Intelligent control method for tool changing robot of shield tunneling machine

Abstract

The invention discloses an intelligent control method for a cutter changing robot of a shield tunneling machine, belongs to the technical field of automatic control, and aims to solve the problems of low hob changing efficiency, inaccurate decision, large positioning error and the like of the shield tunneling machine under a complex stratum. Hob state parameters are collected through multiple sensors, a three-level decision threshold value is constructed, dynamic adjustment is conducted through fuzzy logic, and intelligent hierarchical control over abrasion pre-judgment, early warning and emergency response is achieved. A multi-dimensional weighting model is adopted to calculate the priority of the abraded hob, a cutterhead path is optimized in combination with a dual-objective function, and an efficient execution sequence is generated. In the positioning link, through reverse path planning, PID compensation inertial slippage and cooperation of multispectral imaging and feature matching technologies, hob mark points are accurately extracted, pose deviation is corrected, and it is ensured that the hob mark points enter a robot working space. A double-station conversion table is integrated, synchronous exchange of a new cutter and an old cutter is achieved through hydraulic and servo control, force-position hybrid control is adopted for an end effector, and the disassembly and assembly stability is improved by combining with a torque wrench for staged operation.

Description

Technical Field

[0001] The present invention relates to the field of automation control technology, and more particularly to an intelligent control method for a tool-changing robot of a shield machine. Background Art

[0002] Shield machines are core equipment in tunnel construction, and their cutters play a crucial role in breaking rock during excavation in hard rock formations. The cutters are prone to wear and tear over time, requiring timely replacement to ensure tunneling efficiency and construction safety. In the existing technology, roller cutter replacement mainly relies on manual operation or semi-automated devices, which have significant technical bottlenecks. Manual cutter replacement is limited by the confined space of the shield machine and the complex stratum environment, facing safety risks such as high pressure hypoxia and rock burst collapse. In addition, manual wear detection relies on empirical judgment, which is inefficient and lacks accuracy. Although some automated cutter changing systems have introduced sensor monitoring, they generally have the problem of single-state monitoring. Wear is judged by only a few parameters such as pressure or torque, without forming a multi-dimensional comprehensive assessment, resulting in delayed or misjudgment of replacement decisions. In the positioning and alignment process, traditional systems rely on preset trajectories or simple visual recognition, which makes it difficult to dynamically compensate for shield machine posture changes and equipment errors. The cutter change actuator often deviates from the target roller cutter position, resulting in disassembly and installation failures. These problems make it difficult for the existing technology to meet the engineering requirements of large-diameter shield machines for long-distance excavation in terms of safety, accuracy, and efficiency. There are problems such as inaccurate state monitoring, unintelligent decision-making, large positioning and alignment errors, and insufficient coordination of the cutter change process. Therefore, to overcome these limitations, the present invention proposes an intelligent control method for a shield machine cutter changing robot. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide an intelligent control method for a shield machine cutter changing robot, which solves the technical problems in the existing technology such as low efficiency and poor safety of manual replacement of shield machine cutters, inaccurate status monitoring, unintelligent decision-making, large positioning and alignment errors, and insufficient coordination of cutter changing processes in the automatic cutter changing system. The method realizes real-time monitoring of cutter wear status, decision-making on replacement needs, precise spatial positioning of the cutterhead and collaborative control of the robot, and efficient storage and transmission of cutters, thereby improving the automation level, reliability and construction efficiency of shield machine cutter changing operations.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] An intelligent control method for a shield machine tool changing robot, comprising:

[0006] Collect the shield machine cutter operating status parameters, divide the parameter ranges based on multi-dimensional decision thresholds, trigger the graded warning mechanism, make cutter change decisions based on cutter wear, and build a cutter change sequence;

[0007] The tool change control operation is triggered according to the number of hobs in the tool change sequence:

[0008] Identify the hob type in the tool change sequence, calculate the comprehensive priority index for hob replacement due to wear using a multi-dimensional weighted evaluation model, and construct a wear execution sequence based on the principle of minimizing the cutterhead rotation path; set the comprehensive priority for forced hob replacement, and construct an emergency execution sequence using a dual-objective function. Combined with the available time window length and cutterhead rotation control, generate a tool change instruction set for the tool change sequence;

[0009] Based on the tool change instruction set, the cutterhead rough positioning is completed through reverse rotation path optimization and dynamic planning of the cutterhead motion trajectory. Inertial slip compensation is used to position the target replacement hob in the tool change robot workspace, triggering the tool change operation:

[0010] Extract the target hob replacement marker points, construct a multi-level feature image through multi-scale Gaussian downsampling, obtain the marker point coordinates of the target hob replacement enhanced image to identify the pose deviation, and perform pose correction through a two-level fine-tuning mechanism;

[0011] The collaborative operation of the integrated double-station ring conversion table controls the tool changing robot to perform tool changing operations on the target replacement hob through the coordinated hydraulic lifting station and force-position hybrid control strategy.

[0012] Specifically, the steps for constructing a wear execution sequence include:

[0013] When the worn hob is identified and the tool change sequence is entered, a multi-dimensional weighted evaluation model is used to update the execution order of the tool change sequence, namely:

[0014] Based on the cutter wear of the tool change sequence, a basic priority is established. A regional threshold is configured to divide the cutterhead where each cutter in the tool change sequence is located into local areas. By calculating the weighted average of the cutter wear in the local area, a failure cascading risk index for the cutters in the tool change sequence is generated.

[0015] Linearly superimpose the hob wear and fault chain risk index to output a comprehensive priority index. Use the bubble sort algorithm to dynamically rearrange the tool change sequence and update the execution order of the tool change sequence.

[0016] The cutterhead stop time window and the required buffer time of the cutterhead are obtained, and the available time window is divided. When the available time window is longer than the single tool change benchmark time, the multi-batch replacement mode is started, and the wear execution sequence of the available time window is constructed according to the comprehensive priority index and the principle of minimum cutterhead rotation path.

[0017] Specifically, the specific steps of constructing the wear execution sequence of the available time window include:

[0018] Calculate the number of tool changes that can be accommodated in each available time window based on the single tool change benchmark time and the length of the available time window;

[0019] Configure priority thresholds, including upper and lower priority thresholds, and divide the comprehensive priority index into three execution levels according to the priority thresholds;

[0020] The positions of worn replacement hobs are converted into polar coordinates, and the rotation angle difference between any two worn replacement hobs is calculated. A dual objective function is constructed, including maximizing the comprehensive priority index and minimizing the rotation angle. The dual objective function is balanced by weighted coefficients, and a wear execution sequence is generated for each available time window based on a greedy strategy.

