Double-track automatic tool changing control method and system for full-face tunnel boring machine (TBM)

Abstract

The present application belongs to the field of tool changing robot control, and particularly relates to a double-track automatic tool changing control method and system for a full-face tunnel boring machine (TBM), comprising: in response to a tool changing demand, obtaining a target tool identifier and a tool changing urgency level; collecting real-time state information including tool changing robot state, track occupancy state, and coordinates of a tool to be changed; based on the tool changing urgency level and the real-time state information, generating an operation sequence and a motion trajectory through a motion planning algorithm; controlling the tool changing robot to move to the position of the tool to be changed according to the motion trajectory to perform grabbing and dismounting; after dismounting is completed, controlling the tool changing robot to move the old tool to a tool handover area, and synchronously controlling a tool lifting platform and an automatic transport transfer vehicle to perform new tool transfer and old tool handover according to the operation sequence; and finally controlling the tool changing robot to grab the new tool and move to the position of a tool to be installed to perform installation; the present application realizes automatic collaborative control of the entire tool changing process of the TBM, and improves tool changing efficiency and operation safety.

Description

Technical Field

[0001] This invention belongs to the field of tool-changing robot control, and particularly relates to a dual-track automatic tool-changing control method and system for full-face tunneling machines (TBMs). Background Technology

[0002] In automated tool changing operations of large tunnel boring equipment, a heterogeneous multi-robot system is formed by a tool changing robot, a tool lifting platform, and an automated transport vehicle. These robots share horizontal and vertical tracks within a confined space behind the cutterhead, collaboratively disassembling old tools and installing new ones. Due to varying tool wear levels, dynamically changing task urgency, and real-time changes in track occupancy, the timing of robot actions is highly coupled and prone to right-of-way conflicts. Achieving multi-robot task scheduling and collision-free motion planning is the core challenge for ensuring safe and efficient tool changing.

[0003] Patent CN119458338B discloses a visual servo control method for a tool-changing robot. It generates motion control commands through image compression encoding, feature extraction, and two-level feedback error determination, aiming to improve the end-effector servo accuracy of a single tool-changing robot under varying lighting and occlusion conditions. However, this solution focuses on the visual positioning performance of a single robot, failing to consider prioritizing multiple tools according to their wear urgency, or addressing the temporal coordination and path avoidance between the tool-changing robot and the auxiliary handling robot under shared track constraints. Therefore, it cannot resolve track occupancy conflicts and motion interference caused by multiple robots operating in parallel.

[0004] Patent CN114087240A discloses a hydraulic control system for a tool-changing robot. It employs an integrated axis-mounted proportional servo valve and detection device to implement distributed closed-loop control of the main arm, auxiliary arm, gripping mechanism, and bolt removal / installation motor, thereby improving the dynamic quality of the hydraulic actuators. However, this solution only optimizes the response speed and control accuracy of the drive system of a single tool-changing robot. It does not provide a tool-changing urgency assessment or a multi-task dynamic scheduling mechanism, nor does it address the generation of motion sequence for multi-robot collaboration or motion trajectory planning with avoidance logic. Therefore, it is difficult to achieve conflict-free parallel operation of heterogeneous robot clusters under conditions of limited track resources.

[0005] In summary, existing technologies have improved the local operational capabilities of a single tool-changing robot from the perspectives of visual servoing and hydraulic control, but they are all limited to optimizing single-machine functions and fail to incorporate the dynamic prioritization of multi-robot tasks, the allocation of right-of-way on shared tracks, and the resolution of conflicting action sequences into a unified planning framework. This results in delayed response to highly urgent tasks and frequent interference due to reliance on manual coordination between robots, which not only lengthens the tool-changing cycle but also increases the risk of robot collisions and the safety hazards of human intervention. Therefore, under the constraint of shared track resources, how to achieve integrated planning of task priority-driven multi-robot collaborative action sequences and conflict-free motion trajectories has become a key issue in improving the efficiency and safety of robotic tool changing. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention proposes a dual-track automatic tool changer control method and system for full-face tunneling machines (TBMs). The method includes: responding to tool change requirements by acquiring the target tool identifier and tool change urgency level; collecting real-time status information including the tool changer robot status, track occupancy status, and coordinates of the old tool to be replaced; based on the tool change urgency level and real-time status information, generating an operation sequence and motion trajectory for the tool changer robot, tool lifting platform, and automated guided vehicle (AGV) to collaboratively perform the tool changer operation using a motion planning algorithm. This operation sequence includes the action timing, path segments, and avoidance logic of each device; controlling the tool changer robot to move to the position of the old tool to be replaced according to the motion trajectory to perform gripping and disassembly; after disassembly, controlling the tool changer robot to move the old tool to the tool handover area, and simultaneously controlling the tool lifting platform and AAV to transfer the new tool and hand over the old tool according to the operation sequence; finally, controlling the tool changer robot to grip the new tool and move it to the position to be installed for installation. This invention achieves automated collaborative control of the entire TBM tool change process, improving tool changer efficiency and operational safety.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A dual-rail automatic tool changer control method for full-face tunneling machines (TBMs) includes:

[0009] In response to a tool change request at the coordinate position to be processed, the target tool identifier and tool change urgency level contained in the tool change request are obtained;

[0010] Collect real-time status information, including: the current workstation and joint status of the tool changing robot, the real-time occupancy status of the horizontal and vertical tracks, and the position coordinates of M old tools to be replaced behind the tool head;

[0011] Based on the tool change urgency level and the real-time status information, a motion planning algorithm optimized by particle swarm optimization is used to generate the operation sequence and motion trajectory of the tool change robot, tool lifting platform and automatic transport vehicle to perform the tool change operation in a coordinated manner; wherein, the operation sequence includes the action sequence, path segment and avoidance logic of each device;

[0012] The tool-changing robot is controlled to move according to the motion trajectory by a combination of horizontal and vertical tracks to reach the tool holder where the old tool to be replaced is located, and to perform the old tool grabbing and disassembly.

[0013] Once disassembly is complete, the tool-changing robot is controlled to carry the old tool to the preset tool handover area, and the tool transfer subsystem is activated simultaneously. Based on the operation sequence and motion trajectory, the tool lifting platform and the automatic transport transfer vehicle are controlled to transfer the new and old tools to be loaded.

[0014] When the new blade arrives at the blade exchange area, the tool changing robot is controlled to grab the new blade and move to the blade holder position according to the motion trajectory and operation sequence to perform the new blade installation.

[0015] Specifically, the tool lifting platform and the automated transport vehicle are controlled according to the operation sequence and motion trajectory to transfer new and old tools to be loaded, including:

[0016] The tool lifting platform is controlled to move along the vertical track of the shield to the preset tool handover area, and the automatic transport vehicle is simultaneously controlled to move the target number of new tools to be loaded to the preset tool transfer area within a preset time length. When the tool lifting platform reaches the tool handover area, the tool changing robot is controlled to place the clamped old tool on the tool lifting platform, and the tool lifting platform is controlled to carry the old tool down along the vertical track of the shield to the tool transfer area.

[0017] The preset time length is less than or equal to the sum of the first time length, the second time length, and the third time length; the first time length represents the time it takes for the tool lifting platform to move along the shield vertical track to the preset tool handover area; the second time length represents the time it takes for the tool changing robot to place the clamped old tool on the tool lifting platform; the third time length represents the time it takes for the tool lifting platform to carry the old tool down along the shield vertical track to the tool transfer area.

[0018] Upon arrival at the tool transfer area, the tool lifting platform controls the transfer of old tools to the automated transport vehicle located at the bottom, and grabs the new tools to be loaded, which then rise along the vertical track of the shield to the tool handover area.

[0019] Specifically, a motion planning algorithm optimized by particle swarm optimization generates the operation sequence and motion trajectory for the coordinated execution of the tool changing operation by the tool changing robot, tool lifting platform, and automated transport vehicle, including:

[0020] Obtain the priority weight and time constraint threshold corresponding to the tool changing urgency level, as well as the current pose and joint angle of the tool changing robot, the occupied range of the horizontal and vertical tracks, and the tool holder coordinates of each old tool to be replaced from the real-time status information.

[0021] The particle swarm optimization algorithm is used to optimize the tool change urgency level and the real-time status information to obtain an optimal scheduling parameter set, which includes at least the optimized priority weight and time constraint threshold.

[0022] The multiple tools to be replaced are sorted according to the priority weights to determine the scheduling order of the current task, and the allowable delay window for each device action is set according to the time constraint threshold.

[0023] Based on the occupied interval and the tool holder coordinates, a spatiotemporal state diagram is constructed, which includes track network nodes, path segment capacity and dynamic obstacles. The spatiotemporal state diagram marks the occupied time slice of each track segment with time as the dimension.

[0024] Based on the scheduling order and the spatiotemporal state diagram, the operation sequence and motion trajectory are generated using a preset hierarchical path planning strategy.

[0025] Specifically, the operation sequence and motion trajectory are generated using a preset hierarchical path planning strategy, including a first-layer path planning stage, a second-layer conflict detection and resolution stage, and a third-layer sequence encapsulation stage.

[0026] Specifically, the first-level path planning stage includes:

[0027] Obtain the spatiotemporal state diagram and the priority weights corresponding to the tool change urgency level;

[0028] A path cost function is constructed based on the priority weights, and the path cost function is used to quantify the degree to which time constraints are satisfied in path search;

[0029] Using the spatiotemporal state diagram as a constraint, a heuristic search algorithm is employed to plan a first conflict-free path segment from the current workstation coordinates to the target tool holder coordinates for the tool-changing robot based on the path cost function. The first conflict-free path segment includes the movement sequence along the horizontal and vertical tracks, as well as the entry and exit times in each track segment.

[0030] Obtain the key time slices in the first conflict-free path segment;

[0031] Constrained by the spatiotemporal state diagram and on the premise of not occupying the key time slice, a second path segment is planned for the tool lifting platform from its initial position to the tool handover area, and a third path segment is planned for the automated transport vehicle from its initial position to the tool transfer area. The second path segment and the third path segment are configured to be executed in parallel with the first non-conflicting path segment in the time dimension, so that the tool changing robot, the tool lifting platform, and the automated transport vehicle form an initial non-intersecting path set when sharing the horizontal track and the vertical track.

[0032] Specifically, the second-layer conflict detection and resolution phase consists of:

[0033] Obtain the set of track segment occupancy time slices corresponding to the first conflict-free path segment, the second path segment, and the third path segment, wherein each track segment occupancy time slice includes a track segment identifier, an occupancy start time, and an occupancy end time;

[0034] Intersection detection is performed on any two time slots in the set of track segment occupancy time slots to identify conflicting time slots that overlap in the time slot occupancy interval under the same track segment identifier, and the first device identifier, second device identifier, conflict duration and conflict location coordinates corresponding to each conflicting time slot are recorded.

[0035] For each conflict time slice, the priority weights determined by the tool change urgency levels of the first device and the second device are obtained, and the priority weights of the two are compared.

[0036] If the priority weight of the first device is less than the priority weight of the second device, then the first device is marked as a device to avoid and the second device is marked as a device with priority passage.

[0037] If the priority weight of the first device is greater than the priority weight of the second device, the first device is marked as the priority passage device and the second device is marked as the avoidance device; if the priority weight of the first device is equal to the priority weight of the second device, the avoidance device and the priority passage device are determined according to the preset avoidance rules.

[0038] Specifically, the second-layer conflict detection and resolution stage further includes:

[0039] For each conflict time slot, the duration of the conflict is compared with a preset avoidance time threshold. If the duration of the conflict is less than the avoidance time threshold, the start time of the avoidance device is adjusted so that the time when the avoidance device enters the conflict track section is delayed until after the priority passage device leaves the track section, and the adjusted start time is generated.

[0040] If the duration of the conflict is greater than or equal to the avoidance time threshold, then an avoidance node is inserted for the avoidance device at the track node preceding the conflict location coordinates, and an avoidance path segment containing the waiting time or detour path is generated to replace the conflict part in the original path segment of the avoidance device.

[0041] The path segment adjusted at the start time or the updated path segment after inserting the avoidance path segment is merged with the original path segment that has not conflicted, to obtain the updated first non-conflicting path segment, second path segment and third path segment that satisfy the mutual exclusion occupancy constraint, and the adjusted start time and the avoidance path segment are encapsulated as the avoidance logic in the operation sequence.

[0042] Specifically, the avoidance equipment and priority passage equipment are determined according to preset avoidance rules, including:

[0043] Obtain the track segment type identifier corresponding to the conflict time slice, wherein the track segment type identifier includes the main track segment identifier and the branch track segment identifier;

[0044] If the track segment type identifier is a main track segment identifier, then the first distance traveled by the first device at the start of the conflict time slot and the second distance traveled by the second device at the start of the conflict time slot are obtained, wherein the distance traveled is the cumulative path length along the track from the entry node of the main track segment to the current position of each device; the first distance traveled and the second distance traveled are compared, and the device with the larger distance traveled is marked as the priority passage device, and the other device is marked as the avoidance device;

[0045] If the track section type is identified as a branch track section, then the first remaining path length from the current position of the first device to the exit node of the branch track section and the second remaining path length from the current position of the second device to the exit node of the branch track section are obtained; the first remaining path length and the second remaining path length are compared, and the device with the smaller remaining path length is marked as the priority passage device, and the other device is marked as the avoidance device.

[0046] Specifically, before comparing the duration of the conflict with a preset avoidance time threshold, the method further includes:

[0047] Obtain the attribute parameters of the track segment where the conflict location coordinates are located. The track segment attribute parameters include the track curvature radius, track gradient, and the maximum allowable operating speed of the segment.

[0048] Obtain the braking response time constant and control command transmission delay time corresponding to the avoidance device;

[0049] The maximum safe turning speed corresponding to the track section is calculated based on the track curvature radius, and the maximum safe turning speed is compared with the maximum allowable operating speed. The smaller of the two values ​​is taken as the constrained operating speed of the track section.

[0050] The constrained operating speed is corrected based on the track slope to generate the actual restricted speed for the track section, wherein the actual restricted speed is negatively correlated with the track slope;

[0051] Based on the actual speed limit and the length of the track segment where the conflict location coordinates are located, calculate the theoretical minimum passage time for the track segment;

[0052] The theoretical minimum passage time, the braking response time constant, and the control command transmission delay time are summed to generate the avoidance time threshold.

[0053] Specifically, an avoidance node is inserted at the preceding track node at the conflict location coordinates for the avoidance device, and an avoidance path segment containing the waiting time or detour path is generated, including:

[0054] Obtain the conflict start time and conflict end time corresponding to the conflict time slice, as well as the real-time position coordinates and current running speed of the avoidance device at the current planning time.

[0055] Based on the real-time position coordinates and the current operating speed, calculate the estimated arrival time of the avoidance device as it moves along its original path segment to the previous track node;

[0056] Compare the estimated arrival time with the conflict termination time:

[0057] If the expected arrival time is earlier than the conflict end time, the difference between the conflict end time and the expected arrival time is determined as the waiting time, and a waiting-type avoidance path segment containing the waiting time at the previous track node is generated; at the same time, the entry time of all path segments after the previous track node in the original path segment of the avoidance device is delayed by the waiting time.

