Material process mobile management and control method and system based on AToT

Abstract

The invention relates to the field of industrial automation, and discloses an AToT-based material process mobile management and control method and system, and the method comprises the following steps: collecting a station process state in real time through an AToT network, and constructing a dynamic weight factor generation model; task allocation weights are dynamically optimized based on deep reinforcement learning, and a distributed bidding mechanism is driven to realize cooperative task matching of multiple mobile devices; according to the path curvature and the historical congestion degree, elastic fragments are dynamically divided, and a driving path taking control precision and resource efficiency into consideration is generated; and forming a closed-loop resource scheduling link through a priority-driven fragmentation preemption mechanism and predictive path re-planning. According to the method, quick response of high-priority tasks, dynamic optimal configuration of path resources and real-time resolution of system conflicts can be realized, and the method is suitable for high-dynamic industrial scenes such as automobile manufacturing and electronic assembly.

Description

Technical Field

[0001] The present invention relates to the field of industrial automation technology, and specifically to a method and system for mobile control of material processes based on AToT. Background Art

[0002] In the current industrial automation scenario, as the core link connecting discrete manufacturing processes, the scheduling efficiency of the material transportation system directly affects the overall efficiency of the production line. With the increasing demand for flexible manufacturing, the dynamic fluctuations of process states in the multi-variety and small-batch production mode have intensified, and the traditional logistics control methods based on fixed priority rules or static path partitioning have gradually revealed the defect of insufficient adaptability.

[0003] In the prior art, the task allocation of mobile devices mostly depends on preset weight coefficients, which are difficult to dynamically respond to changes in process urgency and are prone to cause delays in high-priority tasks; the path planning adopts an equal-slice strategy, ignoring the coupled influence of path curvature and real-time load, and is prone to congestion and collision risks in complex scenarios.

[0004] In addition, there is a response delay in resource contention detection and conflict resolution under the centralized scheduling architecture, which is difficult to meet the real-time requirements of high-concurrency logistics tasks.

[0005] How to achieve dynamic optimization of logistics resources and deep coordination of process states has become the key bottleneck restricting the evolution of industrial automation systems towards intelligence. Summary of the Invention

[0006] In view of the deficiencies of the prior art, the present invention provides a method and system for mobile control of material processes based on AToT, which solves the problems of response delay, congestion, and low collaborative efficiency of multiple mobile devices caused by static resource allocation and path planning in the prior art.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for mobile control of material processes based on AToT, comprising the following steps: Obtain the real-time process states of multiple workstations, where the process states include remaining processing time, priority, and fault flag; Dynamically generate a weight factor according to the process states, where the weight factor is used to quantify the time cost of path planning, process waiting penalty, path congestion degree, and conflict risk; In response to the material demand triggered by a workstation, allocate a target mobile device for the material demand through a distributed bidding mechanism; Dynamically divide path slices based on path curvature and real-time congestion degree to generate the driving path of the target mobile device; Monitor the occupancy status of path slices, and trigger slice preemption and path replanning according to the priority.

[0008] Preferably, the real-time process status is broadcast through the AToT network, and the priority of the work station is dynamically adjusted according to the production plan.

[0009] Preferably, the weight factor is dynamically generated by a deep reinforcement learning model, and the input of the model includes a global process status vector, which is composed of the average remaining processing time, the average priority, and the proportion of faulty work stations.

[0010] Preferably, the step of allocating the target mobile device for the material requirement through a distributed bidding mechanism includes: The mobile device calculates the bidding cost, which includes a conflict risk factor that is positively correlated with the number of currently participating bidding devices; If a bidding conflict of multiple mobile devices for the same task is detected, select the mobile device with the lowest bidding cost and the earliest time stamp.

[0011] Preferably, the calculation formula of the bidding cost is:

[0012] Wherein, is the estimated arrival time, is the remaining processing time of the work station, is the path congestion degree, is the number of participating bidding devices, , , , is the dynamic weight factor.

[0013] Preferably, the step of dynamically dividing the path into segments based on the path curvature and the real-time congestion degree to generate the driving path of the target mobile device includes: Calculate the curvature factor according to the turning radius of the current path segment of the mobile device, and the curvature factor is the reciprocal of the turning radius; Obtain the historical congestion degree of the path segment, and the historical congestion degree is the proportion of the time when the segment is occupied; Calculate the dynamic segment length according to the curvature factor and the historical congestion degree, and the calculation formula is: Wherein, is the curvature factor, is the historical congestion degree, is the preset basic segment length, is the maximum allowable segment length; Divide the path segment boundary according to the dynamic segment length to generate the driving path of the target mobile device.