[0021] Specifically, the specific steps of generating a tool change instruction set of a tool change sequence include:

[0022] Obtaining the execution sequence, including the wear execution sequence and the emergency execution sequence; and replacing the hob according to the target of the execution sequence obtained in the execution order;

[0023] When the cutter head stop time window is entered, the target of the execution sequence of the reel cutter replacement is obtained in sequence according to the execution sequence of the available time window of the cutter head stop time window, and the operation reference angle of the tool changing robot is defined;

[0024] According to the execution order of the target hob replacement in the execution sequence, the angle difference between the target hob replacement and the reference angle of the tool changing robot is calculated in sequence;

[0025] The hob comprehensive priority index in the execution sequence of the available time window, the angle difference between the target hob replacement and the operating reference angle of the tool changing robot, and the updated execution order are encapsulated as a tool changing instruction set.

[0026] Specifically, the specific steps of controlling the tool changing robot to perform a tool changing operation on the target replacement hob include:

[0027] During the cutterhead rotation control phase, the target angle for hob replacement is captured in real time, and the initial angle difference is calculated based on the reference angle of the tool change robot. Based on this initial angle difference, it is determined whether to switch to the reverse rotation path.

[0028] The cutterhead motion trajectory is dynamically planned using an S-shaped acceleration and deceleration curve, and the angular velocity and acceleration thresholds are adaptively adjusted based on real-time inertia parameters to achieve rapid and rough positioning of the cutterhead.

[0029] Configure a buffer angle. When the angle difference between the target hob to be replaced and the reference angle of the tool changer robot is less than the buffer angle, the incremental PID control algorithm is activated to compensate for inertial slip and position the target hob to the tool changer robot's workspace.

[0030] Based on multispectral imaging fusion technology, an enhanced image of the target hob replacement is collected and generated. Through multi-scale Gaussian downsampling, a multi-level feature image is constructed. The marker points of the enhanced image of the target hob replacement are located to calculate the position deviation of the target hob replacement.

[0031] Configure the conflict threshold to determine whether to trigger the two-level fine-tuning mechanism for posture correction;

[0032] The end effector of the tool changing robot loosens the target replacement hob and installs the hob to be replaced in stages based on the coordinates of the marking point of the target replacement hob through a force-position hybrid control strategy, so as to collaboratively complete the recovery of the target replacement hob and the assembly of the hob to be replaced through a double-station annular conversion table.

[0033] Specifically, the specific steps of locating the target replacement mark point of the hob enhanced image include:

[0034] After the target replacement hob enters the workspace, multispectral imaging is started to collect infrared and visible light images of the target replacement hob;

[0035] A bilateral filter is used to reduce the noise of the infrared image, and contrast adaptive histogram equalization is performed on the visible light image. The target replacement hob enhanced image is generated by weighted fusion of the infrared image and the visible light image.

[0036] According to the preset marking point of the hob, the theoretical position of the hob enhanced image is replaced at the target, and a dynamic search window is established with the theoretical coordinates as the center;

[0037] The image within the dynamic search window is subjected to multi-scale Gaussian downsampling to construct a multi-level feature image, including a top-level low-resolution image for fast coarse positioning, a middle-level medium-resolution image for marker point template matching optimization, and a bottom-level high-resolution image for feature descriptor matching to locate the marker points of the target replacement hob enhanced image.

[0038] Specifically, the specific steps of determining whether to trigger the two-level fine-tuning mechanism and performing posture correction include:

[0039] According to the theoretical position of the preset mark point in the enhanced image of the target hob replacement, the pose deviation of the coordinates of the target hob replacement mark point is calculated;

[0040] Configure the conflict threshold. If the posture deviation exceeds the conflict threshold, a two-level fine-tuning mechanism is triggered to correct the posture, including:

[0041] In the primary fine-tuning stage, based on the position data of the end effector of the tool change actuator, a compensation path is planned through a path planning algorithm, and the end effector is driven to perform posture correction and compensation. During the correction process, the posture deviation is monitored synchronously.

[0042] Count the number of pose corrections and compensations, and configure a correction threshold. If the number of pose corrections and compensations is greater than the correction threshold, and the pose deviation is greater than the conflict threshold, the secondary fine-tuning process is triggered and secondary fine-tuning is performed; otherwise, the pose correction and compensation is terminated and the cutterhead is locked.

[0043] In the secondary fine-tuning stage, an incremental PID algorithm is used to generate a cutterhead angle correction command, drive the cutterhead to rotate, perform angle correction, and monitor the cutterhead's circumferential displacement in real time. After each angle correction, the image is recaptured to calculate the position deviation of the target hob mark point coordinates. If the position deviation is less than or equal to the conflict threshold, the position correction is terminated and the cutterhead is locked.

[0044] Otherwise, the angle correction times are counted. If the angle correction times are greater than the correction threshold, a posture warning is issued.

[0045] Specifically, the specific steps of collaboratively completing the target replacement hob recovery and the hob assembly to be replaced through the double-station annular conversion table include:

[0046] After the cutterhead is locked, the tool changer robot starts the disassembly and assembly process. Based on the coordinates of the target hob replacement mark, the tool changer robot's end effector uses a force-position hybrid control strategy to drive the torque wrench on the end effector to align with the target hob replacement mark. The target hob is loosened in stages according to the preset torque curve.

[0047] The tool-changing robot's end effector grabs the target replacement hob, and the hydraulic lifting mechanism of the double-station annular conversion table drives the recovery side station to rise vertically, controlling the tool-changing robot to transfer the target replacement hob to the recovery station.

[0048] At the same time, the hob storage bin pushes the hob to be replaced to the supply side station of the double-station annular conversion table through a chain conveyor;

[0049] When the double-station annular conversion table receives the target replacement hob and the hob to be replaced, the conversion operation is triggered to control the servo motor of the double-station annular conversion table to drive the rotary mechanism to rotate and change the position of the target replacement hob and the hob to be replaced;

[0050] The supply-side station of the double-station annular conversion table is raised to the operating height of the tool changing robot. The end adsorption fixture grabs the hob to be replaced and retreats along the mirror path. The force-position hybrid control strategy is used to control the hob to be replaced to move to the installation station for assembly.

