding a certain threshold (e.g., 80%) are assigned a high priority, device pairs with a bandwidth occupancy between 50% - 80% are assigned a medium priority, and device pairs with a bandwidth occupancy below 50% are assigned a low priority. Through the above steps, a device communication priority weight set is obtained.

[0059] Step S273: Based on the spatio-temporal feature fusion model of the device group, perform reliability evaluation on the device communication priority weight set to obtain the device group link reliability index, and based on the device group link reliability index, perform communication weight allocation on the device group to obtain the inter-device communication weight matrix;

[0060] Specifically, the spatio-temporal feature fusion model of the device group and the communication priority weight set can be imported into the MATLAB environment. Based on the spatio-temporal feature fusion model, key features of the device operation state are extracted, such as the temperature, vibration, current, and pressure of the device. At the same time, combined with the communication priority weight set, a reliability evaluation model is used to perform reliability evaluation on each communication link. For example, the Failure Mode and Effects Analysis (FMEA) method is adopted, and combined with the historical failure data and operation state characteristics of the device, the reliability index of each link is calculated. Specifically, for a communication link with high priority, if the corresponding device is in good operation state (such as temperature and vibration within the normal range), a higher reliability index is given; otherwise, if the device operation state is poor, its reliability index is reduced. Finally, communication weight allocation is performed on the device group according to the reliability index.

[0061] Step S274: Dynamically update the topological structure of the initial device group topology diagram according to the inter-device communication weight matrix to obtain the device group topology update strategy, and verify the feasibility of the device group topology update strategy to obtain the device group topology adjustment plan;

[0062] Specifically, the inter-device communication weight matrix can be imported into Gephi and visualized in combination with the initial topology diagram. In Gephi, the initial topology diagram is adjusted according to the communication weight matrix. For example, the bandwidth of the link with high weight is increased, the bandwidth of the link with low weight is reduced, and even some links with extremely low weight are removed. In this way, the device group topology update strategy is generated. Next, MATLAB is used to verify the feasibility of the topology update strategy. In MATLAB, a script is written to simulate the communication traffic between devices, and performance indicators such as communication delay and packet loss rate are calculated according to the updated topological structure. For example, set the simulation scenario: during the peak order period, the communication traffic between AGVs increases significantly, and verify whether the updated topological structure can meet the communication requirements in this case. Through multiple simulations, the performance of the topology update strategy is evaluated. If the simulation results show that the updated topological structure performs well in terms of communication delay and packet loss rate, etc., then the topology update strategy is considered feasible; otherwise, the topological structure needs to be further adjusted. Finally, the device group topology adjustment plan is obtained.

[0063] Step S275: Based on the device group topology adjustment scheme, incrementally update the link structure of the initial device group topology graph to obtain the device group dynamic link structure data, and perform a topology structure reconstruction mapping on the device group dynamic link structure data to obtain the device group dynamic topology graph.

[0064] Specifically, the device group topology adjustment scheme can be imported into the MATLAB environment. This scheme includes link weight adjustment, link addition or deletion operations. According to the adjustment scheme, incrementally update the link structure in the initial topology graph. For example, for a link with an increased weight, improve its communication ability by adjusting the link's bandwidth or communication protocol; for a link with a decreased weight, appropriately reduce its bandwidth or optimize the communication path. During the update process, record the attribute changes of each link, including bandwidth, delay, and reliability parameters, to form the device group dynamic link structure data. Next, use Gephi to visualize the updated link structure data. In Gephi, import the dynamic link structure data and adjust the display style of the links, such as color and thickness, according to the link's weight and attributes. Utilize Gephi's topology analysis function to perform a reconstruction mapping on the updated link structure. For example, through a re-layout algorithm (such as force-directed layout or circular layout), optimize the positions of the devices in the topology graph to make it more in line with the actual operating scenario and communication requirements. Finally, obtain the device group dynamic topology graph, which can reflect the communication link status and dynamic changes between devices in real time.

[0065] Through the collection and analysis of the communication traffic between devices, combined with bandwidth utilization statistics and communication priority allocation, the present invention can accurately evaluate the communication requirements and priorities between devices, so as to allocate higher communication weights to critical tasks and high-traffic devices. Further, based on the spatio-temporal feature fusion model, perform a reliability evaluation on the communication priority weights to ensure the scientificity and reliability of the communication weight allocation, and avoid task failures caused by the unreliability of the communication link. By dynamically updating the topology structure and verifying the feasibility of the topology adjustment scheme, the system can adapt to changes in communication requirements in real time, optimize the link structure, and ensure the communication efficiency and collaborative performance of the device group in a dynamic environment. Through incremental update and topology structure reconstruction, the communication topology of the device group can be flexibly adjusted to cope with complex and changing operating scenarios.

[0066] Preferably, step S3 includes the following steps:

[0067] Step S31: Extract device node features from the device group dynamic topology graph to obtain the device group node feature vector set;

[0068] Specifically, the dynamic topology map of the device group can be imported into the MATLAB environment. The topology map is represented in the form of an adjacency matrix or an edge list, where each node represents a device and each edge represents a communication link between devices. Use the graph theory toolbox of MATLAB to analyze the topology map. By writing scripts, key features of each device node are extracted, such as the degree of the node (i.e., the number of edges connected to the node), the betweenness centrality of the node (indicating the importance of the node in the network, i.e., the number of shortest paths passing through the node), and the clustering coefficient of the node (indicating the degree of aggregation of the nodes around the node). These features can reflect the connection situation and communication ability of the device in the network. Taking a logistics warehouse as an example, assume that there are multiple AGVs (Automated Guided Vehicles) in the warehouse, and they communicate through a wireless network. Calculate the degree, betweenness centrality, and clustering coefficient of each AGV node through MATLAB scripts. For example, the degree of a certain AGV node is 5, indicating that it communicates directly with 5 other AGVs; its betweenness centrality is 0.3, indicating that it has a certain importance in the network and many communication paths pass through this node; its clustering coefficient is 0.6, indicating that the AGV nodes around it are relatively closely connected. Combine these feature values into a feature vector, such as [5, 0.3, 0.6], and generate similar feature vectors for each device node. Finally, summarize the feature vectors of all device nodes to form a device group node feature vector set.

[0069] Step S32: Expand the dimension of the device group node feature vector set to obtain a time-varying device group node feature tensor set;

[0070] Specifically, the device group node feature vector set can be imported into the TensorFlow environment. First, perform a time series expansion on the feature vector of each device node. Assume that the device node feature vector set contains the feature values of each device at multiple time points. For example, the feature vectors of device 1 at time points t1, t2, and t3 are [5, 0.3, 0.6], [6, 0.4, 0.7], and [7, 0.5, 0.8] respectively. Through the tensor operations of TensorFlow, stack these feature vectors in chronological order to form a three-dimensional tensor. Specifically, for device 1, its feature tensor can be expressed as: The dimension of this tensor is 3×3, where the first dimension represents time points and the second dimension represents the feature dimension. Similar processing is performed on all device nodes, and the feature vectors of each device are extended into a three-dimensional tensor. Finally, the three-dimensional tensors of all devices are aggregated to form a time-varying device group node feature tensor set. For example, in a logistics warehouse, assume there are 10 AGVs in the warehouse, each AGV has 3 feature dimensions, and the feature data of each device records information at 100 time points. Then, the dimension of the finally obtained time-varying device group node feature tensor set is 10×100×3. This tensor set can comprehensively reflect the feature changes of the device group at different time points.

[0071] Step S33: Based on the device-to-device communication weight matrix, construct the dynamic link features of the device group dynamic topology graph to obtain the device group link feature tensor set;

[0072] Specifically, the device-to-device communication weight matrix can be imported into the MATLAB environment. Assume there are 10 devices in the device group, and the communication weight matrix is a 10×10 matrix, where each element represents the communication weight between a pair of devices. Next, feature extraction is performed on each link in the dynamic topology graph. The extracted features include the weight, delay, bandwidth utilization, etc. of the link. For example, for link (i, j), extract its weight w ij , delay d ij and bandwidth utilization b ij . Combine these features into a feature vector, such as [w ij , d ij , b ij . Assume the link features between device 1 and device 2 are [0.8, 10ms, 0.7], indicating that the communication weight of this link is 0.8, the delay is 10 milliseconds, and the bandwidth utilization is 70%. Similar processing is performed on all links, and the feature vectors of each link are aggregated to form a three-dimensional tensor. For example, for the dynamic topology graph of 10 devices, the dimension of the link feature tensor is 10×10×3, where the first two dimensions represent the device pair and the third dimension represents the feature dimension. Finally, the device group link feature tensor set is obtained.

