Multi-agent heterogeneous task fusion scheduling method for operating room equipment

Abstract

The application relates to a multi-agent heterogeneous task fusion scheduling method of an operating room device. The method comprises the following steps: acquiring a current operation task, current device state information of each operation device, and generating a heterogeneous task sequence and a device agent model of each operation device; based on the heterogeneous task sequence, an initial scheduling scheme of each target operation device is generated through a multi-objective optimization strategy, and a device execution scheme of each target operation device is generated through the device agent model; based on the device execution scheme of each target operation device, the target operation device is adjusted through a time-space conflict detection adjustment strategy to obtain current execution progress information of each target operation device, and based on the current execution progress information of each target operation device, the device execution scheme of each target operation device is adjusted through the device agent model. The method can improve the applicability of multi-agent heterogeneous coordination of the operation device.

Description

Technical Field

[0001] This application relates to the field of intelligent scheduling technology for medical equipment, and in particular to a multi-subject heterogeneous task fusion scheduling method for operating room equipment. Background Technology

[0002] With the development of precision surgery and intelligent operating rooms, modern operating rooms are equipped with an increasing number of intelligent medical devices, including surgical robots, anesthesia workstations, multi-parameter monitors, intraoperative navigation systems, intraoperative imaging equipment (such as intraoperative CT and intraoperative MRI), auxiliary robotic arms, and intelligent infusion pumps. These devices differ significantly in function, performance, interface protocols, and operating methods, making them typical heterogeneous systems. Therefore, improving the synergistic effect of surgical equipment is a current research focus.

[0003] Traditional surgical equipment coordination is mainly achieved through manual coordination by staff. However, this method not only isolates equipment operation, resulting in low efficiency and poor alignment, but also has a coarse scheduling granularity, causing some medical equipment to be overworked while others remain idle for extended periods. Furthermore, due to the heterogeneous nature of the equipment, coordination and adaptation among multiple devices is difficult, and manual coordination capabilities are poor, resulting in low applicability to multi-entity heterogeneous coordination of surgical equipment. Summary of the Invention

[0004] Based on this, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for multi-subject heterogeneous task fusion scheduling of operating room equipment to address the above-mentioned technical problems.

[0005] Firstly, this application provides a multi-agent heterogeneous task fusion scheduling method for operating room equipment, including: Obtain the current surgical task and the current device status information of each surgical device, and based on the current surgical task and the current device status information of each surgical device, generate a heterogeneous task sequence and a device intelligent agent model of each surgical device; Based on the heterogeneous task sequence, an initial scheduling scheme for each target surgical device is generated through a multi-objective optimization strategy. Based on the initial scheduling scheme for each target surgical device, an execution scheme for each target surgical device is generated through the device agent model. Based on the device execution scheme of each target surgical device, the execution of the current surgical task is adjusted and controlled by a spatiotemporal conflict detection and adjustment strategy to obtain the current execution progress information of each target surgical device. Based on the current execution progress information of each target surgical device, the device execution scheme of each target surgical device is adjusted by the device intelligent agent model of each target surgical device. The process of adjusting and controlling the execution of the current surgical task based on the device execution scheme of each target surgical device, and obtaining the current execution progress information of each target surgical device, is repeated until the current surgical task is completed.

[0006] Optionally, the step of generating a heterogeneous task sequence and a device agent model for each surgical device based on the current surgical task and the current device status information of each surgical device includes: The current surgical task is broken down into the task content of each task node in the task flow, and the equipment capability information required for each task content is identified. Based on the task content of each task node, the task feature information of each task node is identified, and based on the task connection method of each task node in the task flow, the initial heterogeneous task sequence of the current surgical task is constructed. The device capability information required for each task and the task feature information of each task node are filled into the initial heterogeneous task sequence to obtain the heterogeneous task sequence. For each surgical device, based on the current device status information, the device capability information, the current status information, and the workspace structure information of the surgical device are identified, and a device structure model of the surgical device is constructed based on the workspace structure information. The device capability information, current status information, and device structure model of the surgical equipment are added to the initial intelligent agent model to obtain the device intelligent agent model of the surgical equipment.

[0007] Optionally, the step of generating an initial scheduling scheme for each target surgical device based on the heterogeneous task sequence through a multi-objective optimization strategy includes: Based on the heterogeneous task sequence and the current state information of each surgical device, a surgical device population is generated, and based on each surgical device population, a multi-objective evaluation value corresponding to each surgical device population is generated through a multi-dimensional applicability evaluation strategy; the surgical device population includes surgical devices adapted to each task node; Cross-mutation is performed on each of the surgical equipment populations to obtain new surgical equipment populations. Each of the new surgical equipment populations replaces the original surgical equipment populations. The process then returns to the step of generating multi-objective evaluation values ​​for each surgical equipment population based on a multi-dimensional applicability evaluation strategy, thereby obtaining new multi-objective evaluation values ​​for each of the new surgical equipment populations. Based on the multi-objective evaluation values ​​corresponding to each surgical device population and the new multi-objective evaluation values ​​of each new surgical device population, a non-dominated sorting strategy is used to filter each initial target surgical device population, and each initial target surgical device population replaces each surgical device population. The process then returns to the step of generating multi-objective evaluation values ​​corresponding to each surgical device population based on each surgical device population through a multi-dimensional applicability evaluation strategy, until the preset iteration termination condition is met and the iteration process stops. In the last iteration, the initial target surgical equipment group with the highest multi-objective evaluation value is selected as the target surgical equipment group, and the target surgical equipment group is used as the initial scheduling scheme for each target surgical equipment.

[0008] Optionally, the step of generating a device execution plan for each of the target surgical devices based on the initial scheduling scheme of each of the target surgical devices, through the device agent model of each of the devices, includes: For each target surgical device, query the task feature information of the target task node corresponding to the target surgical device in the target surgical device population; Based on the task feature information of the target task node, an initial device execution plan corresponding to each target task node is generated through the device agent model of each target surgical device. Based on the initial device execution plan corresponding to each target task node, a task collaborative execution evaluation result is generated through the heterogeneous task sequence. Based on the evaluation results of the task collaborative execution, the execution schemes of each initial device corresponding to each target task node are iteratively adjusted to obtain the device execution scheme of the target surgical device.

[0009] Optionally, the device execution scheme based on each of the target surgical devices, through a spatiotemporal conflict detection and adjustment strategy, adjusts the control of each of the target surgical devices to execute the current surgical task, and obtains the current execution progress information of each of the target surgical devices, including: Based on the device execution scheme of each target surgical device, the current spatiotemporal conflict detection result of each target surgical device is identified through a spatiotemporal conflict detection strategy; Based on the current spatiotemporal conflict detection results of each target surgical device, the device execution scheme of each target surgical device is adjusted and processed through a three-level conflict resolution strategy to obtain the target device execution scheme of each target surgical device. Based on the target device execution plan of each target surgical device, control each target surgical device to execute the current surgical task, and collect the current sub-execution progress information of each target surgical device; The current sub-execution progress information of each target surgical device is used as the current execution progress information of each target surgical device.

[0010] Optionally, adjusting the device execution plan of each target surgical device based on the current execution progress information of each target surgical device, through the device intelligent agent model of each device, includes: Based on the current execution progress information of each target surgical device, it is determined whether a triggering event exists, and if a triggering event exists, the triggering task node corresponding to the current triggering point is identified; Based on the triggering task node, identify the scope of tasks affected by the triggering event, and based on the scope of tasks affected, identify the repair strategy corresponding to the triggering event; Based on the repair strategy corresponding to the triggering event, adjust the device execution scheme of each target surgical device.

[0011] Secondly, this application also provides a multi-subject heterogeneous task fusion scheduling device for operating room equipment, comprising: The acquisition module is used to acquire the current surgical task and the current device status information of each surgical device, and based on the current surgical task and the current device status information of each surgical device, generate a heterogeneous task sequence and a device intelligent agent model of each surgical device. The generation module is used to generate an initial scheduling scheme for each target surgical device based on the heterogeneous task sequence and through a multi-objective optimization strategy, and to generate a device execution scheme for each target surgical device based on the initial scheduling scheme for each target surgical device and through each device intelligent agent model. The iterative module is used to adjust the control of each target surgical device to execute the current surgical task based on the device execution plan of each target surgical device, through a spatiotemporal conflict detection and adjustment strategy, to obtain the current execution progress information of each target surgical device. Based on the current execution progress information of each target surgical device, the module adjusts the device execution plan of each target surgical device through the device intelligent agent model of each target surgical device, and returns to the step of adjusting the control of each target surgical device to execute the current surgical task based on the device execution plan of each target surgical device, through a spatiotemporal conflict detection and adjustment strategy, to obtain the current execution progress information of each target surgical device, until the current surgical task is completed.

[0012] Optionally, the acquisition module is specifically used for: The current surgical task is broken down into the task content of each task node in the task flow, and the equipment capability information required for each task content is identified. Based on the task content of each task node, the task feature information of each task node is identified, and based on the task connection method of each task node in the task flow, the initial heterogeneous task sequence of the current surgical task is constructed. The device capability information required for each task and the task feature information of each task node are filled into the initial heterogeneous task sequence to obtain the heterogeneous task sequence. For each surgical device, based on the current device status information, the device capability information, the current status information, and the workspace structure information of the surgical device are identified, and a device structure model of the surgical device is constructed based on the workspace structure information. The device capability information, current status information, and device structure model of the surgical equipment are added to the initial intelligent agent model to obtain the device intelligent agent model of the surgical equipment.