[0058] If the estimated arrival time is later than or equal to the conflict end time, it is determined that there is no need to wait, and the original path segment of the avoidance device remains unchanged;

[0059] Obtain a preset maximum waiting time threshold, which is determined based on the time constraint threshold corresponding to the tool change urgency level.

[0060] Specifically, an avoidance node is inserted for the avoidance device at the track node preceding the conflict location coordinates, and an avoidance path segment containing the waiting time or detour path is generated, further including:

[0061] Compare the waiting time with the maximum waiting time threshold:

[0062] If the waiting time is less than or equal to the maximum waiting time threshold, then the waiting-type avoidance path segment is taken as the avoidance path segment.

[0063] If the waiting time exceeds the maximum waiting time threshold, then the path search is performed again with the previous track node as the starting point and the first track node after the conflict part in the original path segment of the avoidance device as the ending point, and the spatiotemporal state diagram as the constraint. A detour path segment that does not pass through the track segment where the conflict position coordinates are located is generated. The entry time of each node in the detour path segment is calculated according to the length of the detour path segment and the actual speed limit. The detour path segment is then spliced ​​in time with the part before the previous track node and the part after the first track node in the original path segment of the avoidance device to obtain a detour-type avoidance path segment containing the detour trajectory.

[0064] The waiting-type avoidance path segment or the detour-type avoidance path segment is used as the updated avoidance device path segment.

[0065] The dual-rail automatic tool changer control system for full-face tunneling machines (TBMs) includes:

[0066] The acquisition module is configured to: in response to a tool change request at the coordinate position to be processed, acquire the target tool identifier and tool change urgency level contained in the tool change request;

[0067] Collect real-time status information, including: the current workstation and joint status of the tool changing robot, the real-time occupancy status of the horizontal and vertical tracks, and the position coordinates of M old tools to be replaced behind the tool head;

[0068] The path planning module, based on the tool change urgency level and the real-time status information, generates the operation sequence and motion trajectory of the tool change robot, tool lifting platform and automatic transport vehicle to perform the tool change operation in a coordinated manner through a motion planning algorithm optimized by particle swarm optimization; wherein, the operation sequence includes the action sequence, path segment and avoidance logic of each device;

[0069] The tool changing control module is configured to control the tool changing robot to move according to the motion trajectory through a combination of horizontal and vertical tracks to reach the tool holder where the old tool to be replaced is located, and to perform the old tool gripping and disassembly.

[0070] Once disassembly is complete, the tool-changing robot is controlled to carry the old tool to the preset tool handover area, and the tool transfer subsystem is activated simultaneously. Based on the operation sequence and motion trajectory, the tool lifting platform and the automatic transport transfer vehicle are controlled to transfer the new and old tools to be loaded.

[0071] When the new blade arrives at the blade exchange area, the tool changing robot is controlled to grab the new blade and move to the blade holder position according to the motion trajectory and operation sequence to perform the new blade installation.

[0072] Compared with the prior art, the beneficial effects of the present invention are:

[0073] To address the shortcomings of existing technologies, this invention constructs a spatiotemporal state diagram that integrates the urgency level of tool changing with the real-time track occupancy status, and employs a hierarchical path planning strategy to generate an operation sequence that includes equipment action timing, path segments, and avoidance logic. This enables precise collaborative control of the tool changing robot, tool lifting platform, and automated transport vehicle on shared horizontal and vertical track resources. Its beneficial effects are as follows: First, based on the priority weight and scheduling order determined by the urgency level of tool changing, the execution sequence of multiple tasks can be dynamically optimized according to the urgency of tool wear, significantly improving the response speed of emergency tool changing tasks and reducing unplanned downtime of TBM caused by tool failure. Second, by accurately marking the time slice occupied by the track section through the spatiotemporal state diagram, combined with the conflict detection and resolution mechanism, right-of-way conflicts between multiple devices at track intersections and shared sections are effectively avoided, ensuring that each device can operate safely and orderly in parallel in a confined space, greatly improving the continuity and efficiency of the tool changing process. Finally, by generating motion trajectories containing avoidance logic, the risk of interference and collision hazards between devices are reduced. While achieving fully automated tool changing, the safety of equipment operation is ensured, the reliance on manual intervention is reduced, and the safety hazards of personnel working behind the tool head are fundamentally eliminated, demonstrating outstanding practicality. Attached Figure Description

[0074] Figure 1 This is a flowchart of the dual-track automatic tool changer control method for a full-face tunnel boring machine (TBM) according to Embodiment 1 of the present invention.

[0075] Figure 2 This is a diagram of the dual-track automatic cutter changer structure of the full-face tunnel boring machine (TBM) according to Embodiment 1 of the present invention;

[0076] Figure 3 This is a schematic diagram of the TBM cutter head tool wear alarm in Embodiment 1 of the present invention;

[0077] Figure 4 This is a schematic diagram of the startup of the TBM tool-changing robot in Embodiment 1 of the present invention;

[0078] Figure 5This is a schematic diagram of the positioning of the TBM tool-changing robot in Embodiment 1 of the present invention;

[0079] Figure 6 This is a schematic diagram of the tool-grabbing TBM tool-changing robot in Embodiment 1 of the present invention;

[0080] Figure 7 This is a schematic diagram of removing the worn-out knife in Embodiment 1 of the present invention;

[0081] Figure 8 This is a schematic diagram of the transfer of worn old blades in Embodiment 1 of the present invention;

[0082] Figure 9 This is a schematic diagram illustrating the transport of worn-out blades to the lifting platform according to Embodiment 1 of the present invention;

[0083] Figure 10 This is a schematic diagram of transporting worn-out tools to a transfer trolley according to Embodiment 1 of the present invention;

[0084] Figure 11 This is a schematic diagram of the lifting platform delivering a new blade in Embodiment 1 of the present invention;

[0085] Figure 12 This is a schematic diagram of Embodiment 1 of the present invention, showing the grasping of a new blade and its transfer to the back of the blade disc;

[0086] Figure 13 This is a schematic diagram of the new tool loading and positioning of the TBM tool changing robot according to Embodiment 1 of the present invention;

[0087] Figure 14 This is a schematic diagram of the new tool loading and positioning of the TBM tool changing robot according to Embodiment 1 of the present invention;

[0088] Figure 15 This is a schematic diagram of the TBM tool changing robot completing the new tool loading in Embodiment 1 of the present invention;

[0089] Figure 16 This is a schematic diagram of the TBM tool-changing robot in Embodiment 1 of the present invention completing its task and returning to the L1 area;

[0090] Figure 17 This is a block diagram of the dual-track automatic tool changer control system for a full-face tunnel boring machine (TBM) according to the present invention.

[0091] Attached reference numerals: ① Tool changing robot, ② Horizontal track, ③ Vertical track, ④ Shield vertical track, ⑤ Tool lifting platform, ⑥ Automated transport vehicle, ⑦ Tool. Detailed Implementation

[0092] Example 1

[0093] Please see Figure 1-16The present invention provides an embodiment of a dual-track automatic tool changer control method for a full-face tunneling machine (TBM), applied to the tool changer control optimization process of the tool changer robot in a full-face tunneling machine, comprising:

[0094] A1. In response to the tool change request at the coordinate position to be processed, obtain the target tool identifier and tool change urgency level contained in the tool change request; collect real-time status information, including: the current station and joint status of the tool changing robot, the real-time occupancy status of the horizontal and vertical tracks, and the position coordinates of M old tools to be replaced behind the tool head.

[0095] For example, in this embodiment, the target tool identification and tool change urgency level are obtained in real time through a TBM tool wear monitoring system. Specifically, a vibration sensor, a temperature sensor, and an encoder are installed at each tool holder on the back of the tool turret. The vibration sensor collects the vibration amplitude of the tool during operation at a sampling frequency of 1 Hz, the temperature sensor simultaneously collects the tool temperature value, and the encoder records the cumulative number of tool rotations. The control system performs a wear assessment every 10 minutes, inputting the collected vibration characteristic values, temperature values, and cumulative number of rotations into a preset tool wear prediction model. This model is trained based on a BP neural network and outputs the current comprehensive tool wear value as a percentage, ranging from 0% to 100%, and the remaining life prediction value in hours. In this embodiment, the tool wear prediction model adopts a three-layer architecture based on a BP neural network, specifically including an input layer, a single hidden layer, and an output layer. The input layer has three neurons, corresponding to vibration feature values ​​(expressed as vibration amplitude, collected once per second), temperature values, and cumulative rotation counts (recorded by the encoder). The hidden layer has eight neurons, using the sigmoid function as the activation function to extract the nonlinear coupling relationship between input features. The output layer has two neurons, outputting the comprehensive wear value (expressed as a percentage, ranging from 0% to 100%) and the predicted remaining lifespan (in hours). During model training, the tool lifecycle data collected and labeled from historical operations is used as the training set. Each sample includes the vibration feature values, temperature values, cumulative rotation counts, and the actual wear amount and actual remaining lifespan as determined by offline detection at the corresponding time. The mean squared error is used as the loss function during training. The weights and biases of each layer of neurons are iteratively updated using the backpropagation algorithm, and gradient descent optimization is performed with a learning rate of 0.01 until the loss function converges to a preset threshold or the maximum number of training rounds is reached. After training, the weight parameters are fixed and deployed in the wear assessment module of the control system. In actual tool changing operations, the control system triggers a wear assessment process every 10 minutes, inputting the real-time collected vibration characteristic values, temperature values, and cumulative rotation counts into the trained BP neural network. The model can then simultaneously output the current tool's comprehensive wear value and remaining life prediction value. For example, the comprehensive wear value of T1 is calculated to be 85%, and the remaining life prediction value is 2.5 hours, providing a quantitative basis for determining subsequent tool changing needs and classifying urgency levels.

[0096] In this embodiment, taking three cutting tools T1, T2, and T3 as examples, the control system uses a trained BP neural network tool wear prediction model to perform forward inference calculations every 10 minutes on the vibration characteristic values, temperature values, and cumulative rotation counts collected at the current moment, obtaining the real-time status of each tool: T1 has a comprehensive wear value of 85% and a predicted remaining life of 2.5 hours; T2 has a comprehensive wear value of 92% and a predicted remaining life of 1.2 hours; and T3 has a comprehensive wear value of 78% and a predicted remaining life of 4.8 hours. The control system internally presets a fixed tool replacement response threshold, which is set at a comprehensive wear value of 80%. This threshold is determined by the statistical fusion of the tool manufacturer's recommended safe operating limits and field operation experience data. When the comprehensive wear value of any tool reaches or exceeds 80%, it is determined that the tool has entered an unreliable wear range, and a tool replacement requirement must be triggered to avoid tunneling interruption or cutterhead damage due to tool failure. Based on this, the T1 overall wear value is 85% ≥ 80%, triggering the tool replacement requirement; the T2 overall wear value is 92% ≥ 80%, triggering the tool replacement requirement; and the T3 overall wear value is 78% < 80%, not triggering the tool replacement requirement.

[0097] For cutting tools that have triggered a tool change requirement, the control system further classifies the urgency level of tool change based on the predicted remaining lifespan: tools with a predicted remaining lifespan of less than or equal to 2 hours are classified as emergency level, as these tools are nearing failure and require priority for tool change, and are assigned the highest priority weight of 0.9; tools with a predicted remaining lifespan greater than 2 hours and less than or equal to 4 hours are classified as normal level, as these tools, although exceeding the safety threshold, still have a certain remaining working window, and are assigned a medium priority weight of 0.7; tools with a predicted remaining lifespan greater than 4 hours are classified as low priority level, as these tools, although triggering a tool change requirement, have a relatively mild wear process, and are assigned a lower priority weight of 0.5. The principles for setting the above time thresholds of 2 hours and 4 hours are as follows: 2 hours corresponds to the average time required for a TBM to complete a complete work cycle under typical tunneling conditions, ensuring that emergency tools have sufficient handling window before failure; 4 hours corresponds to the theoretical maximum time taken from triggering to completion of a conventional tool change operation process. When the remaining lifespan exceeds this time, it indicates that the tool can wait until other more urgent tasks are completed before handling. The priority weights of 0.9, 0.7, and 0.5 for each level represent the quantitative scale of the urgency of the system's response to the task in the scheduling decision. The higher the weight, the higher the priority in path planning and resource allocation. Accordingly, T1 with a remaining lifespan of 2.5 hours belongs to the normal level, with a priority weight of 0.7; T2 with a remaining lifespan of 1.2 hours belongs to the urgent level, with a priority weight of 0.9; T3 does not participate in the urgency level classification and priority weight assignment because it has not triggered a tool change requirement. Finally, the control system encapsulates the tool identifier that triggers the tool change requirement, along with its corresponding urgency level and priority weight, into a tool change requirement signal, which is sent to the motion planning module via industrial Ethernet for subsequent task scheduling and path planning.

[0098] A2. Based on the tool change urgency level and the real-time status information, a motion planning algorithm optimized by particle swarm optimization is used to generate the operation sequence and motion trajectory of the tool change robot, tool lifting platform and automatic transport vehicle to perform the tool change operation in a coordinated manner; wherein, the operation sequence includes the action sequence, path segment and avoidance logic of each device;

[0099] A3. Control the tool changing robot to move according to the motion trajectory by a combination of horizontal and vertical tracks to reach the tool holder where the old tool to be replaced is located, and perform old tool grabbing and disassembly.

[0100] A4. Once disassembly is complete, control the tool changing robot to move the old tool to the preset tool handover area, and simultaneously activate the tool transfer subsystem. Based on the operation sequence and motion trajectory, control the tool lifting platform and the automatic transport transfer vehicle to transfer the new and old tools to be loaded.

[0101] A5. When the new tool to be installed arrives at the tool handover area, the tool changing robot is controlled to grab the new tool and move to the tool holder position according to the motion trajectory and operation sequence to perform the new tool installation. When the new tool is fully installed, the TBM tool changing robot completes the operation, returns to the L1 area, and the optimization simulation ends.