[0073] Step S34: Perform tensor fusion on the time-varying device group node feature tensor set and the device group link feature tensor set to obtain the device group communication characteristic tensor set;

[0074] Specifically, the time-varying device group node feature tensor set and the device group link feature tensor set can be imported into the TensorFlow environment. Assume that the dimension of the time-varying device group node feature tensor set is 10×100×3, representing 3 feature dimensions of 10 devices at 100 time points; the dimension of the device group link feature tensor set is 10×10×3, representing the link features between 10 devices. In TensorFlow, use the tensor fusion operation to merge these two tensors. Specifically, the node feature tensor is extended to the same dimension as the link feature tensor through the broadcasting mechanism, and then the element-wise weighted summation or concatenation operation is performed. For example, the node feature tensor is extended to 10×10×100×3, and the link feature tensor is extended to 10×10×100×3, and then the features at each time point are weighted and summed to obtain the fused tensor. Assume that the weighting coefficients are [0.6, 0.4], indicating that the weight of the node feature is 0.6 and the weight of the link feature is 0.4. In this way, the device group communication characteristic tensor set is obtained, and its dimension is 10×10×100×3. This tensor set can comprehensively reflect the communication characteristics of the device group at different time points, including the comprehensive features of nodes and links.

[0075] Step S35: Identify the temporal correlation of the device group communication characteristic tensor set to obtain the device group temporal correlation feature tensor set, and perform tensor decomposition on the device group temporal correlation feature tensor set to obtain the device communication characteristic data set;

[0076] Specifically, the device group communication characteristic tensor set can be imported into the MATLAB environment. Assume that the dimension of the communication characteristic tensor set is 10×10×100×3, representing 3 communication feature dimensions between 10 devices at 100 time points. First, identify the temporal correlation of the communication characteristic tensor set. Use the autocorrelation function (ACF) and partial autocorrelation function (PACF) to analyze the feature correlation of each device pair at different time points. For example, for the link between device 1 and device 2, calculate the autocorrelation coefficient of its delay feature at 100 time points. By analyzing the autocorrelation graph and partial autocorrelation graph, significant temporal correlations are identified. Assume that when the time delay is 10 time units, the correlation is significant, indicating that the communication delay between device 1 and device 2 still has a strong correlation after 10 time units. Next, perform tensor decomposition on the temporal correlation feature tensor set. Use the CP decomposition (CANDECOMP / PARAFAC decomposition) method to decompose the temporal correlation feature tensor into multiple low-rank tensors. For example, decompose the 10×10×100×3 tensor into 3 low-rank tensors, and each low-rank tensor represents a main communication characteristic pattern. Through tensor decomposition, the key communication characteristics of the device group are extracted, such as the main communication links and key time delay patterns. Finally, organize the decomposed low-rank tensors into the device communication characteristic data set.

[0077] Step S36: Calculate the information flow routing based on the device communication characteristic dataset to obtain an inter-device routing policy table, and perform modeling of the communication delay distribution according to the inter-device routing policy table to obtain a device group interaction characteristic model.

[0078] Specifically, for the detailed implementation process of this embodiment, please refer to the sub-steps of Step S36.

[0079] Through the extraction of node features and dimensionality expansion of the device group dynamic topology graph, as well as the dynamic construction of link features, the present invention can capture multi-dimensional features of device nodes and links. By integrating node features and link features through tensor fusion, it can more comprehensively reflect the communication characteristics between devices, including temporal changes and spatial correlations. Through the identification of temporal correlations and tensor decomposition of the communication characteristic tensor set, key temporal correlation features can be extracted and transformed into a device communication characteristic dataset that can be used for modeling. Finally, based on these data, information flow routing calculation and communication delay distribution modeling are performed to obtain a device group interaction characteristic model, which can provide a scientific basis for communication path planning and delay optimization between devices.

[0080] Preferably, Step S36 includes the following steps:

[0081] Step S361: Construct a routing cost metric model based on the device communication characteristic dataset to obtain a device group routing evaluation model;

[0082] Specifically, the device communication characteristic dataset can be imported into the MATLAB environment. This dataset contains communication delays, bandwidth utilization rates, and link reliability characteristics between devices. Based on these characteristics, a routing cost metric model is defined. For example, the routing cost can be defined as the weighted sum of communication delay and bandwidth utilization, while considering the impact of link reliability. The specific formula is as follows: Routing cost = α × communication delay + β × (1 - bandwidth utilization) + γ × (1 - link reliability), where α, β, and γ are weight coefficients, representing the influence degrees of communication delay, bandwidth utilization, and link reliability on the routing cost, respectively. Assume α = 0.5, β = 0.3, and γ = 0.2, and these parameters can be adjusted according to actual requirements. In MATLAB, write a script to calculate the routing cost between each pair of devices. For example, for the link between Device 1 and Device 2, assume the communication delay is 10 milliseconds, the bandwidth utilization rate is 70%, and the link reliability is 90%, then its routing cost is: Routing cost = 0.5 × 10 + 0.3 × (1 - 0.7) + 0.2 × (1 - 0.9) = 5.06; In this way, calculate the routing costs between all pairs of devices in the device group and store these costs as a matrix to form a device group routing evaluation model.

[0083] Step S362: Perform multi-objective optimization on the device group routing evaluation model to obtain a candidate device group routing set;

[0084] Specifically, the device group routing evaluation model can be imported into the MATLAB environment. Based on the routing cost matrix, a multi-objective optimization problem is defined. For example, the optimization objectives include minimizing communication delay, maximizing bandwidth utilization, and maximizing link reliability. Specifically, the Pareto optimization method can be used to find the optimal solution that balances multiple objectives. In MATLAB, the gamultiobj function is used for multi-objective optimization. First, define the optimization objective function, with communication delay, bandwidth utilization, and link reliability as components of the objective function. For example: The first objective is to minimize communication delay, which is calculated by multiplying the routing selection variable by the communication delay matrix and summing. The second objective is to maximize bandwidth utilization, which is calculated by multiplying the routing selection variable by the bandwidth utilization matrix, summing, and then taking the negative value. The third objective is to maximize link reliability, which is calculated by multiplying the routing selection variable by the link reliability matrix, summing, and then taking the negative value. Next, set the optimization parameters, such as the population size and the number of iterations. Assume the population size is 100 and the number of iterations is 1000. Use the gamultiobj function for optimization, setting the number of variables and the upper and lower bound parameters. Through optimization, a set of Pareto optimal solutions is obtained, and each solution represents a routing scheme.

[0085] Step S363: Measure the routing reliability of the candidate device group routing set to obtain the device group routing reliability index;

[0086] Specifically, the candidate device group routing set can be imported into the MATLAB environment. Assume that 10 candidate routing schemes have been obtained through multi-objective optimization, and each scheme includes the routing selection between devices. Define the routing reliability measurement index. For example, link reliability and path diversity can be used as key indicators for reliability evaluation. Link reliability: For each routing scheme, calculate the product of the reliabilities of all links on the path. Assume the link reliability matrix is known. For path P i , its link reliability R(P i ) can be expressed as: R(P i ) = ∏(u,v)∈P i Link reliability uv ; where, (u, v) represents the path P iLink. Path diversity: Evaluate the path diversity between each pair of devices, that is, whether there are multiple reliable paths to choose from. Path diversity can be quantified by calculating the number of paths between device pairs and the reliability difference of the paths. In MATLAB, write a script to calculate the reliability metrics for each candidate routing scheme. For example, for a system with 10 devices, calculate the product of the link reliabilities between each pair of devices and count the path diversity. Finally, obtain the reliability metrics for each candidate routing scheme.

