Cooperative control method and system for online cutter replacement of shield tunneling machine

Abstract

The invention relates to the technical field of mining and mineral processing, and particularly discloses a cooperative control method and system for online cutter replacement of a shield tunneling machine. The method comprises the steps that multi-source sensing data are fused and analyzed through an artificial intelligence model, and online intelligent diagnosis and automatic tool changing decision making of the tool abrasion state are achieved; in response to the decision, automatically adjusting tunneling parameters to enable the shield tunneling machine to enter a machine stabilizing state, and controlling a cutterhead to perform high-precision indexing system rotary positioning and physical locking according to the decision; an environment-resistant special mechanical arm is controlled to execute automatic tool changing operation, and the action state of the mechanical arm drives the shield tail grouting system to finely adjust the pressure in real time through feedforward control so as to actively offset operation disturbance; and automatic double verification of the installation quality is completed through a torque sensor and machine vision. According to the method, the full-process automation of tool changing from intelligent sensing, decision making to cooperative execution is realized, personnel are liberated from a high-risk tool changing environment, and meanwhile, the stability of an excavation face and a segmental lining during automatic tool changing operation is ensured.

Description

Technical Field

[0001] This invention relates to the field of mining and mineral processing technology, and more specifically, to a collaborative control method and system for online tool changing in tunnel boring machines. Background Technology

[0002] Tunnel boring machines (TBMs) are the core equipment for modern tunnel excavation. The various cutting tools mounted on their cutterheads endure severe abrasion and impact during the breaking of rock and soil. Rapid tool wear is a major factor limiting excavation efficiency and cost. Therefore, timely, safe, and efficient replacement of worn tools is an indispensable key maintenance operation to ensure the continuous and stable construction of TBM tunnel projects. Currently, the industry is committed to promoting the evolution of tool changing technology from the traditional, high-risk, and inefficient model that relies entirely on manual experience to mechanization, automation, and even intelligentization. The development trend is reflected in two aspects: firstly, by adding sensors to monitor excavation parameters to achieve predictive judgment of tool wear; secondly, by developing various auxiliary tooling or mechanical devices to replace or reduce manual operations and improve operational safety.

[0003] However, existing technological explorations still have significant limitations and have failed to form a systematic solution. Firstly, at the intelligent decision-making level, most wear monitoring methods can only provide early warning information or serve as tools for optimizing tunneling parameters. They cannot form a closed-loop cutter-changing decision that drives on-site equipment to execute automatically, including specific cutter identification and sequence information; manual intervention is still required to initiate the operation. Secondly, at the execution level, even with the introduction of cutter-changing robotic arms or specific devices, their design often focuses on single-machine operation and generally lacks deep integration with the tunnel boring machine's main control system and tunnel stability control system. This leads to two prominent problems: firstly, the precise and stable positioning of the cutterhead required for automated cutter changing is difficult to meet; secondly, the mechanical force generated by the cutter-changing action can cause uncontrollable disturbances to the soil chamber pressure and tunnel segments, posing a risk of inducing ground subsidence. In summary, fragmented improvements to existing technologies cannot solve the systemic coordination challenges of the entire cutterhead changing operation process. The industry urgently needs an integrated approach that connects intelligent diagnosis, decision-making, precise positioning, automatic execution, real-time stability control, and a quality closed loop. This approach would truly improve the inherent safety and efficiency of cutterhead changing operations while ensuring the continuity of the tunneling process and the long-term stability of the tunnel. Summary of the Invention

[0004] The purpose of this invention is to provide an online cutter replacement solution for tunnel boring machines that improves the inherent safety and efficiency of cutter replacement operations while ensuring the continuity of the tunneling process and the long-term stability of the tunnel.

[0005] According to a first aspect of the present invention, a collaborative control method for online cutter replacement in a tunnel boring machine is proposed, comprising: S1. Based on multi-source sensor data from the operation of the tunnel boring machine, an artificial intelligence model is used to determine the wear status of the cutting tools and generate a tool replacement decision. S2. In response to the cutterhead change decision, the tunneling parameters are automatically adjusted to bring the tunnel boring machine into a stable state and to establish a pressure buffer for the soil chamber; S3. Control the tool turret to perform indexing positioning and lock according to the tool change decision; S4. After the cutterhead is locked, control the robotic arm to perform the cutter change operation, and control the robotic arm to link with the shield tail grouting system to maintain stable grouting pressure during the cutter change. S5. Verify the tightening torque and perform visual inspection on the newly installed cutting tools; S6. Record the cutterhead change data and guide subsequent operations. Once all operations are completed, resume normal tunneling of the tunnel boring machine.

[0006] According to some embodiments, in the method of the first invention, step S1 includes: Collect multi-source sensor data from the tunnel boring machine; input the multi-source sensor data into a pre-trained artificial intelligence model for analysis to obtain the wear status assessment results of the cutting tools; based on the wear status assessment results, generate a tool replacement decision that includes the identification of the cutting tools to be replaced and the replacement sequence.

[0007] According to some embodiments, in the method of the first invention, the multi-source sensing data includes cutterhead drive current timing data, tunneling rate change data, and vibration spectrum data; the artificial intelligence model is configured to perform feature extraction and fusion analysis on the multi-source sensing data to output wear state assessment results.

[0008] According to some embodiments, in the method of the first invention, feature extraction and fusion analysis specifically includes: The trend features characterizing load fluctuations are extracted from the time-series data of the cutterhead drive current; specific frequency band energy features related to tool wear are extracted from the vibration spectrum data; the trend features, tunneling rate change data and energy features are weighted and fused, and the tool wear level is determined based on the fusion result.

[0009] According to some embodiments, in the method of the first invention, step S2, automatically adjusting the tunneling parameters to bring the tunnel boring machine into a stable state includes the following steps: Based on the geological exploration data and real-time tunneling parameters of the current tunneling section, the current geological type is identified. Based on the identified geological type, the target range for stabilizing the total thrust and cutterhead drive torque is matched and calculated from a pre-set database. The tunnel boring machine's propulsion system and cutterhead drive system are controlled to gradually adjust the total thrust and cutterhead drive torque to the target range for stabilizing the machine. While adjusting the tunneling parameters, a preparatory command is sent to the tail grouting system. The preparatory command is used to switch the grouting system to pressure stabilization mode in preparation for responding to subsequent linkage control. The pressure fluctuations in the soil chamber and the attitude of the tunnel boring machine are monitored. When both are stable within the permissible threshold within a set time, the tunnel boring machine is determined to have entered a stable state.

[0010] According to some embodiments, in the method of the first invention, controlling the tool turret to perform indexing positioning and locking based on a tool change decision includes the following steps: Based on the target tool position determined by the tool change decision, calculate the step-by-step rotation sequence that the tool head needs to execute; according to the step-by-step rotation sequence, control the tool head drive system to execute the step-by-step rotation; after the tool head rotates to a predetermined angle, control the locking mechanism to fix the tool head.

[0011] According to some embodiments, in the method of the first invention, a visual-assisted correction step is included before step S4, including: The actual pose of the target tool mount on the locked tool turret is obtained by a vision sensor; the actual pose is compared with the theoretical position derived from the tool change decision to generate a pose deviation; and the robotic arm is controlled to perform motion trajectory compensation based on the pose deviation.