[0051] Specifically, the steps for constructing a tool change sequence include:

[0052] The initial multi-dimensional decision thresholds for the hob operating state parameters are set based on the physical properties of the hob. The multi-dimensional decision thresholds include a normal operating range, a warning range, and a forced replacement range.

[0053] Combining the construction stratum type, shield machine propulsion speed and cutterhead speed, a fuzzy logic algorithm is used to update the multi-dimensional decision threshold correction coefficient in real time, dynamically adjust the multi-dimensional decision threshold, and update the decision threshold range;

[0054] Compare the received hob operating status parameters with the multi-dimensional decision thresholds dimension by dimension, obtain the decision threshold interval, and trigger the graded warning mechanism:

[0055] If the hob operation state parameter is within the normal operation range, a trend prediction is performed on the hob operation state parameter value;

[0056] If the hob operating state parameter is in the warning range, it will be marked as an abnormal operating state parameter. The abnormal monitoring process will be started for the abnormal hob with abnormal operating state parameters. The number of abnormal operating state parameters and the abnormal duration will be counted, the wear state will be quantified, and the abnormal hob wear degree will be obtained.

[0057] Configure the tool change threshold. If the wear of the abnormal hob is greater than the tool change threshold, it will be included in the tool change sequence; otherwise, the wear calculation result will be continuously monitored and updated.

[0058] If the hob operating status parameter is in the forced replacement range, the hob is included in the tool change sequence.

[0059] Beneficial effects of the present invention:

[0060] The present invention uses multi-dimensional decision thresholds and fuzzy logic algorithms to dynamically divide the hob operating status intervals and adjust the tool change strategy in real time, accurately distinguishing between wear replacement and forced replacement needs, and ensuring the scientificity and timeliness of tool change decisions; with the help of a multi-dimensional weighted evaluation model and a dual-objective function to construct an execution sequence, the tool change priority and the cutter disc rotation path are rationally planned, effectively avoiding inefficient operations caused by priority confusion or path redundancy. Through reverse rotation path optimization, S-shaped acceleration and deceleration curve planning, and incremental PID control, the cutter disc coarse positioning and inertial slip compensation are achieved, significantly improving the stability and accuracy of the positioning process, and reducing rotation energy consumption and time loss. Multi-spectral imaging fusion technology and multi-scale Gaussian downsampling methods are used to accurately extract hob mark points in complex environments such as dust, and a two-stage fine-tuning mechanism is combined to correct posture deviations, solving the problem of insufficient visual positioning and operation accuracy under harsh working conditions. The integrated dual-station annular conversion table and force-position hybrid control strategy enable the coordinated operation of old cutter recovery and new cutter assembly. Through phased operation using hydraulic lifting stations and intelligent torque wrenches, the continuity and reliability of the cutter changing process are guaranteed, and the overall automation level of the cutter changing operation is improved. This effectively responds to construction challenges in complex formations such as hard rock and water-rich formations, and provides stable support for the continuous and efficient excavation of the shield machine. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1This is a structural schematic diagram of an intelligent control method for a tool-changing robot of a shield machine according to the present invention;

[0062] Figure 2 A flowchart of the specific steps for constructing a tool change sequence for the present invention;

[0063] Figure 3 A flow chart for generating a tool change instruction set for replacing a hob cutter in a tool change sequence for the present invention;

[0064] Figure 4 This is a flow chart of the present invention controlling the tool-changing robot to perform a tool-changing operation on a target replacement hob;

[0065] Figure 5 This is a flowchart of triggering the two-stage fine-tuning mechanism of the present invention to perform posture correction. DETAILED DESCRIPTION

[0066] Example 1

[0067] See also Figure 1 ,This embodiment introduces an intelligent control method for a shield machine tool changing robot, including:

[0068] Step S1: The shield machine cutter operating status parameters are collected in real time through multi-sensor fusion, including cutter wear, cutter speed, cutter temperature, cutter vibration frequency, cutter cutting torque and cutter axial load.

[0069] In this embodiment, a combination of contact and non-contact measurement is employed. Contact measurement uses a triaxial accelerometer mounted on the cutter shaft bearing seat to capture tool vibration frequency. A torque sensor integrated into the cutterhead drive spindle captures tool cutting torque. A pressure sensor mounted at the connection between the hob support arm and the shield captures tool axial load. Tool speed is converted from pulse frequency using a magnetic encoder at the cutter shaft end and a Hall effect speed sensor near the cutterhead. Tool temperature is monitored using a miniature wireless temperature sensor in the hob bearing cavity. Non-contact measurement is dynamically triggered based on the contact measurement results. By configuring fluctuation anomaly thresholds for the contact measurement parameters, a preliminary anomaly assessment is performed. Non-contact measurement is triggered if any contact measurement parameter exceeds the fluctuation anomaly threshold. Non-contact measurement is also performed using an industrial vision camera mounted on the inner wall of the cutterhead compartment to capture images of the hob. The camera is equipped with an automatic purge air circuit to mitigate the effects of dust on image quality within the cutterhead compartment. An image recognition algorithm identifies wear features in the hob cutter image and combines this with distance compensation data from a laser rangefinder to calculate tool wear. Specifically, based on the cutterhead position information acquired by the shield machine control system, an industrial camera array is triggered to capture images when the hob cutter enters a capture area pre-calibrated using a 3D model. Camera synchronization control parameters are configured to ensure image capture occurs while the cutterhead is relatively stationary, minimizing motion blur. After image acquisition, an edge detection algorithm is used to extract the hob cutter's contour features, identify the cutter ring's circumferential curve, and locate key points along the hob's edge. Furthermore, structured light stripes emitted by a laser rangefinder are projected onto the hob cutter surface. Triangulation is used to obtain the 3D coordinates of each feature point, allowing the radial wear depth of the cutter blade and the change in cutter ring thickness to be calculated, ultimately yielding quantitative results of tool wear.

[0070] Step S2: Receive the hob operating status parameters and divide them into decision threshold intervals based on multi-dimensional decision thresholds. When the hob operating status parameters exceed the corresponding safety threshold, a hierarchical warning mechanism is automatically triggered to make a tool change decision. A tool change sequence is constructed and, based on the number of hobs in the sequence, tool change control operations are triggered. A tool change instruction set for the sequence is generated, while also supporting manual intervention decisions.