[0087] Step S364: Screen the device group routing according to the device group routing reliability metrics to obtain the optimal device group routing scheme, and perform load balancing on the optimal device group routing scheme to obtain the inter-device routing policy table;

[0088] Specifically, the device group routing reliability metrics can be imported into the MATLAB environment. Based on these metrics, screen out the routing scheme with the highest reliability. For example, set a reliability threshold, such as R min = 0.8, and screen out all routing schemes with a reliability higher than this threshold. Suppose that after screening, there are 5 candidate routing schemes left. Next, perform load balancing on the screened optimal device group routing scheme. Use minimizing the maximum link load as the objective function of load balancing. Specifically, a linear programming method can be used to optimize the link load distribution. In MATLAB, use the linprog function to perform linear programming. Define the objective function: minimize the maximum link load. Define the constraint conditions: ensure that the traffic demand between each pair of devices is met and the link load does not exceed its capacity. Use the linprog function to solve the linear programming problem to obtain the optimal load distribution scheme. For example, assume that the traffic demand matrix between device pairs is known and the link capacity matrix is also known. By solving the linear programming, the load distribution of each link is obtained. Finally, organize the optimal load distribution scheme into an inter-device routing policy table, which details the communication paths and link load distribution between each pair of devices.

[0089] Step S365: Sample the link communication delay of the device group based on the inter-device routing policy table to obtain the device group delay sampling data set, and perform probability distribution fitting on the device group delay sampling data set to obtain the device group delay distribution model;

[0090] Specifically, the communication link of the device group can be sampled for delay according to the inter-device routing policy table. Assuming that there are 10 AGVs in the device group, the routing policy table records the communication path between each device pair in detail. Use a network analyzer or network monitoring tool (such as Wireshark) to sample the delay of each link. For example, sample once every 5 minutes for 24 hours to obtain the delay data of each link. Import these delay data into the MATLAB environment to form a device group delay sampling data set. Next, perform probability distribution fitting on the device group delay sampling data set. In MATLAB, use the fitdist function to fit the distribution of the delay data. Assuming that the delay data conforms to the normal distribution or exponential distribution, the delay distribution model of each link is obtained by fitting. The delay distribution models of all links are summarized to form a device group delay distribution model.

[0091] Step S366: Evaluate the communication link performance of the device group based on the device group delay distribution model to obtain a communication link performance evaluation index, and intelligently model the interaction characteristics of the device group based on the communication link performance evaluation index to obtain a device group interaction characteristic model.

[0092] Specifically, the performance of the communication link of the device group can be evaluated according to the device group delay distribution model. In MATLAB, the performance evaluation indicators of each link, such as average delay, delay jitter and packet loss rate, are calculated. Assume that the link delay distribution model between device 1 and device 2 is a normal distribution with a mean of 15 milliseconds and a standard deviation of 3 milliseconds. The average delay of the link is calculated to be 15 milliseconds and the delay jitter is 3 milliseconds. At the same time, the packet loss rate data is obtained by combining network monitoring tools (such as Wireshark), assuming that the packet loss rate is 0.1%. Next, the interactive characteristics of the device group are intelligently modeled according to the communication link performance evaluation indicators. In MATLAB, the interactive characteristics of the device group are modeled using the system identification toolbox. Assume that the interactive characteristics of the device group can be described by a linear time-invariant system (LTI) model. By inputting the communication traffic data between the devices and the output delay data, the least squares method or the maximum likelihood estimation method is used to estimate the parameters of the system model. For example, for the interactive characteristics between device 1 and device 2, an LTI model is obtained by fitting, and its transfer function is: Among them, s is the complex frequency variable in Laplace transform. This model can describe the relationship between communication traffic and delay between devices. The interaction characteristic models of all device pairs are summarized to form a device group interaction characteristic model.

[0093] By constructing a routing cost measurement model and performing multi-objective optimization solutions, the present invention can comprehensively consider multiple factors (such as communication delay, bandwidth utilization), generate multiple candidate routing solutions, and provide a variety of communication path options for the device group. Through reliability evaluation and screening, combined with load balancing strategies, the optimal routing solution can be selected to ensure the efficiency and stability of the device group during the communication process, and avoid communication bottlenecks or link congestion caused by improper routing selection. By sampling and probability distribution fitting of link communication delays, an accurate delay distribution model can be established. Finally, intelligent modeling of interactive characteristics based on the delay distribution model can more accurately reflect the communication behavior and collaboration needs between devices.

[0094] Preferably, step S4 comprises the following steps:

[0095] Step S41: collecting execution tasks of the device group to obtain a device group execution task set, and performing task complexity analysis on the device group execution task set based on the device group interaction characteristic model to obtain a device task complexity index set;

[0096] Specifically, please refer to the sub-steps of step S41 for the detailed implementation process of this embodiment.

[0097] Step S42: performing task priority division on the device group execution task set according to the device task complexity index set, and obtaining a device group execution task priority matrix;

[0098] Specifically, the device task complexity index set can be imported into the MATLAB environment. The complexity index of each task reflects the difficulty of the task and the demand for system resources. For example, the complexity index of task 1 is 0.8, the complexity index of task 2 is 0.6, and the complexity index of task 3 is 0.9. In MATLAB, a script is written to prioritize tasks. According to the size of the complexity index, tasks are divided into three priorities: high, medium, and low. For example, tasks with a complexity index greater than 0.8 are set to high priority, tasks between 0.6 and 0.8 are set to medium priority, and tasks less than 0.6 are set to low priority. In this way, a priority value is assigned to each task. For example: Task 1: complexity index 0.8, priority is high (2); Task 2: complexity index 0.6, priority is medium (1); Task 3: complexity index 0.9, priority is high (2). The priorities of all tasks are summarized into a device group execution task priority matrix. For example, assuming that there are 3 devices in the device group and each device has 3 tasks, the priority matrix can be expressed as: The rows of the matrix represent devices, the columns represent tasks, and the values ​​in the matrix represent the priorities of the tasks.

[0099] Step S43: Perform hierarchical clustering on the task priority matrix of the device group to obtain the hierarchical structure tree of the tasks executed by the device group;

[0100] Specifically, the task priority matrix of the device group can be imported into the MATLAB environment. The values in the task priority matrix of the device group represent the priorities of the tasks. In MATLAB, the hierarchical clustering method is used to perform clustering analysis on the task priority matrix. Appropriate distance metrics, such as Euclidean distance, and clustering algorithms, such as Ward's method, are selected. By writing a script, hierarchical clustering is performed on the task priority matrix. The specific steps include: using the pdist function to calculate the Euclidean distance between tasks in the task priority matrix. Using the linkage function to perform hierarchical clustering according to Ward's method. Using the dendrogram function to generate the hierarchical structure tree and visualize the results of hierarchical clustering. Through hierarchical clustering, the hierarchical structure tree of the tasks executed by the device group is obtained. For example, assume there are 10 tasks in the device group. Through hierarchical clustering, the tasks are divided into 3 main hierarchical structures. The first layer contains high-priority tasks, the second layer contains medium-priority tasks, and the third layer contains low-priority tasks. The hierarchical structure tree can intuitively display the priority relationship and hierarchical structure between tasks.

[0101] Step S44: Construct the buffer pool in layers according to the hierarchical structure tree of the tasks executed by the device group to obtain the device task buffer pool;

[0102] Specifically, the hierarchical structure tree of the tasks executed by the device group can be imported into the MATLAB environment. The priority levels of the tasks are clearly marked in the hierarchical structure tree of the tasks. For example, the first layer is high-priority tasks, the second layer is medium-priority tasks, and the third layer is low-priority tasks. In MATLAB, the tasks are processed in layers according to the hierarchical structure tree. The tasks are divided into different levels according to the priority levels, and a buffer pool is assigned to each level. For example, high-priority tasks are assigned to the first-layer buffer pool, medium-priority tasks are assigned to the second-layer buffer pool, and low-priority tasks are assigned to the third-layer buffer pool. Each buffer pool has an independent storage space and task management mechanism. Use Visio to draw the hierarchical structure diagram of the buffer pool to intuitively display the distribution of buffer pools at different levels and the task allocation situation. Finally, the hierarchical structure of the device task buffer pool is obtained.