[0012] According to some embodiments, in the method of the first invention, step S4 includes the following steps: Based on the tool change decision, the robot arm generates tool disassembly and installation operation instructions; controls the robot arm to perform tool change operations according to the operation instructions; during the execution of the operation instructions by the robot arm, the grouting pressure adjustment instructions are generated in real time according to the action status and sent to the shield tail grouting system to maintain stable grouting pressure.

[0013] According to some embodiments, in the method of the first invention, generating grouting pressure adjustment instructions in real time based on the action state specifically includes: monitoring the action stage and load state of the end effector of the robotic arm; predicting the disturbance trend of the cutter replacement operation on the soil chamber and segment lining based on the pre-established mapping relationship and the action stage and load state; and generating a time-sequential feedforward control instruction based on the disturbance trend to drive the shield tail grouting system to adjust the grouting pressure in advance.

[0014] According to a second aspect of the present invention, a collaborative control system for online cutter replacement in a tunnel boring machine for implementing the method of the first aspect of the present invention is provided, comprising: The intelligent decision-making module is used to determine the wear status of the cutters and generate cutter replacement decisions based on multi-source sensor data from the operation of the tunnel boring machine through an artificial intelligence model. The collaborative control center is used to respond to tool change decisions and coordinate the control of the following modules; The tunneling and attitude stabilization module is used to automatically adjust the tunneling parameters according to the instructions of the collaborative control center to enable the tunnel boring machine to enter a stable state and to establish a pressure buffer for the soil chamber. The tool turret precision positioning module is used to control the tool turret to perform indexing positioning and lock-up according to the instructions and tool changing decisions of the collaborative control center; The automatic tool changer and linkage execution module is used to control the robotic arm to perform tool changing operations after the cutter head is locked, and to control the robotic arm to link with the shield tail grouting system to maintain stable grouting pressure during tool changing. The quality verification module is used to verify the tightening torque and perform visual verification on newly installed cutting tools; The data management and process control module is used to record cutter change data and guide subsequent operation processes, and to instruct the tunnel boring machine to resume normal tunneling after all operations are completed.

[0015] The technical effects achieved by this invention include: 1. To address the problems of traditional shield tunneling cutter replacement relying on manual shutdown inspection, which is inefficient and cannot predict sudden cutter damage, this invention integrates multi-source data such as cutterhead current, tunneling rate, and vibration spectrum, and uses artificial intelligence models for fusion analysis and predictive judgment. This enables online intelligent diagnosis of cutter wear status and automatic cutter replacement decision generation, transforming the cutter replacement mode from passive response or periodic maintenance to proactive predictive maintenance.

[0016] 2. In response to the problems that the preparation before cutterhead replacement relies on the driver's experience and the process is not standardized, and that the cutterhead replacement operation is carried out in isolation from the tunnel grouting and other stability control systems, which poses a risk of mutual interference, this invention designs a standardized machine stabilization program that includes geological identification, parameter adaptive matching, and pre-linkage of the tail grouting system. This enables the automatic and accurate construction of the shield machine status and multi-system collaborative mode before cutterhead replacement, providing a safe and stable working condition foundation for subsequent automated operations.

[0017] 3. Addressing the core challenge of precise docking between the robotic arm and the rotating cutterhead in automatic cutter change, and the potential impact of mechanical disturbances generated by the cutter change action on the shield tail and affecting the stability of the installed tunnel segments, this invention achieves proactive cancellation of operational disturbances within milliseconds through indexing cutterhead positioning lock-up and feedforward linkage control based on the grouting pressure according to the robotic arm's action status. This overcomes the key technical bottleneck of the inability to coordinate high-precision automated cutter change with real-time stability control of the tunnel strata.

[0018] 4. In response to the lack of reliable verification methods for automated installation quality and the failure to effectively record and optimize tool changing process data, this invention introduces dual quality checkpoints of tightening torque verification and visual inspection of installation in place, combined with automatic recording of full-process data, to achieve closed-loop assurance of tool changing operation quality and continuous optimization and iteration of process parameters, making each operation a data fuel driving system progress. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without exceeding the scope of protection claimed by the present invention.

[0020] Figure 1 This is a flowchart illustrating an embodiment 1000 of the collaborative control method for online cutter replacement in a tunnel boring machine according to the present invention. Figure 2 for Figure 1 A flowchart illustrating step S1 in Example 1000; Figure 3 for Figure 1 A flowchart illustrating sub-step S2A of step S2 in Example 1000; Figure 4 for Figure 1 A flowchart illustrating sub-step S3A of step S3 in Example 1000; Figure 5 for Figure 1 Example 1000 also includes a flowchart of the visual-assisted correction step SA; Figure 6 for Figure 1 A flowchart illustrating step S4 in Example 1000; Figure 7 This is a schematic diagram of an embodiment 2000 of the collaborative control system for online tool changing of a tunnel boring machine according to the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] See Figure 1 , Figure 1This is a schematic flowchart of Embodiment 1000 of the cooperative control method for online tool replacement of a shield machine according to the present invention. As Figure 1 shown, Embodiment 1000 includes Step S1 - Step S6.

[0023] In Step S1, a cooperative control system for online tool replacement of a shield machine (hereinafter referred to as the system for short) judges the tool wear state based on multi-source sensing data of the shield machine operation and generates a tool replacement decision through an artificial intelligence model. Step S1 is the decision starting point of this solution. Its core role is to transform the traditional tool management method that relies on manual experience and offline inspection into an online, automatic, and predictive maintenance mode driven by data. By analyzing the operation state of the shield machine in real time, it actively identifies the tool health and generates precise instructions that can directly drive the subsequent automated processes, which is the prerequisite for realizing fully unmanned intelligent tool replacement.

[0024] Specifically, Step S1 includes: the system collects multi-source sensing data of the shield machine; the system inputs the multi-source sensing data into a pre-trained artificial intelligence model for analysis to obtain the tool wear state evaluation result; the system generates a tool replacement decision based on the wear state evaluation result, including the identification of the tool to be replaced and the replacement sequence.

[0025] In Step S2, in response to the tool replacement decision, the system automatically adjusts the tunneling parameters to make the shield machine enter a stable state and establish a pressure buffer for the soil chamber. Step S2 is the key transition and safety guarantee link from intelligent decision-making to physical execution in this solution. Its core role is to respond to the tool replacement decision generated by S1, actively and accurately switch the shield machine from the dynamic tunneling mode to the static stable state mode designed for tool replacement operations, establish a pressure buffer for the soil chamber, and at the same time wake up the tail seal grouting system to enter the cooperative standby state. This step ensures that all subsequent automated operations are carried out on the basis of a stable, controllable, and safe working condition.

