Hanging system path planning method and system based on digital twinning

Abstract

The invention relates to a hanging system path planning method and system based on digital twinning. The method comprises the following steps: constructing a three-dimensional model of the suspension system based on a three-dimensional modeling tool; acquiring operation data of equipment in the hanging system, and mapping the operation data to the three-dimensional model to obtain a digital twinborn model; generating a target scheduling path of the lifting hook based on the operation state data of the lifting hook, the health state data of the track and the digital twinborn model; and displaying the target scheduling path by using the digital twinborn model. According to the method, through the digital twinborn technology, dynamic planning of the hanging system path can be achieved, and the problems that a traditional two-dimensional plane cannot express the orbit space relation, and path planning optimization is insufficient are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of garment intelligent manufacturing, in particular to a hanging system path planning method and system based on digital twinning. BACKGROUND

[0002] In the field of garment intelligent manufacturing, the hanging system is the core infrastructure connecting the cutting, sewing, quality inspection, warehousing and other links. Through the coordinated operation of the track, hook, sensor and control system, the automatic transmission and production process scheduling of the garment semi-finished product are realized. The path management of the traditional hanging system collects material flow data in real time through RFID, displacement sensor and other devices, and uses a two-dimensional monitoring interface to display the device state. Although this method can realize basic path scheduling, it uses a two-dimensional interface to display static icons and text to express the device state, and the path planning mainly relies on manual experience or fixed logic algorithm.

[0003] In this traditional technical implementation, the following defects exist: the two-dimensional plane cannot express the spatial topological relationship of the three-dimensional track, and lacks the presentation of the dynamic relationship in the three-dimensional space, which makes the scheduling personnel unable to intuitively judge the relative position and motion relationship of the hook in space; the fixed path strategy leads to insufficient path optimization, and the system is prone to idle waste; the path planning based on manual experience lacks data support and verification mechanism, cannot quantitatively evaluate the path advantages and disadvantages, and is prone to path conflicts. SUMMARY

[0004] Therefore, the present application provides a hanging system path planning method and system based on digital twinning, aiming to solve the problem of insufficient path planning optimization based on two-dimensional plane in the traditional hanging system.

[0005] In a first aspect, the present application provides a hanging system path planning method based on digital twinning, comprising:

[0006] constructing a three-dimensional model of the hanging system based on a three-dimensional modeling tool;

[0007] obtaining running data of devices in the hanging system, and mapping the running data to the three-dimensional model to obtain a digital twinning model; the running data of the devices includes running state data of hooks and health state data of tracks;

[0008] generating a target scheduling path of the hook based on the running state data of the hook, the health state data of the track and the digital twinning model;

[0009] displaying the target scheduling path by using the digital twinning model.

[0010] In one embodiment, the three-dimensional model of the hanging system based on the three-dimensional modeling tool comprises:

[0011] acquire three-dimensional point cloud data of the suspension system;

[0012] model the equipment in the suspension system using a LOD hierarchical modeling method based on the three-dimensional point cloud data, to generate the three-dimensional model.

[0013] In one of the embodiments, the generating the target scheduling path of the hook based on the running state data of the hook, the health state data of the track, and the digital twin model comprises:

[0014] predicting collision of the hook based on the running state data of the hook, the health state data of the track, and the digital twin model, to generate a collision risk level;

[0015] if the collision risk level is lower than a preset level, generating a target scheduling path based on the coordinate data of the hook and the health state data of the track through a path planning algorithm;

[0016] if the collision risk level is higher than the preset level, generating a detour path as the target scheduling path through a reinforcement learning algorithm.

[0017] In one of the embodiments, the running state data of the hook comprises coordinate data and motion vector data; and the predicting collision of the hook based on the running state data of the hook and the digital twin model, to generate a collision risk level comprises:

[0018] acquiring predicted coordinate data of the hook within a preset number of future frames based on the coordinate data and the motion vector data of the hook;

[0019] predicting collision of the hook based on distances between the predicted coordinate data and other equipment in the digital twin model, to generate a collision risk level.

[0020] In one of the embodiments, the generating the target scheduling path based on the coordinate data of the hook and the health state data of the track through a path planning algorithm comprises:

[0021] abstracting the track in the digital twin model as a spatial topology graph with all stations in the suspension system as nodes and paths between adjacent stations as edges, wherein the stations comprise sensor layout positions for measuring the coordinate data of the hook and positions of workstations;

[0022] setting a health degree weight of each edge in the spatial topology graph according to the health state data of the track;

[0023] acquiring a congestion level of the track in a future time period based on a traffic prediction model;

[0024] According to the congestion level of the track, a congestion level weight of each edge in the space topology graph is set;

[0025] In the space topology graph, the cost of the path from the starting node to the target node is calculated by an improved shortest path algorithm:

[0026]

[0027] wherein path is a path, for all nodes in the path, L for the physical length between each node in the path, for the weight coefficient of the physical length, for the health degree weight between each node in the path, for the weight coefficient of the health degree weight, for the congestion level weight between each node in the path, for the weight coefficient of the congestion level weight;

[0028] The path with the minimum cost is calculated as the target scheduling path.