[0103] Step S45: Allocate storage space for the device task buffer pool to obtain the device group buffer pool capacity configuration data;

[0104] Specifically, the hierarchical structure of the device task buffer pool can be imported into the MATLAB environment. The hierarchical structure of the device task buffer pool includes the number of tasks and priorities at each level. In MATLAB, the required storage space is calculated based on the number of tasks and priorities at each level. For example, assume that high-priority tasks require an average of 100 MB of storage space, medium-priority tasks require 50 MB, and low-priority tasks require 20 MB. By writing a script, calculate the total storage space required for each level. For example, there are 5 high-priority tasks in the first level, requiring 500 MB; 10 medium-priority tasks in the second level, requiring 500 MB; and 15 low-priority tasks in the third level, requiring 300 MB. Summarize these storage space requirements into the device group buffer pool capacity configuration data. Use Excel to record and manage these configuration data, including the storage space requirements and allocated storage location information for each level. Finally, obtain the device group buffer pool capacity configuration data.

[0105] Step S46: Design the task scheduling rules according to the device group buffer pool capacity configuration data to obtain the device task buffer pool configuration data;

[0106] Specifically, the device group buffer pool capacity configuration data can be imported into the MATLAB environment. The device group buffer pool capacity configuration data includes the storage capacity and the number of tasks at each level. In MATLAB, design the task scheduling rules according to the buffer pool capacity configuration data. Adopt the priority scheduling algorithm to ensure that high-priority tasks are scheduled first. For example, set the rule: high-priority tasks are preferentially allocated to the first-level buffer pool, medium-priority tasks are allocated to the second-level buffer pool, and low-priority tasks are allocated to the third-level buffer pool. If the storage space of a certain level of buffer pool is insufficient, the task will be allocated to the next-level buffer pool, and this adjustment will be recorded. By writing a script, realize the automated design of the task scheduling rules. For example, for a new task, first check its priority, and then allocate it to the appropriate buffer pool according to the priority and buffer pool capacity configuration data. At the same time, record the task allocation situation, including information such as task ID, allocated buffer pool level, and storage location. Finally, summarize these records into the device task buffer pool configuration data to form a complete task scheduling and buffer pool management solution.

[0107] Step S47: Perform task coding compression on the device group according to the device task buffer pool configuration data to obtain the device task allocation plan, and perform a load status evaluation on the device task allocation plan to obtain the system load warning index set.

[0108] Specifically, for the detailed implementation process of this embodiment, please refer to the sub-steps of Step S47.

[0109] By performing complexity analysis and priority division on the tasks executed by the device group, the present invention can reasonably allocate tasks according to the difficulty and importance of the tasks, ensure that high-priority tasks are executed first, thereby improving the task processing efficiency and reducing task backlogs. By constructing a task hierarchy tree through hierarchical clustering and based on this, constructing a buffer pool hierarchy and allocating storage space, the task queue can be managed more efficiently and the task scheduling rules can be optimized. By encoding and compressing tasks and evaluating the load status, the storage and transmission volume of task data can be reduced, the system's resource requirements can be lowered, and the system load condition can be monitored in real time.

[0110] Preferably, step S41 includes the following steps:

[0111] Step S411: Monitor the real-time operating state of the device group to obtain the dynamic operating data stream of the device group, and extract the executed tasks from the dynamic operating data stream of the device group to obtain the device group executed task set;

[0112] Specifically, the device group (such as AGVs, robots, etc.) can be connected to the SINEMA system, and the operating state data of the devices, including information such as position, speed, and load, can be collected in real time through sensors and controllers. This data is transmitted to the monitoring server in the form of a data stream to form the dynamic operating data stream of the device group. In MATLAB, write a script to process the dynamic operating data stream. Through the data parsing module, extract the executed task information of the devices from the data stream. For example, for an AGV, its executed tasks include cargo handling, path planning, etc. By analyzing the position changes and task instructions of the AGV, the specific executed tasks are extracted. Suppose the AGV transports goods from point A in the warehouse to point B. The task extraction module can identify this as a "handling task" and record the starting point, ending point, and execution time of the task. In this way, the dynamic operating data stream of the device group is monitored and tasks are extracted in real time, and finally the device group executed task set is obtained.

[0113] Step S412: Identify the task types of the device group executed task set to obtain the device task type distribution set, and perform spatio-temporal association on the device task type distribution set to obtain the device task spatio-temporal feature vector;

[0114] Specifically, the task set executed by the device group can be imported into the Python environment. First, identify the task types of the executed task set. By analyzing the description information, execution path, and task instructions of the tasks, machine learning algorithms (such as decision trees or support vector machines) are used to classify the task types. For example, the tasks are classified into types such as "carrying tasks", "charging tasks", "maintenance tasks", etc. Assume that a classifier for task types has been obtained through machine learning model training. Each task is classified, and the classification results are stored as the device task type distribution set. Next, by analyzing the execution time and spatial location of the tasks, the spatio-temporal features of the tasks are extracted. For example, for a "carrying task", record the start time and end time of the task, as well as the key location points on the task path. By calculating the duration and path length of the task, the spatio-temporal feature vector of the task is obtained. Assume that the task starts at time point t1 and ends at time point t2, and the path length is L. Then the spatio-temporal feature vector of the task can be expressed as [t1, t2, L]. In this way, spatio-temporal correlation analysis is performed on each task of the device group, and finally the device task spatio-temporal feature vector is obtained.

[0115] Step S413: Construct a task dependency graph based on the device task spatio-temporal feature vector to obtain the device task dependency network, and calculate the topological complexity of the device task dependency network to obtain the device task topological metric index;

[0116] Specifically, the device task spatio-temporal feature vector can be imported into the MATLAB environment. For example, the feature vector of task 1 is [t1, t2, L1], and the feature vector of task 2 is [t3, t4, L2]. Based on these feature vectors, analyze the dependency relationships between the tasks. For example, if the start time t3 of task 2 is after the end time t2 of task 1, and the start location of task 2 is the end location of task 1, it can be considered that task 2 depends on task 1. In MATLAB, use the graph theory toolbox to construct the task dependency graph. Each task is represented as a node in the graph, and the dependency relationship between tasks is represented as a directed edge. For example, the dependency relationship between task 1 and task 2 can be represented as a directed edge pointing from task 1 to task 2. In this way, the device task dependency network is constructed. Next, calculate the topological complexity of the task dependency network. Use graph theory metrics such as betweenness centrality and clustering coefficient to quantify the complexity of the network. For example, calculate the betweenness centrality of each node, which represents the importance of the node in the network; calculate the average clustering coefficient of the network, which represents the degree of aggregation of the nodes in the network. Finally, obtain the device task topological metric index, which can reflect the complexity of the task dependency network and the key task nodes.

[0117] Step S414: Evaluate the computational resource consumption of the device task dependency network to obtain the resource occupancy vector of the device group, and evaluate the load intensity of the device group based on the resource occupancy vector of the device group to obtain the device group load intensity index;

[0118] Specifically, the device task dependency network can be imported into the MATLAB environment. Based on these data, evaluate the computational resource consumption of the tasks. For example, by analyzing the execution time and task type of the tasks, estimate the computational resources required for each task. Assume that the resource consumption of the handling task is 1 unit and the resource consumption of the charging task is 0.5 units. In this way, calculate the resource consumption of each task and summarize the resource consumption of all tasks as the resource occupancy vector of the device group. Next, use the resource utilization rate and task waiting time as evaluation indicators. For example, calculate the resource utilization rate of each device per unit time, that is, the ratio of the resources consumed by the device to complete the task to its total resources. At the same time, count the waiting time of the tasks, that is, the time interval from task submission to start execution. Assume that the resource utilization rate of device 1 is 80% and the average task waiting time is 2 minutes during a certain period; the resource utilization rate of device 2 is 60% and the average task waiting time is 3 minutes. Through these indicators, evaluate the load intensity of the device group. Finally, obtain the device group load intensity index.