[0014] Preferably, the step of triggering segment preemption according to the priority includes: Obtain the estimated arrival time of the high-priority task mobile device at the target shard and the departure time of the currently occupied device; Determine whether the preemption condition is satisfied , and the priority of the high-priority task is greater than the priority of the currently occupied task; where is the estimated arrival time of the high-priority task mobile device at the target shard, is the departure time of the currently occupied device; If the preemption condition is satisfied, trigger the currently occupied device to release the shard and perform path replanning; Update the subsequent task priority of the currently occupied device according to the number of preemption times, and the update formula is:

[0015] where is the updated task priority of the mobile device , is the original task priority of the mobile device , is the cumulative number of preemption times, is the preset compensation coefficient.

[0016] Preferably, the path replanning includes the following steps: Predict the future occupancy rate according to the current shard occupancy data, and the calculation formula is:

[0017] where is the predicted future shard occupancy rate, is the current shard occupancy rate, is the first-order derivative of the current shard occupancy rate with respect to time, is the prediction time window; If , suppress the bidding request to the target shard area, where is the preset maximum allowable occupancy rate threshold; Generate alternative paths that meet the curvature constraints, and the shard division of the alternative paths satisfies: Take the minimum value; where is the dynamic length of the th shard, is the predicted occupancy rate of the th shard, is the total number of shards of the alternative path.

[0018] Preferably, the mobile device is an AGV, an autonomous mobile robot or an automated guided vehicle.

[0019] The present invention also provides a material process movement control system based on AToT, including: A process status perception module for obtaining and broadcasting the real-time process status of workstations; A dynamic decision-making module for generating weight factors and performing distributed bidding; A path coordination module for dynamically dividing path shards and managing shard preemption; An exception handling module for monitoring shard conflicts and triggering path replanning.

[0020] The present invention provides a material process movement control method and system based on AToT. It has the following beneficial effects: 1. Through the process status perception and dynamic weight generation mechanism of the present invention, the system can real-time identify high-priority task requirements. Combining the distributed bidding and shard preemption strategies, it realizes the optimal dynamic allocation of logistics resources. When a sudden high-priority task is inserted, it quickly releases key path resources to ensure the continuous execution of emergency processes and avoid production interruptions.

[0021] 2. Based on the dynamic shard planning algorithm of curvature and congestion degree, the present invention automatically reduces the shard granularity in areas with high path curvature to reduce the risk of motion control errors; and dynamically compresses the shard length in congested areas to improve the path resource reuse rate.

[0022] 3. Through the bidding conflict resolution rule of time sequence priority and the priority compensation mechanism of the present invention, it effectively avoids the deadlock problem caused by resource contention of multiple mobile devices. The low-priority tasks that are frequently preempted gradually increase their competitiveness through the compensation coefficient, forming a task scheduling system that takes into account both efficiency and fairness.

[0023] 4. Using the occupancy change rate prediction model, the present invention actively triggers bidding inhibition and path replanning before the path shards reach the congestion threshold, advancing the congestion prevention window and significantly reducing the risk of congestion spread. Combining with the alternative path generation algorithm with curvature constraints, it ensures the controllability and safety of the detour path.

[0024] 5. Through the distributed architecture design of the AToT network, the present invention decouples the computational loads such as global state perception, local bidding decision-making, and edge conflict verification in layers, while ensuring a millisecond-level response speed, avoiding the computational bottleneck of the centralized server. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of the method flow of the present invention; Figure 2 It is a schematic diagram of the AToT Internet of Things architecture of the present invention; Figure 3 It is a schematic diagram of the system structure of the present invention.

[0026] Among them, 10 is the process status perception module; 20 is the dynamic decision-making module; 30 is the path coordination module; 40 is the exception handling module. Specific implementation mode

[0027] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0028] Please refer to the attached Figure 1 - attached Figure 2 , the present invention provides a method for controlling the movement of materials in processes based on AToT, which can real-time perceive the production process status through Internet of Things technology, dynamically coordinate the path planning and task allocation of mobile devices, and solve the real-time coordination problem between logistics and production processes.