[0102] It should be further explained that this embodiment uses a motion planning algorithm optimized by particle swarm optimization to generate the operation sequence and motion trajectory for the coordinated execution of the tool changing operation by the tool changing robot, the tool lifting platform, and the automated transport vehicle, including:

[0103] A201. Obtain the priority weight and time constraint threshold corresponding to the tool replacement urgency level, as well as the current pose and joint angle of the tool-changing robot, the occupied range of the horizontal and vertical tracks, and the tool holder coordinates of each old tool to be replaced in the real-time status information; use the particle swarm optimization algorithm to optimize the tool replacement urgency level and the real-time status information to obtain the optimal scheduling parameter set, which includes at least the optimized priority weight and time constraint threshold; in this embodiment, the time constraint threshold is set differently according to the remaining life prediction value corresponding to the tool replacement urgency level. The setting principle is: based on the current remaining life prediction value of each tool that triggers the tool replacement requirement, it is converted into the maximum allowable completion time of the tool replacement operation according to a preset ratio to ensure that the entire tool replacement process is safely completed before the tool enters the failure threshold. Specifically, for emergency-level tools (remaining life less than or equal to 2 hours), the time constraint threshold is set to 80% of the predicted remaining life. For example, if the predicted remaining life of T2 is 1.2 hours, then the time constraint threshold T = 1.2 × 0.8 × 3600 = 3456 seconds. For normal-level tools (remaining life greater than 2 hours and less than or equal to 4 hours), the time constraint threshold is set to 60% of the predicted remaining life. For example, if the predicted remaining life of T1 is 2.5 hours, then the time constraint threshold T = 2.5 × 0.6 × 3600 = 5400 seconds. For low-priority-level tools (remaining life greater than 4 hours), the time constraint threshold is set to 50% of the predicted remaining life. For example, if the remaining life of a tool is 5 hours, then the time constraint threshold T = 5 × 0.5 × 3600 = 9000 seconds. The aforementioned conversion ratios of 80%, 60%, and 50% are based on the following: Those skilled in the art, after deducting historical data on the inherent system time consumption such as status acquisition cycles, communication transmission delays, and equipment start-up / stop transition times, reserve reasonable safety margins for tools with different levels of urgency. A 20% buffer time is reserved for emergency levels to handle unforeseen circumstances, a 40% buffer time is reserved for normal levels to balance efficiency and safety, and a 50% buffer time is reserved for low-priority levels to allow for more flexible scheduling windows. This time constraint threshold is encapsulated along with the tool change request signal and sent to the motion planning module via industrial Ethernet, serving as the time-domain boundary condition for constructing the path cost function and setting the allowable delay window for each device's actions in path planning.

[0104] As a specific example, in this embodiment, the motion planning algorithm optimized by the particle swarm optimization algorithm first optimizes the tool change urgency level and real-time status information to obtain the optimal scheduling parameter set. The specific process is as follows: the fitness function is the weighted sum of minimizing the total tool change completion time T_total and minimizing the number of multi-device track conflicts C_conflict. The particle position vector Xi=(kp,kt,ks) is selected by prioritization weight correction coefficient kp (range 0.8~1.2), time constraint tightness coefficient kt (range 0.5~1.0), and avoidance slope influence factor ks (range 0.01~0.03). The population size is initialized to 30. The initial position and velocity of each particle are randomly generated in the solution space. The historical best position Pbest_i of each particle is initialized as its initial position, and the global best position Gbest is initialized as the position of the particle with the lowest fitness. During the iterative optimization process, each generation decodes the particle's kp, kt, and ks into actual scheduling parameters. For tool T1 (original priority weight 0.7, original time constraint threshold 5400 seconds) and tool T2 (original priority weight 0.9, original time constraint threshold 3456 seconds) that have triggered tool change requirements in this embodiment, if a particle's position is kp=1.07, kt=0.88, and ks=0.02, then the optimized priority weight of T1 becomes 0.7×1.07≈0.75, and the time constraint threshold is adjusted to 5400×1.07.

[0105] 0.88 ≈ 4752 seconds (rounded to 4800 seconds in this embodiment); the T2 priority weight becomes 0.9 × 1.07 ≈ 0.96, and the time constraint threshold is adjusted to 3456 × 0.88 ≈ 3041 seconds (rounded to 3000 seconds in this embodiment); simultaneously, the slope correction coefficient is updated to λ = 1 - 0.02 × θ_slope for subsequent avoidance time threshold calculation. The optimized parameters are injected into the rapid simulation evaluation of the motion planning algorithm to obtain the corresponding total completion time and number of conflicts. The fitness value is calculated, and the individual historical best and global best are updated accordingly. The algorithm evolves according to the original velocity and position update formulas of the particle swarm optimization algorithm, where the inertia weight ω decreases linearly from 0.9 to 0.4, and the learning factors c1 = c2 = 2.0. Iteration continues until the maximum number of iterations (100 generations) is reached or the global best fitness remains unchanged for 20 consecutive generations. Finally, the particle swarm optimization algorithm outputs the optimal scheduling parameter set corresponding to the global optimal position, including the optimized T1 priority weight of 0.75 and time constraint threshold of 4800 seconds, T2 priority weight of 0.96 and time constraint threshold of 3000 seconds, and slope correction coefficient λ = 1 - 0.02 × θ_slope. This optimal scheduling parameter set is then passed to the motion planning module, and in the subsequent step A201, it is directly used as the basis for obtaining the priority weight and time constraint threshold values. For example, in A202, the time constraint threshold of 3000 seconds for the highest priority tool T2 is converted into a robot movement time constraint of 300 seconds according to the preset movement time limit ratio of 10%, thereby generating an allowable delay window [180 seconds, 300 seconds].

[0106] A202. Sort multiple tools to be replaced according to the priority weights to determine the scheduling order of the current task, and set the allowable delay window for each device's actions based on the time constraint thresholds. For example, there are currently three old tools T1, T2, and T3 to be replaced, with priority weights corresponding to their tool replacement urgency levels of w1=0.9, w2=0.5, and w3=0.7, respectively, and time constraint thresholds of T11=300 seconds, T21=600 seconds, and T31=450 seconds, respectively. Sort the tools from highest to lowest priority weight, and determine the scheduling order of the task as T11→T31→T21. Based on the time constraint thresholds for each tool, set the allowable delay for the tool changing robot, tool lifting platform, and automated transport vehicle. The window is defined as follows: for the highest priority tool T1, the allowable delay window for the tool changer robot to move to the tool position is [T11-120, T11] = [180 seconds, 300 seconds], meaning that the movement must be completed within 180 seconds after the task starts, otherwise it is considered a timeout. The allowable delay window for the tool lift platform to rise to the handover area is [T11-180, T11-60] = [120 seconds, 240 seconds]. The allowable delay window for the automated transport vehicle to transport the new tool to the transfer area is [T11-240, T11-120] = [60 seconds, 180 seconds]. For the second-priority tool T3, its allowable delay window is recalculated based on the remaining time after the completion of task T11, ensuring that multi-task collaboration meets its respective time constraints.

[0107] A203. Based on the occupied interval and the tool holder coordinates, a spatiotemporal state diagram is constructed, including track network nodes, path segment capacity, and dynamic obstacles. The spatiotemporal state diagram uses time as the dimension to mark the occupied time slices of each track segment. For example, specifically:

[0108] Obtain the track network topology of the horizontal and vertical tracks, wherein the track network topology includes the coordinates of each track intersection point, the length of each track segment, and the connection relationship of each track segment; taking the horizontal track as an example, it is set to include nodes H1 (coordinate (0,0)), H2 (coordinate (5,0)), H3 (coordinate (10,0)), H4 (coordinate (15,0)), and H5 (coordinate (20,0)), the vertical track includes nodes V1 (coordinate (10,0)), V2 (coordinate (10,5)), and V3 (coordinate (10,10)), and the shield vertical track includes nodes S1 (coordinate (10,-5)), S2 (coordinate (10,-10)), and S3 (10,-15), wherein nodes H3 and V1 are the same intersection point;

[0109] The real-time occupancy status of the horizontal and vertical tracks is obtained from the real-time status information. The real-time occupancy status includes the current equipment identifier, equipment type, occupancy start time, and expected occupancy end time for each track segment. The current time is set to 08:00:00. Horizontal track segments H2-H3 are occupied by tool-changing robot A, with an occupancy start time of 07:59:30 and an expected occupancy end time of 08:05:30. Horizontal track segments H4-H5 are occupied by automated transport vehicle B, with an occupancy start time of 07:58:00 and an expected occupancy end time of 08:12:00. Vertical track segments V1-V2 are in an idle state.

[0110] Obtain the coordinates of the tool holders of the M old tools to be replaced, set M=3, the coordinates of tool holder T1 to be replaced are (12, 8), the coordinates of tool holder T2 are (8, 9), and the coordinates of tool holder T3 are (14, 7), where T1 and T3 are located in the tool head area corresponding to the vertical track V2-V3 section, and T2 is located in the tool head area corresponding to the vertical track V1-V2 section;

[0111] Based on the aforementioned track network topology, each track segment is discretized into path segment units with a fixed length (set to 1 meter), and each path segment unit is assigned a unique segment identifier; for example, the horizontal track H2-H3 is 5 meters long and is discretized into five path segment units: H2-1, H2-2, H2-3, H2-4, and H2-5.

[0112] For each path segment unit, the maximum number of devices that can be accommodated at the same time is set as the path segment capacity based on its physical properties (such as track curvature and slope). The capacity of the horizontal straight track segment is set to 2, the capacity of the curved track segment is 1, and the capacity of the vertical track segment is 1.

[0113] Based on the real-time occupancy status, each occupied path segment unit is mapped to the time dimension, generating a two-dimensional occupancy grid with the time axis as the horizontal axis and the track segment identifier as the vertical axis. For the horizontal track H2-3 path segment unit, it is set that it is occupied by the tool changing robot A in the time interval of 07:59:30-08:05:30 and by the automated transport vehicle B in the time interval of 08:00:00-08:12:00 (because B occupies the H4-H5 segment, which contains 10 path segment units from H4-1 to H5-5, they need to be marked separately).

[0114] Project the tool holder coordinates onto the nearest track network node to determine the target track segment and target node corresponding to each old tool to be replaced; the tool holder coordinates (12, 8) of T1 are projected onto the vertical track node V3 (coordinates (10, 10)), and need to move 2 meters down along the vertical track V2-V3 segment to reach it; the tool holder coordinates (8, 9) of T2 are projected onto the vertical track node V2 (coordinates (10, 5)), and need to move 2 meters horizontally and then 4 meters vertically; the tool holder coordinates (14, 7) of T3 are projected onto the vertical track node V3 (coordinates (10, 10)), and need to move 4 meters horizontally and then 3 meters vertically.

[0115] Acquire dynamic obstacle information, including equipment performing tasks, temporarily parked transfer vehicles, and protruding structural components behind the cutterhead; set that at the current moment there is a temporary obstacle under maintenance at the intersection of node H3-V1, which occupies the area from 08:00:00 to 08:20:00, and this obstacle affects the passage of the horizontal track H3-H4 section and the vertical track V1-V2 section;

[0116] All information is integrated to construct a spatiotemporal state diagram, which includes: a set of track network nodes N={H1, H2, H3, H4, H5, V1, V2, V3, S1, S2, S3}, a set of path segment units E={H1-1, ..., H5-5, V1-1, ..., V3-5, S1-1, ..., S3-5}, the capacity C(e) of each path segment unit, and the set of time slices occupied by each path segment unit on the time axis O(e)={[t_start1, t_en... d1, device_id1], [t_start2, t_end2, device_id2], ...}, the set of path segment units D={[e_id, t_start, t_end]} affected by dynamic obstacles within a specific time interval, and the target path segment unit E_target={T1:V2-5, T2:V1-4, T3:V2-5} corresponding to each old knife to be replaced (where V2-5 represents the last path segment unit corresponding to node V2);

[0117] The spatiotemporal state diagram is stored as a three-dimensional data structure, with track segment identification as the first dimension, time axis as the second dimension, and occupancy status and capacity constraints as the third dimension, for use in the generation of conflict detection and avoidance logic in subsequent path planning algorithms.

[0118] A204. Based on the scheduling order and the spatiotemporal state diagram, a preset hierarchical path planning strategy is used to generate the operation sequence and motion trajectory, specifically including a first-layer path planning stage, a second-layer conflict detection and resolution stage, and a third-layer sequence encapsulation stage; the first-layer path planning stage specifically includes:

[0119] A2041. Obtain the priority weight corresponding to the tool change urgency level;

[0120] A2042. Construct a path cost function based on the priority weights. The path cost function is used to quantify the degree of satisfaction of time constraints in path search. Specifically, it involves: first obtaining the priority weight w of the i-th tool to be replaced. i and the corresponding time constraint threshold T i (For example, for emergency tools, w=0.9, T=5 minutes; for regular tools, w=0.5, T=10 minutes); where the path cost function is defined as f(P)=t_base(P)+α w i The path cost function max(0, t_base(P)-T), where t_base(P) is the theoretical travel time along path P, and α is a magnification factor (e.g., 100). This function introduces a timeout penalty term proportional to the priority weight, ensuring that high-priority tasks incur greater costs if they exceed the time constraint during path search. This leads the heuristic algorithm to prioritize paths that can be completed within the constraint time, quantifying the degree to which the time constraint is met. In this embodiment, the magnification factor α is set to 100, which transforms the timeout duration exceeding the time constraint threshold during path planning into a cost comparable to the basic travel time of the path, thus forming an effective gradient guidance in the path search algorithm. The coefficient is set based on the following: the standard moving speed of the tool-changing robot along the horizontal and vertical tracks is 0.5 m / s. In a typical tool-changing scenario, the basic movement time t_base(P) from the current workstation to the farthest tool holder is usually between 30 and 300 seconds, with a numerical order of ten squared. The timeout duration max(0, t_base(P)-T) is expected to be between 0 and tens of seconds under reasonable planning. If directly added to t_base(P), the weight of the timeout penalty in the cost function is too small to guide the search algorithm to prioritize avoiding timeout paths. Setting the amplification coefficient α to 100 can amplify the timeout duration from tens of seconds to thousands of generations, bringing it to the same numerical order of magnitude as the basic movement time, ensuring that the timeout penalty has a substantial impact on the total cost during path search. Taking tool T1 as an example, its priority weight is 0.7, and the time constraint threshold T is 5400 seconds. If the theoretical movement time of a candidate path is 5450 seconds and the timeout duration is 50 seconds, then the timeout penalty is 100 × 0.7 × 50 = 3500. After adding this to the basic movement time of 5450 seconds, the total cost is 8950, which is significantly higher than the total cost of a path without timeout. This guides the heuristic search algorithm to prioritize paths that complete the movement within the time constraint threshold. If the value of α is too small, the cost difference between timeout paths and non-timeout paths is not significant, and the algorithm may not be able to effectively distinguish them. If the value of α is too large, it may cause the algorithm to excessively avoid timeouts and choose suboptimal paths that take too long a detour or wait too long, thus reducing the overall work efficiency. Simulation verification shows that when α is 100, a balance can be achieved between timeout avoidance and path length optimization, so that the path cost function accurately reflects the quantitative requirements of the tool change urgency level on the degree of time constraint satisfaction.

[0121] A2043. Using the spatiotemporal state diagram as a constraint, a heuristic search algorithm is adopted to plan a first conflict-free path segment from the current workstation coordinates to the target tool holder coordinates for the tool changing robot based on the path cost function; the first conflict-free path segment includes the movement sequence along the horizontal and vertical tracks, as well as the entry and exit times in each track segment.