[0029] In one embodiment, the running data further includes the hook throughput of the workstations in the hanging system and image data of the hooks and hangers in the hanging system, and the congestion level of the track in the future target period is obtained based on the traffic prediction model, including:

[0030] Based on the hook throughput and the image data, the current traffic index is determined;

[0031] According to the traffic index in the historical time, a traffic prediction model based on the Prophet algorithm is fitted;

[0032] The traffic prediction model is used to analyze the current traffic index, and the congestion level of the track in the future time period is generated.

[0033] In one embodiment, when the digital twin model displays the target scheduling path, different colors are used to distinguish the equipment and display the target scheduling path.

[0034] In a second aspect, the present application also provides a hanging system path planning system based on digital twin, comprising:

[0035] A model construction module is configured to construct a three-dimensional model of the hanging system based on a three-dimensional modeling tool;

[0036] The data acquisition module is configured to acquire operation data of equipment in the suspension system, and map the operation data to the three-dimensional model to obtain a digital twin model; the operation data of the equipment includes operation state data of a hook and health state data of a track.

[0037] The path planning module is configured to generate a target scheduling path of the hook based on the operation state data of the hook, the health state data of the track, and the digital twin model.

[0038] The data display module is configured to display the target scheduling path through the digital twin model.

[0039] In a third aspect, the present application further provides a computer device, including a memory and a processor, the memory stores a computer program, and the processor implements the steps of the method in any one of the first aspect when executing the computer program.

[0040] In a fourth aspect, the present application further provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the method in any one of the first aspect when executed by a processor.

[0041] The above suspension system path planning method and system based on digital twin can dynamically generate and visualize a scheduling path in a three-dimensional space by constructing a digital twin model and mapping real-time equipment operation data, thereby solving the problems of the traditional two-dimensional plane that cannot express the spatial relationship of the three-dimensional track and the insufficient optimization of the path planning. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the description of the embodiments of the present application or the related art will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0043] Figure 1 It is an application environment diagram of the suspension system path planning method in an embodiment;

[0044] Figure 2 It is a flowchart of the suspension system path planning method in an embodiment;

[0045] Figure 3 It is a flowchart of the step of generating a target scheduling path in an embodiment;

[0046] Figure 4 It is a flowchart of the step of generating a collision risk method in an embodiment;

[0047] Figure 5 a flowchart of steps of a path planning algorithm in an embodiment;

[0048] Figure 6 a flowchart of steps of a congestion level prediction in an embodiment;

[0049] Figure 7 a flowchart of a path planning method in a preferred embodiment;

[0050] Figure 8 a structural block diagram of a path planning system of a suspension system in an embodiment. DETAILED DESCRIPTION

[0051] In order to make the purposes, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0052] The path planning method for a suspension system based on digital twinning provided by the embodiments of the present application can be applied in an environment as shown in Figure 1 The physical device layer 10 includes suspension device entities and controllers, sensor clusters, and communicates with the server 11 through a network. The data storage system 12 stores real-time and historical data, and data to be processed such as path planning models and digital twinning models. The data storage system 12 can be integrated on the server, or placed on a cloud or other network server. The server 11 includes an intelligent analysis layer 110 and a digital twinning layer 111, which can be used to execute the path planning method for the suspension system in the embodiments. The intelligent analysis layer 110 obtains a target scheduling path using data from the data storage system 12 and state information provided by the digital twinning layer 111, and the digital twinning layer 111 maps state information of the physical device layer 10 in real time. The target scheduling path obtained by the intelligent analysis layer 110 and the device three-dimensional model in the digital twinning layer 111 are displayed to the user through the user terminal 13 application layer. The user terminal 13 can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc. The server 11 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0053] In an exemplary embodiment, as Figure 2As shown, a digital twin-based hanging system path planning method is provided. The method is applied to Figure 1 The environment in the method is described by taking steps S202 to S204 as an example:

[0054] S201, constructing a three-dimensional model of the hanging system based on a three-dimensional modeling tool.

[0055] The hanging system is a material transmission system applied to the intelligent manufacturing scene of clothing. Through the coordinated operation of the track and the hook that can slide on the track, the material or clothing is automatically transferred between stations, including but not limited to production stations in the assembly line. Path planning refers to planning the optimal running route of the hanging device from the starting station to the target station through algorithms and control logic. It can be understood that due to the dynamic change characteristics of multi-variety and small-batch of clothing orders, if the hook hanging the clothes adopts a fixed path strategy, it is easy to cause high idle rate of the entire system and collision accidents, so it is necessary to dynamically optimize the running trajectory of the hook and the material transfer sequence through path planning to improve the production line resource utilization rate and fault avoidance ability.

