Artificial Intelligence-Based Command and Dispatch Management Methods and Systems

Abstract

This invention relates to the field of data processing system technology, and discloses a command and dispatch management method and system based on artificial intelligence. The method generates a spatiotemporal occupancy declaration containing the occupied area, time window and task status through edge nodes, and broadcasts it to neighboring nodes. The receiver performs local conflict verification according to the state mutual exclusion rule. When the static observation and dynamic intervention states are detected to coexist in overlapping spatiotemporal states, a priority-based lightweight negotiation is triggered to adjust the occupied time window. This invention realizes autonomous collaborative conflict resolution among distributed nodes, transforming the post-event conflict handling of traditional centralized scheduling into pre-event conflict avoidance, significantly improving the system response speed and operational resilience. At the same time, through physical reachability verification and channel adaptive mechanism, the collaborative reliability in complex environments is guaranteed.

Description

Technical Field

[0001] This invention relates to a command and dispatch management method based on artificial intelligence, belonging to the field of data processing system technology. Background Technology

[0002] The current model has revealed limitations in typical scenarios such as safety inspections and equipment maintenance in chemical industrial parks: First, the state changes of edge nodes require a lengthy chain of perception-upload-decision-distribution, resulting in sluggish system response under high-concurrency tasks and an inability to match dynamic on-site needs; Second, the preset rule base is difficult to cover sudden situations such as the spatiotemporal conflict between high-priority maintenance tasks and routine inspections, forcing the system to frequently rely on manual intervention, reducing decision-making efficiency; Third, the influx of massive edge data overloads central computing resources, triggering a chain reaction and impairing the overall stability of the system.

[0003] In-depth analysis reveals that the aforementioned shortcomings stem from an inherent contradiction in the existing architecture: while attempting to predict and coordinate all edge behaviors through a central node, it is constrained by information transmission delays and insufficient rule completeness. Although the industry has tried to optimize algorithm efficiency or increase redundant computing resources, it has not addressed the core contradiction of the lack of autonomous collaboration mechanisms among distributed units. This fails to eliminate the central bottleneck or achieve conflict avoidance. Therefore, how to construct an autonomous collaboration mechanism for edge nodes to achieve conflict prediction and dynamic resolution of high-concurrency tasks without central intervention becomes the technical problem this invention aims to solve. Summary of the Invention

[0004] This invention provides a command and dispatch management method based on artificial intelligence, the main purpose of which is to solve the problem of distributed edge nodes autonomously predicting and dynamically resolving spatiotemporal conflicts under complex working conditions.

[0005] To achieve the above objectives, the present invention provides an artificial intelligence-based command and dispatch management method, comprising the following steps:

[0006] Step a: The edge node generates a spatiotemporal occupancy declaration based on the task to be executed. The spatiotemporal occupancy declaration encapsulates the occupancy request for a three-dimensional geographic space region in a future time window and includes the occupancy state used to characterize the physical exclusivity requirement of the task. The occupancy state is divided into a static observation state where the environment within the occupied area is constant, and a dynamic intervention state where operations will be performed within the occupied area or environmental changes will be caused.

[0007] Step b: The edge node broadcasts a spatiotemporal occupancy declaration to other neighboring edge nodes via wireless communication;

[0008] Step c: The edge node receives and stores the spatiotemporal occupancy declarations from other nodes, and performs a local state mutual exclusion rule check on its own plan and the received declarations: if the occupancy area and occupancy time window of any two spatiotemporal occupancy declarations overlap, then a declaration in a static observation state and a declaration in a dynamic intervention state are considered to be mutually exclusive and conflicting.

[0009] In step d, if the verification in step c finds a mutual exclusion conflict, the edge node holding the lower priority task will automatically initiate negotiation with the edge node holding the higher priority task. The negotiation is to adjust the occupancy time window in its own spatiotemporal occupancy declaration to avoid the occupancy time window of the higher priority task, thereby resolving the mutual exclusion conflict before the task is physically executed.

[0010] Preferably, the method further includes: the edge node storing a historical location time-series queue of itself and neighboring nodes locally; when the communication interruption duration with a target node exceeds a communication interruption duration threshold, the edge node extrapolates its maximum physical reachability during the communication interruption period based on the historical location time-series queue of the target node using a kinematic model; the maximum physical reachability is compared with the occupied area in the spatiotemporal occupancy declaration broadcast by the target node, and if the occupied area is not covered by the maximum physical reachability, the spatiotemporal occupancy declaration of the target node is deemed invalid.

[0011] Preferably, the step of broadcasting a spacetime occupancy declaration by an edge node further includes: the sending node obtaining a packet loss rate index that quantifies the channel quality between the sending and receiving nodes; and the sending node calculating a dynamic retransmission count based on the obtained packet loss rate index and the task timeliness value corresponding to the spacetime occupancy declaration to be broadcast. The calculation follows the following relationship: ,in, The final number of retransmissions performed; This is a base number of retransmissions; This is a quantified value of the task's timeliness. The packet loss rate is an indicator. It is a positive coefficient; It is a function for rounding up; and broadcasts are performed according to the number of dynamic retransmissions.