[0122] For example, this embodiment uses a heuristic search algorithm to plan a first conflict-free path segment from the current workstation coordinates to the target tool holder coordinates for the tool changing robot based on the path cost function. The specific steps include:

[0123] Set the current planning start time as t0 = 08:00:00, and the current workstation of the tool changing robot corresponds to the track network node H1(0,0); the coordinates of the target tool holder T2 are (8,9), which, when projected to the nearest track node V2(10,5), requires moving 8 meters horizontally and then 4 meters vertically to reach it; obtain the priority weight w = 0.8 corresponding to the urgency level of the tool changing, and the time constraint threshold T = 480 seconds; construct the path cost function f(P) = t_base(P) + α w max(0, t_base(P)-T), where t_base(P) is the theoretical travel time of path P, α=100 is the amplification factor, and the robot's moving speed v=0.5 m / s.

[0124] The starting node H1 and its arrival time t0 are added to the open list as the initial state. The actual cost of the initial state g(H1) = 0. The heuristically estimated cost h(H1) is calculated by dividing the Euclidean distance by the velocity: h(H1) = dist(H1, V2) / v = ≈11.18 / 0.5=22.36 seconds; Total cost f(H1)=g(H1)+h(H1)=22.36 seconds; Close the list initially as empty.

[0125] Select the node with the smallest total cost f from the open list for expansion; the current node is H1.

[0126] Obtain the set of adjacent nodes of node H1. According to the track network topology, the adjacent node is H2(5,0).

[0127] Calculate the travel time Δt1 from H1 to H2 = distance / v = 5 / 0.5 = 10 seconds, and the estimated arrival time t1 = 08:00:10;

[0128] Query the occupancy status of track segment H1-H2 in the time interval [t0, t1] in the spatiotemporal state diagram. There is no occupancy record for this segment, so the path is feasible.

[0129] Calculate the actual cost after reaching node H2: g(H2) = g(H1) + Δt1 = 10 seconds. The heuristic estimate is h(H2) = dist(H2, V2) / v = ≈7.07 / 0.5=14.14 seconds, total cost f(H2)=10+14.14=24.14 seconds;

[0130] Add node H2, its arrival time t1, cost g(H2), and f(H2) to the open list, and move H1 to the closed list.

[0131] Select node H2 (f=24.14 seconds) from the open list with the smallest current f and expand it:

[0132] Get the set of adjacent nodes of node H2, including H1 (closed) and H3 (10, 0);

[0133] For adjacent node H3, the movement time Δt2 = 5 / 0.5 = 10 seconds is calculated, and the estimated arrival time t2 = 08:00:20;

[0134] Query the occupancy status of track segment H2-H3 in the time interval [08:00:10, 08:00:20] in the spatiotemporal state diagram. This segment is occupied by tool changing robot A from 07:59:30 to 08:05:30. Therefore, the time interval overlaps with the occupancy interval, resulting in a conflict. This child node is not feasible.

[0135] Node H2 has no other scalable adjacent nodes and cannot generate any feasible child nodes, so it is moved to the close list and marked as an expansion failure.

[0136] Currently, there are no other nodes in the open list besides the expanded nodes, causing the search to be unable to continue. To resolve the path deadlock, a wait operation is introduced at the node to avoid temporary occupancy conflicts. Specifically, at node H2, the current time is 08:00:10, the occupancy time of the conflict segment H2-H3 ends at 08:05:30, and the waiting time Δt_wait = 08:05:30 - 08:00:10 = 320 seconds. After waiting, the movement time from H2 to H3 is still 10 seconds, and the arrival time at H3 is 08:05:40. Calculate the actual cost after waiting: g(H2_wait) = g(H2) + Δt_wait = 10 + 320 = 330 seconds. Add the node H3 after waiting as a new node to the open list. Its arrival time is 08:05:40, g(H3) = 330 seconds. Heuristically estimate h(H3) = dist(H3, V2) / v = 5 / 0.5 = 10 seconds. The total cost is f(H3) = 330 + 10 = 340 seconds. Record the waiting operation as part of the path.

[0137] Select node H3 (f=340 seconds) from the open list with the smallest current f and expand it:

[0138] The adjacent nodes of node H3 include H4(15,0) and V1(10,0) (H3 and V1 are the same coordinate point, representing the starting point of the vertical track).

[0139] To reach the target, V1 is extended first, that is, moving from H3 along the vertical track;

[0140] Calculate the time Δt3 = 5 / 0.5 = 10 seconds for moving from H3 along the vertical track V1-V2 to V2(10,5). The estimated arrival time is t3 = 08:05:40 + 10 = 08:05:50.

[0141] Query the occupancy status of vertical track segment V1-V2 in the spatiotemporal status diagram during the time interval [08:05:40, 08:05:50]. The segment is free, and the path is feasible.

[0142] Calculate g(V2) = g(H3) + Δt3 = 330 + 10 = 340 seconds. Heuristically estimate h(V2) = dist(V2, target tool holder) / v. The coordinates of the target tool holder T2 are (8, 9), and the coordinates of node V2 are (10, 5). The Euclidean distance between the two points is... The distance is approximately 4.47 meters. The tool-changing robot moves at a speed of 0.5 meters per second along the tool head fine-tuning track. Therefore, the theoretical time to move from V2 to the target tool holder is approximately 8.94 seconds (4.47 / 0.5), rounded down to 9 seconds. The actual cost after reaching V2 is g(V2) = 340 seconds, and the heuristically estimated cost is h(V2) = 9 seconds. Therefore, the total cost of node V2 is f(V2) = 340 + 9 = 349 seconds. Finally, the robot extends from V2 to the target tool holder, with a movement time of 9 seconds and an arrival time of 08:05:59. The total cost is g(Goal) = 349 seconds, h(Goal) = 0, and f(Goal) = 349 seconds.

[0143] Add node V2, its arrival time t3, and cost to the open list;

[0144] The movement from V2 to the fine-tuning track where the target tool holder is located is 2 meters, and the movement time is Δt4 = 4 seconds. The arrival time is t_goal = 08:05:50 + 4 = 08:05:54. The occupancy status of the fine-tuning section is found to be idle. Upon reaching the target node Goal, g(Goal) = g(V2) + Δt4 = 340 + 4 = 344 seconds, h(Goal) = 0, and f(Goal) = 344 seconds.

[0145] The obtained path node sequence and its corresponding entry and exit times are encapsulated into the first conflict-free path segment. The path node sequence is: H1→H2 (waiting)→H3→V2→target node. Specific timestamps are as follows:

[0146] Node H1: Entry time 08:00:00, Exit time 08:00:10;

[0147] Node H2: Entry time 08:00:10, waiting start time 08:00:10, waiting end time 08:05:30, exit time 08:05:40;

[0148] Node H3: Entry time 08:05:40, Exit time 08:05:50;

[0149] Node V2: Entry time 08:05:50, Exit time 08:05:54;

[0150] Goal of the target node: entry time 08:05:54, which is the arrival time.

[0151] The path segment satisfies the occupancy constraints of all track segments in the spatiotemporal state diagram within the corresponding time interval, and the theoretical path movement time t_base = 354 seconds (including movement time of 34 seconds and waiting time of 320 seconds), which is less than the time constraint threshold of 480 seconds. Therefore, the path cost f = 354 seconds (no penalty).

[0152] A2044. Obtain the key time slice in the first conflict-free path segment;

[0153] A2045. Based on the spatiotemporal state diagram and on the premise of not occupying the key time slice, plan a second path segment from the initial position to the tool transfer area for the tool lifting platform, and plan a third path segment from the initial position to the tool transfer area for the automated transport vehicle; wherein, the second path segment and the third path segment are configured to be executed in parallel with the first non-conflicting path segment in the time dimension, so that the tool changing robot, the tool lifting platform, and the automated transport vehicle form an initial non-intersecting path set when sharing the horizontal track and the vertical track;

[0154] For example, in this embodiment, the tool lifting platform plans a second path segment from its initial position to the tool handover area, and the automated transport vehicle plans a third path segment from its initial position to the tool transfer area, specifically including the following steps:

[0155] Step 1: Obtain the initial state and target position of the tool lifting platform and the automated transport vehicle.

[0156] The current planning start time is set to t0 = 08:00:00. The tool lift is initially located at node S3(10, -15) at the bottom of the shield's vertical track, with its target being the preset height point P_lift in the tool handover area. This point is located 2 meters directly below the tool retrieval point (intersection of H3 and V1) on the shield's vertical track, i.e., coordinates (10, -2). The automated guided vehicle (AGVV) is initially located at node B1(0, -15) on the bottom horizontal track, with its target being node B3(10, -15) in the tool transfer area. This point corresponds to position S3 at the bottom of the shield's vertical track and is used to receive the old tools transferred from the lift. The lifting speed of the tool lift is set to v_lift = 0.5 m / s, and the horizontal movement speed of the AVRV is set to v_car = 1.0 m / s.

[0157] Step 2: Obtain the critical time slices in the first conflict-free path segment, specifically:

[0158] The first conflict-free path segment is the path taken by the tool-changing robot from H1 to the target tool holder T2, and its time interval for occupying track resources is as follows:

[0159] H1→H2: 08:00:00-08:00:10 (occupies horizontal track H1-H2);

[0160] H2 waiting: 08:00:10-08:05:30 (does not occupy the track, but is counted in the time);

[0161] H2→H3: 08:05:30-08:05:40 (occupies horizontal track H2-H3);

[0162] H3→V2: 08:05:40-08:05:50 (occupies vertical tracks V1-V2);

[0163] V2→Goal: 08:05:50-08:05:54 (occupies the tool head fine-tuning track).

[0164] The critical time slice is defined as the period that may share the same track resources with the tool lift or automated transport vehicle. Since the shield's vertical track and bottom horizontal track are physically independent of the robot's tracks (H and V tracks) and have no directly shared sections, the critical time slice of the first path segment does not conflict with the second and third path segments. However, to maintain versatility, it is still necessary to check for spatial intersections or the influence of dynamic obstacles.

[0165] Step 3: Using the spatiotemporal state diagram as a constraint, plan the second path segment for the tool lift platform, specifically as follows:

[0166] 3.1 Constructing the moving path of the tool lift platform

[0167] The cutter lift ascends vertically along the shield's vertical track from S3(10, -15) to P_lift(10, -2). The path segments it traverses include S3-S2 (5 meters long, discretely S3-1 to S3-5), S2-S1 (5 meters long, discretely S2-1 to S2-5), and S1 to P_lift (3 meters long, discretely S1-1 to S1-3). The total travel distance is 5 + 5 + 3 = 13 meters, and the travel time Δt_lift = 13 / 0.5 = 26 seconds.

[0168] 3.2 Calculate arrival time and check occupancy.

[0169] The initial time is t0 = 08:00:00, and the arrival time at P_lift is t_lift_arrival = 08:00:26. Query the occupancy status of each segment of the shield's vertical track in the spatiotemporal state diagram within the time interval [08:00:00, 08:00:26]:

[0170] Section S3-S2 (S3-1 to S3-5): The spatiotemporal state diagram shows no occupancy records;

[0171] Section S2-S1 (S2-1 to S2-5): Unoccupied;

[0172] Segment S1-P_lift (S1-1 to S1-3): Unoccupied.

[0173] Therefore, the path is feasible and there is no conflict.

[0174] 3.3 Generate the second path segment

[0175] Encapsulate the path node sequence and timestamps into a second path segment:

[0176] Node S3: Entry time 08:00:00, Exit time 08:00:05 (Move to node S3-2, but the overall path is in segments).

[0177] Section S3→S2: Entry time 08:00:00, exit time 08:00:10 (each meter segment is passed through sequentially, but simplified as the whole section);

[0178] Section S2→S1: Entry time 08:00:10, Exit time 08:00:20;

[0179] Section S1→P_lift: Entry time 08:00:20, Exit time 08:00:26.

[0180] The robot finally reaches the target point P_lift at 08:00:26. This path segment does not overlap or conflict with the first path segment (the robot is moving on the H1-H2 path segment at this time, and the trajectory is different).

[0181] Step 4: Using the spatiotemporal state diagram as constraints, plan the third path segment for the automated transport vehicle, specifically as follows:

[0182] 4.1 Constructing the movement path of the automated transport transfer vehicle

[0183] The automated guided vehicle (AGV) moves horizontally from B1(0, -15) to B3(10, -15) along the bottom horizontal track. The path segments it needs to traverse include B1-B2 (5 meters long, discretely divided into B1-1 to B1-5) and B2-B3 (5 meters long, discretely divided into B2-1 to B2-5). The total travel distance is 10 meters, and the travel time Δt_car = 10 / 1.0 = 10 seconds.

[0184] 4.2 Calculate arrival time and check occupancy

[0185] The initial time is t0 = 08:00:00, and the arrival time at B3 is t_car_arrival = 08:00:10. Query the occupancy status of each segment of the bottom horizontal track in the spatiotemporal state diagram within the range [08:00:00, 08:00:10]:

[0186] Section B1-B2 (B1-1 to B1-5): The spatiotemporal state diagram shows no occupancy records;

[0187] Section B2-B3 (B2-1 to B2-5): Unoccupied.

[0188] Therefore, the path is feasible.

[0189] 4.3 Generate the third path segment

[0190] Encapsulate the path node sequence and timestamps into a third path segment:

[0191] Node B1: Entry time 08:00:00, Exit time 08:00:05;

[0192] Section B1→B2: Entry time 08:00:00, Exit time 08:00:05;

[0193] Section B2→B3: Entry time 08:00:05, Exit time 08:00:10;

[0194] The final arrival time at target point B3 is 08:00:10.

[0195] Step 5: Form an initial set of non-intersecting paths, specifically:

[0196] Align the first, second, and third non-collision path segments on the timeline to obtain the initial set of non-intersecting paths:

[0197] Tool Changer Robot: Path segment [H1→H2(waiting)→H3→V2→Goal], time interval [08:00:00, 08:05:54];

[0198] Tool lift: Path segment [S3→S2→S1→P_lift], time interval [08:00:00, 08:00:26];

[0199] Automated transport vehicle: route segment [B1→B2→B3], time interval [08:00:00, 08:00:10].

[0200] All path segments operate independently on their respective tracks, with no track resource conflicts, and all meet their respective allowable delay windows (for example, the allowable delay window for the tool lift to rise to the handover area is [120 seconds, 240 seconds], and its actual arrival time of 26 seconds is much less than the lower limit, meeting the requirements). This path set provides an initial feasible solution for subsequent conflict detection and resolution.