[0056] With the maturity of three-dimensional modeling technology, the popularization of digital twin architecture, the improvement of hardware performance, and the development of AI algorithms and big data models, it is possible to build a virtual mapping system with high fidelity. The digital twin model integrates real-time sensor data by three-dimensional modeling of physical entities, establishes a two-way mapping mechanism between physical devices and virtual models, forms a virtual space mirror model that can be updated in real time, realizes three-dimensional visual monitoring of the device, and can effectively solve the problem that the traditional two-dimensional interface cannot intuitively display the defects in the three-dimensional track space.

[0057] It can be understood that the digital twin model is composed of a three-dimensional model and a behavior model, and the three-dimensional model represents the three-dimensional structure of the physical entity, and the behavior model represents the real motion state of the physical entity. In specific implementation, the modeling tool for constructing the three-dimensional model can select BIM, Max, Blender, etc. The present example does not make specific limitations.

[0058] S202, obtaining the running data of the device in the hanging system, mapping the running data to the three-dimensional model to obtain a digital twin model; the running data of the device includes the running state data of the hook and the health state data of the track.

[0059] The running data is a necessary condition for constructing the behavior model in the digital twin model, and the digital twin model is driven by real-time running data to realize virtual-real interaction. The source of the running data is diverse. In the embodiment, the RFID sensor deployed on the equipment is used to count the material throughput of each station to realize real-time tracking and counting of the material flow, and the laser displacement sensor is used to collect the hook coordinates in real time. The industrial camera can combine computer vision models such as YOLOv8 to identify scenarios such as conveyor belt congestion and abnormal material placement, which facilitates subsequent statistics and output of data such as congestion area pixel ratio.

[0060] It should be noted that the collected running data can also include a data preprocessing step before being input into the path planning model. In the embodiment, the data preprocessing is processed by the edge node. The edge computing node can be an embedded computing unit deployed in the vicinity of the hanging equipment, which processes data in real time at the source, helps to reduce the latency of data, reduce the occupation of network bandwidth, and reduce the resource pressure of the core computing unit. The specific processing method can include Kalman filtering of displacement data to eliminate noise interference, and timestamp alignment of material throughput data to eliminate time deviation of the RFID sensor.

[0061] Through the preset data mapping rule, the above collected multi-source data is dynamically bound to the corresponding node in the digital twin model, realizing the synchronization of the running state of the virtual model and the physical equipment.

[0062] S203, generating a target scheduling path of the hook based on the running state data of the hook, the health state data of the track, and the digital twin model.

[0063] The path planning model generates a target scheduling path that meets the target optimization condition through comprehensive calculation of multi-dimensional data of the equipment. It can be understood that the optimization target of the path can include path length minimization, collision risk avoidance of the hook, and other dimensions. For different path planning targets, the path planning model can select one or more different data-driven algorithms or models.

[0064] S204, displaying the target scheduling path by using the digital twin model.

[0065] It can be understood that the terminal has a front-end display page, which displays the digital twin model through a three-dimensional scene. In specific implementation, the display content includes but is not limited to the current running state of the equipment, the highlight display of the target path trajectory, the color marking of the load state of the equipment, the highlight color marking of the congestion propagation path, and other multi-dimensional equipment information. And through gesture operations such as zooming and rotating, multi-view observation of the equipment and the path in the digital twin model can be realized.

[0066] It should be noted that in addition to being displayed on the terminal through the digital twin model to provide decision-making basis for the operator, the target scheduling path generated by the path planning model can also be directly used to control the execution equipment in the hanging system to issue control instructions to drive the hooks to run according to the specified planning path.

[0067] In the above-mentioned digital twin-based hanging system path planning method, the visualization of the three-dimensional space of the hanging equipment is realized by constructing a digital twin model, which solves the defect that the traditional two-dimensional interface cannot intuitively present the three-dimensional track relationship, so that the operator can master the hook position and track layout in real time; the target path is formed by driving the path planning model with equipment operation data, which can optimize path scheduling and improve the utilization rate of production line resources; the target path is simulated and displayed in the digital twin model, which makes it easier for the operator to understand the path planning logic and improves the decision-making efficiency.

[0068] In an exemplary embodiment, the above embodiment constructs a digital twin model of the hanging system based on a three-dimensional modeling tool in step S201, which can be constructed in the following manner:

[0069] First, three-dimensional point cloud data of the hanging system is obtained.

[0070] The three-dimensional point cloud data refers to a set of discrete points formed by collecting spatial coordinates of physical objects in the hanging system through oblique photogrammetry or laser scanning equipment. These points are based on three-dimensional coordinates and some also contain attributes such as color and reflectivity, which can accurately restore the spatial position and geometric form of the object.