[0013] Optionally, the generation module is specifically used for: Based on the heterogeneous task sequence and the current state information of each surgical device, a surgical device population is generated, and based on each surgical device population, a multi-objective evaluation value corresponding to each surgical device population is generated through a multi-dimensional applicability evaluation strategy; the surgical device population includes surgical devices adapted to each task node; Cross-mutation is performed on each of the surgical equipment populations to obtain new surgical equipment populations. Each of the new surgical equipment populations replaces the original surgical equipment populations. The process then returns to the step of generating multi-objective evaluation values ​​for each surgical equipment population based on a multi-dimensional applicability evaluation strategy, thereby obtaining new multi-objective evaluation values ​​for each of the new surgical equipment populations. Based on the multi-objective evaluation values ​​corresponding to each surgical device population and the new multi-objective evaluation values ​​of each new surgical device population, a non-dominated sorting strategy is used to filter each initial target surgical device population, and each initial target surgical device population replaces each surgical device population. The process then returns to the step of generating multi-objective evaluation values ​​corresponding to each surgical device population based on each surgical device population through a multi-dimensional applicability evaluation strategy, until the preset iteration termination condition is met and the iteration process stops. In the last iteration, the initial target surgical equipment group with the highest multi-objective evaluation value is selected as the target surgical equipment group, and the target surgical equipment group is used as the initial scheduling scheme for each target surgical equipment.

[0014] Optionally, the generation module is specifically used for: For each target surgical device, query the task feature information of the target task node corresponding to the target surgical device in the target surgical device population; Based on the task feature information of the target task node, an initial device execution plan corresponding to each target task node is generated through the device agent model of each target surgical device. Based on the initial device execution plan corresponding to each target task node, a task collaborative execution evaluation result is generated through the heterogeneous task sequence. Based on the evaluation results of the task collaborative execution, the execution schemes of each initial device corresponding to each target task node are iteratively adjusted to obtain the device execution scheme of the target surgical device.

[0015] Optionally, the iterative module is specifically used for: Based on the device execution scheme of each target surgical device, the current spatiotemporal conflict detection result of each target surgical device is identified through a spatiotemporal conflict detection strategy; Based on the current spatiotemporal conflict detection results of each target surgical device, the device execution scheme of each target surgical device is adjusted and processed through a three-level conflict resolution strategy to obtain the target device execution scheme of each target surgical device. Based on the target device execution plan of each target surgical device, control each target surgical device to execute the current surgical task, and collect the current sub-execution progress information of each target surgical device; The current sub-execution progress information of each target surgical device is used as the current execution progress information of each target surgical device.

[0016] Optionally, the iterative module is specifically used for: Based on the current execution progress information of each target surgical device, it is determined whether a triggering event exists, and if a triggering event exists, the triggering task node corresponding to the current triggering point is identified; Based on the triggering task node, identify the scope of tasks affected by the triggering event, and based on the scope of tasks affected, identify the repair strategy corresponding to the triggering event; Based on the repair strategy corresponding to the triggering event, adjust the device execution scheme of each target surgical device.

[0017] Thirdly, this application provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method described in any one of the first aspects.

[0018] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described in any one of the first aspects.

[0019] Fifthly, this application provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the method described in any one of the first aspects.

[0020] The aforementioned multi-agent heterogeneous task fusion scheduling method for operating room equipment involves: acquiring the current surgical task and the current device status information of each surgical device; generating a heterogeneous task sequence and a device agent model for each surgical device based on the current surgical task and the current device status information of each surgical device; generating an initial scheduling scheme for each target surgical device based on the heterogeneous task sequence using a multi-objective optimization strategy; generating an execution scheme for each target surgical device based on the initial scheduling scheme of each target surgical device using the device agent model of each target surgical device; adjusting the control of each target surgical device to execute the current surgical task based on the device execution scheme of each target surgical device using a spatiotemporal conflict detection and adjustment strategy to obtain the current execution progress information of each target surgical device; adjusting the execution scheme of each target surgical device based on the current execution progress information of each target surgical device using the device agent model of each target surgical device; returning to the execution of the current surgical task based on the device execution scheme of each target surgical device using a spatiotemporal conflict detection and adjustment strategy; and obtaining the current execution progress information of each target surgical device, until the current surgical task is completed. This solution transforms the current surgical task into a heterogeneous task sequence and, combined with the device intelligence model built for each surgical device in this solution, generates an initial scheduling plan for each target surgical device. This avoids the problem of unified scheduling planning for heterogeneous devices and ensures that each target surgical device can be scheduled accurately, uniformly, and collaboratively. This approach avoids the limitations and inefficiencies of manual scheduling, improving the accuracy and efficiency of scheduling various surgical devices. Secondly, after generating the scheduling plan, this solution uses spatiotemporal conflict detection to address coordination anomalies, temporal and spatial anomalies in the scheduling of various surgical devices. This avoids the poor coordination capabilities of manual scheduling while comprehensively considering the rationality, feasibility, and adaptability of scheduling for surgical devices. Finally, during the execution of the surgical task, this solution also collects the current execution progress information of each target surgical device in real time. Based on the task execution status, it adjusts the execution plan of each target surgical device in real time, preventing unforeseen circumstances from affecting the execution process and interfering with the normal operation of the surgery. This enhances the intelligence of the surgical device system control and the technical effect of dynamic and timely adjustment, thereby comprehensively improving the applicability of multi-entity heterogeneous coordination of surgical devices. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating a multi-agent heterogeneous task fusion scheduling method for operating room equipment in one embodiment; Figure 2 This is a flowchart illustrating an example of multi-agent heterogeneous task fusion scheduling for operating room equipment in one embodiment; Figure 3 This is a structural block diagram of a multi-subject heterogeneous task fusion scheduling device for operating room equipment in one embodiment. Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0024] The multi-agent heterogeneous task fusion scheduling method for operating room equipment provided in this application embodiment can be applied to a multi-agent heterogeneous task fusion scheduling system for operating room equipment. This system can be applied to a terminal, which can be, but is not limited to, various personal computers, laptops, mid-range computers, etc. Specifically, the terminal transforms the current surgical task into a heterogeneous task sequence and, combined with the device intelligent agent model constructed for each surgical device according to this solution, generates an initial scheduling scheme for each target surgical device. This avoids the problem of unified scheduling planning for heterogeneous devices and ensures that all target surgical devices can be accurately, uniformly, and collaboratively scheduled. This approach avoids the limitations and inefficiencies of manual scheduling, improving the accuracy and efficiency of scheduling various surgical devices. Secondly, after generating the scheduling plan, this solution uses spatiotemporal conflict detection to address coordination anomalies, temporal and spatial anomalies in the scheduling of various surgical devices. This avoids the poor coordination capabilities of manual scheduling while comprehensively considering the rationality, feasibility, and adaptability of scheduling for surgical devices. Finally, during the execution of the surgical task, this solution also collects the current execution progress information of each target surgical device in real time. Based on the task execution status, it adjusts the execution plan of each target surgical device in real time, preventing unforeseen circumstances from affecting the execution process and interfering with the normal operation of the surgery. This enhances the intelligence of the surgical device system control and the technical effect of dynamic and timely adjustment, thereby comprehensively improving the applicability of multi-entity heterogeneous coordination of surgical devices.

[0025] In one exemplary embodiment, such as Figure 1 As shown, a multi-entity heterogeneous task fusion scheduling method for operating room equipment is provided. Taking the application of this method to a terminal as an example, the method includes the following steps S101 to S103. Wherein: Step S101: Obtain the current surgical task and the current device status information of each surgical device, and generate a heterogeneous task sequence and a device intelligent agent model of each surgical device based on the current surgical task and the current device status information of each surgical device.

[0026] In this embodiment, the terminal responds to the information upload operation of the staff and obtains the surgical task that requires the collaboration of surgical equipment. When obtaining the surgical task, the terminal simultaneously obtains the surgical type (e.g., laparoscopic cholecystectomy, radical prostatectomy, etc.) and extracts the standard surgical step sequence from the surgical procedure knowledge base. The surgical procedure knowledge base stores the standard operating procedures for various surgeries, with each step corresponding to an initial task node. The types of task nodes include, but are not limited to: preoperative preparation, patient positioning, anesthesia induction, endotracheal intubation, surgical area disinfection and draping, incision establishment, pneumoperitoneum establishment, exploration and separation, lesion resection, hemostasis and suturing, specimen removal, incision closure, anesthesia recovery, and postoperative transport. Then, the terminal obtains the current equipment status information of different surgical equipment, including the current spatial location, current operating status, equipment capabilities, and heterogeneous interface information of each surgical equipment. Next, the terminal constructs a sequence of tasks according to task nodes for the current surgical task, thereby obtaining a heterogeneous task sequence, and constructs a device intelligent agent model for each surgical equipment based on the current equipment status information of each surgical equipment. Each device agent model includes the capability vector, state vector, and workspace model of each surgical device. The specific construction process will be explained in detail later.

[0027] Step S102: Based on the heterogeneous task sequence, an initial scheduling scheme for each target surgical device is generated through a multi-objective optimization strategy. Based on the initial scheduling scheme for each target surgical device, an execution scheme for each target surgical device is generated through the intelligent agent model of each device.