[0201] It should be further explained that, in this embodiment, obtaining the key time slices in the first conflict-free path segment in A2044 specifically refers to extracting those time periods from the planned tool-changing robot path that may compete for track resources or intersect with other equipment (tool lifting platform, automated transport vehicle), in order to provide avoidance constraints for subsequent parallel path planning. The identification of key time slices includes two dimensions: one is the time slice that directly occupies the same track segment, and the other is the time slice that, although the tracks are different, has potential conflicts at spatial intersection points (such as the H3-V1 node). For example, taking the first conflict-free path segment obtained by A2043 planning as an example, its path node sequence and timestamp are: H1→H2 (08:00:00-08:00:10), H2 waiting (08:00:10-08:05:30), H2→H3 (08:05:30-08:05:40), H3→V2 (08:05:40-08:05:50), V2→Goal (08:05:50-08:05:54). The H2→H3 section occupies the horizontal track H2-H3, which does not directly overlap with the shield's vertical track. However, node H3 is also the starting point of the vertical track V1. Therefore, there is a risk of spatial intersection at node H3 during this period (08:05:30-08:05:40), and it needs to be marked as a critical time slice. Similarly, the H3→V2 section occupies the vertical track V1-V2, which intersects the shield's vertical track perpendicularly at node V1 (i.e., H3). Therefore, the period 08:05:40-08:05:50 also needs to be marked as a critical time slice. These key time slices and their corresponding track node coordinates are extracted to form a key time slice set K={(H3, [08:05:30, 08:05:40]), (V1, [08:05:40, 08:05:50])}. This set is used to ensure that the tool lifting platform and automated transport vehicle do not occupy or pass through the corresponding intersection nodes during the time period, thereby avoiding potential conflicts.

[0202] It should be further explained that the conflict detection and resolution stage of the second layer in this embodiment is specifically as follows:

[0203] A2046. Obtain the set of track segment occupancy time slices corresponding to the first conflict-free path segment, the second path segment, and the third path segment, wherein each track segment occupancy time slice includes a track segment identifier, an occupancy start time, and an occupancy end time.

[0204] A2047. Perform intersection detection on any two of the time slots occupied by the track segment in the set, identify conflicting time slots that overlap in the time slots occupied under the same track segment identifier, and record the first device identifier, the second device identifier, the duration of the conflict, and the coordinates of the conflict location corresponding to each conflicting time slot.

[0205] A2048. For each conflict time slice, obtain the priority weight determined by the tool change urgency level of the first device and the second device respectively, and compare the priority weights of the two.

[0206] A2049. If the priority weight of the first device is less than the priority weight of the second device, then the first device is marked as a device to avoid and the second device is marked as a device with priority passage.

[0207] A20410. If the priority weight of the first device is greater than the priority weight of the second device, then the first device is marked as the priority passage device and the second device is marked as the avoidance device; if the priority weight of the first device is equal to the priority weight of the second device, then the avoidance device and the priority passage device are determined according to the preset avoidance rules.

[0208] For example, in the conflict detection and resolution stage of the second layer of this embodiment, taking a specific operational scenario as an example: after the tool changing robot completes the disassembly of the old T2 tool, its first conflict-free path segment is planned to move along the H2-H3 section to node H3 from 08:05:30 to 08:05:40; the second path segment of the tool lifting platform rising from the bottom S3 to the tool handover area P_lift is planned to rise along the shield vertical track S1-S2 section from 08:05:35 to 08:05:45; it should be further explained that the conflict detection and resolution stage of the second layer of this embodiment is specifically as follows:

[0209] A2046. Obtain the set of track segment occupancy time slices corresponding to the first conflict-free path segment, the second path segment, and the third path segment, wherein each track segment occupancy time slice includes a track segment identifier, an occupancy start time, and an occupancy end time.

[0210] A2047. Perform intersection detection on any two of the time slots occupied by the track segment in the set, identify conflicting time slots that overlap in the time slots occupied under the same track segment identifier, and record the first device identifier, the second device identifier, the duration of the conflict, and the coordinates of the conflict location corresponding to each conflicting time slot.

[0211] A2048. For each conflict time slice, obtain the priority weight determined by the tool change urgency level of the first device and the second device respectively, and compare the priority weights of the two.

[0212] A2049. If the priority weight of the first device is less than the priority weight of the second device, then the first device is marked as a device to avoid and the second device is marked as a device with priority passage.

[0213] A20410. If the priority weight of the first device is greater than the priority weight of the second device, then the first device is marked as the priority passage device and the second device is marked as the avoidance device; if the priority weight of the first device is equal to the priority weight of the second device, then the avoidance device and the priority passage device are determined according to the preset avoidance rules.

[0214] For example, in the conflict detection and resolution stage of the second layer of this embodiment, taking a specific operation scenario as an example: after the tool changing robot completes the disassembly of the old tool T2, its first conflict-free path segment is planned to move along the H2-H3 section to node H3 from 08:05:30 to 08:05:40; the second path segment of the tool lifting platform rising from the bottom S3 to the tool handover area P_lift is planned to rise along the shield vertical track S1-S2 section from 08:05:35 to 08:05:45; the third path segment of the automated transport vehicle moving from B1 to B3 is planned to move along the bottom horizontal track B2-B3 section from 08:00:05 to 08:00:10. Intersection detection revealed that although the H2-H3 section occupied by the tool-changing robot and the S1-S2 section occupied by the tool lift are on different tracks, they spatially intersect at node (10, 0) H3-V1, and their time intervals overlap from 08:05:35 to 08:05:40, with a conflict duration of 5 seconds. The first device corresponding to this conflict time slot is the tool-changing robot (device ID: R02), and the second device is the tool lift (device ID: L01). The conflict location coordinates are the intersection point of H3-V1. The priority weights of the current tasks of the two devices are obtained: the tool-changing robot is performing the T2 tool change task (priority weight 0.8), and the tool lift is performing the T2 new tool transfer task (priority weight 0.6). After comparison, the priority weight of R02 is greater than that of L01. Therefore, the tool-changing robot R02 is marked as the priority passage device, and the tool lift L01 is marked as the avoidance device. If two devices have the same priority weight, for example, if the tool changing robot and the automated transport vehicle both have a weight of 0.7 in the subsequent T3 task, then according to the preset avoidance rules: at the intersection, if the tool changing robot has entered the section and is about to complete the passage, it is marked as the priority passing device, and the automated transport vehicle waits; if neither of them has entered, the automated transport vehicle is set to avoid the tool changing robot by default according to the device type.

[0215] It should be further explained that this embodiment determines the avoidance device and the priority passage device according to preset avoidance rules, including:

[0216] Obtain the track segment type identifier corresponding to the conflict time slice, wherein the track segment type identifier includes the main track segment identifier and the branch track segment identifier;

[0217] If the track segment type identifier is a main track segment identifier, then the first distance traveled by the first device at the start of the conflict time slot and the second distance traveled by the second device at the start of the conflict time slot are obtained, wherein the distance traveled is the cumulative path length along the track from the entry node of the main track segment to the current position of each device; the first distance traveled and the second distance traveled are compared, and the device with the larger distance traveled is marked as the priority passage device, and the other device is marked as the avoidance device;

[0218] If the track section type is identified as a branch track section, then the first remaining path length from the current position of the first device to the exit node of the branch track section and the second remaining path length from the current position of the second device to the exit node of the branch track section are obtained; the first remaining path length and the second remaining path length are compared, and the device with the smaller remaining path length is marked as the priority passage device, and the other device is marked as the avoidance device.

[0219] For example, the tool-changing robot R1 and the automated transport vehicle C1 have a conflict time slot at the intersection of node H3-V1, and the conflict time is 08:12:30. The track segment type identifier corresponding to the conflict time slot is obtained. After querying the spatiotemporal state diagram, the track segments involved in the intersection point include the main track H3-H4 (horizontal main track) and the branch track V1-V2 (vertical branch track). First, determine the type of track segment where the conflict occurred: If the conflict occurred on the main track H3-H4 segment, then the distance traveled by R1 from the H3 entrance node at the conflict's initiation time (08:12:30) is 3 meters (having passed 3 path segment units), and the distance traveled by C1 from the H4 entrance node is 1 meter (having passed 1 path segment unit). R1, with the larger travel distance, is marked as the priority passing device, and C1 is marked as the avoidance device. If the conflict occurred on the branch track V1-V2 segment, then the remaining path length from R1's current position to the branch track's exit node V2 is 2 meters, and the remaining path length from C1's current position to the exit node V2 is 4 meters. R1, with the smaller remaining path length, is marked as the priority passing device, and C1 is marked as the avoidance device. This rule quantifies the progress of the equipment on the track, ensuring that when priority weights are equal, the equipment closer to completing the current segment's passage has priority, avoiding a decrease in overall operational efficiency due to indiscriminate avoidance.

[0220] It should be further noted that, before comparing the duration of the conflict with a preset avoidance time threshold, this embodiment also includes:

[0221] Obtain the attribute parameters of the track segment where the conflict location coordinates are located. The track segment attribute parameters include the track curvature radius, track gradient, and the maximum allowable operating speed of the segment.

[0222] Obtain the braking response time constant and control command transmission delay time corresponding to the avoidance device;

[0223] The maximum safe turning speed corresponding to the track section is calculated based on the track curvature radius r, and the maximum safe turning speed is compared with the maximum allowable operating speed. The smaller of the two values ​​is taken as the constrained operating speed of the track section.

[0224] The constrained operating speed is corrected based on the track slope to generate the actual restricted speed for the track section, wherein the actual restricted speed is negatively correlated with the track slope. In this embodiment, the correction of the constrained operating speed based on the track slope and the generation of the actual restricted speed are achieved by introducing a slope correction coefficient function. This slope correction coefficient function uses the slope angle of the track section as the independent variable and the correction coefficient as the dependent variable. Its functional form is obtained by fitting the least squares method based on the measured data of safe operation of equipment under different slope conditions in historical operations. Specifically, it is expressed as follows:

[0225] λ(θ_slope)=1-0.02 θ_slope

[0226] Wherein, θ_slope is the slope angle of the track section, in degrees, and is taken as an absolute value. For both uphill and downhill sections, the absolute value of the slope is used for calculation and correction. The coefficient 0.02 is the slope influence factor obtained from the fitting, which characterizes the degree of speed reduction per unit slope angle. This function shows a linear approximation to the measured safe speed reduction law in the range of θ_slope∈[0°, 15°], with a goodness of fit R² of 0.96, which can accurately reflect the negative correlation between the actual speed limit and the track slope.

[0227] Based on this correction coefficient function, multiplying the constrained running speed by the correction coefficient generates the actual limiting speed for this track segment:

[0228] v_limit=v_constraint×λ(θ_slope)=v_constraint×(1-0.02 θ_slope),

[0229] For example, in this embodiment, the horizontal track section H3-H4 where the conflict location is located has a 5-degree downhill slope. The maximum permissible operating speed of this section is compared with the maximum safe turning speed, and the smaller value is taken to obtain the constrained operating speed v_constraint as 1.0 m / s. Substituting the slope angle θ_slope=5° into the correction coefficient function, we get λ(5)=1-0.02×5=0.90; then the actual restricted speed v_limit=1.0×0.90=0.9 m / s. This speed value serves as the actual upper limit of safe passage in this track section and is used for the subsequent calculation of the avoidance time threshold.

[0230] Based on the actual speed limit and the length of the track segment where the conflict location coordinates are located, calculate the theoretical minimum passage time for the track segment;

[0231] The theoretical minimum passage time, the braking response time constant, and the control command transmission delay time are summed to generate the avoidance time threshold.

[0232] For example, before comparing the duration of the conflict with a preset avoidance time threshold, this embodiment takes a specific operational scenario as an example: the tool-changing robot R1 and the tool lift L1 generate a conflict time slice at the intersection of node H3-V1, the conflict duration is 5 seconds, and the conflict location coordinates are (10, 0). First, the attribute parameters of the track segment where the conflict location is located are obtained. This segment is a horizontal track H3-H4, the track curvature radius r is 50 meters (curve), the track slope is 5 degrees downhill, and the maximum allowable running speed of this segment is 1.0 m / s; the braking response time constant of the avoidance device (determined to be the tool lift L1 according to priority comparison) is obtained as 0.3 seconds, and the control command transmission delay time is 0.2 seconds. Based on the track curvature radius of 50 meters, the maximum safe turning speed is calculated according to the formula v_max_safe = (Taking the friction coefficient μ=0.15, g=9.8), the calculated value is v_max_safe≈ The speed is approximately 8.57 m / s, which is much greater than the maximum permissible operating speed of 1.0 m / s. Therefore, the smaller value of 1.0 m / s is taken as the constraint operating speed. Considering the track gradient is 5 degrees downhill, the constraint operating speed is corrected. The speed needs to be reduced on the downhill slope to ensure safety. The correction factor is 0.9, generating the actual limit speed v_limit = 1.0 × 0.9 = 0.9 m / s. The track section H3-H4 where the conflict location is located is 10 meters long (from node H3 to H4). Based on the actual limit speed of 0.9 m / s, the theoretical minimum passage time t_pass = 10 / 0.9 ≈ 11.11 seconds is calculated. The theoretical minimum passage time of 11.11 seconds, the braking response time constant of 0.3 seconds, and the control command transmission delay time of 0.2 seconds are added together to generate the avoidance time threshold t_threshold = 11.11 + 0.3 + 0.2 = 11.61 seconds. This threshold is used for comparison with the conflict duration of 5 seconds. Since the conflict duration is less than the avoidance time threshold, the conflict is resolved by adjusting the start time (delaying the elevator until the robot passes before starting), rather than inserting an avoidance node.

[0233] A20411. For each conflict time slot, the conflict duration is compared with a preset avoidance time threshold. If the conflict duration is less than the avoidance time threshold, the start time of the avoidance device is adjusted so that the time when the avoidance device enters the conflict track section is delayed until after the priority passage device leaves the track section, and the adjusted start time is generated.

[0234] A20412. If the duration of the conflict is greater than or equal to the avoidance time threshold, then an avoidance node is inserted for the avoidance device at the track node preceding the conflict location coordinates, and an avoidance path segment containing the waiting time or detour path is generated to replace the conflict part in the original path segment of the avoidance device.

[0235] It should be further explained that in this embodiment, the preceding track node at the conflict location coordinates is used to insert a avoidance node for the avoidance device, and an avoidance path segment containing the waiting time or detour path is generated, including:

[0236] Obtain the conflict start time and conflict end time corresponding to the conflict time slice, as well as the real-time position coordinates and current running speed of the avoidance device at the current planning time.

[0237] Based on the real-time position coordinates and the current operating speed, calculate the estimated arrival time of the avoidance device as it moves along its original path segment to the previous track node;

[0238] Compare the estimated arrival time with the conflict termination time:

[0239] If the expected arrival time is earlier than the conflict end time, the difference between the conflict end time and the expected arrival time is determined as the waiting time, and a waiting-type avoidance path segment containing the waiting time at the previous track node is generated; at the same time, the entry time of all path segments after the previous track node in the original path segment of the avoidance device is delayed by the waiting time.