[0012] Preferably, the method further includes: edge nodes listening to and identifying spatiotemporal occupancy declarations that are repeatedly broadcast within a set time period due to unresolved conflicts in neighboring areas, as broadcast residuals; in the local virtual geographic grid, the broadcast residuals are spatially density-accumulated according to their declared occupancy areas; if the broadcast residual density growth rate in a certain grid cell exceeds a density growth rate threshold, collaborative congestion warning information about that grid cell is generated and sent to the central management platform.

[0013] Preferably, the negotiation specifically involves: the edge node holding the lower priority task sending a delay proposal message containing its own time window adjustment plan to the edge node holding the higher priority task; and after receiving the acceptance confirmation message returned by the edge node holding the higher priority task, updating the local task execution plan according to the adjustment plan.

[0014] Preferably, the task priority is a static value configured in the edge node based on the task type or a dynamic value issued by the central scheduling system. The edge node generates a spatiotemporal occupancy declaration automatically by the operator selecting a template corresponding to the task type on the terminal device. The template has defined the mapping relationship between task type and occupancy status.

[0015] Preferred, task timeliness value It is mapped to a normalized value based on the maintenance level or the criticality level of the safety inspection.

[0016] Preferably, the density growth rate threshold is adjusted based on the historical average broadcast residual density of the grid cell and in conjunction with the current total number of concurrent global tasks.

[0017] Preferably, the occupied area is defined by a set of vertex coordinates of a polygonal geofence; the occupancy time window is defined by an absolute start time of occupancy and an occupancy duration.

[0018] An AI-based command and dispatch management system includes one or more edge nodes. Each edge node includes a processor, a memory coupled to the processor, and a wireless communication module. The memory stores computer program instructions, and the processor is configured to execute the computer program instructions to enable the edge node to perform the following operations: Based on the task to be executed, generate a spatiotemporal occupancy declaration. The spatiotemporal occupancy declaration encapsulates an occupancy request for a three-dimensional geographic space region within a future time window and includes an occupancy state characterizing the physical exclusivity requirement of the task. The occupancy state is divided into a static observation state where the environment within the occupied area needs to be constant, and a dynamic intervention state where operations will be performed within the occupied area or environmental changes will be caused. The system then communicates wirelessly... The module broadcasts spatiotemporal occupancy declarations to other nearby edge nodes. Through a wireless communication module, it receives and stores spatiotemporal occupancy declarations from other nodes in its memory. It then performs a local state mutual exclusion rule check on its own plan and the received declarations: if the occupancy areas and occupancy time windows of any two spatiotemporal occupancy declarations overlap, a declaration in a static observation state and another in a dynamic intervention state are considered mutually exclusive. If a mutual exclusion conflict is detected, and the edge node holds a lower-priority task, it automatically initiates negotiation. This negotiation involves adjusting the occupancy time window in its own spatiotemporal occupancy declaration to avoid the occupancy time window of a higher-priority task, thereby resolving the mutual exclusion conflict before the task's physical execution.

[0019] Compared with the prior art, the beneficial effects of the present invention are:

[0020] 1. By coupling the lightweight broadcast mechanism of spatiotemporal occupancy declaration with local state mutual exclusion rules, edge nodes can autonomously predict conflicts of overlapping spatiotemporal resources without central intervention. The state division of static observation and dynamic intervention captures the exclusive requirements of different tasks for the physical environment, so that conflict resolution actions are naturally triggered before the physical execution of tasks. This state-based mutual exclusion collaborative mechanism transforms the traditional post-event conflict resolution method into pre-event conflict avoidance, thereby avoiding the decision delay and rule rigidity problems inherent in centralized scheduling.

[0021] 2. Combining historical location data with kinematic extrapolation mechanisms, nodes actively verify the physical reachability of STOD declarations when communication is interrupted. When the actual movement trajectory of the target node cannot cover its declared occupied area, the system automatically unlocks the state of that area. This mechanism transforms ghost intentions in communication interruption scenarios into verifiable physical constraints, enabling distributed collaboration to no longer rely on idealized communication conditions. Instead, it ensures the system's self-consistency in complex environments through cross-verification of physical laws and digital declarations. Furthermore, based on the nonlinear coupling between channel packet loss rate and task timeliness value, the retransmission strategy of STOD is dynamically adjusted. High timeliness value tasks, such as emergency repairs, automatically obtain more transmission resources in poor channels, while routine tasks maintain basic communication overhead. This cross-layer collaboration mechanism achieves adaptive delivery of key information at low cost through the fusion calculation of underlying indicators of communication protocols and business layer logic, enabling the system to maintain reliable transmission of key intentions in dynamic environments.