[0028] In this embodiment, the terminal generates initial scheduling schemes for each target surgical device based on heterogeneous task sequences using a multi-objective optimization strategy. Based on these initial scheduling schemes, and through the intelligent agent models of each device, it generates device execution schemes for each target surgical device. This multi-objective optimization strategy employs an improved non-dominated sorting genetic algorithm (NSGA-II) with an elitist strategy for global task-device matching planning, thereby selecting each target surgical device and generating initial scheduling schemes for each. The improved NSGA-II uses two-dimensional chromosome encoding: the first dimension encodes the task execution sequence, and the second dimension encodes the device allocation scheme. It aims to minimize the maximum completion time, maximize the balance of device utilization, and maximize the total safety margin as three optimization objectives, solving for the Pareto optimal solution set while satisfying task dependency constraints, time window constraints, device capability constraints, and spatial conflict constraints. The method by which each device agent model generates device execution plans is as follows: After the terminal sends the device execution plans for each target surgical device to each device agent, each device agent uses a distributed negotiation mechanism based on Multi-Agent Deep Deterministic Policy Gradient (MADDPG) to perform local dynamic adjustments, thereby generating the device execution plan for the surgical device corresponding to each device agent. During generation, each device agent sets the functions of the surgical device based on the current device state and the task nodes adapted to the surgical device in the heterogeneous task sequence. These settings include task acceptance / rejection, fine-tuning of execution order, and adjustment of resource allocation ratios. The specific process of generating device execution plans will be explained in detail later.

[0029] Step S103: Based on the device execution plan of each target surgical device, the spatiotemporal conflict detection and adjustment strategy is used to adjust the control of each target surgical device to execute the current surgical task, obtain the current execution progress information of each target surgical device, and based on the current execution progress information of each target surgical device, the device execution plan of each target surgical device is adjusted through the intelligent agent model of each device, and the execution of the device execution plan based on each target surgical device, the spatiotemporal conflict detection and adjustment strategy is used to adjust the control of each target surgical device to execute the current surgical task, obtain the current execution progress information of each target surgical device, until the current surgical task is completed.

[0030] In this embodiment, the terminal, based on the device execution plan of each target surgical device, adjusts the control of each target surgical device to execute the current surgical task through a spatiotemporal conflict detection and adjustment strategy, obtains the current execution progress information of each target surgical device, and, based on the current execution progress information of each target surgical device, adjusts the device execution plan of each target surgical device through the intelligent agent model of each device, returns to execute the steps based on the device execution plan of each target surgical device, adjusts the control of each target surgical device to execute the current surgical task through the spatiotemporal conflict detection and adjustment strategy, and obtains the current execution progress information of each target surgical device, until the current surgical task is completed. The spatiotemporal conflict detection strategy is used to detect time conflicts (not idle, running, under maintenance) and spatial conflicts (not in the same space) encountered by each target surgical device during actual operation, and uses a three-level resolution strategy to regulate and resolve the above conflicts, thereby ensuring that there are no time or spatial conflicts when the surgical devices collaboratively execute the target task. The three-level resolution strategy consists of: Level 1, preemptive resolution based on task priority; Level 2, time-off resolution based on buffer time; and Level 3, resolution based on alternative equipment replacement. The current execution progress information is collected in real-time by the terminal during the operation of the surgical equipment. When a triggering event is detected in the execution progress, the execution plan for each surgical equipment is adjusted, and step S102 is re-executed to avoid abnormal equipment coordination or increased surgical risks. These triggering events include task timeout, equipment failure, emergency task insertion, or abnormal patient vital signs. The specific adjustment process will be explained in detail later.

[0031] Based on the above scheme, by transforming the current surgical task into a heterogeneous task sequence and combining it with the device intelligence model built for each surgical device in this scheme, an initial scheduling scheme for each target surgical device is generated, thereby avoiding the problem of unified scheduling planning for heterogeneous devices and ensuring that each target surgical device can be scheduled accurately, uniformly, and collaboratively. This approach avoids the limitations and inefficiencies of manual scheduling, improving the accuracy and efficiency of scheduling various surgical devices. Secondly, after generating the scheduling plan, this solution uses spatiotemporal conflict detection to address coordination anomalies, temporal and spatial anomalies in the scheduling of various surgical devices. This avoids the poor coordination capabilities of manual scheduling while comprehensively considering the rationality, feasibility, and adaptability of scheduling for surgical devices. Finally, during the execution of the surgical task, this solution also collects the current execution progress information of each target surgical device in real time. Based on the task execution status, it adjusts the execution plan of each target surgical device in real time, preventing unforeseen circumstances from affecting the execution process and interfering with the normal operation of the surgery. This enhances the intelligence of the surgical device system control and the technical effect of dynamic and timely adjustment, thereby comprehensively improving the applicability of multi-entity heterogeneous coordination of surgical devices.

[0032] Optionally, based on the current surgical task and the current device status information of each surgical device, a heterogeneous task sequence and a device agent model for each surgical device are generated. This includes: breaking down the current surgical task into the task content of each task node in the task flow, and identifying the device capability information required for each task content; identifying the task feature information of each task node based on its task content, and constructing an initial heterogeneous task sequence for the current surgical task based on the task connection method of each task node in the task flow; filling the initial heterogeneous task sequence with the device capability information required for each task content and the task feature information of each task node to obtain the heterogeneous task sequence; for each surgical device, identifying the device capability information, current status information, and workspace structure information of the surgical device based on its current device status information, and constructing a device structure model of the surgical device based on its workspace structure information; and adding the device capability information, current status information, and device structure model of the surgical device to the initial agent model to obtain the device agent model of the surgical device.

[0033] In this embodiment, the terminal breaks down the current surgical task into the task content of each task node in the task flow and identifies the equipment capability information required for each task content. This equipment capability information is represented as a five-dimensional vector of the surgical equipment: C_req=[c_accuracy,c_force,c_image,c_space,c_bandwidth], where: c_accuracy: the level of operational accuracy, ranging from 1 to 5. Level 1 is low accuracy (e.g., coarse dissection), level 2 is medium accuracy (e.g., routine suturing), level 3 is high accuracy (e.g., fine dissection), level 4 is high accuracy (e.g., vascular anastomosis), and level 5 is extremely high accuracy (e.g., neurosurgical microsurgery). A larger value indicates a higher accuracy requirement. c_force: the level of force feedback requirement, ranging from 0 to 3. 0 indicates no force feedback requirement (e.g., purely visual tasks), 1 indicates basic force feedback requirement (e.g., tissue traction perception), 2 indicates medium force feedback requirement (e.g., suture force control), and 3 indicates high-precision force feedback requirement (e.g., microsurgical force perception). c_image: Image guidance requirement level, an integer ranging from 0 to 4. 0 indicates no image guidance required, 1 indicates 2D image guidance (e.g., ultrasound guidance), 2 indicates 3D image guidance (e.g., CT / MRI navigation), 3 indicates real-time 3D image fusion guidance, and 4 indicates multimodal real-time image fusion guidance. c_space: Space occupancy level, an integer ranging from 1 to 4. Characterizes the minimum workspace radius required for the device to perform the task. Level 1 is for small devices (radius ≤ 200mm), Level 2 is for medium-sized devices (radius ≤ 400mm), Level 3 is for large devices (radius ≤ 600mm), and Level 4 is for extra-large devices (radius > 600mm). c_bandwidth: Communication bandwidth requirement, in Mbps. Characterizes the minimum bandwidth required for data interaction between devices during task execution. For example, simple status monitoring tasks have low bandwidth requirements (approximately 1Mbps), while real-time video transmission and image fusion tasks have higher bandwidth requirements (up to 100Mbps or more). The capability vector C_dev=[c_a,c_f,c_i,c_s,c_b] of the device agent has the same dimension as the task requirement vector. The necessary and sufficient condition for a task to be assigned to a device is that each dimension of the device capability vector is greater than or equal to the corresponding dimension of the task requirement vector.

[0034] The terminal identifies the task characteristic information of each task node based on its task content, and constructs an initial heterogeneous task sequence for the current surgical task based on the task connection method of each task node in the task flow. The task characteristic information includes the task type identifier, time window constraint, estimated execution time, task priority, and safety level attributes of the task node. Specifically, the terminal calculates the task priority P based on the surgical urgency, patient vital sign risk level, and clinical importance of the task content: P = w1·P_urgency + w2·P_risk + w3·P_importance, where w1, w2, and w3 are weighting coefficients satisfying w1 + w2 + w3 = 1, P_urgency is the urgency score, P_risk is the risk level score, and P_importance is the clinical importance score, all ranging from 0 to 1. Secondly, the terminal calculates the Earliest Start Time (EST) and Latest Finish Time (LFT) for each task node. EST is calculated via forward propagation: EST(j) = max{EST(i) + D(i)}, where i is all predecessor tasks of j, and D(i) is the estimated execution time of task i. LFT is calculated via backward propagation: LFT(i) = min{LFT(j) - D(j)}, where j is all successor tasks of i. Specifically, each task node calculates its Earliest Start Time (EST) and Latest Finish Time (LFT).

[0035] EST is calculated via forward propagation: starting from the source node (the task without predecessors), the EST of each task is calculated sequentially along the directed edges. For task j, its EST is the maximum value of the completion times of all predecessor tasks, i.e.: EST(j) = max{EST(i) + D(i) | i ∈ pred(j)} Where pred(j) is the set of all predecessor tasks of task j, and D(i) is the estimated execution time of task i. The EST of the source node is 0.