[0029] As Figure 1 shown, the implementation of the present invention is based on the AToT (All Things on Things) Internet of Things architecture, deployed in an industrial automation scenario, including: Station equipment: The processing unit deployed on the production line, integrated with AToT nodes, which can real-time collect and broadcast the process status.

[0030] Mobile device: Equipped with an edge computing unit, which can receive the station status through the AToT network and execute distributed bidding and dynamic path planning.

[0031] AToT network: Adopting a star topology structure to achieve millisecond-level data synchronization among stations, mobile devices, and edge nodes.

[0032] Edge computing node: Deployed at the workshop level, responsible for local conflict resolution and shard occupancy prediction calculation, reducing the dependence on the cloud.

[0033] Cloud platform: Receiving the global exception events reported by the edge nodes, updating the DRL model parameters, and distributing them to all mobile devices.

[0034] As Figure 2 shown, the method for controlling the movement of materials in processes based on AToT may include the following steps: S1. Obtain the real-time process status of multiple stations; S2. Dynamically generate weight factors according to the process status; S3. Allocate tasks through a distributed bidding mechanism; S4. Dynamically divide path shards to generate a driving path; S5. Monitor the occupancy status of path shards, and perform shard preemption and path replanning.

[0035] The following is a detailed description of each step in the method of the present invention, comprehensively elaborating on the specific implementation principles, technical details, and processes for each step.

[0036] For step S1, in the present invention, the acquisition of process status is achieved through the AToT Internet of Things architecture. Each station device (including production units such as machine tools and assembly tables) integrates an AToT node, and the AToT node includes a multi-source sensor module, a data processing unit, and a wireless communication module.

[0037] Exemplarily, the sensor module collects station operation data, including a remaining processing time measurement sensor, a device status detection sensor, and a production plan interface. The remaining processing time measurement sensor obtains the workpiece processing progress in real time through a laser rangefinder or an encoder, and calculates the remaining processing time in combination with preset processing process parameters. .

[0038] The device status detection sensor monitors the voltage, temperature, and vibration data of the station device, and generates a fault flag when an abnormal value is detected. . The production plan interface is docked with the upper-layer MES system to receive the dynamically adjusted process priority parameters. .

[0039] The data processing unit of the AToT node preprocesses the original data to generate a structured status data MANE packet:

[0040] The data packet is broadcast to the network at a fixed period through the wireless communication module. Preferably, the communication protocol adopts the IEEE802.15.4 standard.

[0041] The mobile device and the edge computing node listen to and parse the data packet through the AToT network to establish a global station status mapping table. The mapping table contains all station IDs and their corresponding real-time status triples. .

[0042] Furthermore, the edge computing unit of the mobile device calculates the global status vector based on the received station status data. . The specific calculation process is as follows: 1. Calculation of average remaining processing time:

[0043] Where, is the total number of valid stations, indicates that the station is in the processing state.

[0044] 2. Calculation of average priority:

[0045] The average priority calculation excludes the faulty workstations, that is, only the priority parameters of ( ) are counted.

[0046]

[0047] Among them, is the number of faulty workstations.

[0048] The global state vector is locally stored by the mobile device and is used for the subsequent dynamic generation of the weight factor. Exemplarily, when is detected, an abnormal state flag is triggered to notify the edge node to start the alternative path planning strategy.

[0049] This embodiment realizes the perception and transmission of the process state through the above technical solution, providing a data basis for subsequent dynamic decision-making. The AToT network architecture has the characteristics of low latency and high reliability, and can adapt to the complex electromagnetic environment in the industrial field. The calculation and storage process of the process state are completed on the edge side of the mobile device, avoiding the single-point failure risk of the centralized server.

[0050] For step S2, in the present invention, the dynamic generation of the weight factor is realized through a deep reinforcement learning (DRL) model. The input of the DRL model is the global state vector calculated in step S1 , and its parameters , , respectively come from the statistical results of the remaining processing time, priority, and fault status of the workstations obtained in real time in step S1. The output is a four-dimensional weight factor vector , where: is the time cost weight factor, which is used to quantify the impact of the travel time of the mobile device on the cost; is the process waiting penalty weight factor, which is used to quantify the penalty of the idle time of the workstation on the cost; is the path congestion degree weight factor, which is used to quantify the amplification effect of path congestion on the cost; is the conflict risk weight factor, which is used to suppress the risk of multiple devices bidding for the same task.