[0022] 3. Spatiotemporal density analysis is performed on suspended STODs that are repeatedly broadcast, transforming local conflict information into early warning signals of regional cooperation pressure. When the accumulation rate of residual intentions within a grid cell exceeds a dynamic threshold, the system identifies resource bottleneck areas that are about to form. This mechanism constructs a mapping channel from the micro-behavior of edge nodes to the macro-state of the system through information byproducts naturally generated during the operation of the main scheme, providing an advanced decision-making basis for resource pre-deployment. STOD broadcasting provides a data foundation for conflict prediction, motion verification ensures the physical authenticity of the data, adaptive retransmission ensures the accessibility of key data, and residual analysis transforms unresolved conflicts into system optimization signals. The synergy of these four elements enables the distributed system to continuously maintain a positive cycle of conflict prediction, physical verification, intention achievement, and state optimization under uncertain conditions such as communication fluctuations and resource competition, realizing a leap in system resilience from passive fault tolerance to active adaptation. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the command, dispatch, and management method of the present invention.

[0024] Figure 2 This is a comparison chart of the average response time of the method of the present invention and traditional centralized scheduling under different numbers of concurrent tasks;

[0025] Figure 3 This is a schematic diagram of the state transition of the edge node in this invention.

[0026] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0028] The AI-based command and dispatch management method provided by this invention is applied to a system composed of multiple edge nodes, each integrating a processor, memory, and wireless communication module. It aims to establish a decentralized autonomous collaboration mechanism to address technical problems such as system response delays and insufficient operational resilience caused by the inherent lengthy information links and rigid decision-making in centralized scheduling models during high-concurrency task scenarios. The core procedure of this method involves prompting each edge node to broadcast its structured declaration encapsulating its future action intentions and autonomously perform conflict verification and negotiation locally. This transforms the traditional centralized post-event conflict handling method into a distributed pre-event conflict avoidance method, achieving dynamic, autonomous, and efficient allocation of spatiotemporal resources without continuous intervention from a central platform. In typical application scenarios such as large chemical industrial parks, multiple types of on-site operations need to be processed in parallel, such as equipment safety inspection tasks and emergency repair tasks. The former typically requires the environmental state within the work area to remain constant to ensure the accuracy of sensor data, while the latter often involves significant changes to the physical environment. Therefore, the two types of tasks require parallel processing. There is an inherent requirement for physical exclusivity. To address this challenge, this solution configures edge nodes to generate a standardized spatiotemporal occupancy declaration based on their pending tasks. The data structure of this declaration is designed to include a set of core fields: first, the occupied area, defined by a set of vertex coordinates of a polygonal geofence; second, the occupancy time window, defined by an absolute start time and a duration; and third, the occupancy status, which is strictly divided into two types: static observation and dynamic intervention. The former represents the need to maintain environmental constancy within the occupied area during task execution, while the latter represents the intention to perform physical operations or induce environmental changes within the occupied area. To ensure the standardization and efficiency of declaration generation, operators can automatically generate spatiotemporal occupancy declarations by calling templates corresponding to specific task types through terminal devices. The templates have pre-defined mapping relationships between task types and occupancy statuses. Thus, by transforming abstract operational requirements into machine-parseable spatiotemporal occupancy declarations with clear physical exclusivity representations, a unified and unambiguous data object is provided for the subsequent autonomous collaborative conflict resolution process.

[0029] In a distributed architecture without a central node, any collaborative behavior relies on mutual awareness among neighboring units. Each edge node needs to effectively inform potential interaction partners of its future action intentions. To this end, edge nodes use their wireless communication modules to broadcast their generated spatiotemporal occupancy declarations to other neighboring edge nodes within their communication coverage area. Correspondingly, each edge node also continuously receives and caches spatiotemporal occupancy declarations from other nodes in its local memory through its wireless communication module, and performs a local state mutual exclusion rule check on its own plans and the received declarations. The logic of this check is deterministically defined as follows: if the occupancy areas and occupancy time windows of any two spatiotemporal occupancy declarations overlap, and one declaration's occupancy state is static observation while the other is dynamic intervention, then these two declarations are considered to constitute a mutual exclusion conflict. Through this lightweight declaration broadcasting and localized mutual exclusion rule check, the system decentralizes the conflict detection function from the center to the network edge, enabling each node to autonomously identify potential physical execution risks based on its local situational information. When the conflict prediction mechanism identifies a potential mutual exclusion... Following a conflict, the system employs a deterministic conflict resolution procedure to prevent operational conflicts or process interruptions during task execution. To this end, this invention introduces a lightweight negotiation mechanism based on task priority. Task priorities can be configured as static values ​​in edge nodes according to task type, or dynamically issued by the central scheduling system based on the global situation. If the local state mutual exclusion rule verification results in a conflict, the edge node holding the lower-priority task automatically initiates negotiation with the edge node holding the higher-priority task. The specific operation of this negotiation is solidified into a standard interaction process: the edge node holding the lower-priority task sends a delay proposal message containing its own time window adjustment plan, aiming to ensure that the adjusted time window no longer overlaps with the time window of the higher-priority task; upon receiving an acceptance confirmation message from the other party, the node updates its local task execution plan and corresponding time-space occupancy declaration according to the adjustment plan. Through this automated negotiation and avoidance process following clearly defined priorities, most disputes over the right to use time and space resources can be resolved before the physical execution of the task.