[0036] LFT is calculated via backpropagation: starting from the sink node (tasks without successors), the LFT of each task is calculated sequentially in the reverse direction of the directed edges. For task i, its LFT is the minimum of the latest start times of all its successor tasks, i.e.: LFT(i) = min{LFT(j) - D(j) | j ∈ succ(i)} Where succ(i) is the set of all successor tasks of task i. The LFT of the sink node is set to the upper limit of the total operation time.

[0037] The terminal constructs an initial heterogeneous task sequence for the current surgical task based on the task connection methods of each task node in the task flow. This task connection method represents the flow logic between each task node. The terminal constructs directed edges based on this flow logic, for example, establishing directed edges based on logical dependencies between steps. If the completion of task A is a necessary condition for the start of task B, then a directed edge e(A,B) is established from A to B. Dependencies include: sequential dependency (the next task can only begin after the previous task is completed), parallel dependency (multiple tasks can be executed simultaneously), and synchronous dependency (the next task can only begin after multiple tasks are completed). The initial heterogeneous task sequence for the current surgical task is obtained, and then the terminal adds the task feature information of each task node to this initial heterogeneous task sequence to obtain the heterogeneous task sequence.

[0038] For each surgical device, the terminal identifies its capabilities, current status, and workspace structure based on its current status information. It then constructs a device structure model based on this workspace structure information. This model is a three-dimensional representation of the device's spatial distribution within the hospital, using the hospital's location as nodes. The device capabilities are also represented as the aforementioned five-dimensional vector. The current status information includes the device's current operating state (idle, running, maintenance, etc.) and its data interface information (i.e., the data format and standards of the received and output data).

[0039] Then, the terminal adds the device capability information, current status information, and device structure model of the surgical equipment to the initial intelligent agent model to obtain the device intelligent agent model of the surgical equipment.

[0040] Based on the above scheme, intelligent agent modeling is performed for each surgical device to ensure the accuracy of individual analysis of each surgical device. Furthermore, the surgical process is broken down to construct heterogeneous sequences, which improves the comprehensiveness of the flow of each task node in the operation and the feature analysis of the task nodes.

[0041] Optionally, based on the heterogeneous task sequence, an initial scheduling scheme for each target surgical device is generated through a multi-objective optimization strategy. This includes: generating a population of surgical devices based on the heterogeneous task sequence and the current state information of each surgical device; and generating a multi-objective evaluation value for each surgical device population based on a multi-dimensional suitability evaluation strategy. The surgical device population includes the surgical device adapted to each task node. Cross-mutation processing is performed on each surgical device population to obtain new surgical device populations. These new surgical device populations replace the existing surgical device populations. The process then returns to the step of generating the multi-objective evaluation value for each surgical device population based on the multi-dimensional suitability evaluation strategy, thus obtaining the new surgical devices. The new multi-objective evaluation value of the population; based on the multi-objective evaluation value corresponding to each surgical device population and the new multi-objective evaluation value of each new surgical device population, the initial target surgical device population is selected through a non-dominated sorting strategy, and each initial target surgical device population replaces the original surgical device population. The process returns to the step of generating the multi-objective evaluation value corresponding to each surgical device population through a multi-dimensional applicability evaluation strategy, until the preset iteration termination condition is met and the iteration process stops; among the initial target surgical device populations obtained in the last iteration, the initial target surgical device population with the highest multi-objective evaluation value is selected as the target surgical device population, and the target surgical device population is used as the initial scheduling scheme for each target surgical device.

[0042] In this embodiment, the terminal generates a population of surgical devices based on the heterogeneous task sequence and the current status information of each surgical device. Based on this population, a multi-objective evaluation value is generated for each surgical device population using a multi-dimensional applicability evaluation strategy. This surgical device population includes the surgical devices adapted to each task node. Specifically, the terminal generates an initial population of size N. In this embodiment, the population size N is set to 100. Each individual is a two-dimensional chromosome, representing a complete scheduling scheme: First dimension (task sequence dimension): a permutation of length M, where M is the total number of tasks. This represents the execution order of tasks and must satisfy topological sorting constraints (i.e., all predecessor tasks are placed before successor tasks). Second dimension (device allocation dimension): a sequence of integers of length M, where each element represents the device number assigned to the corresponding task. This must satisfy device capability matching constraints.

[0043] During initialization, a heuristic strategy is used to generate the initial population to improve the algorithm's convergence speed: Task sequence dimension: Tasks are sorted in descending order of priority while ensuring topological constraints. Specifically, a set of tasks with an in-degree of 0 is maintained. Each time, the highest-priority task is selected from this set and added to the sequence, then the in-degree of its successor tasks is updated. This process is repeated until all tasks are added to the sequence. Equipment allocation dimension: A strategy prioritizing the highest capability matching degree is adopted. For each task, the equipment with the highest overall matching degree is selected from those meeting the capability matching requirements for allocation. The overall matching degree considers factors such as capacity redundancy, equipment load, and health status.

[0044] The multidimensional applicability assessment strategy evaluates each surgical equipment group based on three objectives: minimizing maximum completion time, maximizing equipment utilization balance, and maximizing the total safety margin. The specific assessment method is as follows: Objective 1: Minimize the maximum completion time (makespan): f1 = makespan = max{C_k | k = 1,2,...,K} Where C_k is the completion time of all tasks on the k-th device, and K is the total number of devices. Makespan reflects the total time of the entire operation and is a core indicator for measuring scheduling efficiency.

[0045] Objective 2: Maximize the balance of equipment utilization. f2 = 1 - σ(U) / μ(U) Where U = [U_1, U_2, ..., U_K] is the utilization vector of each device, and U_k = total working time of device k / total scheduling time. σ(U) is the standard deviation of utilization, and μ(U) is the mean of utilization. A larger f2 indicates a more balanced device load, avoiding situations where some devices are overloaded while others are idle.

[0046] Objective 3: Maximize the total safety margin. f3 = Σ(S_i / D_i), i = 1,2,...,M Where S_i is the safety margin time for task i, which is the difference between the latest completion time (LFT) and the actual completion time of task i, S_i = LFT(i) - C(i). D_i is the estimated execution time of task i. The larger f3 is, the more sufficient the safety margin of the scheduling scheme is, and the stronger its ability to cope with unexpected situations.

[0047] Of the three objectives, f1 is a cost-related objective (the smaller the better), while f2 and f3 are benefit-related objectives (the larger the better). Because there are conflicts among the three objectives (for example, pursuing the shortest makespan may lead to uneven equipment load), a multi-objective optimization method is needed to solve for the Pareto optimal solution set.

[0048] The key and difficult point is that the terminal performs cross-mutation on each surgical device population to obtain new surgical device populations. These new populations then replace the existing surgical device populations. The process returns to the previous step, generating multi-objective evaluation values ​​for each surgical device population using a multi-dimensional applicability assessment strategy. This yields new multi-objective evaluation values ​​for each new surgical device population. Then, based on the multi-objective evaluation values ​​of each existing surgical device population and the new multi-objective evaluation values ​​of each new surgical device population, the terminal uses a non-dominated sorting strategy to select initial target surgical device populations. These initial target populations then replace the existing surgical device populations. The process returns to the previous step, generating multi-objective evaluation values ​​for each surgical device population using a multi-dimensional applicability assessment strategy. This process continues until a preset iteration termination condition is met. Specifically, the non-dominated sorting method involves the terminal performing a non-dominated sorting of the surgical device populations based on the aforementioned multi-objective evaluation values, dividing individuals within the population into different Pareto levels (Front 1, Front 2, ...). A lower Pareto rank indicates a better individual. Then, the terminal calculates the crowding distance for individuals within the same Pareto rank. The crowding distance measures the density around an individual and is used to maintain population diversity. A higher crowding distance indicates a sparser population, which is more conducive to maintaining diversity. Next, the terminal uses a binary tournament selection. Each time, two individuals are randomly selected, and their Pareto ranks and crowding distances are compared. The individual with the lower Pareto rank wins; if the ranks are the same, the individual with the higher crowding distance wins. The parent population is selected through this selection operation. The crossover and mutation operations described above involve the terminal performing crossover and mutation operations on the parent population to generate the offspring population. For the task sequence dimension, a topology-based uniform crossover is used: several positions are randomly selected, and the tasks of the two parent populations at these positions are swapped. Then, topology repair ensures that the generated offspring satisfy dependency constraints. Mutation uses exchange mutation: two task positions are randomly selected and swapped, followed by topology repair. For the device allocation dimension, a uniform crossover is used: several positions are randomly selected, and the device numbers of the two parent populations at these positions are swapped. Mutation employs random mutation: a task is randomly selected, and its device ID is replaced with another device ID that meets the capability matching requirements. The crossover probability p_c is set to 0.9, and the mutation probability p_m is set to 0.1. Finally, the terminal merges the parent and offspring populations (size 2N), performs non-dominated sorting, and selects individuals from low to high Pareto rank until N individuals are selected as the new generation population. Within the same rank, selection is based on crowding from highest to lowest. The elite retention strategy ensures that superior individuals are not lost.The preset iteration termination condition is that if the maximum number of iterations G_max (set to 200 generations in this embodiment) is reached, or if the hypervolume index of the Pareto front does not improve significantly (the improvement is less than 1%) for consecutive G_stable generations (set to 30 generations in this embodiment), then the iteration is terminated and the current Pareto optimal solution set is output.