[0051] The generation and optimization of this weight factor rely on the global workstation state vector calculated in step S1, ensuring that in a dynamic production environment, path planning can make optimized decisions based on real-time process priorities, workstation fault conditions, and other key production parameters.

[0052] Exemplarily, the DRL model adopts an Actor-Critic architecture, including a fully connected neural network and a policy gradient optimization algorithm, and its training and inference processes are implemented through the following technical solutions.

[0053] During the model training phase, first, the mapping relationship between the state space and the action space is constructed. The number of neurons in the input layer is 3, corresponding to the average remaining processing time , average priority , and the proportion of faulty workstations in the global state vector respectively. The number of neurons in the output layer is 4, representing the normalized values of the weight factors , , , . On this basis, a reward function is designed to guide the optimization direction of the model:

[0054] where is the average waiting time of the station materials, calculated by the number of stations with unmet demands during the historical data statistics period; is the standard deviation of the mobile device load, reflecting the task allocation balance; and are preset proportionality coefficients, used to balance the optimization weights of the waiting time and load balance.

[0055] The core objective of this reward function design is to balance the station waiting time and the device load balance, ensuring that the generated weight factors can timely reflect the priority weights between different tasks and processes during the production process, while considering the risks of system congestion and conflicts.

[0056] During the model inference phase, the mobile device loads the pre-trained model parameters through the edge computing unit. After inputting the calculated in real time into the network, the hidden layer and output layer calculations are performed in sequence:

[0057]

[0058] where and are the weight matrix and bias vector of the hidden layer, and are the output layer parameters, RelU (Rectified Linear Unit) is the activation function, and the Softmax function normalizes the output into a probability distribution. Through this calculation process, the weight factors can dynamically adapt to the current production environment state.

[0059] To further optimize the model adaptability, an online update mechanism is adopted. The mobile device continuously collects real-time data during operation, including , and the number of path conflicts . These real-time data update the model parameters through the stochastic gradient descent algorithm. After accumulating a preset number of data samples, the model parameters are updated through the stochastic gradient descent algorithm:

[0060] Here, represents the set of model parameters , is the policy gradient loss function, is the learning rate.

[0061] This online update mechanism ensures that the DRL model can continuously adapt to changes in the production environment, such as factors like process priorities, equipment failures, and task conflicts, providing accurate and real-time decision support for subsequent path planning.

[0062] This embodiment realizes the dynamic generation and optimization of the weight factor through the above technical solution, and can adapt to dynamic disturbances such as changes in process priorities and equipment failures in the production environment. Based on the DRL-based multi-objective optimization mechanism, it can achieve the balance of time cost, process waiting penalty, congestion degree, and conflict risk in path planning.

[0063] For step S3, in the present invention, the distributed bidding mechanism realizes the dynamic allocation of material requirements through the following technical solution.

[0064] When the station equipment triggers a material requirement, based on the real-time status data of the station obtained in step S1 , first broadcast a task request packet through the AToT network, and its data structure is defined as:

[0065] Wherein, is the unique identifier of the station, is the remaining processing time, is the task priority, is the latest arrival time of the material, is the global synchronization timestamp based on the IEEE 1588 protocol.

[0066] Mobile devices (including AGVs, autonomous mobile robots, automated guided vehicles, etc.) listen for task requests in the AToT network and trigger the bidding cost calculation process.

[0067] The bidding cost calculation uses the dynamic weight factor generated in step S2 , and the specific formula is:

[0068] In the formula: represents the mobile device from the current location to the work station of the estimated time, is the real-time speed of the device; , is the path congestion degree, representing the real-time occupancy rate of the path shard, which is obtained by statistically analyzing the shard status data synchronized by the AToT network; is the number of mobile devices currently participating in the task bidding, which is statistically analyzed in real time by parsing the number of MAC addresses carrying the task ID in the AToT network.

[0069] After the mobile device generates a bidding message, it broadcasts it to the edge computing node through the AToT network. The data structure of the bidding message is:

[0070] Among them, is the unique identifier of the mobile device (such as MAC address or device ID).