[0071] In this embodiment, the operating status parameters of the hob are received in real time through the industrial bus, including tool wear, tool speed, tool temperature, tool vibration frequency, tool cutting torque and tool axial load. The multi-dimensional decision threshold is constructed based on the fatigue characteristics of the tool material, the distribution law of the stratum stress and the construction safety specifications, and includes three levels of decision threshold intervals: normal operation interval, warning interval and forced replacement interval. When any hob operating status parameter exceeds the warning threshold, the hierarchical warning mechanism is automatically triggered; in the normal operating interval, the wear trend reminder is pushed through the human-computer interaction interface; in the warning interval, a pre-replacement plan is generated, the tool change robot path pre-planning is started, and a warning report is sent; in the forced replacement interval, the shield machine advance is immediately suspended and a tool change instruction set is generated. During the tool change instruction set generation process, an entropy weight method is used to calculate a comprehensive priority based on tool wear and cascading failure risk. The angular difference between the target hob and the tool change robot is calculated in real time using the homogeneous transformation matrix between the cutterhead coordinate system and the tool change robot's base coordinate system. The target hob is the hob identified as performing a tool change according to the tool change instruction set. The operation sequence planning combines the cutterhead rotation period and the tool change robot's single-cycle operation time, using a shortest path first algorithm to avoid path conflicts while also reserving time redundancy to deal with emergencies. A manual intervention interface supports priority adjustment and delayed command execution from the remote control console.

[0072] See also Figure 2 , preferably, the specific steps of constructing the tool change sequence include:

[0073] Based on the physical properties of the hob, including the material fatigue curve and structural strength limit, the initial multidimensional decision threshold of the hob operating status parameters is set; the multidimensional decision threshold includes three-level decision threshold intervals, namely: normal operating interval, warning interval and forced replacement interval, which are used to quantify the operating boundaries of the hob operating status parameters under different working conditions; when the operating status parameter is in the normal operating interval, it means that the operating status parameter value is within the design safety range and the tool performance is stable; when the operating status parameter is in the warning interval, it means that the operating status parameter value is close to the design limit and the pre-maintenance process needs to be initiated; when the operating status parameter is in the forced replacement interval, it means that the operating status parameter value exceeds the safety boundary and there is a risk of failure.

[0074] Real-time geological data is acquired through stratum sensing sensors to determine the type of construction stratum. Based on the construction stratum type and the shield machine's propulsion speed and cutterhead speed, a fuzzy rule database is designed based on historical construction data. A fuzzy logic algorithm is used to update the multi-dimensional decision threshold correction coefficient in real time, dynamically adjust the multi-dimensional decision threshold, and update the decision threshold range.

[0075] The received hob operating status parameters are compared dimension by dimension with the multi-dimensional decision thresholds updated in real time to obtain the decision threshold interval where the hob operating status parameters are located. The hierarchical warning mechanism is triggered according to the decision threshold interval, namely:

[0076] If the hob operating status parameters are within the normal operating range, the trend of the hob operating status parameter values ​​is predicted by combining historical operating status parameter data with a time series analysis method to assess the risk of operating status parameter drift;

[0077] If the hob operating state parameter is in the warning interval, it will be marked as an abnormal operating state parameter, and the abnormal monitoring process will be started for the abnormal hob with abnormal operating state parameters. By counting the number of abnormal operating state parameters and the duration of the abnormality, the abnormal duration refers to the length of time that one or more operating state parameters of the hob remain in the abnormal state after entering the warning interval and triggering the abnormal mark. Combined with the number of abnormal operating state parameters and the duration of the abnormality of the abnormal hob, the operating state parameter values ​​of the abnormal hob are evaluated as a whole, the wear state of the abnormal hob is quantified, and the wear degree of the abnormal hob is obtained; for example, when calculating the wear degree of the abnormal hob, the number of abnormal operating state parameters that enter the warning interval is first counted and differentiated weights are assigned according to the parameter type, and then the time decay or accumulation coefficient is set according to the duration of the abnormality. The number of abnormal parameters, duration and degree of parameter abnormality are coupled and calculated by weighted summation or fuzzy comprehensive evaluation, and finally the wear value in the range of 0 to 1 is output to quantify the tool wear state.

[0078] Configure the tool change threshold. If the wear of the abnormal hob exceeds the tool change threshold, it is determined that operation is unsustainable, marked as worn and replaced, and included in the tool change sequence. If it does not exceed the threshold, abnormal monitoring will continue and the wear calculation results will be updated in real time until the tool change conditions are met or the operating status parameters return to normal.

[0079] If the hob operation status parameter is in the forced replacement range, it will be marked as a forced hob replacement and included in the tool change sequence.

[0080] Preferably, the specific steps of generating a tool change instruction set of a tool change sequence include:

[0081] Operations such as starting and stopping the shield machine cutterhead and initializing the robot have a fixed time consumption. If a tool change is triggered when a single tool is worn, frequent starts and stops will result in an excessively high proportion of fixed time consumption, reducing overall construction efficiency. Configure the tool change control threshold. When the number of cutters in the tool change sequence exceeds the tool change control threshold, the tool change control is triggered. Based on the cutter type in the tool change sequence, a tool change instruction set is generated, namely:

[0082] See also Figure 3,When it is recognized that the hob is worn and needs to be replaced and enter the tool change sequence, a multi-dimensional weighted evaluation model is used to update the execution order of the tool change sequence, namely:

[0083] A basic priority is established based on the wear of the hob in the tool change sequence, and the fault cascading risk index is evaluated by analyzing the health status of adjacent tools in the cutterhead partition where the hob is located.

[0084] Specifically, by configuring regional thresholds, the cutterhead surface is divided into local areas where each disc is located according to the radial spacing and circumferential angle. The weighted average value of the disc wear in the local area where the disc is located during the cutter change sequence is calculated in real time. This is normalized based on the formation rock hardness parameters to generate a fault chain risk index in the range of 0 to 1.

[0085] By linearly superimposing the hob wear degree and the fault chain risk index, the comprehensive priority index of the worn hob replacement is output. The bubble sort algorithm is used to dynamically rearrange the tool change sequence and update the execution order of the tool change sequence.