[0240] If the estimated arrival time is later than or equal to the conflict end time, it is determined that there is no need to wait, and the original path segment of the avoidance device remains unchanged;

[0241] A preset maximum waiting time threshold is obtained. In this embodiment, the maximum waiting time threshold is used to constrain the longest allowable time for the avoidance device to pause and wait at the avoidance node during the conflict resolution process. If this time is exceeded, the waiting strategy is abandoned and a detour strategy is adopted instead. This threshold is determined by calculating the time constraint threshold of the tool change urgency level corresponding to the current task according to a preset ratio. Specifically, the time constraint threshold T corresponding to the tool change urgency level is used as the benchmark, and a preset percentage of this threshold is taken as the maximum waiting time threshold, that is: T_wait_max=β×T; where β is the waiting time ratio coefficient, which is set to 10% in this embodiment. The basis for setting this coefficient is that within the allowable time window of the entire tool change process, sufficient execution time should be reserved for necessary actions such as robot track movement, tool grabbing and disassembly, and new tool installation; the waiting time is only used as an auxiliary means of conflict resolution and should not excessively squeeze the time resources of normal operation. Therefore, the maximum waiting time is limited to within 10% of the total time constraint threshold to ensure that even in the most unfavorable conflict scenario, the device still has more than 90% of its time budget to complete the core operation actions. For example, tool T1 belongs to the normal level, with a time constraint threshold T = 5400 seconds, so the maximum waiting time threshold T_wait_max = 0.10 × 5400 = 540 seconds; tool T2 belongs to the emergency level, with a time constraint threshold T = 3456 seconds, so the maximum waiting time threshold T_wait_max = 0.10 × 3456 ≈ 346 seconds. During the conflict resolution phase, when the calculated waiting time is less than or equal to the maximum waiting time threshold, the system adopts a waiting-type avoidance strategy; when the waiting time exceeds the maximum waiting time threshold, the system abandons waiting and instead performs a new path search at the previous track node, generating a detour path segment that does not pass through the conflict section, to avoid the overall tool change operation exceeding the time constraint threshold due to prolonged waiting.

[0242] The path segment adjusted at the start time or the updated path segment after inserting the avoidance path segment is merged with the original path segment that has not conflicted, to obtain the updated first non-conflicting path segment, second path segment and third path segment that satisfy the mutual exclusion occupancy constraint, and the adjusted start time and the avoidance path segment are encapsulated as the avoidance logic in the operation sequence.

[0243] For example, a conflict time slice occurs between the tool changer robot R02 and the tool lift L01 at the intersection of node H3-V1. The conflict starts at 08:12:30 and ends at 08:12:45, lasting for 15 seconds. After priority weight comparison, the tool changer robot R02 is the priority passage device, and the tool lift L01 is the avoidance device. The conflict duration of 15 seconds is compared with a preset avoidance time threshold, which is calculated to be 12.61 seconds based on the track segment attributes. Since the conflict duration (15 seconds) is greater than the avoidance time threshold (12.61 seconds), an avoidance node needs to be inserted at the track node preceding the conflict location coordinates (intersection of H3-V1) where the avoidance device L01 is located. According to the query, L01 is currently located at the shield vertical track node S2(10, -10). Its original path segment was planned to move along S2-S1-P_lift. The conflict section is located in the S1-P_lift segment (which needs to pass through the intersection point H3-V1). The previous track node is S1(10, -5). The real-time position coordinates of L01 at the conflict start time of 08:12:30 are obtained as S2(10, -10), and the current running speed is 0.5 m / s. The distance it takes to move along the original path segment to the previous track node S1 is calculated to be 5 meters. The estimated arrival time is 08:12:30 + 5 / 0.5 = 08:12:40. Comparing the estimated arrival time of 08:12:40 with the conflict end time of 08:12:45, the estimated arrival time is earlier than the conflict end time. Therefore, the waiting time = conflict end time - estimated arrival time = 08:12:45 - 08:12:40 = 5 seconds. Obtain the preset maximum waiting time threshold, which is determined based on the time constraint threshold corresponding to the tool change urgency level. The time constraint threshold for the current task corresponding to the T2 tool is 480 seconds, and the maximum waiting time threshold is set to 10% of the time constraint threshold, i.e., 48 seconds. Since the waiting time of 5 seconds is less than the maximum waiting time threshold of 48 seconds, a waiting-type avoidance path segment is generated: L01 pauses and waits for 5 seconds at node S1, without occupying the track during the waiting period. After the waiting period ends, it continues to move along the S1-P_lift segment at 08:12:45, and the entry time of the path segment after S1-P_lift in the original path segment is uniformly delayed by 5 seconds. The second path segment after the waiting adjustment is merged with the first and third path segments that do not conflict to obtain the updated second path segment: L01 arrives at S1 at 08:12:40, waits until 08:12:45, and moves to P_lift from 08:12:45 to 08:12:55. The adjusted start time and waiting avoidance path segments are encapsulated into avoidance logic in the operation sequence for subsequent execution. If the waiting time exceeds 48 seconds, a detour path segment needs to be generated at S1, such as detouring via an alternate track to avoid conflict. However, in this example, the waiting time is within the threshold, so a waiting strategy is adopted.

[0244] This embodiment constructs a spatiotemporal state diagram by integrating the urgency level of tool changing with the real-time track occupancy status, and generates a multi-device collaborative operation sequence using a hierarchical path planning strategy. This achieves precise collaborative control of the tool changing robot, tool lifting platform, and automated transport vehicle on shared track resources. Its beneficial effects are as follows: First, task sorting based on priority weights and the setting of allowable delay windows ensures that urgent tools are processed first, avoiding high-priority tools from timed out due to task disorder. Second, by accurately marking the time slices occupied by track sections through the spatiotemporal state diagram, combined with key time slice identification and conflict detection and resolution mechanisms, right-of-way conflicts between multiple devices at track intersections and shared sections are effectively avoided, ensuring the safety and continuity of parallel operations. Finally, by introducing dynamic adjustment of waiting time and encapsulation of avoidance logic, unnecessary detours are minimized while meeting time constraints, significantly improving the overall efficiency of the tool changing process, while reducing equipment collision risks and the need for manual intervention, providing reliable support for the intelligent and automated operation of TBM tool changing.

[0245] It should be further explained that this embodiment uses a combination of horizontal and vertical tracks to move to the location of the old tool holder to be replaced, and performs the old tool gripping and disassembly, including:

[0246] A301. The control system acquires the first conflict-free path segment generated in step S2. This path segment contains the node sequence and corresponding timestamps of the tool-changing robot from the current station H1 to the target tool holder T2: H1 departs at 08:00:00, arrives at H2 at 08:00:10 and waits until 08:05:30, moves to H3 from 08:05:30 to 08:05:40, moves along the vertical track to V2 from 08:05:40 to 08:05:50, and moves along the tool head fine-tuning track to the target tool holder T2 from 08:05:50 to 08:05:54. The control system converts the path node sequence into the position reference input of the proportional-integral-derivative controller. Combining the robot's current joint angle and the preset moving speed of 0.5 m / s, it calculates the speed and direction of the drive motor of the walking mechanism in real time, and generates horizontal track movement commands and vertical track lifting commands to ensure that the robot passes through each track segment strictly according to the planned time.

[0247] A302. As the robot moves along the horizontal track from H1 to H2, the control system continuously monitors the occupancy status of the track segment H2-H3 ahead using a magnetic induction sensor. When the robot arrives at H2 at 08:00:10, it detects that the H2-H3 segment is still occupied by the tool-changing robot A until 08:05:30. Based on the pre-set waiting instructions in the path planning, the control system controls the robot to pause at node H2, the drive motors enter standby mode, and a timer starts to wait for 320 seconds. During the waiting period, the control system maintains monitoring of the track status. If it detects that the occupancy has ended prematurely, it immediately wakes up the robot to continue moving; otherwise, it strictly waits until 08:05:30.

[0248] After the A303 waiting period, the control system activated the walking mechanism at 08:05:30, driving the robot to move along the H2-H3 section at a speed of 0.5 meters per second, reaching node H3 at 08:05:40. It then switched to vertical drive mode, ascending along the V1-V2 vertical track at the same speed, reaching node V2 at 08:05:50. Finally, the cutter head fine-tuning track drive mechanism was activated, moving the robot horizontally for 2 meters at a speed of 0.5 meters per second, reaching the target tool holder T2 position at 08:05:54. Upon arrival, the laser rangefinder sensor confirmed that the relative distance between the robot and the tool holder was less than 5 millimeters, generating a positioning confirmation signal.

[0249] A304. After the robot reaches its position, the control system immediately activates the vision positioning subsystem mounted on the robot's gripper. This vision system includes two industrial cameras with a resolution of 1920 x 1080 and a frame rate of 30 frames per second, as well as a ring-shaped LED light source. The control system first controls the cameras to capture images of the four bolts on the tool holder from multiple angles, acquiring an image dataset of the bolt area. Then, the Canny edge detection algorithm is used to extract the edge features of the bolts, obtaining the contour data of each bolt. The extracted edge features are matched with a preset bolt template database. Through feature comparison and spatial coordinate transformation, the three-dimensional spatial coordinates and attitude parameters of the four bolts are calculated. The attitude parameters include the angle between the bolt axis and the tool holder plane. Finally, bolt positioning parameter information is generated, including the X-axis coordinate, Y-axis coordinate, Z-axis coordinate, yaw angle, and pitch angle values ​​for each bolt. In this embodiment, the preset bolt template database is a set of reference features pre-built and stored in the vision positioning module of the control system through offline calibration before the tool change operation. The database was constructed as follows: First, a standard bolt with specifications exactly matching those used for the actual tools mounted on the cutter head was selected as the calibration sample. For example, an M24 high-strength bolt with a nominal diameter of 24 mm, a thread pitch of 3 mm, and a hexagonal head width across the flats of 36 mm was used. The calibration bolt was installed on a standard tool holder fixture. Under standard lighting conditions, two industrial cameras on the end effector of the tool-changing robot, with a resolution of 1920×1080 and a frame rate of 30 frames per second, were used in conjunction with a ring LED light source to photograph the calibration bolt from multiple preset angles (including 0° directly above, 15° to the left, 15° to the right, and ±10° pitch). At least 50 frames were captured from each angle, resulting in a total of over 200 bolt sample images. For each captured image frame, the Canny edge detection algorithm is used to extract the bolt's edge contour features. The circular boundary of the bolt head is located using Hough transform circle detection, and the vertex features of the hexagonal head are extracted using a corner detection algorithm, forming a multi-dimensional feature descriptor that includes the bolt's edge contour, the hexagonal head's apex corner, and the bolt's axial direction. Simultaneously, a laser rangefinder sensor acquires the bolt's three-dimensional spatial coordinates and attitude parameters relative to the camera coordinate system, including yaw and pitch angles. Each set of feature descriptors is associated and bound with its corresponding three-dimensional spatial coordinates and attitude parameters to construct a bolt template data record. All bolt template data records are categorized by bolt specification and stored in the control system's non-volatile storage medium, forming a bolt template database.In the actual tool changing operation of the A305, after the robot arrives at the tool holder where the old tool to be removed is located, the vision positioning subsystem takes real-time pictures of the four fixing bolts on the tool holder, extracts the edge features and posture information of the current bolts, and performs feature matching calculations one by one with each template record stored in the bolt template database. The matching algorithm adopts the normalized cross-correlation matching criterion, and a matching score exceeding the preset threshold of 0.85 is judged as a successful match. After a successful match, the system can directly read the precise three-dimensional spatial coordinates and posture parameters of the bolt from the corresponding template record, and generate positioning compensation commands based on the real-time detection results for the posture adjustment and precise alignment of the tool changing robot's gripper.

[0250] A305. The control system inputs bolt positioning parameter information into the tool retrieval command generation module. This module first reads the current robot gripper hardware parameters from the equipment parameter library, including the maximum gripper opening angle range of 0 to 45 degrees, the gripping force output range of 100 N to 800 N, and the attitude adjustment accuracy of 0.1 degrees. Then, based on the spatial position and attitude of the bolt, it selects parameter combinations suitable for the current bolt state from the preset mapping relationship between gripper attitude, gripper gripping force, and tool disengagement risk probability. In this embodiment, the mapping relationship between gripper attitude, gripper gripping force, and tool disengagement risk probability is realized through an offline tool disengagement risk probability calculation model. This model uses the gripper opening angle and gripper gripping force as input variables and the tool disengagement risk probability as the output variable. Its internal calculation logic is determined by regression analysis based on the mechanical response data accumulated in historical tool changing operations.

[0251] Specifically, the execution process of the grid search algorithm is as follows:

[0252] The first step is to discretize the gripper opening angle within its effective range of 0° to 45° into 10 candidate values ​​in 5° increments: 0°, 5°, 10°, 15°, 20°, 25°, 30°, 35°, 40°, and 45°. The second step is to discretize the gripper force within its effective range of 100N to 800N into 8 candidate values ​​in 100N increments: 100N, 200N, 300N, 400N, 500N, 600N, 700N, and 800N. This results in 10 × 8 = 80 combinations of candidate parameters.

[0253] The second step involves, for each candidate parameter combination (θ_grip_i, F_j), where θ_grip_i is the i-th candidate opening angle and F_j is the j-th candidate gripping force, calling a preset tool disengagement risk probability calculation function to calculate the normalized risk probability value P_risk(θ_grip_i, F_j) of the old tool accidentally slipping or deflecting from the gripper under that parameter combination. The value ranges from 0 to 1, with a higher value indicating a higher risk of disengagement. This calculation function is based on the statistical frequency of actual disengagement events recorded in historical tool changing operations under different gripper opening angles and gripping force combinations, and mechanical simulation data. It is obtained through multivariate nonlinear regression fitting, and its expression is: P_risk(θ_grip, F) = 1 / (1+e (-(a×θ_grip+b×F+c×θ_grip×F+d) ));

[0254] Where a, b, c, and d are regression coefficients. In this embodiment, after fitting and calibration, a = -0.12, b = -0.008, c = 0.0001, and d = 2.5. This function adopts a Sigmoid-type nonlinear mapping, which can smoothly describe the coupled influence of the gripper opening angle and gripping force on the risk of tool detachment: when the opening angle is too large or the gripping force is too small, the exponential term tends to be negative, and the risk probability approaches 1; when the opening angle is moderate and the gripping force is sufficient, the exponential term tends to be positive, and the risk probability approaches 0.

[0255] For example, substituting the candidate combination (θ=15°, F=500N) into the above function, we calculate the exponential term = -0.12×15 - 0.008×500 + 0.0001×15×500 + 2.5 = -1.8 - 4.0 + 0.75 + 2.5 = -2.55; P_risk(15, 500) = 1 / (1+e 2.55 )

[0256] ≈1 / (1+12.8)≈0.072, meaning that the probability of tool detachment corresponding to this parameter combination is about 0.072, which is a low risk level.

[0257] The third step involves iterating through all 80 candidate parameter combinations, comparing the P_risk values ​​for each combination, and selecting the parameter combination with the smallest P_risk value as the optimal solution. If multiple parameter combinations have the same smallest P_risk value, the combination with a smaller gripping force and a moderate opening angle is prioritized to balance the energy consumption of the gripper drive and the safety margin of the mechanical limit.