[0071] The edge node performs conflict detection on the bidding message. When it detects that there are multiple bidding messages for the same task satisfying the timestamp conflict condition:

[0072] When, the conflict resolution rule is triggered. Among them, , represent different mobile devices and for the same task of the bidding message timestamps, is the timestamp conflict threshold.

[0073] The conflict resolution rule is executed according to the following priorities: Cost priority principle: Select the mobile device with the lowest bidding cost, that is ; Timing priority principle: When the costs of multiple devices are the same, select the bidding message with the earliest global timestamp earliest.

[0074] The mobile device that wins the bid receives a confirmation instruction through the AToT network and locks the task execution right. The devices that are not selected update the bidding parameters locally, including adjusting the predicted occupancy rate of the path shard and the conflict risk factor , and re - participate in the bidding for other tasks.

[0075] Through the above - mentioned technical solution, this embodiment realizes the dynamic task allocation of multiple mobile devices, and can adjust the bidding strategy according to the real - time working conditions. The conflict resolution mechanism based on time stamps is combined with the cost optimization model, which has the technical effects of avoiding resource contention and improving the task response efficiency.

[0076] For step S4, in the present invention, the dynamic path sharding planning is used to divide the driving path of the mobile device into manageable dynamic shards, and its core role is to balance the path control accuracy and the system scheduling efficiency.

[0077] First, the mobile device calculates the geometric curvature of the path segment based on the station position data obtained in step S1 and its own real - time position . The curvature factor The turning radius in is obtained through the lidar point cloud fitting circular arc algorithm, and its calculation depends on the relative position relationship between the station and the mobile device. On this basis, when the priority of the station is relatively high, the mobile device will select a low - curvature path to reduce the error caused by path bending.

[0078] Subsequently, the mobile device obtains the historical congestion degree of the target path shard from the AToT network . The historical congestion degree is the proportion of the time occupied by the shard within the recent time window:

[0079] In the formula, is a binary function, which takes 1 when the shard is occupied at time and 0 otherwise; is the total number of sampling times within the statistical period. The historical congestion degree is updated through the real - time transmitted AToT network status data to ensure that the congestion status of the shard accurately reflects the current production line load.

[0080] Based on the curvature factor and the historical congestion degree , the mobile device calculates the dynamic shard length:

[0081] Among them, is the preset basic shard length, is the maximum shard length allowed by the system, and the exponential term realizes the attenuation of the shard length with the increase of curvature and congestion degree. This attenuation mechanism is related to the path congestion degree weight Combined, it can adjust the slicing accuracy of the path according to the current production status.

[0082] The exponential term in is actually equivalent to , which reflects the adjustment effect of the path congestion degree weight on slicing division. The slicing length decreases as increases, thereby improving the path control accuracy in high congestion scenarios. According to this dynamic adjustment mechanism, the mobile device optimizes the path planning, so as to better adapt to the real-time changes of the production environment.

[0083] The slicing boundary coordinates are dynamically generated by accumulating the slicing length. Let the path starting point be , then the end coordinates of the th slice are:

[0084] where is the dynamic length of the th slice. The slicing coordinate sequence constitutes the driving path of the mobile device. This path planning is associated with the task bidding information in step S3, enabling the path planning to adapt to the priorities of each task and the device load status in real time.

[0085] After the mobile device generates the slicing sequence of the driving path, it encapsulates the slicing boundary data into an AToT message for broadcasting. This message includes the slice ID, length and the associated task ID (from the winning task in step S3 ), enabling other mobile devices to predict the slice conflict risk based on the task priority. By broadcasting these slice information through the AToT network, the mobile devices can share the path information, thereby reducing path conflicts and improving the efficiency of path planning.

[0086] This embodiment realizes the dynamic optimization of path slicing through the above technical solutions, and can adjust the slicing granularity according to the real-time road conditions. The non-linear coupling calculation of curvature and congestion degree has the effect of suppressing path conflicts in high curvature regions, and the dynamic attenuation characteristic of the slicing length can reduce the control difficulty of complex path segments.

[0087] For step S5, in the present invention, slice preemption and path replanning are the core links of closed-loop control, and their role is to coordinate the conflicts between high-priority tasks and path resources in real time, and maintain the overall efficiency of the system through dynamic replanning.