[0049] In the last iteration, the initial target surgical equipment population with the highest multi-objective evaluation value is selected as the target surgical equipment population, and the target surgical equipment population is used as the initial scheduling scheme for each target surgical equipment.

[0050] Based on the above scheme, the multi-objective optimization, cross-iteration, and elite retention strategies constructed in this scheme ensure that the generated target surgical equipment population is the surgical equipment population that best uses the current target task.

[0051] Optionally, based on the initial scheduling scheme of each target surgical device, an execution scheme for each target surgical device is generated through the intelligent agent model of each device. This includes: for each target surgical device, querying the task feature information of the target task node corresponding to the target surgical device in the target surgical device population; based on the task feature information of the target task node, generating each initial device execution scheme corresponding to the target task node through the intelligent agent model of each target surgical device; and based on each initial device execution scheme corresponding to each target task node, generating a task collaborative execution evaluation result through a heterogeneous task sequence; and based on the task collaborative execution evaluation result, iteratively adjusting each initial device execution scheme corresponding to each target task node to obtain the execution scheme of the target surgical device.

[0052] In this embodiment, for each target surgical device, the terminal queries the task feature information of the target task node corresponding to the target surgical device in the target surgical device population. Then, based on the task feature information of the target task node, the terminal generates initial device execution schemes corresponding to each target task node through the device agent model of each target surgical device. Specifically, when generating the initial device execution schemes corresponding to each target task node, the terminal uses the TOPSIS (Technique for Order Preference by Similarity to an IdealSolution) method based on entropy weight to select the optimal compromise scheme as the initial scheduling scheme from the Pareto optimal solution set. This selection method is as follows: the terminal assumes that there are a total of n permutations and combinations in which each target surgical device in the target surgical device population can complete the task of its corresponding target task node. Then, the target surgical device population corresponds to n scheduling schemes, and each scheduling scheme has m attribute values ​​(m=3). Then, the terminal constructs an n×m decision matrix X = [x_ij]. Finally, the terminal performs vector standardization on the decision matrix to eliminate the influence of dimensions. r_ij = x_ij / √(Σ(x_ij)²), i = 1,2,...,n Next, the terminal uses information entropy to calculate the objective weights of each target attribute. The entropy value of the j-th attribute is: e_j = -k · Σ(p_ij · ln p_ij) Where p_ij = r_ij / Σr_ij, k = 1 / ln n. The weight of the j-th attribute is: w_j = (1 - e_j) / Σ(1 - e_j) The basic idea of ​​the entropy weight method is that the greater the difference in attribute values, the more information is provided, and the greater the weight.

[0053] Next, the terminal multiplies the normalized matrix by the weight vector to obtain the weighted normalized matrix: v_ij = w_j · r_ij S295. Determine the positive and negative ideal solutions: = [max(v_ij) | j is a benefit-type attribute, min(v_ij | j is a cost-type attribute]; = [min(v_ij) | j is a benefit-type attribute, max(v_ij) | j is a cost-type attribute].

[0054] Next, the terminal calculates the distance from each solution to the ideal solution. Distance to the negative ideal solution : = √(Σ(v_ij - )²) = √(Σ(v_ij - )²) Then, the terminal calculates the relative proximity of each solution as follows: C_i = / ( + ) Finally, the terminal selects the scheme with the largest C_i as the optimal compromise, and uses the scheduling scheme of each target surgical device in this optimal compromise as the initial device execution scheme for each target surgical device.

[0055] Next, for each target surgical device, the terminal, based on the task feature information of the target task node corresponding to the target surgical device and the five-dimensional vector corresponding to the device capability information of the target surgical device, uses a three-dimensional judgment strategy to determine the task collaborative execution evaluation result of the initial device execution plan for that surgical device. This task collaborative execution result includes the task acceptance probability, execution time offset, and resource allocation ratio. The specific strategy execution method is as follows: 1. Task acceptance probability p_accept: Value range [0,1]. Represents the degree of willingness of the device agent to accept the currently assigned task. p_accept > 0.5 indicates acceptance, otherwise rejection.

[0056] 2. Execution time offset Δt: Value range [-Δt_max, Δt_max], unit is seconds. It represents the amount by which the device agent wants to advance or delay the task execution time. Δt_max is set to 20% of the estimated task duration.

[0057] 3. Resource allocation ratio r_alloc: Value range [0.5, 1.0]. Represents the proportion of resources allocated to the current task by the device agent. For devices that support multi-task parallelism, some resources can be allocated to the current task, and the remaining resources can be used for other tasks. For single-task devices, r_alloc is always 1.0.

[0058] Based on the evaluation results of the collaborative execution of this task and the preset reward function, the initial device execution schemes corresponding to each target task node are adjusted to make the device execution schemes more suitable for the target task node, thereby obtaining the device execution scheme for the target surgical device. The preset reward function is a reward function with four objectives: task completion, efficiency, safety, and collaboration. The immediate reward for the i-th device agent model is: r_i = w1·r_task + w2·r_efficiency + w3·r_safety + w4·r_cooperation The meanings and calculation methods of each component are as follows: 1. r_task: Task completion reward. Successfully completing the task earns +10 points, while failing (e.g., due to severe timeout or device malfunction causing task interruption) earns -20 points. This component encourages agents to actively complete tasks.

[0059] 2. r_efficiency: Efficiency bonus. The calculation formula is: r_efficiency = -|Δt| / Δt_max This component encourages agents to reduce time skew and execute according to the scheduling scheme as much as possible, thereby improving overall execution efficiency.

[0060] 3.r_safety: Safety penalty. A penalty is applied when the task completion time exceeds the latest completion time (LFT). r_safety = -(Completion Time - LFT) / D_i × 50 Where D_i is the estimated execution time of task i. The longer the timeout, the greater the penalty. This component ensures that the task is completed within the safe time window.

[0061] 4. r_cooperation: Cooperation reward. A +5 cooperation reward is given when an agent successfully collaborates with neighboring devices to complete a complex task (such as a surgical robot cooperating with a navigation system to complete a navigation-guided operation). This component encourages cooperation between devices.

[0062] The weighting coefficients w1, w2, w3, and w4 are set to 0.4, 0.2, 0.3, and 0.1, respectively. The weighting reflects the relative importance of different objectives: task completion is the primary objective, safety is a crucial constraint, and efficiency and collaboration are the directions for optimization.

[0063] When adjusting the above technical solution, this solution utilizes a Centralized Training with Decentralized Execution (CTDE) framework. This framework works as follows: during the training phase, a centralized critic network is trained using global information (observations and actions of all agents) to evaluate the value of joint actions; during the execution phase, each agent makes decisions based solely on its local observations through an independent policy network, without the need for communication or global information. The architecture of this distributed execution framework includes a policy network structure and a critic network structure, specifically: Policy Network (Actor Network) Structure: Each agent has an independent Actor network μ_i(o_i | θ_μi) used to output actions based on the observed state. The network structure is a three-layer fully connected network: Input Layer: Dimension equal to the observed state dimension dim(o_i). Hidden Layer 1: Dimension 256, ReLU activation function. Hidden Layer 2: Dimension 256, ReLU activation function. Output layer: Dimension equals the action dimension dim(a_i), activation function Tanh (maps the output to [-1,1], then scales it to the actual action range). All agents have the same Actor network structure, but their parameters are independent.

[0064] The Critic Network structure: A globally shared Critic Network Q(o_1, o_2, ..., o_N, a_1, a_2, ..., a_N | θ_Q) is used to evaluate the value of the joint actions of all agents. The network structure is a three-layer fully connected network: Input layer: dimension Σdim(o_i) + Σdim(a_i), representing the concatenation of observations and actions from all agents. Hidden layer 1: dimension 512, activation function ReLU. Hidden layer 2: dimension 512, activation function ReLU. Output layer: dimension 1, outputting the Q-value estimate of the joint action.

[0065] Based on the above approach, a compromise solution is developed that is applicable to all target surgical devices and can also be applied to multi-target execution of heterogeneous task sequences, effectively improving the device execution performance of each target surgical device.

[0066] Optionally, based on the device execution scheme of each target surgical device, a spatiotemporal conflict detection and adjustment strategy is used to adjust and control the execution of the current surgical task of each target surgical device, thereby obtaining the current execution progress information of each target surgical device. This includes: based on the device execution scheme of each target surgical device, identifying the current spatiotemporal conflict detection result of each target surgical device through a spatiotemporal conflict detection strategy; based on the current spatiotemporal conflict detection result of each target surgical device, adjusting and processing the device execution scheme of each target surgical device through a three-level conflict resolution strategy to obtain the target device execution scheme of each target surgical device; based on the target device execution scheme of each target surgical device, controlling each target surgical device to execute the current surgical task, and collecting the current sub-execution progress information of each target surgical device; and using the current sub-execution progress information of each target surgical device as the current execution progress information of each target surgical device.

[0067] In this embodiment, the terminal, based on the device execution scheme of each target surgical device, identifies the current spatiotemporal conflict detection results of each target surgical device through a spatiotemporal conflict detection strategy. Specifically, the spatiotemporal conflict detection method is as follows: 1. Time Conflict Detection: A time conflict refers to a situation where the same target surgical device is occupied by multiple tasks simultaneously. For each target surgical device, maintain a list of its task execution time intervals T_list = [(t_s1, t_e1), (t_s2, t_e2), ...], where t_sk and t_ek are the start and end times of the k-th task, respectively.