[0026] In some specific embodiments, in Step S2, the process of the system automatically adjusting the tunneling parameters to make the shield machine enter a stable state includes: According to the geological exploration data and real-time tunneling parameters of the current tunneling section of the shield machine, identify the current geological type; based on the identified geological type, match and calculate the stable state target range of the total thrust and cutterhead drive torque from a preset database; control the propulsion system and cutterhead drive system of the shield machine to gradually adjust the total thrust and cutterhead drive torque within the stable state target range; while adjusting the tunneling parameters, send a preparatory instruction to the tail seal grouting system, and the preparatory instruction is used to make the grouting system switch to the pressure stability maintenance mode to prepare for responding to subsequent linkage control. Monitor the pressure fluctuation of the soil chamber and the attitude of the shield machine. When both are stable within the permitted threshold within a set time, it is determined that the shield machine enters a stable state.

[0027] In step S2, establishing a pressure buffer for the soil chamber is a proactive and precise ground stabilization measure that runs parallel to adjusting tunneling parameters. Its purpose is to construct a uniform and stable mechanical cushion layer in front of the cutterhead before the cutterhead operation, with a pressure slightly higher than the static earth pressure, to actively support the excavation face and isolate it from external disturbances.

[0028] Optionally, in step S2, the innovation of this scheme compared to traditional single-pipe pressurization lies in introducing a multi-channel, zoned, and adaptive pressure buffering system that is intelligently linked to the wear status of the cutterhead and geological information in step S2B, which establishes a pressure buffer for the soil chamber. The implementation process is as follows: S2B1. Intelligent Decision-Making and Parameter Distribution: The collaborative control center calculates the pressure buffer target based on the tool change decision generated by S1 and the current geological type identified by S2A1. The tool change decision includes the location of severely worn tools.

[0029] The system generates a pressure distribution matrix based on geological permeability and the distribution area of ​​the cutters to be replaced on the cutterhead. For example, for a fan-shaped area with severe cutter wear, the target pressure at the injection point behind the corresponding soil chamber baffle will be set slightly higher to provide stronger local support; other areas will use the base pressure value. These target values, along with the corresponding microbubble bentonite mixing ratio and injection flow rate parameters, are sent to the soil chamber pressure buffer control system.

[0030] S2B2. System Start-up and Media Preparation: After receiving the command, the pressure buffer system starts the bentonite slurry station and microbubble generator.

[0031] The slurry station prepares bentonite slurry at a set concentration, and then synthesizes it into a microbubble bentonite medium in a dynamic mixer with uniform, fine bubbles generated by a microbubble generator. Compared with pure air or pure mud, this mixed medium has the advantages of compressibility, good buffering, and stability that prevents escape, making it a key innovative material in this embodiment.

[0032] S2B3. Precise injection and closed-loop control in zones: The medium is injected into the soil chamber through independent pipelines.

[0033] The system controls multi-channel electric regulating valves to adjust the injection flow to different zones within the soil chamber. Each zone is equipped with a high-precision pressure sensor, which feeds real-time data back to the controller, comparing it with the target value in the pressure distribution matrix issued by S2B1. The controller employs a fuzzy PID algorithm to perform independent closed-loop regulation of each channel, ensuring that the pressure in each zone quickly and smoothly reaches its set target and maintains dynamic stability throughout the tool change process.

[0034] S2B4. Pressure Maintenance and System Readiness: Once the target pressure is reached, the system switches to a low-power pressure maintenance mode.

[0035] The main injection pump operates at a reduced frequency, switching to a pressure-maintaining circuit consisting of a precision metering pressure-replenishing pump and an automatic venting valve. This system automatically compensates for pressure losses caused by minute formation seepage or temperature changes, controlling pressure fluctuations within ±0.02 MPa. Once the system self-checks and confirms pressure stability, it sends a soil chamber pressure buffer ready signal to the collaborative control center.

[0036] According to the above preferred embodiment, the process of "establishing a pressure buffer for the soil chamber" in step S2 of the present invention achieves three technical effects: (1) Active and precise support: The passive pressure monitoring is transformed into active pressure construction. By applying differentiated pressure in different zones, targeted reinforcement support is achieved for the excavation face, especially in areas with severe tool wear, which greatly improves the local stability of the face during tool replacement. (2) State Adaptation: The parameters of the pressure buffer are not fixed, but are intelligently linked with the geological type identification results and tool wear decision, realizing the transition from experience-based preset to state adaptation, which is a significant innovation that is different from the existing fixed-program pressurization method.

[0037] (3) Provide a steady-state foundation for automated operations: The established microbubble bentonite pressure buffer layer has excellent stability and anti-disturbance ability, providing a reliable mechanical environment for subsequent cutter head positioning and precise automated operation of robotic arm.

[0038] In some specific embodiments, in step S3, the system controls the tool head to perform indexing positioning and locking according to the tool change decision. Specifically, this includes: calculating the step-by-step rotation sequence to be executed by the tool head based on the target tool position determined by the tool change decision; controlling the tool head drive system to execute the step-by-step rotation according to the step-by-step rotation sequence; and controlling the locking mechanism to fix the tool head after the tool head has rotated to a predetermined angle.

[0039] Optionally, step S3 is the core physical execution basis for the automatic tool change in this solution. Its core innovation lies in transforming the traditional continuously rotating tool head control into a discrete, point-to-point indexing and positioning control mode, creating an absolutely static, zero-displacement rigid working reference for subsequent robotic arm operations.

[0040] Traditional tunnel boring machine cutterhead control uses a continuous rotation speed servo or torque servo mode, with the goal of maintaining rotational motion. This solution, however, borrows the indexing principle from the control concepts of high-precision CNC machine tool rotary tables. Its core is point-to-point control, and the specific principles include: (1) Discretization of the target: The system no longer regards the cutter head as a continuously moving object, but discretizes its circumference into a series of specific angle points determined by the position of the target cutter.

[0041] (2) Motion-stationary separation: The motion of the cutter head is strictly divided into two stages: a high-speed, controlled rotation stage and an absolutely stationary, rigidly locked positioning stage. The robotic arm only operates during the positioning stage.

[0042] (3) Closed-loop accuracy guarantee: The position of the cutter head is fed back in real time through a high-resolution angle encoder and compared with the target position to form a closed-loop position control.

[0043] Step S3 of this invention solves the primary problem of automated tool changing through the above principle: how to provide a perfectly stationary operating platform for a precision robotic arm from a rotating giant system that is several meters in diameter, weighs hundreds of tons, and bears complex loads.

[0044] In step S4, after the cutterhead is locked, the system controls the robotic arm to perform the cutter replacement operation and controls the robotic arm to work in conjunction with the tail grouting system to maintain stable grouting pressure during the cutter replacement. The core task of step S4 is to safely and accurately complete the physical operation of cutter replacement based on the fact that the cutterhead has been precisely positioned and locked, and to ensure that the disturbance to the tunnel segments and the strata during this physical operation is minimized.

[0045] Optionally, step S4 includes: generating tool disassembly and installation operation instructions for the robotic arm based on the tool change decision; controlling the robotic arm to perform tool change operations according to the operation instructions; and generating grouting pressure adjustment instructions in real time according to the action status during the execution of the operation instructions by the robotic arm, and sending them to the tail grouting system to maintain stable grouting pressure. In step S5, the system verifies the tightening torque and performs visual inspection on the newly installed tool. Utilizing the end effector of the robotic arm in S4 and the system's independent vision resources, the system objectively assesses the physical results of the operation, specifically including: (1) Tightening torque verification: After the robotic arm completes the bolt tightening, its end-effector six-dimensional force sensor records the actual tightening torque curve. The system automatically compares the peak torque and final angle of this curve with the standard torque-angle process window issued in the S1 tool change decision for this tool model. If the data falls within the window, the verification is successful.