[0088] When a high-priority task triggers slice preemption, first obtain the slice time window data of the currently occupied device, and the time window is defined as:

[0089] Among them, is the current system time, is the dynamic length of the occupied shard, is the driving speed of the currently occupied device. The estimated arrival time of the high-priority task mobile device is calculated as:

[0090] In the formula, is the current position of the high-priority device, is its real-time speed, is the starting coordinate of the target shard. This calculation is associated with the shard length and path dynamic update process in step S4 to ensure real-time response of the path.

[0091] The shard preemption condition judgment needs to meet the following two points simultaneously: Time window overlap condition:

[0092] Priority condition:

[0093] Among them, the smaller the priority value, the higher the priority. This priority condition is determined based on the task bidding mechanism in step S3 to ensure that high-priority tasks can preempt path resources in a timely manner when needed.

[0094] If the preemption condition is met, the currently occupied device immediately releases the shard control right and broadcasts a shard release instruction through the AToT network. The device updates the subsequent task priorities according to the number of preemption times:

[0095] Among them, is the updated task priority of the mobile device , is the mobile device 's original task priority, is the cumulative number of preemption times (the initial value is 0, incrementing by 1 each time of preemption), is the preset compensation coefficient, used to prevent task starvation.

[0096] The path replanning process includes the following steps: 1. Future occupancy prediction: Based on the current shard occupancy (real-time statistical value) and its change rate:

[0097] Calculate the predicted value:

[0098] Among them, is the prediction time window. The calculation of the predicted occupancy rate is associated with the dynamic path sharding generation in step S4 to ensure that the path occupancy rate can be adjusted and predicted at any time.

[0099] 2. Bidding request suppression: When the predicted occupancy rate ( is a preset threshold), the edge node sends a suppression instruction to the mobile device to prohibit bidding for the target sharding area. This strategy effectively avoids excessive competition for path resources and ensures the stability of the overall system operation.

[0100] 3. Alternative path generation: The A* algorithm is combined with dynamic sharding constraints to generate alternative paths. The optimization objective is to minimize the total congestion cost:

[0101] Among them, is the dynamic length of the th shard (calculated through step S4), is the predicted occupancy rate of the corresponding shard, is the total number of shards of the alternative path. The path planning scheme generated by the A* algorithm combined with sharding constraints can dynamically respond to the arrival of high-priority tasks and optimize the congestion situation on the path.

[0102] After the path is generated, the sharding division result is broadcast and updated through the AToT network. Through this mechanism, the mobile device can obtain the latest path planning result in real time and adjust according to the current path load situation, improving the operation efficiency of the system and the response speed of task scheduling.

[0103] This embodiment realizes the dynamic scheduling of sharding resources and path optimization through the above technical solutions. The priority compensation mechanism can prevent the resource starvation problem of low-priority tasks caused by frequent preemption. The combination of the prediction model and the path optimization objective has the effect of reducing the overall congestion risk of the system.

[0104] Generally speaking, the present invention realizes intelligent logistics scheduling through a multi-level cooperation mechanism: first, it perceives the processing status and priority of each work station in real time, and dynamically generates decision weights that integrate multiple factors such as time cost and congestion degree; based on this, a distributed bidding network is constructed to enable the mobile device to autonomously calculate the optimal task matching path; during the driving process, elastic shards are dynamically divided according to the path curvature and historical congestion to achieve refined spatial resource management; finally, through the shard preemption mechanism driven by priority and the replanning strategy based on trend prediction, a closed-loop control link of perception - decision - execution - optimization is formed, ensuring the response of high-priority tasks while maintaining the overall operation efficiency of the system.

[0105] The AToT-based material process mobile control system described below can be correspondingly referred to the AToT-based material process mobile control method described above.

[0106] Please refer to the appendix Figure 3 , the present invention also provides an AToT-based material process mobile control system, including: A process status perception module 10, configured to obtain and broadcast the real-time process status of a work station; A dynamic decision-making module 20, configured to generate weight factors and execute distributed bidding; A path coordination module 30, configured to dynamically divide path shards and manage shard preemption; An exception handling module 40, configured to monitor shard conflicts and trigger path replanning.

[0107] The system of this embodiment can be used to execute the above method embodiment, and its principle and technical effect are similar, which will not be elaborated here.