[0258] Based on the above grid search, the optimal parameter combination determined in this embodiment is: gripper opening angle 15°, gripper gripping force 500N, and the corresponding minimum probability of tool detachment risk is 0.072. The control system generates the gripper attitude control value and gripping force control value in the tool retrieval command accordingly.

[0259] The control system selects the parameter combination with the lowest risk probability: a gripper opening angle of 15 degrees, a gripper contact angle perpendicular to the bolt axis, and a gripping force of 500 Newtons. The probability of tool disengagement corresponding to this combination is 0.072. Finally, the optimal parameters are converted into angle control values ​​for each joint of the gripper and torque control values ​​for the drive motor. Combined with the bolt positioning information, a tool retrieval command is generated, which includes attitude adjustment instructions, gripping force application instructions, and action timing instructions.

[0260] A306. The control system sends the tool-retrieving command to the robot gripper drive subsystem. First, the gripper adjusts its posture according to the command, aligning the flexible electric wrenches on the upper and lower sides with the two bolts respectively, and the gripper body adjusts to a position parallel to the tool holder. Then, the flexible electric wrenches activate, tightening the bolts with a torque of 200 Nm and loosening them by rotating clockwise, while the gripper clamps the old tool from both sides of the tool holder with a gripping force of 500 Nm. The control system monitors the loosening angle of each bolt and the gripping force feedback of the gripper in real time. When it detects that all four bolts have been loosened more than 360 degrees and the gripping force of the gripper stabilizes at 500 Nm, the control system will proceed. When the force reaches within 10 Newtons, the gripping is confirmed successful. Finally, the control system instructs the telescopic arm to slowly retract, smoothly extracting the old blade from the blade holder, completing the old blade disassembly operation, and generating a disassembly completion signal at 08:06:00. Throughout the process, if abnormal fluctuations in gripping force or bolt jamming are detected, the system immediately pauses and triggers an alarm, awaiting manual intervention.

[0261] It should be further explained that, in this embodiment, the tool lifting platform and the automatic transport vehicle are controlled according to the operation sequence and motion trajectory to transfer the new and old tools to be loaded, specifically as follows:

[0262] A401. Control the tool lifting platform to move along the vertical track of the shield to the preset tool handover area, and simultaneously control the automatic transport vehicle to move the target number of new tools to be loaded to the preset tool transfer area within a preset time length. When the tool lifting platform reaches the tool handover area, control the tool changing robot to place the clamped old tool on the tool lifting platform, and control the tool lifting platform to carry the old tool down along the vertical track of the shield to the tool transfer area.

[0263] The preset time length is less than or equal to the sum of the first time length, the second time length, and the third time length; the first time length represents the time it takes for the tool lifting platform to move along the shield vertical track to the preset tool handover area; the second time length represents the time it takes for the tool changing robot to place the clamped old tool on the tool lifting platform; the third time length represents the time it takes for the tool lifting platform to carry the old tool down along the shield vertical track to the tool transfer area.

[0264] For example, in this embodiment, after receiving the disassembly completion signal at 08:06:00, the control system synchronously activates the tool transfer subsystem. The control system generates start commands for the tool lifting platform and the automated transport vehicle according to the pre-planned timing in the operation sequence. The control system first sends a lifting command to the tool lifting platform, controlling it to rise vertically from its initial docking position, i.e., node S3 at the bottom of the shield vertical track, with coordinates (10, -15), to the preset tool handover area. The tool handover area is set as point P_lift, located 2 meters directly below the intersection of nodes H3 and V1, with coordinates (10, -2). The lifting platform's rising speed is 0.5 meters per second, the rising distance is 13 meters, and the required rising time is 26 seconds. Therefore, the lifting platform starts rising at 08:06:00, and the expected arrival time is 08:06:26.

[0265] Simultaneously, the control system sends a movement command to the automated transport vehicle, controlling it to move from the initial node B1 on the bottom horizontal track (coordinates (0, -15)) to the preset tool transfer area node B3 (coordinates (10, -15)). The vehicle's horizontal movement speed is 1.0 meter per second, the movement distance is 10 meters, and the required movement time is 10 seconds. Therefore, the vehicle starts moving at 08:06:00 and is expected to arrive at 08:06:10. This preset movement time of 10 seconds is less than the sum of the first, second, and third time lengths. The first time length is the time for the lifting platform to rise (26 seconds), the second time length is the time required for the tool-changing robot to place the old tool (estimated at 5 seconds), and the third time length is the time for the lifting platform to descend (26 seconds). The sum of these three times is 57 seconds, satisfying the condition that the preset time length is less than or equal to the sum of the three.

[0266] During the ascent of the lifting platform, the control system, according to the operation sequence instructions, moves the tool-changing robot carrying the old tool from its current disassembly position to above the tool handover area. The robot starts from the T2 tool holder position (Goal), coordinates (8, 5), and first returns to node V2 (coordinates (10, 5)) at a speed of 0.5 m / s along the tool head fine-tuning track, a distance of 2 meters, taking 4 seconds, arriving at V2 at 08:06:04. Then, it descends along the vertical track V2-V1 at a speed of 0.5 m / s to the intersection of node H3 and V1, coordinates (10, 0), a distance of 5 meters, taking 10 seconds, arriving at H3 at 08:06:14. At this point, the robot is directly above the tool handover area and generates a positioning signal via its position sensor, awaiting the lifting platform's arrival.

[0267] At 08:06:26, the tool lifting platform arrived at the handover area P_lift on time. The control system confirmed the platform's positioning via limit switches and laser rangefinders, generating a platform positioning signal. Subsequently, the vision positioning subsystem was activated to precisely locate the platform's bearing surface, acquiring its three-dimensional spatial coordinates and attitude parameters, and calculating the relative positional deviation between the robot gripper and the platform's bearing surface. Based on this deviation, the robot adjusted the gripper's attitude and smoothly placed the old tool onto the platform between 08:06:26 and 08:06:31, a process that took 5 seconds. Upon completion of placement, the control system generated a signal indicating that the old tool placement was complete.

[0268] A402. Upon arrival at the tool transfer area, control the tool lifting platform to transfer the old tool to the automatic transport vehicle located at the bottom, and grab the new tool to be loaded, and rise along the vertical track of the shield to the tool handover area.

[0269] For example, after receiving the signal that the old tool placement was complete at 08:06:31, the tool lift immediately initiated the descent procedure, descending along the vertical track of the shield from the handover area P_lift to the bottom tool transfer area node S3 at coordinates (10, -15) at a speed of 0.5 meters per second. The descent distance was 13 meters, taking 26 seconds, and it arrived at S3 at 08:06:57. At this time, the automated guided vehicle had already arrived at the transfer area B3 at 08:06:10 and generated a standby signal through the proximity switch.

[0270] Assuming that at 08:06:57, the tool lift reaches S3, the control system confirms the docking position between the lift and the transport vehicle via a proximity switch. The lift activates the transfer mechanism, pushing the old tool from the lift's support surface to the transport vehicle's support area according to pre-calibrated pushing parameters. The pushing process takes 4 seconds, and the old tool transfer is completed at 08:07:01, generating an old tool transfer completion signal. After the transfer is complete, the transport vehicle, carrying the old tool, awaits subsequent dispatch instructions.

[0271] Subsequently, the tool lifting platform grabs the new tool from the transfer vehicle. The control system, based on the preset mapping relationship between the gripper posture, gripping force, and the probability of damage to the first tool, combined with the current position of the new tool on the transfer vehicle, generates a grabbing command with the goal of minimizing the risk probability. The lifting platform grabbing device completes the grabbing of the new tool between 08:07:01 and 08:07:06, taking 5 seconds, and generates a new tool grabbing completion signal. After grabbing, the lifting platform restarts its ascent program, carrying the new tool along the shield's vertical track at a speed of 0.5 meters per second to the tool handover area P_lift, taking 26 seconds to ascend. It arrives at the handover area at 08:07:32 and generates a limit switch signal indicating that the lifting platform has arrived at the handover area, ready for the tool changing robot to grab and install it.

[0272] For example, in this embodiment, the control of the tool-changing robot to grasp the new tool and move it to the tool holder position according to the motion trajectory and operation sequence, and the specific process of installing the new tool includes:

[0273] At 08:07:32, after the tool lift platform, carrying the new tool to be installed, arrives at the tool transfer area P_lift, the control system uses limit switches and a laser rangefinder to dual-verify the platform's positioning accuracy, ensuring that the relative distance between the platform's bearing surface and the tool changer's gripper is less than or equal to 5 mm. A new tool positioning signal is then generated and sent to the tool changer control module. Simultaneously, the tool changer control module receives new tool parameter verification information from the tool transfer subsystem, confirming that the new tool's model and specifications match the tool holder (e.g., T2 tool holder), that the new tool's surface is undamaged, and that the bolt holes are precisely aligned with the tool holder bolts. If the verification passes, the tool changer robot is activated to grasp the tool; if the verification fails, an alarm is triggered and the installation process is paused.

[0274] A502, the tool changing robot remains positioned above node H3 in the tool handover area, at coordinates (10, 0), ready to operate. Upon receiving the gripping command, the robot adjusts the gripper to above the lifting platform's bearing surface via its telescopic arm, activates the vision positioning subsystem, and captures images of the new tool's bolt holes and the gripper's mating area. Using Canny edge detection and template matching algorithms, the robot calculates the new tool's 3D coordinates, posture parameters, and bolt hole position deviation, generating a gripping positioning compensation command. Based on this command, the robot adjusts the gripper's posture to an opening angle of 15 degrees, adapting it to the new tool's gripping surface. It controls the gripper to clamp the new tool with a gripping force of 500 Newtons, monitoring the gripping force stability in real time to ensure fluctuations are within ±10 Newtons. After confirming a secure grip and a risk of release of less than or equal to 0.02, a gripping completion signal is generated, and the tool lifting platform is simultaneously controlled to release the new tool's gripping mechanism, preparing for platform retraction.

[0275] The A503 tool change control module calls the path planning module to generate a dedicated motion trajectory for installing the new tool. This trajectory is the reverse of the old tool removal path, reusing horizontal, vertical, and fine-tuning track resources while avoiding conflicts with other equipment. The trajectory parameters are converted into a proportional-integral-derivative (PID) controller position reference input. Combined with the robot's current joint angles and a movement speed of 0.5 m / s, horizontal and vertical track drive commands are generated. The robot, carrying the new tool, starts from node H3, ascends along the vertical track V1-V2 to node V2 (coordinates (10, 5)) in 10 seconds, and then moves along the tool head fine-tuning track to the tool holder T2 position (coordinates (8, 9)) in 4 seconds. The entire process strictly adheres to the timing constraints in the operation sequence. A magnetic induction sensor monitors the track occupancy status in real time. If a temporary conflict occurs, preset avoidance logic is executed, including delayed start or temporary stop, ensuring arrival at the target tool holder at the planned time.

[0276] A504. After the robot reaches the position of the tool holder, the laser rangefinder sensor confirms again that the relative distance between the robot and the tool holder is less than or equal to 5 mm, generating a positioning signal. The vision positioning subsystem initiates secondary positioning, taking images of the tool holder bolts and the new tool bolt holes, calculating the alignment deviation between the new tool and the tool holder, including horizontal deviation, vertical deviation, and posture deviation, and generating alignment adjustment commands. The robot fine-tunes the gripper posture and telescopic arm position according to the commands, gradually aligning the new tool with the tool holder mounting position, ensuring that the new tool bolt holes are completely aligned with the tool holder bolts, with a deviation of less than or equal to 0.1 mm. After alignment is completed, the telescopic arm is controlled to slowly push the new tool, smoothly embedding it into the tool holder mounting slot until the new tool is in contact with the tool holder positioning surface, generating an alignment completion signal.

[0277] A505. After alignment, the robot activates the flexible electric wrench. Based on preset bolt tightening parameters, including a torque of 200 Nm and a tightening angle greater than or equal to 360 degrees, it tightens the bolts in a symmetrical order, first diagonally and then adjacently. During the tightening process, the torque value of the electric wrench and the bolt rotation angle are monitored in real time. After each bolt is tightened, relevant parameters are recorded and the tightening effect is verified. If abnormalities such as insufficient torque or bolt jamming occur, the operation is immediately paused and an alarm is triggered; if all bolts meet the tightening requirements and the torque deviation is less than or equal to... If the bolt tightening is completed within 5 Nm and the rotation angle is greater than or equal to 360 degrees, a tightening completion signal will be generated.

[0278] A506. After the bolts are tightened, the tool change control module initiates the installation effect verification process. The vision positioning subsystem captures an overall image of the new tool after installation, detecting the fit between the new tool and the tool holder to ensure the fit gap is less than or equal to 0.2 mm. The laser rangefinder monitors the installation height of the new tool to ensure it matches the preset installation height. Simultaneously, the joint status of the tool change robot, the initial vibration amplitude and temperature value of the new tool are collected and compared with the preset standard values ​​to confirm no abnormalities. If the verification passes, the new tool installation is deemed qualified; if the verification fails, the electric wrench is controlled to loosen the bolts, and the alignment and tightening process is re-executed until the verification passes. In this embodiment, the preset standard values ​​are a set of benchmark parameters pre-established and stored in the control system verification module before the tool change operation is performed, through a combination of offline calibration, theoretical calculation, and historical data statistics. These parameters are used to quantitatively compare and determine the installation quality after the new tool is installed. The specific construction process is as follows: For the tool changing robot joint status standard values, during the initial calibration process after the tool changing robot completes factory debugging and on-site deployment, each joint is controlled to return to its initial standby position according to the reset procedure. The absolute value encoder of each joint records the current joint angle value as the mechanical zero-position reference, and this zero-position reference is solidified and stored as the joint status standard value. For the new tool initial vibration amplitude standard value, the average vibration amplitude is calculated by statistically analyzing historical vibration monitoring data of the same model of new tools operating under no-load conditions in the same installation environment. Take the standard deviation σ as the reference. The ±3σ range is considered the normal fluctuation range; vibration exceeding this range is considered abnormal. The temperature standard value is set based on the ambient temperature baseline behind the tool shroud and the empirical value of steady-state temperature rise during normal tool operation. The upper limit standard value is set to the ambient temperature plus 15℃, i.e., T_standard = T_env + 15℃. During actual verification, the control system collects the current joint angle value of the tool-changing robot, the initial vibration amplitude of the new tool, and the tool temperature value, and compares them with the above preset standard values. If the joint angle deviation is ≤0.5°, the vibration amplitude falls within the μ±3σ range, and the temperature value is ≤T_standard, the installation status is considered normal, and the new tool installation is confirmed to be qualified.