[0119] Step S415: Perform feature fusion on the device task topology metric index and the device group load intensity index to obtain the device task complexity feature set;

[0120] Specifically, the device task topology metric index (such as betweenness centrality and clustering coefficient) and the device group load intensity index (such as resource utilization rate and task waiting time) can be imported into the MATLAB environment. Assume that the device task topology metric index includes the betweenness centrality BC i and the clustering coefficient CC i , and the device group load intensity index includes the resource utilization rate RU j and the task waiting time WT j . In MATLAB, write a script to perform feature fusion on these indicators. For example, perform a weighted sum of the topology metric index of each task and the load intensity index of the related device to obtain the device task complexity feature. Assume that the weight assignment is: the weight of betweenness centrality is 0.4, the weight of clustering coefficient is 0.3, the weight of resource utilization rate is 0.2, and the weight of task waiting time is 0.1. For task i, its complexity feature CF i can be expressed as:

[0121] CF i = 0.4×BC i + 0.3×CC i + 0.2×RU related+0.1×WT related ;

[0122] wherein, RU related and WT related respectively represent the resource utilization rate of the device related to task i and the task waiting time. In this way, the complexity characteristics of each task are calculated, and the complexity characteristics of all tasks are summarized into a device task complexity characteristic set. For example, in a logistics warehouse, assume there are 10 AGVs in the warehouse, and each AGV has multiple tasks. The complexity characteristics of each task are obtained through feature fusion.

[0123] Step S416: Identify the influencing factors for the device task complexity characteristic set based on the device group interaction characteristic model to obtain a device task influencing factor set, and adjust the weights of the device task influencing factor set to obtain a device task weight vector;

[0124] Specifically, the device task complexity characteristic set can be imported into the MATLAB environment. The complexity characteristics of each task in the device task complexity characteristic set include multiple dimensions, such as topological metric indicators and load intensity indicators. In MATLAB, use the device group interaction characteristic model (assumed to be a linear time-invariant system model) to identify the influencing factors for the task complexity characteristic set. By analyzing the relationship between the task complexity characteristics and the device interaction characteristics, the factors that have a greater impact on the task complexity are identified. For example, through regression analysis or correlation analysis, determine the influence degrees of betweenness centrality, clustering coefficient, resource utilization rate, and task waiting time on the task complexity. Assume the analysis result shows that betweenness centrality has the greatest impact on the task complexity, followed by resource utilization rate, and the impacts of clustering coefficient and task waiting time are relatively small. Next, adjust the weights according to the importance of the influencing factors. For example, adjust the weight of betweenness centrality to 0.5, the weight of resource utilization rate to 0.3, the weight of clustering coefficient to 0.1, and the weight of task waiting time to 0.1. Through the above operations, a device task weight vector is obtained. For example, for task i, its weight vector W i can be expressed as: W i = [0.5, 0.1, 0.3, 0.1]; In this way, the weight vectors of each task are adjusted, and finally a device task weight vector set is obtained.

[0125] Step S417: Quantify the complexity according to the device task weight vector to obtain a device task complexity index set.

[0126] Specifically, the device task weight vector can be selected and imported into the MATLAB environment. The weight vector of each task in the device task weight vector set includes weight values of multiple dimensions, such as betweenness centrality weight, clustering coefficient weight, resource utilization weight, and task waiting time weight. In MATLAB, a script is written to perform a weighted sum of the complexity characteristics of each task to obtain the complexity index of the task. For example, for task i, its complexity characteristic is CF i =[BC i , CC i , RU related , WT related , and the weight vector is W i =[0.5, 0.1, 0.3, 0.1]. The complexity index CD i of task i can be expressed as: i = BC i ×0.5 + CC i ×0.1 + RU related ×0.3 + WT related ×0.1; In this way, the complexity index of each task is calculated, and the complexity indexes of all tasks are aggregated into a device task complexity index set. For example, in a logistics warehouse, assuming there are 10 AGVs in the warehouse and each AGV has multiple tasks, the complexity index of each task is obtained through complexity quantification.

[0127] Through real-time runtime monitoring and task extraction, the present invention can dynamically obtain the execution task set of the device, ensuring the timeliness and accuracy of task information. Through task type recognition and spatio-temporal correlation analysis, a task dependency graph can be constructed to quantify the dependency relationship and topological complexity between tasks, thereby comprehensively evaluating the complexity of tasks. By combining the evaluation of computational resource consumption and load intensity evaluation, the demand of tasks for resources and the load situation of the device can be comprehensively considered. By identifying influence factors and adjusting weights, the complexity of tasks can be accurately quantified.

[0128] Preferably, step S47 includes the following steps:

[0129] Step S471: Perform compressed feature encoding on the device task buffer pool configuration data to obtain a device task encoding dictionary;

[0130] Specifically, the device task buffer pool configuration data can be imported into the Python environment. The configuration data includes information such as task ID, buffer pool level, storage location, etc. Use LabelEncoder or OneHotEncoder in the Scikit-learn library to encode categorical features (such as buffer pool level), converting text information into numerical features. For example, encode the buffer pool levels "high", "medium", and "low" as 1, 2, and 3 respectively. For continuous features (such as storage location), normalization or standardization methods can be used for processing. In this way, each feature in the configuration data is converted into a numerical form and summarized into a device task encoding dictionary. For example, the encoding dictionary for task 1 is: {task ID: 1, buffer pool level: 1, storage location: 0.5}.

[0131] Step S472: Use the device task encoding dictionary to adaptively compress the task set executed by the device group to obtain a compressed device task data set;

[0132] Specifically, the device task encoding dictionary can be imported into the Python environment. Use the zlib or gzip library to compress the task data. For example, for the data of task 1, first convert its encoded features into a byte stream, and then use the zlib.compress() function for compression. Suppose the original data size of task 1 is 100KB, and after compression, the data size is reduced to 50KB. In this way, the data of each task is compressed, and the compressed data size and compression ratio are recorded. Finally, all the compressed task data is summarized into a compressed device task data set. For example, assume there are 10 tasks in the device group, and the compressed data of each task is stored in a file, and the file name contains the task ID and the compression ratio.

[0133] Step S473: Perform device capability matching on the compressed device task data set to obtain a device task adaptation matrix;

[0134] Specifically, the compressed device task dataset can be imported into the MATLAB environment. Assume that the dataset contains the compression features and size information of each task. At the same time, import the device capability data, which describe parameters such as the processing capacity, storage capacity, and task execution efficiency of each device. For example, the processing capacity of device 1 is 100 tasks per hour, and the storage capacity is 1 GB; the processing capacity of device 2 is 80 tasks per hour, and the storage capacity is 500 MB. In MATLAB, write a script to match the capabilities of each task with each device. By calculating the matching degree between the compressed data size of the task and the storage capacity of the device, and the matching degree between the processing requirements of the task and the processing capacity of the device, a matching score is obtained. For example, for task 1 (compressed data size is 50 KB, processing requirement is high priority), the matching score of device 1 is 0.9 (storage matching degree 0.8, processing capacity matching degree 0.9), and the matching score of device 2 is 0.7 (storage matching degree 0.6, processing capacity matching degree 0.8). In this way, a matching score matrix, that is, a device-task adaptation matrix, is generated for the matching relationship between each task and each device. For example, for 3 tasks and 2 devices, the adaptation matrix can be expressed as: Among them, the rows of the matrix represent tasks, the columns represent devices, and the values in the matrix represent the adaptation degree between the tasks and the devices.

[0135] Step S474: Based on the device-task adaptation matrix, perform task allocation for the device group to obtain a device-task allocation plan, and perform load prediction on the device-task allocation plan to obtain a device load prediction curve;

[0136] Specifically, the device task adaptation matrix can be imported into the MATLAB environment, and the values in the matrix represent the degree of adaptation between tasks and devices. In MATLAB, optimization algorithms (such as linear programming or integer programming) are used to allocate tasks. The objective function is defined as maximizing the total adaptation score between tasks and devices, while considering the processing capacity and storage capacity limitations of the devices. For example, the linprog function is used to solve the linear programming problem to obtain the optimal task allocation scheme. Suppose the task allocation results are: task 1 is allocated to device 1, task 2 is allocated to device 2, and task 3 is allocated to device 1. These allocation results are recorded as the device task allocation scheme. Next, load prediction is performed on the device task allocation scheme. According to the task allocation situation of each device, the load intensity of the device is calculated. For example, device 1 is allocated 2 tasks, and device 2 is allocated 1 task. Suppose the processing time of task 1 and task 3 is 1 hour, and the processing time of task 2 is 0.5 hour. The load intensity of device 1 is 2 tasks / hour, and the load intensity of device 2 is 0.5 tasks / hour. Through time series analysis or machine learning models (such as linear regression), the load situation of the device in the future is predicted to obtain the device load prediction curve. For example, the load intensity change curve of device 1 in the next 24 hours and the load intensity change curve of device 2 are predicted.