[0086] During shield construction, the cutterhead cannot be changed without stopping. The cutterhead operation cycle parameters are obtained in real time from the shield machine main control system to obtain the cutterhead stop time window. Due to the inertia of the cutterhead, there is an unstable buffer stage during the stop process. Direct tool change is likely to cause positioning deviation. By establishing a correlation between the cutterhead angular displacement and time, the buffer time required for the cutterhead is obtained, and the available time window of the stop time window is divided. The cutterhead is frequently started and stopped, which is time-consuming and wears out the equipment. When the available time window is longer than the single tool change benchmark time, the multi-batch replacement mode is started. The degree of tool wear is different. Considering the relative position of the cutterhead and the tool changing robot, according to the comprehensive priority index and the principle of minimum cutterhead rotation path, the wear execution sequence of each available time window is constructed in turn to make the tool change time as short as possible, that is:

[0087] Calculate the number of tool changes that can be accommodated in each available time window based on the single tool change benchmark time and the length of the available time window;

[0088] Configure priority thresholds, including upper and lower priority thresholds, and divide the comprehensive priority index into three execution levels according to the priority thresholds;

[0089] The positions of worn-out replacement hobs are converted into polar coordinates, the rotation angle difference between any two worn-out replacement hobs is calculated, and a dual-objective function is constructed, including maximizing the comprehensive priority index and minimizing the rotation angle. The dual-objective function is balanced by a weighted coefficient. Based on a greedy strategy, the next worn-out replacement hob closest to the current one and with the highest priority is selected in turn to generate a wear execution sequence for each available time window.

[0090] When the cutter head stops in the time window, the wear execution sequence of the wear execution sequence is obtained in sequence according to the available time window, and the tool change robot operation reference angle is defined, that is, the cutter head angle when the hob rotates to the front of the robot;

[0091] According to the execution order of the wear execution sequence for replacing the hob, a bidirectional optimization algorithm is used to select the rotation path. The angle difference between the target replacement hob to be replaced and the operating reference angle of the tool changing robot is calculated in sequence to ensure that the cutter head rotation angle does not exceed 180°, thereby reducing rotation energy consumption and time loss.

[0092] The hob comprehensive priority index, the angle difference between the target hob replacement and the tool-changing robot's operating reference angle, and the updated execution order in the wear execution sequence of each available time window are encapsulated into a tool-changing instruction set that complies with industrial control standards.

[0093] When it is recognized that the forced replacement of the cutter disc is included in the cutter change sequence, the comprehensive priority of the forced replacement of the cutter disc is set to the highest level, and an emergency stop is performed. An emergency stop signal is sent to the shield machine control system through the industrial bus, suspending the cutter disc rotation and keeping the current power system on standby to retain the cutter disc fine-tuning capability;

[0094] Obtain the current available time window for emergency shutdown. If it is greater than the single tool change benchmark time, calculate the number of tool changes that can be accommodated in the available time window for emergency shutdown. Based on the angle difference between the hob in the tool change sequence and the forced hob replacement, and the hob comprehensive priority index, construct an emergency execution sequence including forced hob replacement through a dual-objective function:

[0095] When the available time window for emergency stop of the cutter disc is entered, the angle difference between the target hob replacement of the emergency execution sequence and the operating reference angle of the tool changing robot is calculated in sequence according to the emergency execution sequence to generate the tool changing instruction set of the available time window for emergency stop.

[0096] Step S3: Based on the tool change instruction set in the available time window, the target replacement hob is determined, and the tool change robot is controlled to perform the tool change operation on the target replacement hob. The angle difference between the target replacement hob and the operating reference angle of the tool change robot is obtained in real time. The cutter head is controlled to rotate so that the target replacement hob is in the working space of the tool change robot. When the target replacement hob is in the working space of the tool change robot, the end effector of the tool change robot is controlled to complete the hob removal and installation operations, and a new hob is provided through the double-station annular tool conversion table, and the old hob is recovered.

[0097] In this embodiment, upon receiving a command, an absolute encoder is used to obtain the current cutterhead angle in real time. The minimum rotation path to the target reference angle is calculated, and the cutterhead is rotated. PID control is initiated to compensate for inertial slip when the cutterhead approaches the target reference angle. The tool-changing robot then grabs the new hob using a dual-station circular transfer table. The end effector removes and installs the new cutter based on coordinates calculated using a homogeneous transformation matrix. The entire process triggers triple safety mechanisms, including mechanical locking and force sensor calibration.

[0098] See also Figure 4 Preferably, the specific steps of controlling the tool-changing robot to perform a tool-changing operation on the target replacement hob include:

[0099] During the cutter head rotation control phase, the absolute encoder is used to capture the current angle of the target hob in real time, and the initial angle difference is calculated based on the reference angle of the tool change robot. The initial angle difference is used to determine whether to switch to the reverse rotation path, that is:

[0100] If the initial angle difference exceeds 180°, the reverse rotation path will be automatically switched to optimize the actual rotation angle to 360° minus the initial difference, thus minimizing the cutter head rotation stroke.

[0101] The cutterhead motion trajectory is dynamically planned using an S-shaped acceleration and deceleration curve, and the angular velocity and acceleration thresholds are adaptively adjusted based on real-time inertia parameters to ensure rapid and coarse positioning of the cutterhead. Real-time inertia parameters are calculated by integrating the torque sensor and angular accelerometer.

[0102] Configure the buffer angle. When the angle difference between the target hob replacement and the operating reference angle of the tool changing robot is less than the buffer angle, start the incremental PID control algorithm to control the cutter disc rotation. Use the laser displacement sensor to monitor the inertial slip amount during the cutter disc braking process in real time, and compensate with the servo motor's reverse micro-pulse so that the target hob replacement is in the workspace of the tool changing robot when the cutter disc enters the available time window.

[0103] When the target hob is replaced and enters the workspace, the tool change operation is triggered and multispectral imaging is started. Near-infrared light and visible light are used for illumination. The near-infrared light highlights the reflective characteristics of the oxide layer of the laser-etched mark, and the visible light captures the surface texture details of the tool. Infrared and visible light images of the target hob are collected.