[0108] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A material process movement control method based on AToT, characterized in that: The following steps are involved: Obtaining real-time process status of multiple workstations, the process status including remaining processing time, priority and fault flag; Dynamically generate a weight factor according to the process status, the weight factor is used to quantify the time cost of path planning, process waiting penalty, path congestion and conflict risk; In response to material requirements triggered by workstations, a target mobile device is allocated for the material requirements through a distributed bidding mechanism; Dynamically divide the path into slices based on path curvature and real-time congestion to generate a driving path for the target mobile device; Monitor the occupancy status of path shards and trigger shard preemption and path replanning based on priority.

2. The material process movement control method based on AToT according to claim 1 is characterized in that: The real-time process status is broadcast via the AToT network, and the priority of the workstation is dynamically adjusted according to the production plan.

3. The material process movement control method based on AToT according to claim 1 is characterized in that: The weight factor is dynamically generated through a deep reinforcement learning model, and the input of the model includes a global process state vector, which is composed of an average remaining processing time, an average priority, and a proportion of faulty workstations.

4. The material process movement control method based on AToT according to claim 1 is characterized in that: The step of allocating target mobile devices for the material demand through a distributed bidding mechanism includes: The mobile device calculates a bidding cost, wherein the bidding cost includes a conflict risk factor, and the conflict risk factor is positively correlated with the number of current bidding participating devices; If a bidding conflict for the same task among multiple mobile devices is detected, the mobile device with the lowest bidding cost and the earliest timestamp is selected.

5. The material process movement control method based on AToT according to claim 4 is characterized in that: The calculation formula for the bidding cost is: ， in, To estimate the arrival time, is the remaining processing time of the workstation, is the path congestion, To bid for the number of participating devices, , , , is the dynamic weight factor.

6. The material process movement control method based on AToT according to claim 1 is characterized in that: The step of dynamically dividing the path into segments based on the path curvature and the real-time congestion degree to generate the driving path of the target mobile device comprises: Calculating a curvature factor according to a turning radius of a current path segment of the mobile device, wherein the curvature factor is the inverse of the turning radius; Obtaining the historical congestion of the path segment, where the historical congestion is the percentage of time the segment is occupied; The dynamic shard length is calculated according to the curvature factor and the historical congestion degree, and the calculation formula is: , in, is the curvature factor, is the historical congestion level, To preset the basic fragment length, is the maximum allowed fragment length; The path segmentation boundaries are divided according to the dynamic segmentation length to generate a driving path for the target mobile device.

7. The material process movement control method based on AToT according to claim 1 is characterized in that: The step of triggering shard preemption according to priority includes: Obtain the estimated arrival time of the high priority task mobile device to the target slice and the departure time of the currently occupied device; Determine whether the preemption conditions are met , and the priority of the high-priority task is greater than the priority of the currently occupied task; among them, The estimated arrival time of the mobile device for the high priority task to reach the target shard, The departure time of the currently occupied device; If the preemption conditions are met, the currently occupied device is triggered to release the slice and perform path replanning; The priority of subsequent tasks of the currently occupied device is updated according to the number of preemption times, and the updating formula is: ， in, For mobile devices The updated task priority of For mobile devices The original task priority, is the cumulative number of preemptions. is the preset compensation coefficient.

8. The material process movement control method based on AToT according to claim 7 is characterized in that: The path replanning comprises the following steps: The future occupancy rate is predicted based on the current shard occupancy data. The calculation formula is: ， in, is the predicted future shard occupancy, is the current shard occupancy rate, is the first-order derivative of the current shard occupancy rate over time, is the prediction time window; like , suppress bidding requests to the target shard region, where is the preset maximum allowed occupancy threshold; Generate an alternative path that satisfies the curvature constraint, where the slice division of the alternative path satisfies: Take the minimum value; in, For the The dynamic length of the slices, For the The predicted occupancy of the shards, The total number of shards for the candidate path.

9. The material process movement control method based on AToT according to claim 1 is characterized in that: The mobile device is an AGV, an autonomous mobile robot or an unmanned guided vehicle.

10. A material process movement control system based on AToT, used to execute the material process movement control method based on AToT according to any one of claims 1 to 9, characterized in that: include: Process status perception module, used to obtain and broadcast the real-time process status of the workstation; Dynamic decision module for generating weight factors and executing distributed bidding; Path coordination module, used to dynamically divide path fragments and manage fragment preemption; The exception handling module is used to monitor shard conflicts and trigger path replanning.

Images