[0137] Step S475: Based on the device load prediction curve, identify the bottlenecks in the device group to obtain the device system bottleneck analysis report;

[0138] Specifically, the device load prediction curve can be imported into the MATLAB environment, and these curves describe the load intensity change of each device in the next 24 hours. For example, the load prediction curve of device 1 shows that it reaches the peak at the 10th hour, and the load intensity is 3 tasks / hour; the load prediction curve of device 2 shows that it reaches the peak at the 15th hour, and the load intensity is 1.2 tasks / hour. In MATLAB, a script is written to analyze the load prediction curve to identify potential bottlenecks. The bottleneck is defined as the time point when the device load intensity exceeds 80% of its processing capacity. For example, the processing capacity of device 1 is 100 tasks / hour. When the load intensity exceeds 80 tasks / hour, it is considered that device 1 has a bottleneck. By analyzing the load prediction curve, it is found that the load intensity of device 1 at the 10th hour is 3 tasks / hour, which exceeds 80% of its processing capacity. Therefore, it is identified that device 1 has a bottleneck at the 10th hour. Similarly, by analyzing the load prediction curve of device 2, it is found that the load intensity of device 2 at the 15th hour is 1.2 tasks / hour, which does not exceed 80% of its processing capacity. Therefore, device 2 does not have a bottleneck. Finally, the bottleneck identification results are summarized into the device system bottleneck analysis report.

[0139] Step S476: Define thresholds for the device system bottleneck analysis report to obtain a set of device group load threshold metrics, and construct control system warning rules based on the set of device group load threshold metrics to obtain a set of system load warning metrics.

[0140] Specifically, the device system bottleneck analysis report can be imported into the MATLAB environment. In MATLAB, load thresholds are defined according to the bottleneck analysis report. For example, the device load threshold is set to 80% of the device processing capacity. For device 1, the processing capacity is 100 tasks per hour, and the load threshold is 80 tasks per hour; for device 2, the processing capacity is 80 tasks per hour, and the load threshold is 64 tasks per hour. Finally, a set of device group load threshold metrics is obtained. For example, the load threshold of device 1 is 80 tasks per hour, and the load threshold of device 2 is 64 tasks per hour. Next, use MATLAB to write a script to monitor the load intensity of the device in real time and compare it with the load threshold. When the load intensity of the device exceeds its load threshold, the warning rule is triggered. For example, when the load intensity of device 1 exceeds 80 tasks per hour, the system sends a warning signal to prompt the operator that device 1 faces an overload risk. Record these warning rules in Excel to form a set of system load warning metrics.

[0141] Through compression feature encoding and adaptive compression, the present invention can effectively reduce the storage and transmission volume of task data, reduce the resource requirements, and at the same time ensure the integrity and availability of task data. Through device capacity matching, the system can reasonably allocate tasks according to the actual capabilities of the devices. Through bottleneck identification based on the load prediction curve, potential load bottlenecks can be discovered in advance. Through threshold definition and warning rule construction for the bottleneck analysis report, the device load status can be monitored in real time.

[0142] Preferably, step S5 includes the following steps:

[0143] Step S51: Configure the simulation environment for the device group based on the device group interaction characteristic model to obtain a set of device group simulation scenario parameters, and initialize the operating conditions for the set of device group simulation scenario parameters according to the set of system load warning metrics to obtain an initial configuration plan for the device group simulation.

[0144] Specifically, the device group interaction characteristic model can be imported into the Simulink environment. Using the modeling tools of Simulink, a simulation environment can be constructed according to the interaction characteristic model. For example, a simulation module is created for each device, and the modules are connected through communication links. The parameters of the modules include the processing capacity, storage capacity, and task allocation of the devices. In Simulink, the initial load status of each device is set. For example, the initial load of device 1 is set to 60 tasks per hour (lower than its load threshold of 80 tasks per hour), and the initial load of device 2 is set to 50 tasks per hour (lower than its load threshold of 64 tasks per hour). According to the warning rules, a monitoring mechanism for the simulation environment is set. When the device load exceeds the threshold, a warning signal is triggered. Finally, an initial configuration plan for the device group simulation is obtained, which includes the initial states of the devices, the communication link configuration, and the monitoring mechanism.

[0145] Step S52: Perform multi-threaded parallel simulation on the initial configuration plan for the device group simulation to obtain a set of device group simulation operation trajectories, and extract performance indicators from the set of device group simulation operation trajectories to obtain a device group performance characteristic tensor;

[0146] Specifically, the initial configuration plan for the device group simulation can be imported into the Simulink environment. In Simulink, multi-threaded parallel simulation is started. The parfor function in the Parallel Computing Toolbox is used to perform parallel processing on the simulation tasks of each device. For example, assume there are 10 devices in the device group, and the simulation tasks of each device can run independently. Through the parfor function, these tasks are assigned to multiple threads for parallel execution. During the simulation process, the operation trajectories of each device are recorded, including the load changes, task execution status, and usage of communication links of the devices. For example, the operation trajectory of device 1 shows that its load gradually increases during the simulation and reaches 70 tasks per hour at the 10th hour, and the operation trajectory of device 2 shows that its load reaches 60 tasks per hour at the 15th hour. Next, using the data analysis function of MATLAB, key performance indicators are extracted from the operation trajectories, such as the average load intensity, task execution time, and communication delay of the devices. For example, the average load intensity of device 1 is calculated to be 70 tasks per hour, the task execution time is 1 hour, and the communication delay is 10 milliseconds; the average load intensity of device 2 is 60 tasks per hour, the task execution time is 0.5 hour, and the communication delay is 15 milliseconds. These performance indicators are summarized into a device group performance characteristic tensor.

[0147] Step S53: Evaluate the performance of the device group performance characteristic tensor to obtain a set of device group performance evaluation indicators, and generate a device simulation performance evaluation report according to the set of device group performance evaluation indicators;

[0148] Specifically, the device group performance characteristic tensor can be imported into the MATLAB environment. For example, the performance characteristics of device 1 are [70, 1, 10], indicating that its average load intensity is 70 tasks per hour, the task execution time is 1 hour, and the communication delay is 10 milliseconds; the performance characteristics of device 2 are [60, 0.5, 15], indicating that its average load intensity is 60 tasks per hour, the task execution time is 0.5 hour, and the communication delay is 15 milliseconds. In MATLAB, define performance evaluation metrics, such as load efficiency (the ratio of average load intensity to processing capacity), task response time (task execution time), and communication efficiency (the reciprocal of communication delay). Calculate the performance evaluation metrics for each device. For example, the load efficiency of device 1 is 70 / 100 = 0.7, the task response time is 1 hour, and the communication efficiency is 1 / 10 = 0.1; the load efficiency of device 2 is 60 / 80 = 0.75, the task response time is 0.5 hour, and the communication efficiency is 1 / 15 ≈ 0.067. Aggregate these evaluation metrics into a device group performance evaluation metric set. Use MATLAB's report generation tool (such as the publish function) or manually write a report in Microsoft Word. Record in detail the performance evaluation metrics of each device in the report, including load efficiency, task response time, and communication efficiency.