[0071] Then, according to the obtained three-dimensional point cloud data, the equipment in the hanging system is modeled using the LOD hierarchical modeling method to generate a digital twin model. The LOD (Level of Detail) hierarchical modeling method is a technology that differentiates the precision of three-dimensional models according to the importance of objects in the scene and the observation distance, which can optimize model performance while ensuring visual accuracy. The modeling accuracy index is reflected by the number of facets, which is the basic unit of a three-dimensional model. The more facets, the higher the model detail, but the higher the computational load.

[0072] Specifically, in this embodiment, the equipment is classified according to importance, such as: the main track and the lifting arm are core equipment, high-precision modeling is adopted, and the number of facets is selected as 500,000; the workshop environment is a secondary key part, and the number of facets is 100,000; and the hanger, station, and other auxiliary equipment have a facet number of 2000. During modeling, the UV unfolding technology is used to map the surface of the three-dimensional model to a two-dimensional plane, and the normal map algorithm is used to compress the texture data of the model. In the above embodiment, the above method can improve the compression rate of model data while maintaining visual accuracy.

[0073] In some embodiments, in step S203, based on the running state data of the hooks, the health state data of the track, and the digital twin model, a target scheduling path of the hooks is generated, as shown in Figure 3 Specifically, the method comprises the following steps:

[0074] S301, based on the running state data of the hooks, the health state data of the track, and the digital twin model, collision prediction is performed on the hooks, and a collision risk level is generated.

[0075] It can be understood that, since the hanging system often contains a cross stereo track network, when multiple hooks move synchronously in three-dimensional space, if the path planning is not fully optimized, the probability of collision between the hooks and other hooks will be greatly increased. Moreover, the local bending of the track caused by long-term load operation and the unstable loading of materials will also cause the running track of the hooks to deviate and collide with the edges of the track and other obstacles.

[0076] Specifically, the running state data of the hooks is collected in real time by laser displacement sensors deployed in the hanging system, for accurately tracking the moving track of the hooks in the track. The track health state data is the basis for judging whether the track has potential faults such as wear and overload, and is obtained by evaluating the running state of the track equipment. In specific implementation, the running state of the track equipment can be obtained by vibration sensors, current sensors and other devices. In path planning, this data can be used to avoid fault risk areas, ensuring the safety and feasibility of path planning.

[0077] It can be understood that, since the track of the hanging equipment usually needs to cover the workshop, it is reasonable to first segment the track and then evaluate the health degree of each segment of the track. Further, the sensors for detecting the running state of the equipment should also be distributed in each segment of the track. For specific calculation of the health state data, a reasonable evaluation method can be used, such as combining vibration anomaly analysis, current and voltage anomaly analysis, etc., to detect the health of the track, and the detection results are used as the health state data.

[0078] The three-dimensional track model in the digital twin model is used to input the real-time hook coordinate data into the model, and the moving track of the hooks in the future period of time is predicted through spatial geometric calculation. It is calculated that the track overlaps with other hooks, track obstacles or workstations in the future time, and a risk level of predicted collision is generated, such as a first level of slight risk and a third level of serious risk.

[0079] S302, it is judged whether the collision risk level is higher than a preset level.

[0080] S303, if the collision risk level is lower than the preset level, a target scheduling path is generated based on the operation state data of the hook and the health state data of the track through a path planning algorithm.

[0081] It can be understood that when the collision risk level is lower than the pre-set threshold, it means that the probability of collision of the hook in the future time is small, and the hook can be moved to the target station according to the normal transportation path. At this time, the system starts the path planning algorithm, and generates the optimal scheduling path combining the current coordinates of the hook and the track health state parameters. Specifically, the path planning algorithm can select the shortest path planning algorithm such as Dijkstra algorithm, Floyd algorithm, SPFA algorithm, or other artificial intelligence algorithms with path planning ability, which can be adjusted through different specific business objectives.

[0082] S304, if the collision risk level is higher than the preset level, a detour path is generated as the target scheduling path through a reinforcement learning algorithm.

[0083] When the collision risk level is higher than the preset value, it means that the hook has a high probability of collision with other hooks or obstacles in the current moving path, and the traditional fixed path strategy usually cannot cope with the dynamically changing risk scene. On this basis, reinforcement learning, as a machine learning method that learns the optimal strategy through the interaction between the agent and the environment in a trial-and-error way, can dynamically adaptively generate strategies according to real-time data. In this embodiment, a reinforcement learning algorithm is used to simulate a detour scheme in the digital twin model, such as detouring left on the main track or detouring right on the standby track, and the optimal detour path is selected to make the hook reach the target work station.

[0084] In the path planning method of the above embodiment, the hook trajectory is predicted in real time through the digital twin model, and the strategy is dynamically switched according to the risk level. When the risk is low, the shortest path algorithm is used to ensure efficiency, and when the risk is high, the reinforcement learning is used to generate a detour path, which can maximize efficiency under the premise of safety, reduce collision accidents, and improve the reliability of the system.