[0068] When a new task is assigned, let the time interval of the new task be [t_s_new, t_e_new]. Check if this interval overlaps with any existing interval in the list. The necessary and sufficient condition for overlap is: t_s_new < t_e_k and t_e_new > t_s_k If there exists a task k that satisfies this condition, it is determined to be a time conflict.

[0069] 2. Spatial Conflict Detection: Spatial conflict refers to the overlapping of the workspaces of multiple target surgical devices operating simultaneously, which may lead to physical interference. Spatial conflict detection is based on the bounding box model of the workspace of the target surgical device.

[0070] The workspace of each surgical device is represented by an axis-aligned bounding box (AABB), defined by the minimum corner point P_min = (x_min, y_min, z_min) and the maximum corner point P_max = (x_max, y_max, z_max). For devices with degrees of freedom of motion (such as robotic arms), the AABB is the outer envelope of the entire reachable space of the end effector; for stationary devices, the AABB is the space occupied by the target surgical device body.

[0071] The necessary and sufficient condition for two target surgical devices i and j to have a spatial conflict is that their AABB projections on all three coordinate axes overlap. X-axis: x_min_i < x_max_j and x_max_i > x_min_j Y axis: y_min_i < y_max_j and y_max_i > y_min_j Z axis: z_min_i < z_max_j and z_max_i > z_min_j When all three conditions are met, it is determined to be a spatial conflict.

[0072] Spatial conflict detection needs to consider the time factor; that is, a true spatiotemporal conflict only occurs when two target surgical devices are both operational within the same time period and their workspaces overlap. Therefore, spatial conflict detection must be combined with time scheduling information.

[0073] Then, based on the current spatiotemporal conflict detection results of each target surgical device, the terminal adjusts the device execution plan of each target surgical device through a three-level conflict resolution strategy to obtain the target device execution plan for each target surgical device. The execution method of this three-level conflict resolution strategy is as follows: First-priority preemptive resolution: When a conflict is detected, the priorities P of the two conflicting tasks are first compared. The task with the higher priority continues to execute, while the task with the lower priority is preempted and enters a waiting queue, awaiting rescheduling after the preempted task is completed.

[0074] If two tasks have the same priority, their safety levels are compared. Tasks with higher safety levels are performed first. Safety levels are determined based on the clinical risk of the task. For example, tasks involving the maintenance of vital signs (such as anesthesia maintenance and cardiopulmonary resuscitation) have the highest safety level; tasks involving critical procedural steps (such as lesion resection and vascular anastomosis) have the next highest safety level; and auxiliary tasks (such as specimen processing and instrument counting) have the lowest safety level.

[0075] Preemptive resolution is the most direct and fastest resolution method, but it may cause delays for low-priority tasks. Therefore, it should only be used when high-priority tasks truly need to be executed immediately.

[0076] Second-level time offset resolution: If preemption is not feasible (e.g., both tasks are high-priority and cannot be easily preempted), then an attempt is made to resolve the conflict through time offset. The basic idea of ​​time offset is to postpone (or advance) the execution time of one task so that it does not overlap with the execution time of the other task.

[0077] The principle for selecting which task to time offset is: choose the task with the larger buffer time. The task's buffer time B is: B = LFT - EST - D Where EST is the earliest start time, LFT is the latest finish time, and D is the estimated execution time. Buffer time B represents the time margin within which the task can be flexibly adjusted during the time window.

[0078] Calculation of time offset: Let the conflict overlap time be Δ_overlap, then the offset Δ_shift = Δ_overlap + Δ_safe, where Δ_safe is the safety interval time (set to 5 seconds in this embodiment) to ensure that there is enough transition time between the two tasks.

[0079] After the offset, it is necessary to re-verify whether the task meets the time window constraint (i.e., the completion time after the offset does not exceed the LFT). If it does not meet the constraint, the time offset is not feasible, and a third-level resolution strategy needs to be tried.

[0080] Third-level equipment replacement and resolution: If the time offset still cannot resolve the conflict (e.g., the buffer time for both tasks is insufficient), then try to assign one of the conflicting tasks to the alternative device, that is, resolve the conflict by replacing the device.

[0081] Selection criteria for alternative equipment: From among currently available equipment that meets the capability matching requirements (i.e., all dimensions of the capability vector are not lower than the task requirements), the equipment with the highest overall matching degree is selected. The formula for calculating the overall matching degree M_match is: M_match = α·C_match + β·S_proximity + γ·H_status in: C_match: Capability matching degree. Calculates the cosine similarity between the device capability vector and the task requirement vector, C_match = (C_dev · C_req) / (|C_dev| |C_req|). The higher the matching degree, the more suitable the device is for performing the task.

[0082] S_proximity: Spatial proximity. The closer the device is to the surgical area, the higher the proximity. The formula for calculating spatial proximity is S_proximity = 1 / (1 + d / d_ref), where d is the distance from the device to the center of the surgical area, and d_ref is the reference distance (set to 1000mm in this embodiment).

[0083] H_status: Device health status score. The better the health status, the higher the score.

[0084] The weighting coefficients α, β, and γ are set to 0.5, 0.3, and 0.2, respectively, reflecting that ability matching is the primary factor, while spatial location and health status are secondary considerations.

[0085] If a suitable alternative device is found, the conflicting task is transferred to the alternative device for execution; if no suitable alternative device is found, the conflict cannot be resolved and a dynamic rescheduling mechanism needs to be triggered (step S5).

[0086] Based on the target device execution plan for each target surgical device, the terminal controls each target surgical device to execute the current surgical task and collects the current sub-execution progress information of each target surgical device. Finally, the terminal uses the current sub-execution progress information of each target surgical device as the current execution progress information of each target surgical device.

[0087] Based on the above scheme, by detecting time and space conflicts and resolving them through a three-level resolution strategy, various non-cooperative conflict problems existing when multiple target surgical devices perform tasks are effectively eliminated, and the collaborative efficiency of multiple heterogeneous devices is effectively improved.

[0088] Optionally, based on the current execution progress information of each target surgical device, the device execution plan of each target surgical device is adjusted through the intelligent agent model of each device, including: determining whether there is a triggering event based on the current execution progress information of each target surgical device, and identifying the triggering task node corresponding to the current triggering point when a triggering event exists; identifying the scope of the affected task corresponding to the triggering event based on the triggering task node, and identifying the repair strategy corresponding to the triggering event based on the scope of the affected task; and adjusting the device execution plan of each target surgical device based on the repair strategy corresponding to the triggering event.

[0089] In this embodiment, the terminal determines whether a triggering event exists based on the current execution progress information of each target surgical device, and identifies the triggering task node corresponding to the current triggering point when a triggering event exists. The triggering methods for different triggering events are as follows: The terminal monitors four types of triggering events in real time, and the occurrence of any one of these events may trigger dynamic rescheduling: 1. Task Timeout Event: A task is considered to have timed out when its actual execution time exceeds 120% of the estimated time. Task timeouts may cause delays in subsequent tasks, requiring adjustments to the scheduling scheme. The timeout threshold of 120% is set to account for normal fluctuations in surgical time and to avoid frequent rescheduling.

[0090] 2. Device Failure Event: When the device health status score s_health falls below the threshold H_threshold (default 0.3), it is considered a device failure. Device failure means that the device cannot continue to perform tasks and its assigned tasks need to be transferred to other devices.

[0091] 3. Emergency Task Insertion Event: When an emergency surgical task needs to be inserted into the current schedule (such as emergency trauma surgery, emergency cesarean section, etc.), the emergency task needs to be included in the schedule and re-planned. Emergency tasks usually have the highest priority.

[0092] 4. Abnormal Patient Status Events: When a patient's vital signs (heart rate, blood pressure, blood oxygen saturation, body temperature, etc.) deviate from the normal range and exceed a set threshold, the patient's status is considered abnormal. Abnormal patient status may require suspending the current operation, increasing monitoring efforts, or taking emergency measures, thus affecting the dispatch plan.

[0093] The terminal identifies the scope of tasks affected by a triggered event based on the triggering task node, and then identifies the corresponding repair strategy based on the scope of affected tasks. The identification of the scope of affected tasks is performed using a forward propagation method: starting from the task node corresponding to the trigger point, all subsequent tasks are traversed along the directed edges of the Directed Acyclic Graph (DAG), and these tasks are marked as affected tasks. The size of the affected task set A_affected determines the scope of rescheduling. Specifically, for equipment failure events, affected tasks are all incomplete tasks on that device and their successors; for task timeout events, affected tasks are the timed-out task itself and all its successors; for emergency task insertion events, affected tasks are tasks with lower priority than emergency tasks that compete for resources with emergency tasks; and for abnormal patient status events, affected tasks are the currently executing tasks and their successors.

[0094] Finally, based on the repair strategy corresponding to the triggered event, the terminal adjusts the device execution plan for each target surgical device. This repair strategy includes local repair strategies and global readjustment strategies. The execution method of each strategy is as follows: If the number of affected tasks does not exceed 30% of the total number of tasks (i.e., |A_affected| / M ≤ 0.3), a local repair strategy will be adopted.

[0095] The approach to partial repair is to keep the scheduling scheme for unaffected tasks unchanged and only re-optimize the scheduling of the affected task set. This maximizes the stability of the scheduling scheme, reduces disturbances caused by adjustments, lowers computational overhead, and improves response speed.

[0096] The execution process of partial repair: Taking the set of affected tasks as input, the distributed negotiation scheduling in step S3 is re-executed. Each device agent, based on the current system state (completed tasks, tasks in progress, and current device state), reallocates and reorders the affected tasks through the MADDPG negotiation mechanism.