[0030] In complex electromagnetic environments such as industrial sites, the reliability of wireless communication becomes a key constraint on the accurate transmission of intended information. To address this challenge, this invention designs a channel-adaptive dynamic retransmission mechanism. This mechanism couples the underlying channel quality with the upper-layer task value for calculation, thereby achieving intelligent scheduling of communication resources. Specifically, before broadcasting a spatial occupancy declaration, the transmitting node first obtains a quantized packet loss rate index for the channel quality between itself and the target receiving node. At the same time, it extracts a parameter from the task attributes corresponding to the time and space occupancy declaration to be broadcast, which quantifies its timeliness value. This parameter is mapped to a normalized value based on the maintenance level or criticality level of the task or safety inspection. Subsequently, the sending node calculates a dynamic retransmission count according to the following formula. , ,in, This is a base number of retransmissions used to ensure the basic reachability of routine tasks under normal channel conditions. It is a positive coefficient used to calibrate the system's sensitivity to changes in mission value and channel quality, while As a rounding function, the sending node ultimately rounds up according to the calculated number of dynamic retransmissions. By performing a broadcast operation, high-time-value mission intents can automatically obtain more retransmission resources in poor channels, thereby improving the success rate of transmitting critical intent information in uncertain network environments.

[0031] Communication interruptions are another common challenge in distributed systems, potentially leading to nodes misinterpreting the intentions of other nodes, resulting in so-called ghost intentions. This occurs when a target node is physically unreachable from its declared region, yet its outdated declaration still ineffectively locks up spatiotemporal resources. To address this issue, this invention introduces a physical reachability verification mechanism. Each edge node maintains a historical location time-series queue of itself and its neighboring nodes in its local memory. When the communication interruption duration with a target node exceeds a preset communication interruption duration threshold, the edge node triggers a verification procedure based on the target node's records prior to the communication interruption. The historical location time-series queue is used to extrapolate the theoretical maximum physical reachability during communication interruption using a kinematic model. Subsequently, the calculated maximum physical reachability is geometrically compared with the occupied area in the last successfully broadcast spatiotemporal occupancy declaration of the target node. If the occupied area is not covered by the maximum physical reachability, the system determines that the spatiotemporal occupancy declaration is invalid due to physical inaccessibility and actively removes the occupancy mark of the resource in local storage. This mechanism introduces physical laws to cross-validate the digital declaration, providing a reliable basis for the system's state consistency and resource release in communication interruption scenarios.

[0032] While the aforementioned mechanisms aim to resolve most individual conflicts, in high-density work areas, unresolved local and short-term conflicts can still accumulate into systemic congestion risks. This macro-level situation is difficult for individual nodes to perceive. To identify such risks, this invention proposes a collaborative congestion early warning method based on broadcast residuals. Edge nodes are configured to listen for and identify spatiotemporal occupancy declarations repeatedly broadcast within a set time period in neighboring areas due to unresolved conflicts, identifying these as broadcast residuals. Within a virtual geographic grid maintained locally by the node, the spatial density of the identified broadcast residuals is accumulated based on the declared occupancy area. Simultaneously... The system calculates the growth rate of broadcast residual density within a grid cell and compares it with a dynamically adjusted density growth rate threshold. This threshold is adaptively adjusted based on the historical average broadcast residual density of the grid cell during the same period and the current total number of concurrent global tasks. If the calculated density growth rate exceeds this dynamic threshold, it indicates that the collaborative pressure in the area is abnormally accumulating. The node will generate a collaborative congestion warning for the grid cell and send it to the central management platform. This mechanism transforms the byproduct of the micro-interactions between edge nodes into predictive information on the macro-level operation of the system, providing a basis for decision-making for higher-level resource pre-deployment and traffic control.

[0033] Example 1: In a large chemical industrial park, an emergency online welding repair task was initiated for a critical reactant transport pipeline. This task had a set high priority, and its work area was defined by an edge node carrying an operating terminal. Because the welding operation would generate heat and smoke disturbance to the surrounding environment, the resulting spatiotemporal occupancy declaration was marked as dynamically intervened. At the same time, an autonomous inspection drone equipped with a high-precision volatile organic compound (VOC) sensor was performing a routine safety inspection task according to a predetermined route. This task had a lower priority than the aforementioned welding task. To obtain stable air samples, its flight path would pass through the adjacent area affected by the welding operation, and the corresponding area had been broadcast. The static observation state spatiotemporal occupancy declaration of the domain and time window; in the initial stage of the task, the inspection UAV, as an edge node holding a lower priority task, identifies an intersection between its static observation path and the dynamic intervention area of ​​the welding task in the future time window when its processor performs local state mutual exclusion rule verification, constituting a mutual exclusion conflict; accordingly, the UAV automatically initiates negotiation with the welding task node and actively avoids the conflict by sending a delay proposal message containing the adjusted flight path and time window. At the stage where negotiation is completed and the tasks of all parties are about to be executed, the work area encounters localized severe convective weather, causing short-term heavy rainfall, which leads to the degradation of the wireless communication channel quality between the two edge nodes, resulting in a high packet loss rate. The value increases instantaneously.