[0046] (2) Visual verification: Images of newly installed tools are captured under dedicated lighting using a dirt-resistant industrial camera fixed in the tool changing chamber or carried by a robotic arm. A feature matching algorithm is used to compare the tool outline, bolt head orientation, and other features in the captured images with a standard installation status template image rendered from the tool head digital model. Verification includes checking whether the installation is in place and whether there is any obvious misalignment or foreign object, achieving non-contact visual confirmation.

[0047] (3) Decision-making and fault tolerance: The results of dual verification complement each other. If both pass, success is recorded. If either fails, the system will issue an alarm and decide whether to retry, skip, or transfer to manual intervention based on preset strategies. This step will transform the final check that relies on human experience into a quantifiable and traceable automated quality inspection.

[0048] In step S6, the system records the cutterhead replacement data and guides subsequent operations. Once all operations are completed, the tunnel boring machine resumes normal tunneling.

[0049] In some specific embodiments, the system completes structured data recording in step S6 and automatically generates a report for this tool change operation, which fully includes: the decision ID and trigger data of S1, the final stabilization parameters of S2, the angle and lock-up status of each positioning point of S3, the robotic arm motion log and linkage commands of S4, and the verification results and image snapshots of S5. All data is timestamped and bound to a unique ID for this operation.

[0050] Subsequently, in step S6, the system core maintains a process state machine. Based on the queue of tools that have not yet been completed in the tool change decision, S6 guides the system to cyclically execute steps S3 to S5. When the queue is cleared, the state machine enters a recovery preparation state.

[0051] Optionally, in step S6, the system safely resumes tunneling. First, the robotic arm is controlled to return to the safe position and lock; second, the cutterhead locking mechanism is released; then, according to the original tunneling parameters recorded in S2 before stabilization, the thrust and cutterhead torque are gradually and in stages restored, while simultaneously instructing the grouting system to switch back from pressure stabilization mode to synchronous grouting mode. Throughout the process, attitude and pressure are continuously monitored to ensure a smooth transition. Finally, the tunnel boring machine resumes automatic tunneling, and the system awaits the next decision trigger in S1.

[0052] Figure 2 for Figure 1 A flowchart illustrating step S1 in Example 1000. (See attached diagram.) Figure 2 As shown, step S1 includes steps S11-S13.

[0053] In step S11, the system acquires multi-source sensor data from the tunnel boring machine (TBM). Optionally, in step S11, the system utilizes a sensor network deployed at key parts of the TBM to simultaneously acquire various physical signals that can indirectly or directly reflect the working status of the cutterhead. These signals characterize the complex interaction process between the cutterhead, the cutter, and the soil from different dimensions.

[0054] In some specific embodiments, in step S11, the multi-source sensing data includes cutterhead drive current timing data, tunneling rate variation data, and vibration spectrum data, and the specific acquisition process includes: In step S11, the system first senses signals through a sensor network, including: (1) The current sensor is installed on the power cable of the cutter head drive motor to collect the instantaneous value of the three-phase current of the motor at a high frequency (such as 1000Hz) to form the timing data of the cutter head drive current. The fluctuation of the current directly reflects the load change caused by the cutter cutting rock and soil of different hardness during the rotation of the cutter head.

[0055] (2) The encoder of the propulsion system reads the displacement signal of the propulsion cylinder, and the control system calculates the real-time tunneling rate change data accordingly. An abnormal decrease in tunneling rate is a macroscopic manifestation of reduced cutting efficiency caused by tool wear.

[0056] (3) The vibration acceleration sensor is mounted on the main bearing seat of the cutter head drive unit or on the back structure of the cutter head via a magnetic base to collect high-frequency vibration signals. After being processed by fast Fourier transform, the signal is converted into vibration spectrum data. Defects such as tool wear and chipping will change the frequency characteristics of its impact with the rock and soil, and generate energy changes in specific frequency bands (such as the harmonics of the cutter head passing frequency).

[0057] Subsequently, the system transmits the aforementioned sensor signals to the system's industrial control computer in real time via fieldbus. The data acquisition software filters, aligns, and formats the raw data, packaging it into standardized data frames for subsequent analysis.

[0058] In step S12, the system inputs multi-source sensor data into a pre-trained artificial intelligence model for analysis to obtain the tool wear status assessment result.

[0059] Optionally, in step S12, the artificial intelligence model is configured to: perform feature extraction and fusion analysis on multi-source sensor data to output wear status assessment results. The feature extraction and fusion analysis specifically includes: extracting trend features characterizing load fluctuations from the cutterhead drive current time series data; extracting specific frequency band energy features related to tool wear from vibration spectrum data; weightedly fusing the trend features, tunneling rate change data, and energy features; and determining the tool wear level based on the fusion results.

[0060] Optionally, in some specific embodiments, the artificial intelligence model in step S12 employs a CNN-LSTM-Attention hybrid neural network model. In the offline phase, the model is pre-trained using massive amounts of historical tunneling data to learn the complex mapping relationship between data and wear status. Optionally, the historical tunneling data includes sensor data and corresponding manually measured wear labels.

[0061] Optionally, the hybrid neural network model accepts a multi-channel time series sample as input, with each sample corresponding to a fixed time window: for the cutterhead drive current time series data and tunneling rate variation data, the model directly processes their time series form; for vibration spectrum data, since it is already frequency domain information, the model treats the spectrum of each time slice as a special image. Based on the characteristics of the tunnel boring machine's working process, the model of this invention specifically designs a non-uniform sampling data alignment module. Because the tunnel boring machine may be in different working conditions such as advancing, stopping, or changing steps, the data flow is non-uniform. This module can normalize the intermittent, non-equal interval sampled raw data into an equal interval sequence that the model can process through interpolation and resampling, and label it with working condition labels so that the model can distinguish the data patterns under different working conditions.

[0062] In some specific embodiments, the hybrid neural network model in step S12 of the present invention extracts features through the following parallel feature extraction branches: The vibration spectrum branch, CNN path, employs a one-dimensional convolutional neural network to process spectral data. The convolutional kernel is designed to automatically learn and identify key resonant frequency bands or sidebands related to tool wear (such as hob wear or scraper tooth breakage). This path outputs a frequency domain energy feature vector.

[0063] Current timing branch, LSTM path: Employs a Long Short-Term Memory (LSTM) network to process current timing data. LSTM's memory cells are particularly well-suited for capturing the periodic fluctuations in trend caused by load variations at different tool positions during one revolution of the tool turret, as well as the trend baseline drift due to increased wear. This path outputs a time-domain trend feature vector.

[0064] The tunneling rate branch has the following auxiliary input: the tunneling rate is a scalar sequence that is directly concatenated with the aforementioned feature vector.