[0097] The time complexity of local repair is O(M_affected × K), where M_affected is the number of affected tasks and K is the number of devices. Since M_affected is usually much smaller than M, local repair is computationally very fast, and can be completed in seconds.

[0098] If the number of affected tasks exceeds 30% of the total number of tasks (i.e., |A_affected| / M > 0.3), or if there are still irresolvable conflicts after partial repair, a global rescheduling will be triggered.

[0099] Global rescheduling uses the current system state as the initial condition and re-executes the global optimization (improved NSGA-II algorithm) in step S2. The current system state includes: completed tasks, currently executing tasks and their remaining time, the current state of each device, and newly inserted urgent tasks, etc.

[0100] The number of global rescheduling iterations, G_max_global, can be dynamically adjusted based on the urgency level. (1) General case: Use the default number of iterations (200 generations) to ensure scheduling quality; (2) Emergency situations (such as abnormal patient vital signs or equipment failure leading to surgical interruption): reduce the number of iterations (e.g., 50 generations) to achieve a faster response speed at the cost of sacrificing some optimization quality; (3) In extremely urgent situations (such as cardiac arrest or massive bleeding): use heuristics to generate scheduling schemes directly without iterative optimization to ensure millisecond-level response.

[0101] The time complexity of global rescheduling is O(G_max × N × M × K), where G_max is the number of iterations, N is the population size, M is the number of tasks, and K is the number of devices. In typical scenarios, global rescheduling can be completed within tens of seconds.

[0102] Based on the above scheme, by monitoring the triggered events in real time, the execution plans of each target surgical device can be effectively adjusted in a timely manner. This not only ensures the accuracy of the collaborative work of each surgical device and improves the efficiency of completing the target task, but also effectively enhances the timely control of abnormalities of each surgical device, greatly reducing the risk of further impact on the target task.

[0103] This application also provides an example of multi-agent heterogeneous task fusion scheduling for operating room equipment, such as... Figure 2 As shown, the specific processing procedure includes the following steps: Step S201: Obtain the current surgical task and the current device status information of each surgical device.

[0104] Step S202: The current surgical task is broken down into the task content of each task node in the task flow, and the equipment capability information required for each task content is identified.

[0105] Step S203: Based on the task content of each task node, identify the task feature information of each task node, and construct the initial heterogeneous task sequence of the current surgical task based on the task connection method of each task node in the task flow.

[0106] Step S204: Fill the initial heterogeneous task sequence with the equipment capability information required for each task content and the task feature information of each task node, respectively, to obtain the heterogeneous task sequence.

[0107] Step S205: For each surgical device, based on the current device status information, identify the device capability information, current status information, and workspace structure information of the surgical device, and construct a device structure model of the surgical device based on the workspace structure information.

[0108] Step S206: Add the device capability information, current status information, and device structure model of the surgical equipment to the initial intelligent agent model to obtain the device intelligent agent model of the surgical equipment.

[0109] Step S207: Based on the heterogeneous task sequence and the current status information of each surgical device, generate a population of surgical devices, and based on each population of surgical devices, generate a multi-objective evaluation value corresponding to each population of surgical devices through a multi-dimensional applicability evaluation strategy.

[0110] Step S208: Perform crossover and mutation processing on each surgical device population to obtain each new surgical device population, and replace each surgical device population with each new surgical device population. Return to execute the step of generating multi-objective evaluation values ​​for each surgical device population based on each surgical device population and through a multi-dimensional applicability assessment strategy, to obtain new multi-objective evaluation values ​​for each new surgical device population.

[0111] Step S209: Based on the multi-objective evaluation values ​​corresponding to each surgical device group and the new multi-objective evaluation values ​​of each new surgical device group, each initial target surgical device group is screened through a non-dominated sorting strategy, and each initial target surgical device group replaces each surgical device group. The process returns to the step of generating multi-objective evaluation values ​​corresponding to each surgical device group through a multi-dimensional applicability evaluation strategy, until the preset iteration termination condition is met and the iteration process stops.

[0112] Step S210: Among the initial target surgical equipment groups obtained in the last iteration, select the initial target surgical equipment group with the highest multi-objective evaluation value as the target surgical equipment group, and use the target surgical equipment group as the initial scheduling scheme for each target surgical equipment.

[0113] Step S211: For each target surgical device, query the task feature information of the target task node corresponding to the target surgical device in the target surgical device population.

[0114] Step S212: Based on the task feature information of the target task node, generate the initial device execution plan corresponding to each target task node through the device intelligent agent model of each target surgical device, and generate the task collaborative execution evaluation result based on the initial device execution plan corresponding to each target task node through the heterogeneous task sequence.

[0115] Step S213: Based on the task collaborative execution evaluation results, iteratively adjust the execution schemes of each initial device corresponding to each target task node to obtain the device execution scheme of the target surgical device.

[0116] Step S214: Based on the device execution scheme of each target surgical device, the current spatiotemporal conflict detection result of each target surgical device is identified through the spatiotemporal conflict detection strategy.

[0117] Step S215: Based on the current spatiotemporal conflict detection results of each target surgical device, the device execution scheme of each target surgical device is adjusted and processed through a three-level conflict resolution strategy to obtain the target device execution scheme of each target surgical device.

[0118] Step S216: Based on the target device execution plan of each target surgical device, control each target surgical device to execute the current surgical task, and collect the current sub-execution progress information of each target surgical device.

[0119] Step S217: Use the current sub-execution progress information of each target surgical device as the current execution progress information of each target surgical device.

[0120] Step S218: Based on the current execution progress information of each target surgical device, determine whether there is a triggering event, and if there is a triggering event, identify the triggering task node corresponding to the current triggering point.

[0121] Step S219: Based on the triggering task node, identify the scope of the affected tasks corresponding to the triggering event, and based on the scope of the affected tasks, identify the repair strategy corresponding to the triggering event.

[0122] Step S220: Based on the repair strategy corresponding to the triggering event, adjust the device execution plan of each target surgical device.

[0123] Step S221: Return to step S216 until the current surgical task is completed.

[0124] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0125] Based on the same inventive concept, this application also provides a multi-subject heterogeneous task fusion scheduling device for operating room equipment to implement the multi-subject heterogeneous task fusion scheduling method for operating room equipment described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the multi-subject heterogeneous task fusion scheduling device for operating room equipment provided below can be found in the limitations of the multi-subject heterogeneous task fusion scheduling method for operating room equipment described above, and will not be repeated here.

[0126] In one exemplary embodiment, such as Figure 3 As shown, a multi-agent heterogeneous task fusion scheduling device for operating room equipment is provided, comprising: an acquisition module 310, a generation module 320, and an iteration module 330, wherein: The acquisition module 310 is used to acquire the current surgical task and the current device status information of each surgical device, and generate a heterogeneous task sequence and a device intelligent agent model of each surgical device based on the current surgical task and the current device status information of each surgical device. The generation module 320 is used to generate an initial scheduling scheme for each target surgical device based on the heterogeneous task sequence and through a multi-objective optimization strategy, and to generate a device execution scheme for each target surgical device based on the initial scheduling scheme for each target surgical device and through each device intelligent agent model. The iteration module 330 is used to adjust the control of each target surgical device to execute the current surgical task based on the device execution plan of each target surgical device, through a spatiotemporal conflict detection and adjustment strategy, to obtain the current execution progress information of each target surgical device, and based on the current execution progress information of each target surgical device, adjust the device execution plan of each target surgical device through the device intelligent agent model of each target surgical device, and return to the step of adjusting the control of each target surgical device to execute the current surgical task based on the device execution plan of each target surgical device, through a spatiotemporal conflict detection and adjustment strategy, to obtain the current execution progress information of each target surgical device, until the current surgical task is completed.

[0127] Optionally, the acquisition module 310 is specifically used for: The current surgical task is broken down into the task content of each task node in the task flow, and the equipment capability information required for each task content is identified. Based on the task content of each task node, the task feature information of each task node is identified, and based on the task connection method of each task node in the task flow, the initial heterogeneous task sequence of the current surgical task is constructed. The device capability information required for each task and the task feature information of each task node are filled into the initial heterogeneous task sequence to obtain the heterogeneous task sequence. For each surgical device, based on the current device status information, the device capability information, the current status information, and the workspace structure information of the surgical device are identified, and a device structure model of the surgical device is constructed based on the workspace structure information. The device capability information, current status information, and device structure model of the surgical equipment are added to the initial intelligent agent model to obtain the device intelligent agent model of the surgical equipment.

[0128] Optionally, the generation module 320 is specifically used for: Based on the heterogeneous task sequence and the current state information of each surgical device, a surgical device population is generated, and based on each surgical device population, a multi-objective evaluation value corresponding to each surgical device population is generated through a multi-dimensional applicability evaluation strategy; the surgical device population includes surgical devices adapted to each task node; Cross-mutation is performed on each of the surgical equipment populations to obtain new surgical equipment populations. Each of the new surgical equipment populations replaces the original surgical equipment populations. The process then returns to the step of generating multi-objective evaluation values ​​for each surgical equipment population based on a multi-dimensional applicability evaluation strategy, thereby obtaining new multi-objective evaluation values ​​for each of the new surgical equipment populations. Based on the multi-objective evaluation values ​​corresponding to each surgical device population and the new multi-objective evaluation values ​​of each new surgical device population, a non-dominated sorting strategy is used to filter each initial target surgical device population, and each initial target surgical device population replaces each surgical device population. The process then returns to the step of generating multi-objective evaluation values ​​corresponding to each surgical device population based on each surgical device population through a multi-dimensional applicability evaluation strategy, until the preset iteration termination condition is met and the iteration process stops. In the last iteration, the initial target surgical equipment group with the highest multi-objective evaluation value is selected as the target surgical equipment group, and the target surgical equipment group is used as the initial scheduling scheme for each target surgical equipment.