[0104] A bilateral filter is used to reduce the noise of the infrared image and preserve the edge sharpness of the marker points; contrast adaptive histogram equalization is performed on the visible light image to enhance the visibility of features in low-light areas. The infrared image and the visible light image are weightedly fused to generate an enhanced image of the target replacement hob;

[0105] The theoretical position of the hob enhanced image is replaced on the target according to the preset marking point of the hob, and a dynamic search window is established with the theoretical coordinates as the center. The dynamic search window dynamically adjusts the window range according to the historical positioning deviation and environmental disturbance;

[0106] The image in the dynamic search window is subjected to multi-scale Gaussian downsampling to construct a multi-level feature image, including a top-level low-resolution image for fast coarse positioning, a middle-level medium-resolution image for marker point template matching optimization, and a bottom-level high-resolution image for feature descriptor matching to locate the marker points of the target replacement hob enhanced image.

[0107] In the top-level low-resolution feature image, the background noise is suppressed by morphological top-hat transformation, and the candidate marker regions are extracted by adaptive threshold segmentation. The preliminary marker regions are screened out by circularity and area constraints.

[0108] Locate the preliminary marker area in the middle-resolution image. Use the normalized cross-correlation matching algorithm to search within the preliminary marker area based on the marker template. Identify the response peak point within the preliminary marker area based on the normalized cross-correlation index. Then, divide the candidate marker area by fitting a quadratic surface to the response peak point.

[0109] Locate the candidate regions of the marker points in the underlying high-resolution image, extract the feature descriptors of the candidate regions of the marker points, perform similarity comparison with the marker point template, and output the coordinates of the marker points of the target replacement hob enhanced image based on the similarity.

[0110] According to the theoretical position of the preset mark point in the enhanced image of the target hob replacement, the pose deviation of the coordinates of the target hob replacement mark point is calculated;

[0111] See also Figure 5 , configure the conflict threshold. If the posture deviation is greater than the conflict threshold, the two-level fine-tuning mechanism is triggered to correct the posture, including:

[0112] During the primary fine-tuning stage, based on the position data of the end effector of the tool change actuator fed back by the six-dimensional force sensor, a compensation path is planned in real time through a path planning algorithm, such as the improved RRT* algorithm, to drive the end effector to perform posture correction compensation along the axial, radial and tangential directions of the tool, and synchronously monitor the posture deviation during the correction process; the number of posture correction compensations is counted, and a correction threshold is configured. If the number of posture correction compensations is greater than the correction threshold, and the posture deviation is greater than the conflict threshold, the secondary fine-tuning process is triggered and secondary fine-tuning is performed; otherwise, the posture correction compensation is terminated, and the tool disc is locked to provide a stable benchmark for subsequent disassembly and assembly operations.

[0113] In the secondary fine-tuning stage, the incremental PID algorithm is used to generate a micro-angle correction command for the cutterhead, drive the cutterhead to rotate, perform angle correction, and monitor the cutterhead's circumferential displacement in real time. After each angle correction, the image is recaptured to calculate the position deviation of the target hob mark point coordinates.

[0114] If it is less than or equal to the conflict threshold, the posture correction is terminated. Otherwise, the number of angle corrections is counted. If the number of angle corrections is greater than the correction threshold, a posture warning is issued, a manual intervention request is pushed, and the cutter head posture rollback function is started at the same time to restore to the most recent stable posture.

[0115] When the posture deviation is less than or equal to the conflict threshold, the cutter head is locked to provide a stable benchmark for subsequent disassembly and assembly operations.

[0116] After the cutterhead is locked, the tool changer robot initiates the disassembly and assembly process. Based on the coordinates of the target hob replacement mark, the tool changer robot's end effector integrates multispectral vision and posture sensor data to control the alignment of the torque wrench through a force-position hybrid control strategy. During the process, it relies on visual sub-pixel feature matching and posture inertial measurement to calculate posture residuals in real time and dynamically compensate for them, ensuring that the coaxiality of the tool and the threaded hole meets assembly requirements. After alignment, the target hob is loosened in stages according to the preset torque curve. Pre-loosening releases thread engagement, static friction is overcome in the transition phase, and axial constraints are released in the final stage.

[0117] The axial pressure and lateral offset of the end effector are monitored in real time to detect any abnormal resistance. If so, vibration mode is activated to assist in loosening the target hob during replacement. The end effector applies micro-axial vibrations while injecting specialized cutting fluid to reduce friction. During the vibration process, the frequency and phase are dynamically adjusted to match the bolt's resonance characteristics for efficient loosening.

[0118] The tool-changing robot's end-effector gripper grasps the target and replaces the hob. After the gripper grasps the target, it uses contact pressure and a six-axis force sensor to verify grip stability. The robot then moves along a pre-planned obstacle-avoidance path, integrating the cutterhead cabin's point cloud map to avoid fixed obstacles. During transport, a laser scanner updates the local point cloud in real time, triggering joint trajectory replanning when encountering unexpected obstacles to dynamically maintain a safe distance. Furthermore, based on the hob's center of gravity data fed back by the force sensor, if a center of gravity shift is detected, the robot automatically switches to a low-inertia trajectory, preventing the robot from reaching dangerous positions such as joint limits or center of gravity instability.

[0119] When the target replacement hob becomes loose, the hydraulic lifting mechanism of the double-station annular conversion platform drives the recovery side station to rise vertically, and the tool changing robot transfers the target replacement hob to the recovery station along the pre-planned obstacle avoidance path. During the transfer process, the laser scanner monitors the obstacles around the path in real time and dynamically maintains a safe distance.

[0120] Simultaneously, the hob storage bin pushes the hob to be replaced to the supply-side station of the double-station ring-type transfer station via a chain conveyor. When the double-station ring-type transfer station receives the target replacement hob and the hob to be replaced, a switching operation is triggered, controlling the servo motor of the double-station ring-type transfer station to drive the rotary mechanism to rotate, swapping the target replacement hob with the hob to be replaced.

[0121] The supply-side station of the dual-station circular transfer table is raised to the tool-changing robot's operating height. The end-effector suction fixture grabs the hob to be replaced and retracts along a mirrored path. Path planning incorporates the return load changes to dynamically distribute joint speeds, controlling the hob's movement to the installation station. A hybrid force-position control strategy is employed during the installation phase: Using marker coordinates for positioning, when the hob is located at the installation station, the system switches to impedance control mode, adaptively adjusting and monitoring the end-effector's press-fitting force. A cross-sequential loading strategy is employed, controlling the torque wrench to apply incremental force, pausing after each application to release stress and ensure even thread force, thus completing the assembly of the hob.