[0085] In the above embodiment, the collision of the hook is predicted based on the operation state data of the hook and the digital twin model, and a collision risk level is generated, as shown in Figure 4 , which specifically includes steps S401 to S403:

[0086] S401, based on the coordinate data and the motion vector data of the hook, the predicted coordinate data of the hook within a preset number of frames in the future is obtained.

[0087] It can be understood that the running state data of the hook includes coordinate data and motion vector data, which are collected in real time by the laser displacement sensor, wherein the motion vector includes the speed and acceleration of the hook movement. The above data can be smoothed by Kalman filtering algorithm after collection, and the specific process will not be described in this embodiment.

[0088] It should be noted that in this embodiment, a BIM model is used to construct a digital twin model, and the model is divided into a cubic grid of 0.5 meters x 0.5 meters x 0.5 meters to divide the model space. Therefore, the physical coordinate values collected need to be mapped to the unified space coordinate system preset in the BIM model. In order to improve the calculation efficiency while ensuring the accuracy of collision detection, according to the physical size of the hook and the hangers on the hook, an Oriented Bounding Box (OBB) of the hook is modeled in the digital twin system.

[0089] According to the coordinates and motion vectors of the current and future predicted frames of the OBB bounding box, a prediction coordinate sequence of the future 3 frames is generated, wherein the frame number is determined by collecting the hook video stream through an industrial camera.

[0090] S402, based on the distance between the predicted coordinate data and other devices in the digital twin model, the hook is collided and the collision risk level is generated.

[0091] First, detect the obstacles in the surrounding environment of the predicted hook coordinates that will collide with the hook. Similar to step S401, the track, workstation equipment, other hooks and other entities in the digital twin model are converted into OBB bounding boxes. The spatial hash table is used to locate the grid elements intersected with the hook bounding box, and the BIM model objects in the grid are detected. In specific implementation, since the BIM model decomposes the track, workstation and other entities into triangular patches, a ray can be emitted from the predicted position to the motion direction to detect the intersection point of the ray and the BIM model triangular patch. If the intersection point exists and the distance is less than a safety threshold, it means that there is an obstacle in the focus position that will collide with the future trajectory of the hook.

[0092] Next, based on the distance between the obstacle and the predicted hook coordinates, the collision risk level is determined. By calculating the distance between the detected obstacle and the predicted hook coordinates, the collision risk can be graded. In addition to the interval to which the distance belongs, the collision risk level can also be further calculated in combination with the track health state parameter.

[0093] In the above embodiment, by predicting the collision of the hook in the BIM model and calculating the risk level based on the distance of the obstacle and the track health score, the accuracy of the hook collision risk prediction can be effectively improved.

[0094] In some embodiments, based on the coordinate data of the hooks, the health state data of the tracks, a target scheduling path is generated by a path planning algorithm, as shown in Figure 5 The steps S501 to S504 are shown as follows:

[0095] S501, taking all the stations in the hanging system as nodes and the paths between adjacent stations as edges, the tracks in the digital twin model are abstracted into a spatial topology graph.

[0096] The stations include the workstations in the production line, such as the production workstations, buffer areas, and quality inspection points, and the collection points of the coordinate data of the hooks.

[0097] It can be understood that the spatial topology graph can efficiently express the spatial connection logic, and the path search is simplified from three-dimensional space calculation to the shortest path problem in graph theory. Each workstation in the hanging system is taken as a node in graph theory, and the track path between adjacent workstations is defined as an edge to construct a three-dimensional spatial topology graph for subsequent path planning algorithm processing. Specifically, each node records the physical coordinates of the workstation, each edge marks the physical length of the path (accurate to 0.1m), and the connection relationship represents the passing attribute, such as one-way passing between node A (workstation A) and node B (workstation B).

[0098] S502, according to the health state data of the tracks, setting the health degree weight of each edge in the spatial topology graph.

[0099] The specific setting of the health state weight can adopt a reasonable evaluation method, such as converting the health state data to a suitable scoring system, which is not limited in this embodiment.

[0100] S503, based on the traffic prediction model, obtaining the congestion level of the tracks in the future time period.

[0101] The traffic prediction model includes but is not limited to models that can output regression prediction, such as time series models and deep learning models. Then, a mapping relationship between the congestion level and the weight coefficient is established, and the congestion level data is converted to a suitable scoring system, which is not limited in this embodiment. It can be understood that the congestion level on the tracks will affect the order priority, which includes the delivery deadline urgency, customer level, production process priority, etc. For example, when the clothes order transported by the hook has high priority, a route with higher idle degree and lower congestion level should be selected for it.

[0102] S504, in the spatial topology graph, the cost of the path from the starting node to the target node is calculated by the improved shortest path algorithm.