[0279] A507. After the new tool is installed successfully, the tool-changing robot's gripper releases the new tool, the opening angle is adjusted to 45 degrees, and the telescopic arm retracts to its initial position. Then, following the motion trajectory, it returns to the preset standby station, such as node H1, and generates a reset completion signal via a sensor. Simultaneously, the tool-changing control module encapsulates the new tool installation completion signal and installation verification parameters, sending them via industrial Ethernet to the path planning module and the tool transport subsystem. This notifies the path planning module to update the track occupancy status, release the track segment occupied during the new tool installation, and notifies the tool transport subsystem to complete the subsequent process of transporting the old tool. Thus, the entire tool-changing process for a single tool, including disassembly, transport, and installation, is completed. If there are multiple tools to be replaced, such as T1 and T3, steps A3 to A5 are repeated according to the scheduling order in the operation sequence until all tools to be replaced are installed.

[0280] Example 3

[0281] Please see Figure 17 Another embodiment of the present invention provides: a dual-rail automatic tool changer control system for a full-face tunneling machine (TBM), comprising:

[0282] The acquisition module is configured to: in response to a tool change request at the coordinate position to be processed, acquire the target tool identifier and tool change urgency level contained in the tool change request;

[0283] Collect real-time status information, including: the current workstation and joint status of the tool changing robot, the real-time occupancy status of the horizontal and vertical tracks, and the position coordinates of M old tools to be replaced behind the tool head;

[0284] The path planning module, based on the tool change urgency level and the real-time status information, generates the operation sequence and motion trajectory of the tool change robot, tool lifting platform and automatic transport vehicle to perform the tool change operation in a coordinated manner through a motion planning algorithm optimized by particle swarm optimization; wherein, the operation sequence includes the action sequence, path segment and avoidance logic of each device;

[0285] The tool changing control module is configured to control the tool changing robot to move according to the motion trajectory through a combination of horizontal and vertical tracks to reach the tool holder where the old tool to be replaced is located, and to perform the old tool gripping and disassembly.

[0286] Once disassembly is complete, the tool-changing robot is controlled to carry the old tool to the preset tool handover area, and the tool transfer subsystem is activated simultaneously. Based on the operation sequence and motion trajectory, the tool lifting platform and the automatic transport transfer vehicle are controlled to transfer the new and old tools to be loaded.

[0287] When the new tool arrives at the tool exchange area, the tool changing robot is controlled to grab the new tool and move to the tool holder position according to the motion trajectory and operation sequence to perform the new tool installation. When the new tool is fully installed, the TBM tool changing robot completes the operation and returns to the L1 area.

[0288] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of the present invention without departing from the spirit and scope of the claims. All of these variations are within the protection scope of the present invention.

Claims

1. A dual-rail automatic tool changer control method for full-face tunneling machines (TBMs), characterized in that, include: In response to a tool change request at the coordinate position to be processed, the target tool identifier and tool change urgency level contained in the tool change request are obtained; Collect real-time status information, including: the current workstation and joint status of the tool changing robot, the real-time occupancy status of the horizontal and vertical tracks, and the position coordinates of M old tools to be replaced behind the tool head; Based on the tool change urgency level and the real-time status information, a motion planning algorithm optimized by particle swarm optimization is used to generate the operation sequence and motion trajectory of the tool change robot, tool lifting platform and automatic transport vehicle to perform the tool change operation in a coordinated manner; wherein, the operation sequence includes the action sequence, path segment and avoidance logic of each device; The tool-changing robot is controlled to move according to the motion trajectory by a combination of horizontal and vertical tracks to reach the tool holder where the old tool to be replaced is located, and to perform the old tool grabbing and disassembly. Once disassembly is complete, the tool-changing robot is controlled to carry the old tool to the preset tool handover area, and the tool transfer subsystem is activated simultaneously. Based on the operation sequence and motion trajectory, the tool lifting platform and the automatic transport transfer vehicle are controlled to transfer the new and old tools to be loaded. When the new blade arrives at the blade exchange area, the tool changing robot is controlled to grab the new blade and move to the blade holder position according to the motion trajectory and operation sequence to perform the new blade installation.

2. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 1, characterized in that, The process of controlling the tool lifting platform and the automated transport vehicle to transfer new and old tools according to the operation sequence and motion trajectory includes: The tool lifting platform is controlled to move along the vertical track of the shield to the preset tool handover area, and the automatic transport vehicle is simultaneously controlled to move the target number of new tools to be loaded to the preset tool transfer area within a preset time length. When the tool lifting platform reaches the tool handover area, the tool changing robot is controlled to place the clamped old tool on the tool lifting platform, and the tool lifting platform is controlled to carry the old tool down along the vertical track of the shield to the tool transfer area. The preset time length is less than or equal to the sum of the first time length, the second time length, and the third time length; the first time length represents the time it takes for the tool lifting platform to move along the shield vertical track to the preset tool handover area; the second time length represents the time it takes for the tool changing robot to place the clamped old tool on the tool lifting platform; the third time length represents the time it takes for the tool lifting platform to carry the old tool down along the shield vertical track to the tool transfer area. Upon arrival at the tool transfer area, the tool lifting platform controls the transfer of old tools to the automated transport vehicle located at the bottom, and grabs the new tools to be loaded, which then rise along the vertical track of the shield to the tool handover area.

3. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 2, characterized in that, The motion planning algorithm optimized by particle swarm optimization generates the operation sequence and motion trajectory for the coordinated execution of the tool changing operation by the tool changing robot, tool lifting platform, and automated transport vehicle, including: Obtain the priority weight and time constraint threshold corresponding to the tool changing urgency level, as well as the current pose and joint angle of the tool changing robot, the occupied range of the horizontal and vertical tracks, and the tool holder coordinates of each old tool to be replaced from the real-time status information. The particle swarm optimization algorithm is used to optimize the tool change urgency level and the real-time status information to obtain an optimal scheduling parameter set, which includes at least the optimized priority weight and time constraint threshold. The multiple tools to be replaced are sorted according to the priority weights to determine the scheduling order of the current task, and the allowable delay window for each device action is set according to the time constraint threshold. Based on the occupied interval and the tool holder coordinates, a spatiotemporal state diagram is constructed, which includes track network nodes, path segment capacity and dynamic obstacles. The spatiotemporal state diagram marks the occupied time slice of each track segment with time as the dimension. Based on the scheduling order and the spatiotemporal state diagram, the operation sequence and motion trajectory are generated using a preset hierarchical path planning strategy.

4. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 3, characterized in that, The method of generating the operation sequence and motion trajectory using a preset hierarchical path planning strategy includes a first-layer path planning stage, a second-layer conflict detection and resolution stage, and a third-layer sequence encapsulation stage; the first-layer path planning stage specifically includes: Obtain the spatiotemporal state diagram and the priority weights corresponding to the tool change urgency level; A path cost function is constructed based on the priority weights, and the path cost function is used to quantify the degree to which time constraints are satisfied in path search; Using the spatiotemporal state diagram as a constraint, a heuristic search algorithm is employed to plan a first conflict-free path segment from the current workstation coordinates to the target tool holder coordinates for the tool-changing robot based on the path cost function. The first conflict-free path segment includes the movement sequence along the horizontal and vertical tracks, as well as the entry and exit times in each track segment. Obtain the key time slices in the first conflict-free path segment; Constrained by the spatiotemporal state diagram and on the premise of not occupying the key time slice, a second path segment is planned for the tool lifting platform from its initial position to the tool handover area, and a third path segment is planned for the automated transport vehicle from its initial position to the tool transfer area. The second path segment and the third path segment are configured to be executed in parallel with the first non-conflicting path segment in the time dimension, so that the tool changing robot, the tool lifting platform, and the automated transport vehicle form an initial non-intersecting path set when sharing the horizontal track and the vertical track.

5. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 4, characterized in that, The second layer of conflict detection and resolution phase specifically includes: Obtain the set of track segment occupancy time slices corresponding to the first conflict-free path segment, the second path segment, and the third path segment, wherein each track segment occupancy time slice includes a track segment identifier, an occupancy start time, and an occupancy end time; Intersection detection is performed on any two time slots in the set of track segment occupancy time slots to identify conflicting time slots that overlap in the time slot occupancy interval under the same track segment identifier, and the first device identifier, second device identifier, conflict duration and conflict location coordinates corresponding to each conflicting time slot are recorded. For each conflict time slice, the priority weights determined by the tool change urgency levels of the first device and the second device are obtained, and the priority weights of the two are compared. If the priority weight of the first device is less than the priority weight of the second device, then the first device is marked as a device to avoid and the second device is marked as a device with priority passage. If the priority weight of the first device is greater than the priority weight of the second device, then the first device is marked as the priority passage device and the second device is marked as the avoidance device; If the priority weight of the first device is equal to the priority weight of the second device, then the device to be avoided and the device to be given priority are determined according to the preset avoidance rules.

6. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 5, characterized in that, The second-layer conflict detection and resolution phase specifically includes: For each conflict time slot, the duration of the conflict is compared with a preset avoidance time threshold. If the duration of the conflict is less than the avoidance time threshold, the start time of the avoidance device is adjusted so that the time when the avoidance device enters the conflict track section is delayed until after the priority passage device leaves the track section, and the adjusted start time is generated. If the duration of the conflict is greater than or equal to the avoidance time threshold, then an avoidance node is inserted for the avoidance device at the track node preceding the conflict location coordinates, and an avoidance path segment containing the waiting time or detour path is generated to replace the conflict part in the original path segment of the avoidance device. The path segment adjusted at the start time or the updated path segment after inserting the avoidance path segment is merged with the original path segment that has not conflicted, to obtain the updated first non-conflicting path segment, second path segment and third path segment that satisfy the mutual exclusion occupancy constraint, and the adjusted start time and the avoidance path segment are encapsulated as the avoidance logic in the operation sequence.

7. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 6, characterized in that, The process of determining the avoidance device and the priority passage device according to the preset avoidance rules includes: Obtain the track segment type identifier corresponding to the conflict time slice, wherein the track segment type identifier includes the main track segment identifier and the branch track segment identifier; If the track segment type identifier is a main track segment identifier, then the first distance traveled by the first device at the start of the conflict time slot and the second distance traveled by the second device at the start of the conflict time slot are obtained, wherein the distance traveled is the cumulative path length along the track from the entry node of the main track segment to the current position of each device; the first distance traveled and the second distance traveled are compared, and the device with the larger distance traveled is marked as the priority passage device, and the other device is marked as the avoidance device; If the track section type is identified as a branch track section, then the first remaining path length from the current position of the first device to the exit node of the branch track section and the second remaining path length from the current position of the second device to the exit node of the branch track section are obtained; the first remaining path length and the second remaining path length are compared, and the device with the smaller remaining path length is marked as the priority passage device, and the other device is marked as the avoidance device.

8. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 7, characterized in that, Before comparing the duration of the conflict with a preset avoidance time threshold, the method further includes: Obtain the attribute parameters of the track segment where the conflict location coordinates are located. The track segment attribute parameters include the track curvature radius, track gradient, and the maximum allowable operating speed of the segment. Obtain the braking response time constant and control command transmission delay time corresponding to the avoidance device; The maximum safe turning speed corresponding to the track section is calculated based on the track curvature radius, and the maximum safe turning speed is compared with the maximum allowable operating speed. The smaller of the two values ​​is taken as the constrained operating speed of the track section. The constrained operating speed is corrected based on the track slope to generate the actual restricted speed for the track section, wherein the actual restricted speed is negatively correlated with the track slope; Based on the actual speed limit and the length of the track segment where the conflict location coordinates are located, calculate the theoretical minimum passage time for the track segment; The theoretical minimum passage time, the braking response time constant, and the control command transmission delay time are summed to generate the avoidance time threshold.

9. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 8, characterized in that, The step of inserting a avoidance node for the avoidance device at the preceding track node at the conflict location coordinates and generating an avoidance path segment containing a waiting time or detour path includes: Obtain the conflict start time and conflict end time corresponding to the conflict time slice, as well as the real-time position coordinates and current running speed of the avoidance device at the current planning time. Based on the real-time position coordinates and the current operating speed, calculate the estimated arrival time of the avoidance device as it moves along its original path segment to the previous track node; Compare the estimated arrival time with the conflict termination time: If the expected arrival time is earlier than the conflict end time, the difference between the conflict end time and the expected arrival time is determined as the waiting time, and a waiting-type avoidance path segment containing the waiting time at the previous track node is generated. At the same time, the entry time of all path segments in the original path segment of the avoidance device after the previous track node will be delayed by the waiting time. If the estimated arrival time is later than or equal to the conflict end time, it is determined that there is no need to wait, and the original path segment of the avoidance device remains unchanged; Obtain a preset maximum waiting time threshold, which is determined based on the time constraint threshold corresponding to the tool change urgency level.

10. The dual-rail automatic tool changer control method for a full-face tunneling machine (TBM) as described in claim 9, characterized in that, The step of inserting a avoidance node for the avoidance device at the preceding track node at the conflict location coordinates and generating an avoidance path segment containing a waiting time or detour path also includes: Compare the waiting time with the maximum waiting time threshold: If the waiting time is less than or equal to the maximum waiting time threshold, then the waiting-type avoidance path segment is taken as the avoidance path segment. If the waiting time exceeds the maximum waiting time threshold, then the path search is performed again with the previous track node as the starting point and the first track node after the conflict part in the original path segment of the avoidance device as the ending point, and the spatiotemporal state diagram as the constraint. A detour path segment that does not pass through the track segment where the conflict position coordinates are located is generated. The entry time of each node in the detour path segment is calculated according to the length of the detour path segment and the actual speed limit. The detour path segment is then spliced ​​in time with the part before the previous track node and the part after the first track node in the original path segment of the avoidance device to obtain a detour-type avoidance path segment containing the detour trajectory. The waiting-type avoidance path segment or the detour-type avoidance path segment is used as the updated avoidance device path segment.

11. A dual-track automatic tool changer control system for a full-face tunneling machine (TBM), implemented based on any one of claims 1-10, characterized in that... include: The acquisition module is configured to: in response to a tool change request at the coordinate position to be processed, acquire the target tool identifier and tool change urgency level contained in the tool change request; Collect real-time status information, including: the current workstation and joint status of the tool changing robot, the real-time occupancy status of the horizontal and vertical tracks, and the position coordinates of M old tools to be replaced behind the tool head; The path planning module, based on the tool change urgency level and the real-time status information, generates the operation sequence and motion trajectory of the tool change robot, tool lifting platform and automatic transport vehicle to perform the tool change operation in a coordinated manner through a motion planning algorithm optimized by particle swarm optimization; wherein, the operation sequence includes the action sequence, path segment and avoidance logic of each device; The tool changing control module is configured to control the tool changing robot to move according to the motion trajectory through a combination of horizontal and vertical tracks to reach the tool holder where the old tool to be replaced is located, and to perform the old tool gripping and disassembly. Once disassembly is complete, the tool-changing robot is controlled to carry the old tool to the preset tool handover area, and the tool transfer subsystem is activated simultaneously. Based on the operation sequence and motion trajectory, the tool lifting platform and the automatic transport transfer vehicle are controlled to transfer the new and old tools to be loaded. When the new blade arrives at the blade exchange area, the tool changing robot is controlled to grab the new blade and move to the blade holder position according to the motion trajectory and operation sequence to perform the new blade installation.

Images