[0149] Step S54: Extract the error characteristics from the device simulation performance evaluation report to obtain the device group error characteristic matrix, and perform sensitivity measurement on the device group error characteristic matrix to obtain the device group parameter sensitivity vector;

[0150] Specifically, the performance evaluation metrics in the device simulation performance evaluation report can be imported into the MATLAB environment. At the same time, the performance metrics in the actual operation data are imported for comparison with the simulation results. In MATLAB, write a script to extract the error characteristics of the performance evaluation metrics. Calculate the error between the simulation results and the actual operation data. For example, the actual load efficiency of device 1 is 0.65, the task response time is 1.2 hours, and the communication efficiency is 0.09; the actual load efficiency of device 2 is 0.72, the task response time is 0.6 hours, and the communication efficiency is 0.06. The calculated errors are as follows: For device 1: Load efficiency error: 0.7 - 0.65 = 0.05; Task response time error: 1 - 1.2 = -0.2; Communication efficiency error: 0.1 - 0.09 = 0.01; For device 2: Load efficiency error: 0.75 - 0.72 = 0.03; Task response time error: 0.5 - 0.6 = -0.1; Communication efficiency error: 0.067 - 0.06 = 0.007; Next, use the sensitivity analysis tool (such as the fmincon function) in the optimization toolbox of MATLAB to evaluate the influence degree of each parameter on the error. For example, assume that the sensitivity of the load efficiency is 0.8, the sensitivity of the task response time is 0.5, and the sensitivity of the communication efficiency is 0.3, and finally obtain the device group parameter sensitivity vector.

[0151] Step S55: Solve the simulation parameters of the device group based on the device group parameter sensitivity vector to obtain an optimized device simulation parameter solution;

[0152] Specifically, the device group parameter sensitivity vector can be imported into the MATLAB environment. For example, the sensitivity vector of device 1 is [0.8, 0.5, 0.3], indicating that the sensitivities of load efficiency, task response time, and communication efficiency are 0.8, 0.5, and 0.3 respectively; the sensitivity vector of device 2 is [0.8, 0.5, 0.3]. In MATLAB, an optimization algorithm (such as the fmincon function) is used to define the optimization objective function, for example, to minimize the error value in the error feature matrix. According to the sensitivity vector, weights are set for each parameter, and higher weights are assigned to parameters with higher sensitivities. For example, the weight of load efficiency is 0.8, the weight of task response time is 0.5, and the weight of communication efficiency is 0.3. Through the optimization algorithm, the simulation parameters are adjusted to minimize the optimization objective function. Assume that the initial simulation parameters are: for device 1, the load efficiency is 0.7, the task response time is 1 hour, and the communication efficiency is 0.1; for device 2, the load efficiency is 0.75, the task response time is 0.5 hour, and the communication efficiency is 0.067. Through optimization, the optimized simulation parameters are obtained. For example, for device 1, the load efficiency is adjusted to 0.68, the task response time is adjusted to 1.1 hours, and the communication efficiency is adjusted to 0.09; for device 2, the load efficiency is adjusted to 0.73, the task response time is adjusted to 0.55 hours, and the communication efficiency is adjusted to 0.065. Finally, an optimized device simulation parameter scheme is obtained, which includes the optimized parameter values for each device.

[0153] Step S56: Obtain the spatio-temporal feature fusion model of the device group, and perform incremental learning and update on the spatio-temporal feature fusion model of the device group according to the optimized device simulation parameter scheme to obtain the device control simulation model.

[0154] Specifically, the spatio-temporal feature fusion model obtained in step S23 can be obtained. In MATLAB, incremental learning and update are performed on the spatio-temporal feature fusion model according to the optimized device simulation parameter scheme. An incremental learning algorithm (such as an online learning algorithm) is used to update the model. Assume that the optimized device simulation parameter scheme includes: for device 1, the load efficiency is adjusted to 0.68, the task response time is adjusted to 1.1 hours, and the communication efficiency is adjusted to 0.09; for device 2, the load efficiency is adjusted to 0.73, the task response time is adjusted to 0.55 hours, and the communication efficiency is adjusted to 0.065. These optimized parameters are used as new training data and input into the spatio-temporal feature fusion model. Through the incremental learning algorithm, the model gradually adapts to the new parameter settings and updates the weights and biases of the model. For example, the fitcnsvm function in MATLAB is used to perform incremental learning and update on the SVM model. Assume that the initial weights of the model are [w1, w2, w3] and the bias is b. Through incremental learning, the weights and biases of the model are adjusted according to the new training data, and the updated weights are [w1′, w2′, w3′] and the bias is b′. Finally, the device control simulation model is obtained.

[0155] By constructing a simulation environment configuration and initializing it according to the system load warning index set, the present invention can generate a simulation initial configuration plan that conforms to the actual operating conditions, ensuring the accuracy and reliability of the simulation process. Through multi-threaded parallel simulation, it can quickly generate a set of simulation operation trajectories of the device group and extract performance feature tensors, providing comprehensive data support for performance evaluation. By evaluating the performance feature tensors and generating a performance evaluation report, it can clearly reflect the operating state of the device group during the simulation process. Through error feature extraction and sensitivity measurement, parameters that have a greater impact on the simulation accuracy can be identified. By incrementally learning and updating the spatio-temporal feature fusion model of the device group, the model parameters can be dynamically adjusted according to the optimization plan.

[0156] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to encompass all changes that fall within the meaning and scope of the equivalent elements of the application documents within the present invention.

[0157] The above description is only a specific implementation manner of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A simulation method for an automatic equipment control system, characterized in that: The following steps are involved: Step S1: collect multi-source sensor data from the device group to obtain the original data set of device operation; Perform time stamp synchronization on the original data set of device operation to obtain the device operation synchronization data set; Perform heterogeneous information fusion on the equipment operation synchronization data set to obtain the equipment operation feature matrix; Step S2: Calculate the spatial relationship of the device group based on the device operation feature matrix to obtain an initial topological diagram of the device group; Perform communication strength weight analysis on the initial topology diagram of the device group to obtain a communication weight matrix between devices; dynamically update the initial topology diagram of the device group according to the communication weight matrix between devices to obtain a dynamic topology diagram of the device group; Step S3: constructing a communication characteristic tensor for the dynamic topology diagram of the device group to obtain a device communication characteristic data set; performing information flow routing calculation based on the device communication characteristic data set to obtain a routing strategy table between devices; The communication delay distribution model is performed according to the inter-device routing strategy table to obtain the device group interaction characteristic model; Step S4: constructing a task buffer pool hierarchically for the device group interaction characteristic model to obtain device task buffer pool configuration data; According to the configuration data of the device task buffer pool, the device group is task-encoded and compressed to obtain a device task allocation plan; Evaluate the load status of the equipment task allocation plan to obtain a set of system load warning indicators; Step S5: Based on the device group interaction characteristic model and the system load early warning indicator set, the device group is simulated and run to obtain a device simulation performance evaluation report; Conduct error analysis on the equipment simulation performance evaluation report to obtain equipment simulation parameter optimization solutions; The spatiotemporal feature fusion model of the equipment group is obtained, and the spatiotemporal feature fusion model of the equipment group is updated according to the equipment simulation parameter optimization plan to obtain the equipment control simulation model.

2. The simulation method for an automatic equipment control system according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: monitor the thermal field distribution of the equipment group during operation to obtain a data set of equipment temperature field; Step S12: collecting vibration characteristics of the equipment group to obtain a time series data set of equipment vibration signals; Step S13: sampling the current of each device execution component in the device group to obtain a device current characteristic data set; Step S14: monitoring the pressure distribution of the equipment group during operation to obtain an equipment pressure field data set; Step S15: Record the device temperature field data set, the device vibration signal time series data set, the device current characteristic data set, and the device pressure field data set as the device operation original data set, and synchronize the timestamp of the device operation original data set to obtain the device operation synchronized data set; Step S16: Perform heterogeneous information fusion on the device operation synchronization data set to obtain a device operation feature matrix.

3. The simulation method for an automatic equipment control system according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: Perform a three-dimensional spatial scan on the device group to obtain point cloud data of the device operation scene; Step S22: collecting multi-angle images of the operation scene of the equipment group to obtain a multi-view image set of the operation scene, and registering and fusing the point cloud data of the equipment operation scene with the multi-view image set of the operation scene to obtain a three-dimensional mapping model of the operation scene; Step S23: Perform feature space mapping projection on the device operation feature matrix and the operation scene stereo mapping model to obtain a device group spatiotemporal feature fusion model; Step S24: Calculate the spatial distance of the device group based on the device group spatiotemporal feature fusion model to obtain the inter-device communication distance matrix, and perform spatial clustering on the device group according to the inter-device communication distance matrix to obtain the device group spatial distribution data; Step S25: constructing a minimum spanning tree for the spatial distribution data of the device group to obtain a basic connection diagram of the device group, and supplementing the device group with redundant links based on the basic connection diagram of the device group to obtain a network connectivity diagram of the device group; Step S26: reconstructing the topology of the device group network connectivity diagram to obtain an initial topology diagram of the device group; Step S27: performing communication strength weight analysis on the initial topology map of the device group to obtain a communication weight matrix between devices, and dynamically updating the initial topology map of the device group according to the communication weight matrix between devices to obtain a dynamic topology map of the device group.