[0034] Under this condition, the channel adaptation mechanism of the edge node holding a high-priority welding task is triggered. The processor of this node is based on a higher task timeliness value. Quantitative values ​​and the current rising packet loss rate Through relational formulas Calculate an increased dynamic retransmission count. It continuously broadcasts its spatiotemporal occupancy declaration of dynamic intervention status at this frequency, ensuring that its core operational intent remains effectively propagated even under adverse channel conditions; simultaneously, the inspection drone, due to its mission timeliness value... The quantization value is at a normal level, and its dynamic retransmission count is... The increase was relatively slow, thus ensuring the propagation of its own intentions while avoiding excessive occupation of communication channels. Furthermore, if a drone is forced to make a large detour to avoid severe convective weather, causing the communication interruption time with a fixed monitoring node in the park to exceed the preset communication interruption time threshold, the fixed monitoring node will activate physical reachability verification. By extrapolating the maximum physical reachability range of the drone during this period through a kinematic model, once it is determined that it cannot reach a pre-locked observation area within the originally declared time, the spatiotemporal resources occupied by the failed intention will be released locally in advance. Ultimately, the welding and maintenance task received a continuous and reliable broadcast due to its high-priority intentions and was completed on time without interference, avoiding interruption of the production process. The inspection drone, based on the successfully received high-priority intentions, performed effective avoidance and path adjustment, and maintained the effectiveness of its own task and the global consistency of the system state under the dual uncertainties of communication quality fluctuations and large physical position shifts. The overall operation state of the system transitioned from a critical state on the verge of physical conflict and communication failure to an orderly operation state in which each unit autonomously coordinates and operates in parallel without conflict based on priority and physical constraints.

[0035] Example 2: To objectively verify the reliability of the channel adaptive mechanism of the present invention in delivering differentiated delivery of different value task intentions under dynamically changing communication environments, a software-based simulation test platform was established. This platform aims to quantitatively evaluate the change in the success rate of key intention transmission of the system compared to the static retransmission strategy after introducing a dynamic retransmission strategy that couples channel quality with task timeliness value. The test platform consists of two simulated edge nodes, which are connected by a programmable controllable packet loss rate index. The experiment uses virtual channels for communication to simulate signal fluctuations and attenuation characteristics commonly found in industrial settings; the core of the experiment lies in observing and recording the timeliness and effectiveness of tasks under different channel conditions. In combination, the spatiotemporal occupancy declaration broadcast by the sending node is reflected in the actual reception success rate of the receiving node.

[0036] During the experimental parameter setting phase, the formula for calculating the number of dynamic retransmissions was established. positive coefficients in The value of this parameter needs to be determined by a trade-off between the system's communication guarantee strength for high-value tasks and the overall efficiency of network spectrum resource utilization. A value that is too large... This will cause the system to allocate more retransmission resources to high-value tasks under high packet loss rates, while a smaller value will result in a lower retransmission rate. This may prevent the necessary communication redundancy for critical tasks under poor channel conditions. Therefore, an offline calibration procedure is adopted to determine... The value of is: In the preset scenario with the highest packet loss rate, find the minimum value that enables the successful delivery rate of the highest value task's intent to reach the predetermined target of 99.5%. In this experiment, the base retransmission count is set. The value is 2, and this procedure is followed to select... These are non-limiting example values ​​used in subsequent experiments; the experimental process was divided into a control group and an experimental group, each running for a simulation duration of 600 seconds, during which the packet loss rate of the virtual channel was measured. The system was configured to periodically switch between a low level of 0.05 and a high level of 0.30. The control group used a static retransmission strategy with a constant retransmission count of 3; the experimental group used a channel adaptive mechanism. In both experiments, the transmitting nodes alternately broadcast two types of spatiotemporal occupancy declarations at the same frequency: one for high-time-value tasks, and the other for high-time-value tasks. One is set to version 1.0; the other is a regular time-sensitive value task, which... The threshold was set to 0.2. During the experiment, when the channel switched from a low packet loss rate to a high packet loss rate, the average number of retransmissions for high-time-value tasks in the experimental group automatically increased from 2.0 to 4.0, while the average number of retransmissions for regular time-value tasks remained at 2.0. In contrast, the number of retransmissions in the control group remained at 3.0. Table 1 shows the intent delivery success rate data recorded under different channels and strategies.

[0037] Table 1: Comparison of Success Rates of Intent Delivery under Different Strategies

[0038]

[0039] Referring to Table 1, the experimental data shows that when the channel condition deteriorates to When the value is 0.30, the success rate of intent delivery for all tasks under the static strategy drops to 91.3%, while the success rate of high-value tasks in the experimental group using the adaptive strategy remains at 99.5%. This difference is directly attributed to the higher success rate of the dynamic retransmission calculation mechanism. Values ​​and higher Value coupling automatically compensates for additional communication resources for the transmission of critical mission intents, a mechanism that does not exist in static strategies. At the same time, although the success rate of regular value tasks in the test group decreased, it did not consume additional retransmission resources, thus prioritizing the allocation of limited channel capacity to more critical intents.