[0103] Specifically, the path cost calculation formula is as follows:

[0104]

[0105] wherein path is a path, L is all nodes in a path, L is the physical length between nodes in a path, L is the weight coefficient of the physical length, L is the health degree weight between nodes in a path, L is the weight coefficient of the health degree weight, L is the congestion level weight between nodes in a path, L is the weight coefficient of the congestion level weight.

[0106] It is worth noting that in the present embodiment, the Dijkstra algorithm is improved by assigning a health degree and a congestion level to each workstation node as a weight. Among them, the physical length weight is , The item can make the path planning prefer to choose a short distance route, The item makes the algorithm avoid low health degree track sections, It can make the algorithm tend to follow the order priority, and the higher the priority, the faster the route selection. For example, the node with high health degree weight and high congestion level weight will significantly increase the path cost, although the actual physical path of this path is short, but it is not the best transportation path.

[0107] S505, calculate the path with the minimum cumulative cost as the target scheduling path.

[0108] By traversing all possible paths in the topology graph, the cumulative cost of each path is calculated, and the path with the minimum cost is selected as the target scheduling path. For example, when a hook needs to go from workstation A to workstation D, the algorithm compares the paths A→B→D (cost 75), A→C→D (cost 68), A→B→C→D (cost 92), etc. After comparison, A→C→D with a cost of 68 is selected as the optimal path. The path generated in this way can not only ensure a relatively short physical distance, but also avoid low health degree track sections, achieving a balance between efficiency and safety at the algorithm level, and solving the high failure rate problem caused by ignoring the device state in the traditional fixed path strategy.

[0109] In step S304 of the above example, the generation of the detour route by the reinforcement learning algorithm can be implemented in the following way:

[0110] First, collect running data from multiple sensors, including hook state data and environment state data, wherein the hook state data includes hook position, hook speed, and hook load, and the environment state data includes the position and speed of the obstacle.

[0111] Preferably, the above operation data specifically includes: hook three-dimensional coordinates (X, Y, Z), velocity vector (Vx, Vy, Vz), load weight, target work station coordinates obtained by a laser displacement sensor; an industrial camera combined with a target detection algorithm identifies the positions and speeds of three obstacles (such as other hooks, work station equipment) around the hook. After normalization processing, these data are constructed into a multi-dimensional state vector.

[0112] The hook state data and the environment state data are input into a reinforcement learning obstacle avoidance model to obtain the bypass path. In this embodiment, the reinforcement learning algorithm obstacle avoidance model selects a TwinDelayed Deep Deterministic Policy Gradient (TD3) algorithm, which is a deterministic deep reinforcement learning algorithm combining a deep deterministic policy gradient algorithm and double Q learning. The training and deployment process of the model includes the following steps:

[0113] (1) Offline pre-training part: the TD3 model is trained using historical operation record data containing the above multi-dimensional state vector. The historical record contains a four-tuple of (state, action, reward, next state). In addition, using the pre-trained model enables the algorithm to quickly master the obstacle avoidance rules.

[0114] (2) The trained TD3 model is deployed in a digital twin environment. When the system detects a potential collision risk, the current multi-dimensional state vector is input into the trained TD3 model, and the model outputs the optimal action. It is worth noting that the optimal action can be converted into a control instruction through an edge computing node and drive the hook to execute bypass.

[0115] In the above embodiment, the TD3 deep reinforcement learning algorithm can effectively improve the accuracy of the obstacle avoidance strategy compared with the traditional reinforcement learning algorithm. Combined with offline pre-training and online optimization of digital twins, it can quickly learn historical obstacle avoidance experience and adapt to the dynamic changes of the track environment, thereby improving the safety and efficiency of the hanging system operation.

[0116] In some exemplary embodiments, in step S503, the operation data further includes the hook throughput of the work station in the hanging system and the image data of the hooks and hangers in the hanging system, and the specific steps are as shown in Figure 6 , including steps S601 to S603:

[0117] S601, based on the hook throughput and the image data of the hooks and hangers in the hanging system, determine the current traffic index.

[0118] The hook throughput of the work station is the number of hooks in or out of each work station recorded by the RFID sensor. For the image data of the hooks and hangers in the hanging system, as preferred, in this embodiment, an industrial camera is used in combination with a YOLOv8 target detection algorithm to calculate the ratio of the number of pixels in the congestion area to the total number of pixels in each frame of image of the hooks and hangers suspended by the hooks in the track, to obtain the congestion pixel ratio, and define the congestion level of the conveying belt according to the ratio. The data can also be processed by data cleaning, data normalization, etc., and converted into a standard format that can be processed by the algorithm, and the above two types of data are fused to obtain comprehensive flow index data.

[0119] It should be further noted that in order to train the Prophet model, the comprehensive flow index data in the historical time is needed, and the above data are all time series data, each data needs to include its corresponding time stamp.

[0120] S602, fitting a flow prediction model based on the Prophet algorithm according to the comprehensive flow index in the historical time.