[0065] Optionally, to suit the actual working scenarios of the tunnel boring machine in this invention, the model introduces a geological context attention mechanism. During training and inference, the model receives the geological type code of the current tunneling loop as auxiliary input. The attention mechanism uses this information to dynamically adjust the weights of different feature branches in the final decision. For example, in hard rock formations, vibration spectrum characteristics may be more sensitive; in soft soil formations, the stability of current changes may be more indicative.

[0066] Finally, after the feature vectors from each branch are concatenated in step S12, they are fed into a fully connected network for fusion and abstraction. The wear status assessment result for each tool is then output through the Softmax output layer of the wear level classifier, typically categorized into four levels: normal, light wear, moderate wear, and severe wear requiring replacement. A confidence probability can be output for each level.

[0067] In step S13, the system generates a tool change decision based on the wear condition assessment results, which includes the identifier of the tool to be replaced and the replacement sequence. Step S13 is the key link between intelligent diagnosis and automatic execution. Its core function is to transform and encapsulate the abstract wear condition assessment results output in S12 into a set of structured and operable instructions that can be directly understood and executed by subsequent steps.

[0068] Optionally, the triggering conditions for step S13 include the following: when the artificial intelligence model in step S12 outputs that the wear level of one or more tools is determined to be severely worn or needs to be replaced, or when its predicted remaining life is lower than a preset absolute safety threshold, the system automatically triggers the decision generation process.

[0069] Specifically, the input data for step S13 includes a list of wear condition assessment results for all cutting tools (including tool ID, wear level, confidence level, predicted remaining life, etc.). Simultaneously, in step S13, the system also reads the current cutterhead attitude parameters and the tool database of the tunnel boring machine.

[0070] Optionally, the tool change decision generated in step S13 is not a simple tool list, but a structured digital instruction package that includes job path planning and resource scheduling.

[0071] Optionally, a specific embodiment of the tool change decision includes: { "decision_id": "TC_20231027_001", "trigger_time": "2023-10-27T14:30:15Z", "overall_strategy": "MINIMIZE_INDEXING", "tool_change_sequence": [ { "step": 1, "operation": "REMOVE", "tool_id": "CUTTER_R3_A12", "location": {"radius": 3.2, "angle": 125.5}, "target_state": "REMOVED", "next_optimal_stopping_angle": 280.0 }, { "step": 2, "operation": "INSTALL", "tool_id": "CUTTER_NEW_001", "location": {"radius": 3.2, "angle": 125.5}, "target_state": "INSTALLED", "torque_spec": 650.0 }, { "step": 3, "operation": "REMOVE", "tool_id": "SCRAPER_S1_B05", "location": {"radius": 1.8, "angle": 34.2}, "target_state": "REMOVED", "next_optimal_stopping_angle": 34.2 } / / ... More replacement steps ], "interlock_requirements": { "grouting_pressure_mode": "ACTIVE_COMPENSATION", "expected_max_disturbance": 0.05 } } The synergistic meaning of each field is as follows: tool_change_sequence: Specifies the change order. Optimizing this order is key to reducing tool turret idle time and improving efficiency.

[0072] Location information: Provides absolute coordinate targets for the indexing positioning of S3. The tool turret drive system will calculate the rotation sequence based on this.

[0073] `next_optimal_stopping_angle`: This is a forward-looking optimization field. It indicates the angle the tool turret should rotate to immediately after completing the current tool change so that the robotic arm can move to the position of the next tool in the sequence via the shortest path. This reflects deep collaboration with the S3 step.

[0074] torque_spec: Provides a preset standard value for verifying the tightening torque of the S5.

[0075] interlock_requirements: This explicitly informs step S4 that the grouting system needs to enter active compensation mode for this cutter change operation, and estimates the maximum possible disturbance value, providing a parameter benchmark for linkage control.

[0076] In some specific embodiments, the decision generation step in step S13 is not a simple listing of worn tools, but an optimization process incorporating domain knowledge, including: (1) Initial screening and clustering: The system first screens out all the tools that need to be replaced. Then, it performs cluster analysis based on the angular position of these tools on the tool head, and groups the tools that are adjacent to each other.

[0077] (2) Sequence optimization (core): When generating the change order, the following principles are followed to ensure coherence and rationality with subsequent steps: Minimize tool head rotation principle: Prioritize replacing tools within the same cluster. When replacing tools within a group, sort them by calculating the shortest rotation path of the tool head (considering clockwise / counterclockwise) to ensure that after the robotic arm completes a tool replacement, the tool head only needs to rotate slightly or not at all to start the next tool, directly supporting the efficient indexing and positioning of S3.

[0078] Safety and stability are prioritized: tools with abnormally severe wear are given the highest priority and replaced first, even if their position is not optimal. This aligns with the primary goal of S2 in establishing a stable operating state.

[0079] Tool type coordination principle: Consider the mutual support effect of different types of tools. Sometimes, tools that are not yet at the threshold but are heavily worn may be proactively included in the replacement sequence to maintain the dynamic balance of the tool head load, which is beneficial for maintaining the stable machine state after S2.

[0080] Resource binding and instruction encapsulation: The system assigns a new tool entity with a unique ID to each change location from the virtual tool library and binds its physical storage location information in the warehouse to the decision. Finally, all optimized information is encapsulated into the aforementioned structured instruction package.

[0081] According to such Figure 2The embodiment shown in this invention integrates and analyzes multi-dimensional data such as current, vibration, and tunneling rate, and performs deep feature extraction and weighted fusion using a domain-customized AI model. This step enables accurate and early online assessment and prediction of tool wear status. Furthermore, based on this assessment, a structured optimal tool change sequence instruction is automatically generated. This instruction not only clarifies "which tool to change to" and "when to change to", but also proactively plans the optimal path and parameters for coordination with subsequent positioning and execution steps. This provides a reliable and accurate scheduling blueprint for the entire automated tool change process, laying the intelligent foundation for full-process collaborative control.

[0082] Figure 3 for Figure 1 A flowchart illustrating sub-step S2A of step S2 in Example 1000. (See attached diagram.) Figure 3 The diagram shows the sub-step S2A for automatically adjusting tunneling parameters to bring the tunnel boring machine into a stable state, specifically including steps S2A1-S2A5.

[0083] In step S2A1, the system identifies the current geological type based on the geological exploration data and real-time tunneling parameters of the current tunneling section of the tunnel boring machine.

[0084] In some specific embodiments, in step S2A1, the system calls two data sources. The first is pre-entered engineering geological exploration report data, which, based on the tunnel boring machine's current position calculated from the segment ring number and tunneling mileage, matches the dominant geological type of the section. The second is real-time tunneling parameters, mainly the fluctuation characteristics of cutterhead torque and propulsion speed over the past few minutes. The system inputs a data vector integrating geological description and real-time parameter features into a lightweight classification model to perform online identification and verification of the actual micro-geological conditions faced at the current tunnel face, ultimately outputting a standardized geological type code. This identification result provides a crucial basis for subsequent adaptive parameter matching.

[0085] In step S2A2, based on the identified geological type, the system matches and calculates the target range for stable operation of total thrust and cutterhead drive torque from a preset database.

[0086] Optionally, in step S2A2, a "Geological-Stable Machine Parameter" knowledge base is pre-set in the system. This knowledge base is built based on historical construction data and expert experience, and defines the recommended range of total thrust and the range of cutterhead drive torque required to maintain the stability of the excavation face when stopping the machine to change cutters in each geological type.