[0040] Example 3: This example combines Figures 1 to 3 Description of AI-based command and dispatch management methods and systems, such as Figure 1 As shown, the process begins with task triggering. A node then generates a spatiotemporal occupancy declaration, which includes an occupancy area time window and static observation / dynamic intervention of the occupancy status. Subsequently, the node broadcasts the spatiotemporal occupancy declaration to neighboring nodes to announce its action intent. Simultaneously, a parallel channel-adaptive dynamic retransmission mechanism is activated. This mechanism dynamically calculates the number of retransmissions by acquiring the channel packet loss rate and task timeliness value to ensure reliable transmission of critical intents. During the receiving and local verification phase, the node needs to determine if there is a state mutual exclusion between itself and the received declaration. A parallel physical reachability verification mechanism operates at this stage. This mechanism verifies the validity of the declaration by monitoring communication interruptions and extrapolating the reachability range based on a kinematic model, avoiding phantom intents. After verification, the system enters a decision point to detect mutual exclusion conflicts. If there is no conflict, the task is executed in an orderly manner. If a conflict is detected, the system proceeds to a priority-based negotiation phase. This phase follows the principle that low-priority nodes adjust their time windows to avoid high-priority nodes, ultimately achieving conflict resolution. This avoids conflicts before the task's physical execution, ensuring the task can be executed in an orderly manner.

[0041] like Figure 2 As shown in the figure, the average response time performance of the method of the present invention and traditional centralized scheduling under different task loads is compared. The horizontal axis of the figure represents the number of concurrent tasks, ranging from 10 to 100, and the vertical axis represents the average response time in milliseconds. The solid curves in the figure represent the distributed autonomous collaboration method used in the present invention, while the dashed curves represent the traditional centralized scheduling method. As can be seen from the data in the figure, with the increase of the number of concurrent tasks, the average response time of traditional centralized scheduling shows a sharp nonlinear increase, while the average response time of distributed autonomous collaboration shows a gentle near-linear increase. Moreover, under all tested concurrent task numbers, its response time is significantly lower than that of traditional centralized scheduling, thus verifying the beneficial effect of the present invention in improving system response speed and coping with high-concurrency task scenarios.

[0042] like Figure 3As shown, the process begins with a node in the standby state. After receiving a task to be executed, it transitions to the declaration generation state. Once the declaration is complete, the node enters the core broadcast and listen state. In this state, if a declaration from a neighboring node is received, a local verification procedure is triggered. The verification result will lead to different branches: if there is no conflict or a conflict is found and the node has a higher priority, it will return to the broadcast and listen state; if a conflict is found and the node has a lower priority, it will enter the negotiation and avoidance state. After successful negotiation and updating the plan, the node returns to the declaration generation state to broadcast the updated intent. When the node's occupied time window is reached and there is no conflict, it transitions from the broadcast and listen state to the task execution state. After the task is completed, the node finally returns to the standby state, forming a complete closed-loop operation process.

[0043] Example 4: In the initial calibration phase of deploying the command and dispatch management system of this invention in a newly built large-scale chemical industrial park or after a major upgrade, it is necessary to configure the key parameters and algorithms used in the system to ensure operational resilience and macro-situational awareness, so as to adapt it to the topology, node movement characteristics and expected task density of the specific site. To perform this configuration, the system engineer first calibrates the communication interruption duration threshold in the physical reachability verification mechanism. The setting of this threshold aims to balance the system's locking time for the intention of disconnected nodes with the risk of erroneous unlocking. The calibration procedure is as follows: Collect the maximum operating speed of various types of edge nodes in the site under typical operating modes. And set a maximum tolerance time for resource error locking in the system based on business needs. Then the communication interruption duration threshold The value of is defined as ,in This is a safety distance margin; if the communication interruption between an edge node and the target node lasts longer than this margin... The edge node then initiates the calculation of the maximum physical reachability of the target node. This calculation procedure is defined as follows: taking the location coordinates of the target node at the time of its last broadcast spatiotemporal occupancy declaration as the center, and its maximum running speed... Duration since the start of the communication interruption The product of and is used as the radius, i.e. Construct a circular reachable region and perform a geometric intersection operation on this region with its declared but not yet started occupied regions. If the two do not intersect, the spatiotemporal occupancy declaration is considered physically unreachable and becomes invalid.