4. The simulation method for an automatic equipment control system according to claim 1, characterized in that: Step S27 includes the following steps: Step S271: collecting inter-device communication traffic of the device group to obtain an inter-device communication traffic data set, and performing data packet statistics on the inter-device communication traffic data set to obtain an inter-device communication frequency matrix; Step S272: performing bandwidth utilization statistics on the inter-device communication frequency matrix to obtain a device communication bandwidth occupancy matrix, and assigning communication priorities to the device groups according to the device communication bandwidth occupancy matrix to obtain a device communication priority weight set; Step S273: performing reliability evaluation on the device communication priority weight set based on the device group spatiotemporal feature fusion model to obtain the device group link reliability index, and performing communication weight allocation on the device group based on the device group link reliability index to obtain the inter-device communication weight matrix; Step S274: dynamically updating the topology structure of the initial topology diagram of the device group according to the inter-device communication weight matrix, obtaining a device group topology update strategy, and verifying the feasibility of the device group topology update strategy to obtain a device group topology adjustment plan; Step S275: incrementally update the link structure of the initial topology map of the device group based on the device group topology adjustment plan to obtain dynamic link structure data of the device group, and reconstruct the topology structure of the dynamic link structure data of the device group to obtain a dynamic topology map of the device group.

5. The simulation method for an automatic equipment control system according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: extracting device node features from the device group dynamic topology graph to obtain a device group node feature vector set; Step S32: Dimensionally expand the device group node feature vector set to obtain a time-varying device group node feature tensor set; Step S33: constructing dynamic link features for the dynamic topology graph of the device group based on the inter-device communication weight matrix to obtain a device group link feature tensor set; Step S34: performing tensor fusion on the time-varying device group node feature tensor set and the device group link feature tensor set to obtain a device group communication characteristic tensor set; Step S35: performing timing correlation identification on the device group communication characteristic tensor set to obtain a device group timing correlation feature tensor set, and performing tensor decomposition on the device group timing correlation feature tensor set to obtain a device communication characteristic data set; Step S36: performing information flow routing calculation according to the device communication characteristic data set to obtain an inter-device routing strategy table, and performing communication delay distribution modeling according to the inter-device routing strategy table to obtain a device group interaction characteristic model.

6. The simulation method for an automatic equipment control system according to claim 5, characterized in that: Step S36 includes the following steps: Step S361: constructing a routing cost measurement model according to the device communication characteristic data set to obtain a device group routing evaluation model; Step S362: performing multi-objective optimization on the device group routing evaluation model to obtain a candidate device group routing set; Step S363: performing routing reliability measurement on the candidate device group routing set to obtain a device group routing reliability index; Step S364: Screening device group routes according to the device group route reliability index to obtain an optimal device group route solution, and load balancing the optimal device group route solution to obtain an inter-device routing policy table; Step S365: sampling the link communication delay of the device group based on the inter-device routing policy table to obtain a device group delay sampling data set, and performing probability distribution fitting on the device group delay sampling data set to obtain a device group delay distribution model; Step S366: Evaluate the communication link performance of the device group based on the device group delay distribution model to obtain a communication link performance evaluation index, and intelligently model the interaction characteristics of the device group based on the communication link performance evaluation index to obtain a device group interaction characteristic model.

7. The simulation method for an automatic equipment control system according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: collecting execution tasks of the device group to obtain a device group execution task set, and performing task complexity analysis on the device group execution task set based on the device group interaction characteristic model to obtain a device task complexity index set; Step S42: performing task priority division on the device group execution task set according to the device task complexity index set, and obtaining a device group execution task priority matrix; Step S43: performing hierarchical clustering on the device group execution task priority matrix to obtain a device group execution task hierarchy tree; Step S44: constructing a buffer pool hierarchically according to the device group execution task hierarchy structure tree to obtain a device task buffer pool; Step S45: Allocate storage space to the device task buffer pool to obtain device group buffer pool capacity configuration data; Step S46: Designing task scheduling rules according to the device group buffer pool capacity configuration data to obtain device task buffer pool configuration data; Step S47: Perform task encoding compression on the device group according to the device task buffer pool configuration data to obtain a device task allocation plan, and perform load status evaluation on the device task allocation plan to obtain a system load warning indicator set.

8. The simulation method for an automatic equipment control system according to claim 7, characterized in that: Step S41 includes the following steps: Step S411: monitor the real-time operation status of the device group to obtain a dynamic operation data stream of the device group, and extract execution tasks from the dynamic operation data stream of the device group to obtain an execution task set of the device group; Step S412: performing task type identification on the task set executed by the device group to obtain a device task type distribution set, and performing spatiotemporal correlation on the device task type distribution set to obtain a spatiotemporal feature vector of the device task; Step S413: constructing a task dependency graph according to the spatiotemporal feature vector of the equipment task to obtain an equipment task dependency network, and performing topological complexity calculation on the equipment task dependency network to obtain an equipment task topology metric; Step S414: evaluating the computing resource consumption of the device task dependent network to obtain a device group resource occupancy vector, and evaluating the load intensity of the device group according to the device group resource occupancy vector to obtain a device group load intensity index; Step S415: performing feature fusion on the device task topology metric index and the device group load intensity index to obtain a device task complexity feature set; Step S416: Identify influencing factors of the equipment task complexity feature set based on the equipment group interaction characteristic model to obtain an equipment task influencing factor set, and adjust the weight of the equipment task influencing factor set to obtain an equipment task weight vector; Step S417: quantify the complexity according to the device task weight vector to obtain a device task complexity index set.

9. The simulation method for an automatic equipment control system according to claim 7, characterized in that: Step S47 includes the following steps: Step S471: compressing feature encoding of the device task buffer pool configuration data to obtain a device task encoding dictionary; Step S472: Adaptively compressing the task set executed by the device group using the device task coding dictionary to obtain a compressed device task data set; Step S473: performing device capability matching on the compressed device task data set to obtain a device task adaptation matrix; Step S474: performing task allocation on the equipment group based on the equipment task adaptation matrix to obtain an equipment task allocation plan, and performing load prediction on the equipment task allocation plan to obtain an equipment load prediction curve; Step S475: Identify bottlenecks of the equipment group based on the equipment load prediction curve to obtain an equipment system bottleneck analysis report; Step S476: define thresholds for the equipment system bottleneck analysis report to obtain an equipment group load threshold indicator set, and construct control system warning rules based on the equipment group load threshold indicator set to obtain a system load warning indicator set.

10. The simulation method for an automatic equipment control system according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: constructing a simulation environment configuration for the device group based on the device group interaction characteristic model to obtain a device group simulation scenario parameter set, and initializing the operating conditions of the device group simulation scenario parameter set according to the system load warning indicator set to obtain an initial configuration plan for the device group simulation; Step S52: performing multi-threaded parallel simulation on the initial configuration scheme of the device group simulation to obtain a device group simulation operation trajectory set, and extracting performance indicators from the device group simulation operation trajectory set to obtain a device group performance feature tensor; Step S53: performing performance evaluation on the equipment group performance characteristic tensor to obtain an equipment group performance evaluation index set, and generating an equipment simulation performance evaluation report according to the equipment group performance evaluation index set; Step S54: extracting error features from the equipment simulation performance evaluation report to obtain an equipment group error feature matrix, and performing sensitivity measurement on the equipment group error feature matrix to obtain an equipment group parameter sensitivity vector; Step S55: solving simulation parameters of the device group based on the device group parameter sensitivity vector to obtain a device simulation parameter optimization solution; Step S56: Acquire the spatiotemporal feature fusion model of the equipment group, and perform incremental learning and update on the spatiotemporal feature fusion model of the equipment group according to the equipment simulation parameter optimization scheme to obtain the equipment control simulation model.

Images