[0122] Working principle and its effect:

[0123] The present invention is based on multi-dimensional decision thresholds and fuzzy logic algorithms to perform real-time dynamic analysis of the hob operating status parameters, trigger tool change decisions through a hierarchical early warning mechanism, and realize intelligent closed-loop control from status monitoring to command generation.

[0124] A multi-dimensional weighted evaluation model comprehensively considers hob wear and the cascading failure risk index, and combines a dual-objective function to optimize the cutterhead rotation path. This ensures that high-priority tools are replaced first while reducing ineffective rotation paths, significantly improving the scientific nature and efficiency of tool change decisions. In the positioning phase, reverse rotation path planning, S-shaped acceleration and deceleration curve control, and incremental PID compensation enable the cutterhead to quickly and accurately position itself within the tool change workspace.

[0125] Multispectral imaging fusion technology and multi-scale Gaussian downsampling algorithm are used to effectively overcome dust interference, accurately extract hob marking points and calculate posture deviations. A two-stage fine-tuning mechanism is used to correct errors and ensure operational accuracy.

[0126] The collaborative design of the dual-station ring transfer table enables the synchronous exchange of old and new cutters through hydraulic lifting and servo rotation mechanisms. Combined with a force-position hybrid control strategy and an intelligent torque wrench for phased operation, this achieves a continuous and automated cutter-changing process. This enables adaptive decision-making in complex strata, effectively coping with harsh working conditions such as hard rock and water-rich soils. This significantly improves the safety, efficiency, and reliability of cutter-changing operations, providing strong technical support for the shield machine's sustained and stable tunneling.

[0127] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An intelligent control method for a shield machine tool changing robot, characterized in that: include: Collect shield machine cutter operating status parameters; Divide parameter intervals based on multi-dimensional decision thresholds, trigger a graded warning mechanism, make tool change decisions based on hob wear, and build a tool change sequence; The tool change control operation is triggered according to the number of hobs in the tool change sequence: Identify the hob type in the tool change sequence, calculate the comprehensive priority index for hob replacement due to wear using a multi-dimensional weighted evaluation model, and construct a wear execution sequence based on the principle of minimizing the cutterhead rotation path; set the comprehensive priority for forced hob replacement, and construct an emergency execution sequence using a dual-objective function. Combined with the available time window length and cutterhead rotation control, generate a tool change instruction set for the tool change sequence; Based on the tool change instruction set, the cutterhead rough positioning is completed through reverse rotation path optimization and dynamic planning of the cutterhead motion trajectory. Inertial slip compensation is used to position the target replacement hob in the tool change robot workspace, triggering the tool change operation: Extract the target hob replacement marker points, construct a multi-level feature image through multi-scale Gaussian downsampling, obtain the marker point coordinates of the target hob replacement enhanced image to identify the pose deviation, and perform pose correction through a two-level fine-tuning mechanism; The collaborative operation of the integrated double-station ring conversion table controls the tool changing robot to perform tool changing operations on the target replacement hob through the coordinated hydraulic lifting station and force-position hybrid control strategy.

2. The intelligent control method for a shield machine tool changing robot according to claim 1, characterized in that: The specific steps of constructing the wear execution sequence include: When the worn hob is identified and the tool change sequence is entered, a multi-dimensional weighted evaluation model is used to update the execution order of the tool change sequence, namely: Establish a basic priority based on the wear of the hob cutters in the tool change sequence, configure regional thresholds, divide the cutterhead where each hob cutter in the tool change sequence is located into local areas, and calculate the weighted average wear of the hob cutters in the local areas to generate the failure cascading risk index of the hob cutters in the tool change sequence; Linearly superimpose the hob wear and fault chain risk index to output a comprehensive priority index. Use the bubble sort algorithm to dynamically rearrange the tool change sequence and update the execution order of the tool change sequence. The cutterhead stop time window and the required buffer time of the cutterhead are obtained, and the available time window is divided. When the available time window is longer than the single tool change benchmark time, the multi-batch replacement mode is started, and the wear execution sequence of the available time window is constructed according to the comprehensive priority index and the principle of minimum cutterhead rotation path.

3. The intelligent control method for a shield machine tool changing robot according to claim 2, characterized in that: The specific steps of constructing the wear execution sequence of the available time window include: Calculate the number of tool changes that can be accommodated in each available time window based on the single tool change benchmark time and the length of the available time window; Configure priority thresholds, including upper and lower priority thresholds, and divide the comprehensive priority index into three execution levels according to the priority thresholds; The positions of worn replacement hobs are converted into polar coordinates, and the rotation angle difference between any two worn replacement hobs is calculated. A dual objective function is constructed, including maximizing the comprehensive priority index and minimizing the rotation angle. The dual objective function is balanced by weighted coefficients, and a wear execution sequence is generated for each available time window based on a greedy strategy.

4. The intelligent control method for a shield machine tool changing robot according to claim 3, characterized in that: The specific steps of generating the tool change instruction set of the tool change sequence include: Obtaining an execution sequence, including a wear execution sequence and an emergency execution sequence; and obtaining a target of the execution sequence to replace the hob according to the execution order; When the cutter head stop time window is entered, the target of the execution sequence of the reel cutter replacement is obtained in sequence according to the execution sequence of the available time window of the cutter head stop time window, and the operation reference angle of the tool changing robot is defined; According to the execution order of the target hob replacement in the execution sequence, the angle difference between the target hob replacement and the reference angle of the tool changing robot is calculated in sequence; The hob comprehensive priority index in the execution sequence of the available time window, the angle difference between the target hob replacement and the operating reference angle of the tool changing robot, and the updated execution order are encapsulated as a tool changing instruction set.