[0129] Optionally, the generation module 320 is specifically used for: For each target surgical device, query the task feature information of the target task node corresponding to the target surgical device in the target surgical device population; Based on the task feature information of the target task node, an initial device execution plan corresponding to each target task node is generated through the device agent model of each target surgical device. Based on the initial device execution plan corresponding to each target task node, a task collaborative execution evaluation result is generated through the heterogeneous task sequence. Based on the evaluation results of the task collaborative execution, the execution schemes of each initial device corresponding to each target task node are iteratively adjusted to obtain the device execution scheme of the target surgical device.

[0130] Optionally, the iteration module 330 is specifically used for: Based on the device execution scheme of each target surgical device, the current spatiotemporal conflict detection result of each target surgical device is identified through a spatiotemporal conflict detection strategy; Based on the current spatiotemporal conflict detection results of each target surgical device, the device execution scheme of each target surgical device is adjusted and processed through a three-level conflict resolution strategy to obtain the target device execution scheme of each target surgical device. Based on the target device execution plan of each target surgical device, control each target surgical device to execute the current surgical task, and collect the current sub-execution progress information of each target surgical device; The current sub-execution progress information of each target surgical device is used as the current execution progress information of each target surgical device.

[0131] Optionally, the iteration module 330 is specifically used for: Based on the current execution progress information of each target surgical device, it is determined whether a triggering event exists, and if a triggering event exists, the triggering task node corresponding to the current triggering point is identified; Based on the triggering task node, identify the scope of tasks affected by the triggering event, and based on the scope of tasks affected, identify the repair strategy corresponding to the triggering event; Based on the repair strategy corresponding to the triggering event, adjust the device execution scheme of each target surgical device.

[0132] The modules in the multi-entity heterogeneous task fusion scheduling device of the aforementioned operating room equipment can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0133] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a multi-agent heterogeneous task fusion scheduling method for operating room equipment. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0134] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0135] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of a multi-agent heterogeneous task fusion scheduling method for operating room equipment.

[0136] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program being executed by a processor to implement the steps of a multi-agent heterogeneous task fusion scheduling method for operating room equipment.

[0137] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of a multi-agent heterogeneous task fusion scheduling method for operating room equipment.

[0138] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0139] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0140] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0141] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A multi-agent heterogeneous task fusion scheduling method for operating room equipment, characterized in that, The method includes: Obtain the current surgical task and the current device status information of each surgical device, and based on the current surgical task and the current device status information of each surgical device, generate a heterogeneous task sequence and a device intelligent agent model of each surgical device; Based on the heterogeneous task sequence, an initial scheduling scheme for each target surgical device is generated through a multi-objective optimization strategy. Based on the initial scheduling scheme for each target surgical device, an execution scheme for each target surgical device is generated through the device agent model. Based on the device execution scheme of each target surgical device, the execution of the current surgical task is adjusted and controlled by a spatiotemporal conflict detection and adjustment strategy to obtain the current execution progress information of each target surgical device. Based on the current execution progress information of each target surgical device, the device execution scheme of each target surgical device is adjusted by the device intelligent agent model of each target surgical device. The process of adjusting and controlling the execution of the current surgical task based on the device execution scheme of each target surgical device, and obtaining the current execution progress information of each target surgical device, is repeated until the current surgical task is completed.

2. The method according to claim 1, characterized in that, The process of generating a heterogeneous task sequence and a device agent model for each surgical device based on the current surgical task and the current device status information of each surgical device includes: The current surgical task is broken down into the task content of each task node in the task flow, and the equipment capability information required for each task content is identified. Based on the task content of each task node, the task feature information of each task node is identified, and based on the task connection method of each task node in the task flow, the initial heterogeneous task sequence of the current surgical task is constructed. The device capability information required for each task and the task feature information of each task node are filled into the initial heterogeneous task sequence to obtain the heterogeneous task sequence. For each surgical device, based on the current device status information, the device capability information, the current status information, and the workspace structure information of the surgical device are identified, and a device structure model of the surgical device is constructed based on the workspace structure information. The device capability information, current status information, and device structure model of the surgical equipment are added to the initial intelligent agent model to obtain the device intelligent agent model of the surgical equipment.

3. The method according to claim 2, characterized in that, The process of generating an initial scheduling scheme for each target surgical device based on the heterogeneous task sequence through a multi-objective optimization strategy includes: Based on the heterogeneous task sequence and the current state information of each surgical device, a surgical device population is generated, and based on each surgical device population, a multi-objective evaluation value corresponding to each surgical device population is generated through a multi-dimensional applicability evaluation strategy; the surgical device population includes surgical devices adapted to each task node; Cross-mutation is performed on each of the surgical equipment populations to obtain new surgical equipment populations. Each of the new surgical equipment populations replaces the original surgical equipment populations. The process then returns to the step of generating multi-objective evaluation values ​​for each surgical equipment population based on a multi-dimensional applicability evaluation strategy, thereby obtaining new multi-objective evaluation values ​​for each of the new surgical equipment populations. Based on the multi-objective evaluation values ​​corresponding to each surgical device population and the new multi-objective evaluation values ​​of each new surgical device population, a non-dominated sorting strategy is used to filter each initial target surgical device population, and each initial target surgical device population replaces each surgical device population. The process then returns to the step of generating multi-objective evaluation values ​​corresponding to each surgical device population based on each surgical device population through a multi-dimensional applicability evaluation strategy, until the preset iteration termination condition is met and the iteration process stops. In the last iteration, the initial target surgical equipment group with the highest multi-objective evaluation value is selected as the target surgical equipment group, and the target surgical equipment group is used as the initial scheduling scheme for each target surgical equipment.

4. The method according to claim 3, characterized in that, The initial scheduling scheme based on each of the target surgical devices, through the intelligent agent model of each device, generates a device execution scheme for each of the target surgical devices, including: For each target surgical device, query the task feature information of the target task node corresponding to the target surgical device in the target surgical device population; Based on the task feature information of the target task node, an initial device execution plan corresponding to each target task node is generated through the device agent model of each target surgical device. Based on the initial device execution plan corresponding to each target task node, a task collaborative execution evaluation result is generated through the heterogeneous task sequence. Based on the evaluation results of the task collaborative execution, the execution schemes of each initial device corresponding to each target task node are iteratively adjusted to obtain the device execution scheme of the target surgical device.

5. The method according to claim 4, characterized in that, The device execution scheme based on each of the target surgical devices, through a spatiotemporal conflict detection and adjustment strategy, adjusts the control of each of the target surgical devices to execute the current surgical task, and obtains the current execution progress information of each of the target surgical devices, including: Based on the device execution scheme of each target surgical device, the current spatiotemporal conflict detection result of each target surgical device is identified through a spatiotemporal conflict detection strategy; Based on the current spatiotemporal conflict detection results of each target surgical device, the device execution scheme of each target surgical device is adjusted and processed through a three-level conflict resolution strategy to obtain the target device execution scheme of each target surgical device. Based on the target device execution plan of each target surgical device, control each target surgical device to execute the current surgical task, and collect the current sub-execution progress information of each target surgical device; The current sub-execution progress information of each target surgical device is used as the current execution progress information of each target surgical device.

6. The method according to claim 5, characterized in that, The step of adjusting the execution plan of each target surgical device based on the current execution progress information of each target surgical device, through the intelligent agent model of each device, includes: Based on the current execution progress information of each target surgical device, it is determined whether a triggering event exists, and if a triggering event exists, the triggering task node corresponding to the current triggering point is identified; Based on the triggering task node, identify the scope of tasks affected by the triggering event, and based on the scope of tasks affected, identify the repair strategy corresponding to the triggering event; Based on the repair strategy corresponding to the triggering event, adjust the device execution scheme of each target surgical device.

7. A multi-entity heterogeneous task fusion scheduling device for operating room equipment, characterized in that, The device includes: The acquisition module is used to acquire the current surgical task and the current device status information of each surgical device, and based on the current surgical task and the current device status information of each surgical device, generate a heterogeneous task sequence and a device intelligent agent model of each surgical device. The generation module is used to generate an initial scheduling scheme for each target surgical device based on the heterogeneous task sequence and through a multi-objective optimization strategy, and to generate a device execution scheme for each target surgical device based on the initial scheduling scheme for each target surgical device and through each device intelligent agent model. The iterative module is used to adjust the control of each target surgical device to execute the current surgical task based on the device execution plan of each target surgical device, through a spatiotemporal conflict detection and adjustment strategy, to obtain the current execution progress information of each target surgical device. Based on the current execution progress information of each target surgical device, the module adjusts the device execution plan of each target surgical device through the device intelligent agent model of each target surgical device, and returns to the step of adjusting the control of each target surgical device to execute the current surgical task based on the device execution plan of each target surgical device, through a spatiotemporal conflict detection and adjustment strategy, to obtain the current execution progress information of each target surgical device, until the current surgical task is completed.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images