[0121] The Prophet algorithm is a time series prediction model, and the Prophet is based on an additive model, which is particularly suitable for processing data affected by multiple periods with seasonal, holiday effects and trend changes. As a preferred embodiment, the formula of the Prophet model can be written as:

[0122]

[0123] wherein, is the trend term, which captures the inflection point of the flow caused by the sudden surge of orders through piecewise linear regression; is the seasonal term, which uses Fourier series to fit the daily / weekly production cycle, such as a 20% increase in material flow during the morning rush hour; is the holiday term, such as a 30% decrease in the predicted flow of the conveying belt on special dates. Alternatively, the noise term is used to represent unpredictable random fluctuations.

[0124] The processed comprehensive flow index data in the historical event is input into the Prophet model for fitting, and the model automatically identifies the time points where the trend changes significantly and uses an optimization algorithm to fit the parameters of the model. For the adjustable parameters in the Prophet model, such as the trend model type, the number of change points, whether to enable annual, weekly, and daily seasonality, etc., can be adjusted according to actual needs, and the prediction trend output by the model is more consistent through parameter adjustment.

[0125] S603, using the flow prediction model to analyze the current flow index to generate the congestion level of the track in the future time period.

[0126] After the Prophet model is trained, the model can be used to predict the congestion trend curve of the future time period, such as the next 24 hours, with one data point predicted every 15 minutes. The predicted value is compared with the preset threshold to generate a congestion risk level. For example, according to the model prediction result, the comprehensive flow index at a certain time is 0.78, which exceeds the preset value 0.7 of high risk, and is marked as high congestion risk, with a congestion level of three, and the risk duration is calculated at the same time.

[0127] The congestion risk level can not only be used for path planning, but also be displayed in the digital twin model interface with the prediction result. Specifically, the high-risk area can be marked with a red area and the predicted congestion time can be labeled.

[0128] In some exemplary embodiments, when the digital twin model displays the target scheduling path, different colors are used to distinguish the equipment and display the target scheduling path.

[0129] Specifically, different color mappings can be set for equipment types and states in the system, such as gray for the main track and blue lines for the target scheduling path. When the hook is in the transportation process, the path is colored according to the equipment passed. In addition, the operation interface of the digital twin system is also used to display the congestion trend curve and congestion risk obtained by the flow prediction model analysis, as well as the execution result feedback after the detour route is executed.

[0130] Through the above visualization method, the complex scheduling logic is converted into intuitive colors and dynamic pictures, so that the management personnel can quickly identify the transportation path and path conflict, and improve the decision-making efficiency.

[0131] In order to facilitate understanding, a preferred embodiment is provided to explain the path planning method of the hanging system based on digital twin in the present application, as shown in Figure 7 The flow of the method includes the following steps:

[0132] S701, a three-dimensional model of the hanging system is constructed based on a three-dimensional modeling tool.

[0133] S702, the running data of the equipment in the hanging system is obtained, the running data is input into the three-dimensional model, and a digital twin model is obtained.

[0134] S703, the current flow index is determined based on the hook throughput and the image data of the hook and the hanger.

[0135] S704, a flow prediction model based on the Prophet algorithm is fitted according to the flow index in the historical time.

[0136] S705, analyzing the current traffic index by using the traffic prediction model to generate a congestion level of the track in a future time period.

[0137] S706, predicting a predicted hook coordinate in a future preset number of frames based on the coordinate data of the hook and the motion vector.

[0138] S707, performing collision prediction on the hook based on the predicted coordinate data and the distance between the hook and other devices in the digital twin model, and generating a collision risk level.

[0139] S708, determining whether the collision risk level is higher than a preset level.

[0140] S709, if the collision risk level is lower than the preset level, generating a target scheduling path based on the coordinate data of the hook, the health state number of the track, and the congestion level of the track by using a path planning algorithm.

[0141] S710, if the collision risk level is higher than the preset level, generating a detour path as the target scheduling path by using a reinforcement learning algorithm.

[0142] S711, displaying the target scheduling path by using the digital twin model.

[0143] It should be understood that, although each step in the flowchart involved in each of the above-described embodiments is displayed in sequence according to the direction of the arrow, these steps are not necessarily executed in sequence according to the direction of the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in each of the above-described embodiments can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0144] Based on the same inventive concept, the embodiments of the present application also provide a digital-twin-based suspension system path planning system. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme described in the above method, so the specific limitations in one or more embodiments provided below can refer to the limitations of the digital-twin-based suspension system path planning method described above, and will not be repeated here.

[0145] In one exemplary embodiment, as shown in Figure 8 a digital-twin-based suspension system path planning system 800 is provided, which includes a model construction module, a data acquisition module, a path planning module, and a data display module, wherein:

[0146] a model construction module 801, configured to construct a digital twin model of the suspension system based on a three-dimensional modeling tool;

[0147] a data acquisition module 802, configured to acquire operation data of equipment in the suspension system, and input the operation data into the digital twin model;

[0148] a path planning module 803, configured to generate a target scheduling path in the digital twin model based on the operation data and using a path planning model;

[0149] a data display module 804, configured to display the target scheduling path through the digital twin model.