5. The intelligent control method for a shield machine tool changing robot according to claim 1, characterized in that: The specific steps of controlling the tool-changing robot to perform a tool-changing operation on the target replacement hob include: During the cutterhead rotation control phase, the target angle for hob replacement is captured in real time, and the initial angle difference is calculated based on the reference angle of the tool change robot. Based on this initial angle difference, it is determined whether to switch to the reverse rotation path. The cutterhead motion trajectory is dynamically planned using an S-shaped acceleration and deceleration curve, and the angular velocity and acceleration thresholds are adaptively adjusted based on real-time inertia parameters to achieve rapid and rough positioning of the cutterhead. Configure a buffer angle. When the angle difference between the target hob to be replaced and the reference angle of the tool changer robot is less than the buffer angle, the incremental PID control algorithm is activated to compensate for inertial slip and position the target hob to the tool changer robot's workspace. Based on multispectral imaging fusion technology, an enhanced image of the target hob replacement is collected and generated. Through multi-scale Gaussian downsampling, a multi-level feature image is constructed. The marker points of the enhanced image of the target hob replacement are located to calculate the position deviation of the target hob replacement. Configure the conflict threshold to determine whether to trigger the two-level fine-tuning mechanism for posture correction; The end effector of the tool changing robot loosens the target replacement hob and installs the hob to be replaced in stages based on the coordinates of the marking point of the target replacement hob through a force-position hybrid control strategy, so as to collaboratively complete the recovery of the target replacement hob and the assembly of the hob to be replaced through a double-station annular conversion table.

6. The intelligent control method for a shield machine tool changing robot according to claim 5, characterized in that: The specific steps of positioning the target to replace the marker point of the hob enhanced image include: After the target replacement hob enters the workspace, multispectral imaging is started to collect infrared and visible light images of the target replacement hob; A bilateral filter is used to reduce noise on the infrared image, and contrast adaptive histogram equalization is performed on the visible light image, and a target replacement hob enhanced image is generated by weighted fusion of the infrared image and the visible light image; According to the preset marking point of the hob, the theoretical position of the hob enhanced image is replaced at the target, and a dynamic search window is established with the theoretical coordinates as the center; The image within the dynamic search window is subjected to multi-scale Gaussian downsampling to construct a multi-level feature image, including a top-level low-resolution image for fast coarse positioning, a middle-level medium-resolution image for marker point template matching optimization, and a bottom-level high-resolution image for feature descriptor matching to locate the marker points of the target replacement hob enhanced image.

7. The intelligent control method for a shield machine tool changing robot according to claim 5, characterized in that: The specific steps of determining whether to trigger the two-stage fine-tuning mechanism and performing posture correction include: According to the theoretical position of the preset mark point in the enhanced image of the target hob replacement, the pose deviation of the coordinates of the target hob replacement mark point is calculated; Configure the conflict threshold. If the posture deviation exceeds the conflict threshold, a two-level fine-tuning mechanism is triggered to correct the posture, including: In the primary fine-tuning stage, based on the position data of the end effector of the tool change actuator, a compensation path is planned through a path planning algorithm, and the end effector is driven to perform posture correction and compensation. During the correction process, the posture deviation is monitored synchronously. Count the number of pose corrections and compensations, and configure a correction threshold. If the number of pose corrections and compensations is greater than the correction threshold, and the pose deviation is greater than the conflict threshold, the secondary fine-tuning process is triggered and secondary fine-tuning is performed; otherwise, the pose correction and compensation is terminated and the cutterhead is locked. In the secondary fine-tuning stage, an incremental PID algorithm is used to generate a cutterhead angle correction command, drive the cutterhead to rotate, perform angle correction, and monitor the cutterhead's circumferential displacement in real time. After each angle correction, the image is recaptured to calculate the position deviation of the target hob mark point coordinates. If the position deviation is less than or equal to the conflict threshold, the position correction is terminated and the cutterhead is locked. Otherwise, the angle correction times are counted. If the angle correction times are greater than the correction threshold, a posture warning is issued.

8. The intelligent control method for a shield machine tool changing robot according to claim 5, characterized in that: The specific steps of collaboratively completing the target replacement hob recovery and the hob assembly to be replaced by the double-station annular conversion table include: After the cutterhead is locked, the tool changer robot starts the disassembly and assembly process. Based on the coordinates of the target hob replacement mark, the tool changer robot's end effector uses a force-position hybrid control strategy to drive the torque wrench on the end effector to align with the target hob replacement mark. The target hob is loosened in stages according to the preset torque curve. The tool-changing robot's end effector grabs the target replacement hob, and the hydraulic lifting mechanism of the double-station annular conversion table drives the recovery side station to rise vertically, controlling the tool-changing robot to transfer the target replacement hob to the recovery station. At the same time, the hob storage bin pushes the hob to be replaced to the supply side station of the double-station annular conversion table through a chain conveyor; When the double-station annular conversion table receives the target replacement hob and the hob to be replaced, the conversion operation is triggered to control the servo motor of the double-station annular conversion table to drive the rotary mechanism to rotate and change the position of the target replacement hob and the hob to be replaced; The supply-side station of the double-station annular conversion table is raised to the operating height of the tool changing robot. The end adsorption fixture grabs the hob to be replaced and retreats along the mirror path. The force-position hybrid control strategy is used to control the hob to be replaced to move to the installation station for assembly.

9. The intelligent control method for a shield machine tool changing robot according to claim 1, characterized in that: The specific steps of constructing the tool change sequence include: Setting an initial multi-dimensional decision threshold for the hob operating state parameter based on the physical properties of the hob, wherein the multi-dimensional decision threshold includes a normal operating range, a warning range, and a forced replacement range; Combining the construction stratum type, shield machine propulsion speed and cutterhead speed, a fuzzy logic algorithm is used to update the multi-dimensional decision threshold correction coefficient in real time, dynamically adjust the multi-dimensional decision threshold, and update the decision threshold range; Compare the received hob operating status parameters with the multi-dimensional decision thresholds dimension by dimension, obtain the decision threshold interval, and trigger the graded warning mechanism: If the hob operating state parameter is within the normal operating range, a trend prediction is performed on the hob operating state parameter value; If the hob operating state parameter is in the warning range, it is marked as an abnormal operating state parameter, and an abnormal monitoring process is started for the abnormal hob with abnormal operating state parameters, and the number of abnormal operating state parameters and abnormal duration are counted, the wear state is quantified, and the wear degree of the abnormal hob is obtained; Configure the tool change threshold. If the wear of the abnormal hob is greater than the tool change threshold, it will be included in the tool change sequence; otherwise, the wear calculation result will be continuously monitored and updated. If the hob operating status parameter is in the forced replacement range, the hob is included in the tool change sequence.

Images