[0044] Secondly, to ensure the effective operation of the collaborative congestion early warning function, the logic for calculating the density of broadcast residuals and dynamically adjusting the early warning threshold needs to be configured. The system divides a uniform virtual geographic grid in the geographic space. When an edge node detects a spatiotemporal occupancy declaration that has been repeatedly broadcast within a set time period due to unresolved conflicts (i.e., a broadcast residual), it performs a spatial density accumulation operation: for each grid cell covered by the occupancy area declared by the broadcast residual, the residual count value of that cell is... Increasing by 1, the node's processors at fixed time intervals Calculate the residual density growth rate for each grid cell. Its calculation formula is This growth rate will be correlated with a dynamic density growth rate threshold. By comparison, the dynamic adjustment procedure for this threshold is defined as follows: ,in The baseline threshold was derived through statistical analysis of the broadcast residual density data from the same historical period of this grid cell. This represents the current total number of concurrent global tasks. This represents the average total number of concurrent tasks in the same historical period, while It is a sensitivity adjustment positive coefficient used to control the magnitude of threshold changes with global task load; by executing the above calibration and algorithm definition procedures, several key parameters and internal processing logic that originally required on-site engineers' experience to judge are transformed into reproducible configuration processes. The system's two major functions, physical reachability verification and collaborative congestion early warning, have changed from an uncertain state in the early stages of deployment to a stable operating state in which internal parameters match external environmental characteristics, core algorithm paths are clear, and functional responses are predictable.

[0045] Example 5: Before the command and dispatch management system is officially put into operation, to ensure the objectivity and consistency of the decision-making basis of its built-in priority arbitration and adaptive communication mechanisms, an offline data filling procedure is executed. In this procedure, the system administrator, in collaboration with process and safety engineers in the application field, conducts a systematic risk and value assessment of all preset task types. This assessment is based on a standardized matrix of influence factors, which include the potential impact level of task interruption on safe production, the estimated loss of economic benefits, and the risk exposure to the surrounding environment. The assessment results of each task type under each factor are quantified into scores, and then a fixed configuration table is generated through weighted summation and normalization. This table establishes the relationship between specific task types and their corresponding static task priority values ​​and task timeliness values. The deterministic mapping relationship between the coefficients is loaded into the memory of each edge node.

[0046] Meanwhile, to improve the accuracy of physical reachability verification in heterogeneous edge node environments, the system also includes a pre-calibration procedure for kinematic models during the deployment phase. For edge nodes with different motion characteristics configured within the system, a series of standardized maneuverability tests are performed in a controlled area, including maximum acceleration / deceleration and maximum turning rate tests, and the position and time series data of the entire process are recorded. Then, the collected data is fitted and analyzed with a built-in library containing kinematic models of various complexities. A kinematic model identifier that can describe the motion boundary is selected and associated for this type of node. This identifier, along with the node's category information, is stored in the system configuration so that other nodes can call the matching model for extrapolation calculation when performing physical reachability verification on it.

[0047] Example 6: In a high-density concurrent operation area, when an edge node holding a lower-priority task attempts to avoid a higher-priority task by adjusting its own time window, its generated candidate delay proposal, after local verification, is found to create a new mutual exclusion conflict with another third-party time-space occupancy declaration. The system then enters a local negotiation deadlock. Furthermore, when two conflicting time-space occupancy declarations have the same task priority value held by their publishing nodes, the priority-based adjudication mechanism cannot be directly applied. To address the negotiation deadlock caused by the first type of multi-party conflict, the system has a built-in mechanism for escalation after negotiation failure. The procedure is as follows: when an edge node fails to generate a postponement proposal that does not create any new conflicts within a preset number of attempts, the node stops performing further avoidance calculations locally. Instead, it encapsulates its own spacetime occupancy declarations, along with the spacetime occupancy declarations of all other related nodes that have direct or indirect conflicts with it, into a collaborative conflict set. The node then sends a high-level adjudication request to the central management platform. Upon receiving the request, the central management platform uses its global information and computing resources to replan or serialize all tasks within the conflict set and finally issues the conflict-free scheduling instructions to the relevant edge nodes.

[0048] For the second type of conflict scenario where task priority values ​​are the same, the system employs a deterministic equal-priority resolution procedure. Each edge node is assigned a globally unique and immutable hardware device identifier during manufacturing. When the spatiotemporal occupancy declarations of two edge nodes conflict and their task priority values ​​are the same, both nodes automatically execute this equal-priority resolution procedure. That is, the node with the smaller unique hardware device identifier in lexicographical order automatically gains priority for its task intent, while the other node accordingly fulfills its obligation to yield or negotiate. Through this final decision-making rule based on inherent device attributes, the system possesses predictable handling capabilities in all potential conflict scenarios, thereby avoiding decision-making stagnation or execution deadlock that may result from equal priorities.

[0049] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A command and dispatch management method based on artificial intelligence, characterized in that, Includes the following steps: Step a: The edge node generates a spatiotemporal occupancy declaration based on the task to be executed. The spatiotemporal occupancy declaration encapsulates the occupancy request for a three-dimensional geographic space region in a future time window and includes the occupancy status used to characterize the physical exclusivity requirement of the task. Occupancy status is divided into static observation status, where the environment within the occupied area is constant, and dynamic intervention status, where operations will be carried out within the occupied area or environmental changes will be caused. Step b: The edge node broadcasts a spatiotemporal occupancy declaration to other neighboring edge nodes via wireless communication; Step c: The edge node receives and stores the spatiotemporal occupancy declarations from other nodes, and performs a local state mutual exclusion rule check on its own plan and the received declarations: if the occupancy area and occupancy time window of any two spatiotemporal occupancy declarations overlap, then a declaration in a static observation state and a declaration in a dynamic intervention state are considered to be mutually exclusive and conflicting. In step d, if the verification in step c finds a mutual exclusion conflict, the edge node holding the lower priority task will automatically initiate negotiation with the edge node holding the higher priority task. The negotiation is to adjust the occupancy time window in its own spatiotemporal occupancy declaration to avoid the occupancy time window of the higher priority task, thereby resolving the mutual exclusion conflict before the task is physically executed.