[0087] Specifically, in step S2A2, the system queries the knowledge base based on the geological code identified in S2A1 to obtain basic target parameter values. Subsequently, the system fine-tunes the target values ​​by considering the actual torque value of the current cutterhead: for example, if the current torque is already high, the target torque value for stabilization may be set to a stable range slightly lower than the current value to avoid sudden load drops during shutdown. Finally, a specific [lower thrust limit, upper thrust limit] and [lower torque limit, upper torque limit] are calculated and locked as the target range for this stabilization operation.

[0088] In step S2A3, the system controls the tunnel boring machine's propulsion system and cutterhead drive system to gradually adjust the total thrust and cutterhead drive torque to within the target range for stable operation. The control commands in step S2A3 are issued by the collaborative control center. The system does not change all parameters simultaneously, but follows a gradual transition strategy of "first reducing torque, then adjusting thrust, with a double closed loop." First, a command is sent to the cutter head drive system to smoothly adjust the cutter head torque to the target range and maintain it at a rate of 1%-2% per second to reduce the rated torque.

[0089] As the cutterhead torque begins to decrease, a command is sent to the propulsion system to gradually reduce the propulsion speed to zero with a small deceleration, while simultaneously adjusting the total thrust to the target range.

[0090] Throughout the adjustment process, the propulsion controller and the cutterhead drive controller feed back the current thrust and torque values ​​to the main control system in real time, forming two independent closed-loop controls to ensure that the parameters converge to the target range accurately and without overshoot.

[0091] In step S2A4, while adjusting the tunneling parameters, the system sends a preparatory command to the tail grouting system. The preparatory command is used to switch the grouting system to pressure stabilization mode in preparation for responding to subsequent linkage control.

[0092] While adjusting the tunneling parameters, the collaborative control center sends a structured preparatory instruction to the control PLC of the tail grouting system via the industrial network. This instruction includes at least: a mode switching command, the expected operation type, and pressure monitoring priority parameters. Upon receiving the instruction, the grouting system immediately switches its main control logic from "synchronous grouting" mode to "pressure stabilization" mode. In this new mode, the start and stop of the grouting pump will no longer prioritize the injection volume, but rather focus on maintaining the tail grout pressure at the set value. Simultaneously, its control interface is opened to receive real-time feedforward compensation signals from the robotic arm system.

[0093] In step S2A5, the system monitors the soil chamber pressure fluctuations and the tunnel boring machine's (TBM) attitude. When both stabilize within permissible thresholds within a set time, the TBM is determined to have entered a stable operating state. Specifically, in some embodiments, in step S2A5, after the tunneling parameters are adjusted, the system initiates a stable operating state monitoring window. Within this window, the system monitors two core indicators at high frequency: Pressure fluctuation in the soil chamber: The deviation between the real-time pressure value and the target value, as well as the standard deviation of the fluctuation, are calculated using pressure sensors on the soil chamber partition.

[0094] Tunnel boring machine attitude: The rate of change of the tunnel boring machine's pitch and yaw angles is monitored through tilt sensors and a guidance system.

[0095] The system presets permissible thresholds for these two indicators. Only when both indicators remain consistently within the thresholds for the entire monitoring window will the system ultimately determine that a stable operating state has been achieved and generate a ready signal. This signal is a necessary interlock condition for allowing the process to proceed to the next step, S3, tool head positioning. Optionally, specific embodiments of the permissible thresholds include: pressure fluctuation ±0.05 bar, and attitude angle change rate <0.001 degrees / second.

[0096] According to such Figure 3 In the implementation shown, step S2 of the present invention creates a stable and system-ready static working environment for the entire automated tool changing operation, which fully reflects the deep collaboration of multiple subsystems in terms of timing and logic in this solution. This is the fundamental premise for ensuring the safe and smooth implementation of subsequent high-precision and high-risk operations.

[0097] Figure 4 for Figure 1 A flowchart illustrating step S3 in Example 1000. (See attached diagram.) Figure 4 As shown, step S3 includes steps S31-S33.

[0098] In step S31, the system calculates the step-by-step rotation sequence required by the tool head based on the target tool position determined by the tool change decision. Optionally, step S1 is a path planning process with the goal of maximizing efficiency. The system reads the tool change decision generated in S1 and obtains the polar coordinate angles of the tool head for all tools to be changed.

[0099] Optionally, the algorithm in step S31 starts from the current tool head angle and optimizes by minimizing the total rotational travel. It sorts and calculates the angles of all tools to be changed, and calculates the shortest path to all target points. It also considers the rotation direction and mechanical interference to ensure that the tool head rotation does not collide with the fixed structure inside the chamber. Optionally, the rotation direction is usually chosen to minimize the total rotation angle, for example, not exceeding 180°.

[0100] Optionally, step S31 outputs a structured step-by-step rotation sequence instruction, for example: [Sequence 1: Rotate from the current position to 45.2°; Sequence 2: Rotate from 45.2° to 127.8°; ...].

[0101] In step S32, the system controls the cutter head drive system to perform step-by-step rotation according to the step-by-step rotation sequence. In step S32, the system converts the planned point sequence into smooth and precise physical motion. The specific operation process includes: (1) Command issuance: The collaborative control center sends the step-by-step rotation sequence command to the main controller of the tool head drive system. Optionally, the main controller is usually a high-performance frequency converter or servo drive.

[0102] (2) S-curve motion planning: The drive does not simply accelerate and decelerate, but uses an S-shaped speed curve for planning. This ensures that the acceleration of the cutterhead changes continuously when it starts and stops, eliminates rigid impact, and makes the motion extremely smooth, which is crucial for protecting large bearings and maintaining stable pressure in the soil chamber.

[0103] (3) Closed-loop position servo: During the rotation process, the driver receives real-time position feedback from the high-precision encoder of the cutter head spindle, performs closed-loop adjustment, and dynamically corrects the angle error that may be caused by load fluctuations, ensuring that the cutter head reaches each preset target position accurately and smoothly.

[0104] In step S33, after the cutter head rotates to a predetermined angle, the system controls the locking mechanism to engage and fix the cutter head. Achieving the transition from dynamic to ultra-static motion at the end of the movement is crucial for ensuring the reliability of the operational reference. The specific operation process includes: (1) Position trigger: When the encoder confirms that the cutter head has entered the allowable error range of the target position, the main controller immediately sends a [positioning complete, ready to lock] signal.

[0105] (2) Rigid locking: This signal triggers an independent hydraulic or electric servo locking mechanism. Under servo control, the locking pin of this mechanism is precisely inserted into the corresponding precision locking hole on the cutter head drive flange or a special brake disc. Optionally, this process is usually completed within 1-2 seconds.

[0106] (3) Status Confirmation and Interlocking: The displacement and pressure sensors on the locking mechanism provide real-time feedback on whether the locking is in place and reaches the preset locking force. The system will only deactivate the cutter head drive system and send the [Cutter Head Locked] status signal to the entire system after receiving the [Lock Confirmation] signal. This signal is the highest priority safety interlocking condition for the robotic arm system to start operation in step S4.