[0150] The various modules in the above suspension system path planning system based on digital twin can be all or partially implemented through software, hardware, and combinations thereof. The various modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so as to be called and executed by a processor to perform operations corresponding to the various modules.

[0151] Those skilled in the art can understand that, Figure 8 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. A specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0152] In one embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0153] In one embodiment, a computer readable storage medium is also provided, storing a computer program, and the computer program is executed by a processor to implement the steps in the above method embodiments.

[0154] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., without being limited thereto.

[0155] The technical features of the above embodiments can be combined in any manner. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present application.

[0156] The above-described embodiments are merely illustrative of several embodiments of the present application, and the description is relatively specific and detailed, but should not be understood as a limitation on the scope of the patent. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of protection of the present application should be subject to the appended claims.

Claims

1. A method for path planning of a suspension system based on digital twinning, characterized in that, The method comprises: constructing a three-dimensional model of the hanging system based on a three-dimensional modeling tool; obtaining running data of equipment in the hanging system, and mapping the running data to the three-dimensional model to obtain a digital twin model; the running data of the equipment comprises running state data of a hook and health state data of a track; generating a target scheduling path of the hook based on the running state data of the hook, the health state data of the track, and the digital twin model; displaying the target scheduling path by using the digital twin model.

2. The method of claim 1, wherein, The method comprises: obtaining three-dimensional point cloud data of the hanging system; based on the three-dimensional point cloud data, using an LOD hierarchical modeling method to model the equipment in the hanging system, and generating the three-dimensional model.

3. The method of claim 1, wherein, The method comprises: based on the running state data of the hook, the health state data of the track, and the digital twin model, performing collision prediction on the hook to generate a collision risk level; if the collision risk level is lower than a preset level, generating a target scheduling path by a path planning algorithm based on the running state data of the hook and the health state data of the track; if the collision risk level is higher than the preset level, generating a detour path as the target scheduling path by a reinforcement learning algorithm.

4. The method of claim 3, wherein, The running state data of the hook comprises coordinate data and motion vector data; the collision prediction on the hook based on the running state data of the hook and the digital twin model to generate a collision risk level comprises: based on the coordinate data and the motion vector data of the hook, obtaining predicted coordinate data of the hook within a preset number of future frames; based on the distance between the predicted coordinate data and the coordinate data of other equipment in the digital twin model, performing collision prediction on the hook to generate a collision risk level.

5. The method of claim 3, wherein, The method comprises: taking all stations in the hanging system as nodes, paths between adjacent stations as edges, and the tracks in the digital twin model as a spatial topology graph, wherein the stations comprise workstations in a production line and coordinate data acquisition points of the hook; setting a health degree weight of each edge in the spatial topology graph according to the health state data of the track; obtaining a congestion level of the track in a future time period based on a traffic prediction model; setting a congestion level weight of each edge in the spatial topology graph according to the congestion level of the track; in the spatial topology graph, calculating a cost of a path from a starting node to a target node by an improved shortest path algorithm: wherein, path is a path, L is all nodes in the path, L is a physical length between nodes in the path, L is a weight coefficient of the physical length, L is a health weight between nodes in the path, L is a weight coefficient of the health weight, L is a congestion level weight between nodes in the path, L is a weight coefficient of the congestion level weight. calculating the path with the minimum cost as the target scheduling path.

6. The method of claim 5, wherein, The running data further comprises a hook throughput of a workstation in the hanging system and image data of the hook and a hanger in the hanging system; the method comprises: obtaining a congestion level of the track in a future target time period based on a traffic prediction model. determine a current traffic index based on the hook throughput and the image data; fit a traffic prediction model based on a Prophet algorithm according to the traffic indexes in a historical time; analyze the current traffic index by using the traffic prediction model to generate a congestion level of the track in a future time period.

7. The method of claim 1, wherein, The digital twin model distinguishes the equipment by using different colors when displaying the target scheduling path.

8. A path planning system for a suspended system based on digital twins, characterized in that, Comprise: a model construction module, configured to construct a three-dimensional model of a suspension system based on a three-dimensional modeling tool; a data acquisition module, configured to acquire operation data of equipment in the suspension system, and map the operation data to the three-dimensional model to obtain a digital twin model; the operation data of the equipment comprises running state data of a hook and health state data of a track; a path planning module, configured to generate a target scheduling path of the hook based on the running state data of the hook, the health state data of the track, and the digital twin model; a data display module, configured to display the target scheduling path through the digital twin model. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor executes the computer program to realize the steps of the method in any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method in any one of claims 1 to 7.

Images