2. The command and dispatch management method based on artificial intelligence according to claim 1, characterized in that, The method also includes: edge nodes storing their own and neighboring nodes' historical location time-series queues locally; when the communication interruption duration with a target node exceeds a communication interruption duration threshold, the edge node extrapolates its maximum physical reachability during the communication interruption based on the target node's historical location time-series queue using a kinematic model; the maximum physical reachability is compared with the occupied area in the target node's broadcast spatiotemporal occupancy declaration, and if the occupied area is not covered by the maximum physical reachability, the target node's spatiotemporal occupancy declaration is deemed invalid.

3. The command and dispatch management method based on artificial intelligence according to claim 1, characterized in that, The steps for edge nodes to broadcast a space-time occupancy declaration further include: the sending node obtaining a packet loss rate index that quantifies the channel quality between the sending and receiving nodes; and the sending node calculating a dynamic retransmission count based on the obtained packet loss rate index and the task timeliness value corresponding to the space-time occupancy declaration to be broadcast. The calculation follows the following relationship: ,in, The final number of retransmissions performed; This is a base number of retransmissions; This is a quantified value of the timeliness of the task. The packet loss rate is an indicator. It is a positive coefficient; and broadcasting is performed according to the number of dynamic retransmissions.

4. The command and dispatch management method based on artificial intelligence according to claim 1, characterized in that, The method also includes: edge nodes listening to and identifying spatiotemporal occupancy declarations that are repeatedly broadcast within a set time period due to unresolved conflicts in neighboring areas, as broadcast residuals; in the local virtual geographic grid, the broadcast residuals are spatially density-accumulated based on their declared occupancy areas; if the broadcast residual density growth rate in a certain grid cell exceeds a density growth rate threshold, collaborative congestion warning information about that grid cell is generated and sent to the central management platform.

5. The command and dispatch management method based on artificial intelligence according to claim 1, characterized in that, The negotiation process is as follows: the edge node holding the lower priority task sends a delay proposal message containing its own time window adjustment plan to the edge node holding the higher priority task; and after receiving the acceptance confirmation message from the edge node holding the higher priority task, it updates its local task execution plan according to the adjustment plan.

6. The command and dispatch management method based on artificial intelligence according to claim 1, characterized in that, Task priority is a static value configured in the edge node based on the task type or a dynamic value issued by the central scheduling system. The edge node generates a spatiotemporal occupancy declaration automatically by the operator selecting a template corresponding to the task type on the terminal device. The template has defined the mapping relationship between task type and occupancy status.

7. The command and dispatch management method based on artificial intelligence according to claim 3, characterized in that, Task timeliness value It is mapped to a normalized value based on the maintenance level or the criticality level of the safety inspection.

8. The command and dispatch management method based on artificial intelligence according to claim 1, characterized in that, The occupied area is defined by the set of vertex coordinates of a polygonal geofence; the occupancy time window is defined by an absolute start time of occupancy and an occupancy duration.

9. A command and dispatch management system based on artificial intelligence, characterized in that, The system includes one or more edge nodes. Each edge node includes a processor, a memory coupled to the processor, and a wireless communication module. The memory stores computer program instructions, and the processor is configured to execute the computer program instructions to cause the edge node to perform the following operations: Based on the task to be executed, generate a spatiotemporal occupancy declaration. The spatiotemporal occupancy declaration encapsulates an occupancy request for a three-dimensional geographic space region within a future time window and includes an occupancy state characterizing the physical exclusivity requirement of the task. The occupancy state is divided into a static observation state requiring environmental constancy within the occupied area, and a dynamic intervention state that will involve operations within the occupied area or cause environmental changes. The edge node communicates with neighboring edge nodes via the wireless communication module. Edge nodes broadcast spacetime occupancy declarations; via wireless communication modules, they receive and store spacetime occupancy declarations from other nodes in their memory, and perform a local state mutual exclusion rule check on their own plans and the received declarations: if the occupancy areas and occupancy time windows of any two spacetime occupancy declarations overlap, then a declaration in a static observation state and another declaration in a dynamic intervention state are considered mutually exclusive; if a mutual exclusion conflict is found during the check, when the edge node holds a lower priority task, it automatically initiates negotiation. This negotiation involves adjusting the occupancy time window in its own spacetime occupancy declaration to avoid the occupancy time window of a higher priority task, thereby resolving the mutual exclusion conflict before the task is physically executed.

Images