[0107] According to such Figure 4The illustrated implementation method, with its indexing, positioning, and locking mechanism, provides the following technical advantages: 1. Improved baseline stability of automated operations: By converting continuous motion into step-by-step positioning and supplementing it with physical locking, a highly stable pose reference is provided for robotic arm operations that is far superior to that under continuous rotation conditions, effectively addressing the engineering challenge of high-precision docking of dynamic targets.

[0108] 2. Improved safety and controllability of the operation process: Step-by-step rotation optimizes the motion path and reduces unnecessary idling; rigid locking actively suppresses possible drift or residual movement of the cutter head, enhancing the overall system stability and safety during tool changing operations.

[0109] 3. This step constitutes a key link in the overall system coordination: The precise and stable tool head status achieved in this step, as well as the output [tool head locked] confirmation signal, provides an important state premise and logical basis for the safe and orderly coordinated development of subsequent processes, reflecting the coordinated control design concept that runs through this solution.

[0110] Figure 5 for Figure 1 Example 1000 also includes a flowchart illustrating the visual-assisted correction step SA. (See attached diagram.) Figure 5 As shown, the visual-assisted correction steps SA include steps SA1-SA3.

[0111] Step SA is a precision assurance step performed between the tool head locking process in step S3 and the robotic arm tool change in step S4. Its core function is to serve as a high-precision digital calibration process, using machine vision technology to measure and compensate for minute deviations between the actual physical pose of the tool mount after the tool head is locked and the theoretical model of the control system, thereby ensuring that every gripping and installation operation of the robotic arm is completed accurately.

[0112] In some specific embodiments, in step SA1, the system acquires the actual pose of the target tool mount on the locked tool turret using a vision sensor. The process includes: First, in step SA1, after the system confirms receipt of the [tool head locked] signal, the collaborative control center immediately sends an image acquisition command to the vision subsystem. An environmentally resistant industrial camera deployed in a fixed position within the tool changer or on the robotic arm base is activated. The camera quickly captures a high-resolution two-dimensional image of the area where the target tool mount on the currently locked tool head is located. Simultaneously, if a laser contour scanner or binocular stereo vision system is configured, it will acquire three-dimensional point cloud data of the mount surface. Optionally, the environmentally resistant industrial camera is equipped with an anti-fog, dustproof, and vibration-resistant housing and dedicated lighting.

[0113] Subsequently, the vision processing unit runs the recognition algorithm. First, the algorithm locates the features of the mounting base in the image based on the tool ID in the tool change decision. Then, using the camera calibration model and hand-eye calibration matrix, combined with 3D point cloud data, it calculates the six-degree-of-freedom pose of the mounting base in the current global coordinate system (i.e., three translation coordinates [X, Y, Z] and three rotation angles [Rx, Ry, Rz]), as the actual pose.

[0114] In step SA2, the system compares the actual pose with the theoretical workstation derived from the tool change decision to generate a pose deviation.

[0115] Optionally, in step SA2, the system retrieves the theoretical position pose of the current target tool mount from the tool change decision data package and the tool head digital twin model. This theoretical value is derived based on the tool head design drawings and ideal positioning results.

[0116] Subsequently, the actual pose calculated by the vision system and the theoretical pose of the workstation are transformed to the same reference coordinate system. The system performs matrix operations to directly calculate the pose transformation matrix between the two. This transformation matrix is ​​deconstructed into easily understandable deviation quantities, outputting a six-dimensional deviation vector: [ΔX, ΔY, ΔZ, ΔRx, ΔRy, ΔRz]. This vector clearly indicates the specific offset of the mounting base in position and angle relative to the theoretical values.

[0117] In step SA3, the robotic arm is controlled to perform motion trajectory compensation based on the pose deviation, including: Compensation command generation: The deviation vector [ΔX,ΔY,ΔZ,ΔRx,ΔRy,ΔRz] is immediately sent to the robotic arm controller.

[0118] Trajectory correction: The robotic arm controller does not modify the original work path planned based on the theoretical workstation. Instead, it uses the received deviation vector as a real-time offset and performs positive kinematic superposition at each path point on the original path. This means that when the robotic arm performs each action, the target pose of its end effector is automatically increased by this compensation amount.

[0119] Execution and Verification: The robotic arm moves according to the corrected trajectory. The effect is that even if the actual position of the tool mount deviates from the theoretical position due to factors such as tool head load deformation or slight slippage during locking, the end effector of the robotic arm can still accurately align with and fit the actual mount. After the initial compensation action, the vision system can perform a quick check to ensure that the deviation has been eliminated.

[0120] According to such Figure 5 The embodiment shown in this invention provides a crucial adaptive accuracy layer for the entire automated tool changer system, and its technical effects are mainly reflected in: 1. It effectively solves the common engineering problem of inconsistency between theoretical models and physical reality caused by deformation, thermal expansion, assembly tolerances, and cumulative positioning errors in large mechanical structures. The system no longer relies purely on the absolute accuracy of the initial positioning, but ensures the relative accuracy of the final operating end through real-time measurement and compensation.

[0121] 2. Improved system robustness and job success rate: This step enables the system to withstand minor disturbances. Even if there is a micron-level or sub-millimeter-level drift after the cutter head is locked, visual correction can capture and correct it in time, greatly reducing the risk of job failure due to misalignment and ensuring the smoothness and reliability of the automated process.

[0122] Figure 6 for Figure 1 A flowchart illustrating step S4 in Example 1000. (See attached diagram.) Figure 6 As shown, step S4 includes steps S41-S43.

[0123] In step S41, based on the tool change decision, tool disassembly and installation operation instructions for the robotic arm are generated. Optionally, step S41 converts the high-level task instructions into executable robot motion scripts, including the following processes: First, the system receives and parses the structured tool change decision from S1. The tool_change_sequence included in the decision is a direct input. Then, the motion sequence planning is executed. The planning module generates a specific robotic arm motion sequence based on the operation (REMOVE or INSTALL) of each step in the sequence, the tool_id, and the precise position information provided by S3 / S4. This includes a series of atomic motion commands such as moving to a safe approach point, tool alignment, performing disassembly / installation (with planned torque and speed curves), and moving back to the safe point.

[0124] Subsequently, the system completes resource and security binding, binds the entity information of the new tool from the virtual library, loads the process parameters required for the operation, and finally generates a digital work order that can be directly executed by the robotic arm controller.

[0125] In step S42, the robotic arm is controlled to perform a tool change operation according to the work instructions. Step S42 is the execution of the physical operation, requiring extremely high reliability and environmental adaptability. The specific operation process includes: (1) Environmental confirmation and start-up: The system only allows the robotic arm controller to load and start the work order generated by S41 after receiving the confirmation signals of [Cutter head locked] and [Visual correction completed]; (2) Environmentally resistant operation: The robotic arm moves according to the planned trajectory. Its end effector, including the integrated hydraulic torque wrench and quick-change tool head, actively opens the local air curtain cleaning when approaching the target to blow away the mud and dirt on the mounting surface; Optionally, the robotic arm is a six-axis or seven-axis serial robot that is resistant to dirt and explosion-proof; (3) Force-position hybrid control: During the critical contact stage, the controller switches to the force-position hybrid control mode to find the correct position while maintaining a certain contact force to avoid rigid collision; During disassembly, the system intelligently judges whether the thread is loosened smoothly according to the monitored torque and angle; During installation, the system strictly tightens the thread step by step according to the preset torque curve.

[0126] In step S43, during the execution of the operation command by the robotic arm, a grouting pressure adjustment command is generated in real time according to the action status and sent to the tail grouting system to maintain the stability of the grouting pressure.

[0127] Optionally, the real-time generation of grouting pressure adjustment instructions based on the action status in step S43 specifically includes: monitoring the action stage and load status of the end effector of the robotic arm; predicting the disturbance trend of the cutter replacement operation on the soil chamber and segment lining based on the pre-established mapping relationship and the action stage and load status; and generating a time-sequential feedforward control instruction based on the disturbance trend to drive the shield tail grouting system to adjust the grouting pressure in advance.

[0128] In some specific embodiments, in step S43, the static feedforward control embodiment based on the rule-instruction lookup table generates the grouting pressure adjustment command. Specifically, this includes: real-time monitoring of the action stage and load state of the robotic arm end effector; a pre-built lookup table of (action stage, load state level) -> feedforward command parameters within the system; this table is derived from a large amount of experimental data, and its internal logic is: the combination of a specific action and load state is mapped to a preset standard disturbance trend that is considered to most effectively counteract the typical disturbances in that state; based on the real-time monitored state combination, the corresponding command parameters are directly obtained from the lookup table, and a feedforward control command with a fixed timing shape is immediately generated to drive the grouting system to execute.

[0129] In some specific embodiments, in step S43, adaptive feedforward control based on parameterized transfer functions generates grouting pressure adjustment commands, including: real-time monitoring of the robotic arm's action phase and load status; the system pre-builds transfer function models for different action phases as mapping relationships, such as G(s) = K * e^(-τs) / (Ts+1). The initial values ​​of parameters such as gain K, time constant T, and delay τ are determined experimentally. During online operation, the system uses recently monitored load change data and actual grouting pressure response data, employing algorithms such as recursive least squares, to identify and update the most suitable K, T, and τ parameters under the current working condition in real time. Using the updated model, with the real-time load dF / dt as input, the system dynamically calculates the detailed pressure disturbance trend curve ΔP_pred(t) for the next few seconds. Based on the dynamically predicted disturbance trend ΔP_pred(t), its reverse curve is generated as a time-sequential feedforward command -Kp *ΔP_pred(t) (Kp is the feedforward gain), driving the grouting system to act in advance.

[0130] According to such Figure 6 The embodiments shown in the present invention achieve the following technical effects: By translating high-level decisions into reliable robotic action sequences and equipping them with environmental perception and adaptive control, it becomes possible to complete high-precision assembly operations such as tool changing within the damp, dusty, and space-constrained tunnel boring machine chamber, fundamentally replacing high-risk manual labor. Through a feedforward linkage control mechanism, the robot can anticipate the environmental disturbances caused by its own actions and initiate compensation in advance. This successfully resolves the long-standing inherent contradiction in automated tool changing: mechanical operations inevitably generate disturbances, while tunnel stability requires the avoidance of disturbances. This mechanism unifies the two, achieving static stability during dynamic operations.

[0131] Figure 7 This is a schematic diagram of an embodiment 2000 of a collaborative control system for online cutter replacement in a tunnel boring machine according to the present invention. Figure 7 As shown, the system of embodiment 2000 includes an intelligent decision-making module 201, a collaborative control center 202, a tunneling and attitude stabilization module 203, a cutterhead precision positioning module 204, an automatic tool change and linkage execution module 205, a quality verification module 206, and a data management and process control module 207.

[0132] In some specific embodiments: The intelligent decision-making module 201 uses multi-source sensor data from the tunnel boring machine (TBM) operation to determine the cutter wear status and generate cutter replacement decisions through an artificial intelligence model. The collaborative control center 202 responds to the cutter replacement decisions and coordinates the control of the following modules. The tunneling and attitude stabilization module 203 automatically adjusts tunneling parameters according to the instructions from the collaborative control center to bring the TBM into a stable operating state and establishes a pressure buffer for the soil chamber.

[0133] The cutterhead precision positioning module 204 controls the cutterhead to perform indexing, positioning, and locking based on instructions and cutter change decisions from the collaborative control center. The automatic cutter change and linkage execution module 205 controls the robotic arm to perform the cutter change operation after the cutterhead is locked, and controls the robotic arm to link with the tail grouting system to maintain stable grouting pressure during cutter change. The quality verification module 206 verifies the tightening torque and performs visual verification on the newly installed cutters. The data management and process control module 207 records cutter change data, guides subsequent operation processes, and, upon completion, instructs the tunnel boring machine to resume normal tunneling.

[0134] The embodiments of the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, any changes or modifications made by those skilled in the art based on the ideas of the present invention, its specific implementation methods, and its application scope, are all within the scope of protection of the present invention. In summary, the content described in this specification should not be construed as a limitation of the present invention.

Claims

1. A collaborative control method for online cutter replacement in a tunnel boring machine, characterized in that, include: S1. Based on multi-source sensor data from the operation of the tunnel boring machine, an artificial intelligence model is used to determine the wear status of the cutting tools and generate a tool replacement decision. S2. In response to the cutterhead change decision, the tunneling parameters are automatically adjusted to bring the tunnel boring machine into a stable state and to establish a pressure buffer for the soil chamber; S3. Control the tool turret to perform indexing positioning and lock according to the tool change decision; S4. After the cutterhead is locked, control the robotic arm to perform the cutter replacement operation, and control the robotic arm to link with the shield tail grouting system to maintain stable grouting pressure during cutter replacement; S5. Verify the tightening torque and perform visual inspection on the newly installed cutting tools; S6. Record the cutterhead change data and guide subsequent operations. Once all operations are completed, resume normal tunneling of the tunnel boring machine.

2. The method according to claim 1, characterized in that, Step S1 includes: Collect multi-source sensor data from the tunnel boring machine; The multi-source sensor data is input into a pre-trained artificial intelligence model for analysis to obtain the tool wear state assessment result; Based on the wear condition assessment results, a tool replacement decision is generated, which includes the identifier of the tool to be replaced and the replacement order.

3. The method according to claim 2, characterized in that, The multi-source sensor data includes cutterhead drive current timing data, tunneling rate variation data, and vibration spectrum data; the artificial intelligence model is configured to perform feature extraction and fusion analysis on the multi-source sensor data to output the wear state assessment result.

4. The method according to claim 3, characterized in that, The feature extraction and fusion analysis specifically includes: Extract trend features characterizing load fluctuations from the timing data of the cutter head drive current; Extract specific frequency band energy features related to tool wear from the vibration spectrum data; The trend characteristics, the tunneling rate change data, and the energy characteristics are weighted and fused, and the wear level of the cutting tool is determined based on the fusion result.

5. The method according to claim 1, characterized in that, In step S2, automatically adjusting the tunneling parameters to bring the tunnel boring machine into a stable state includes the following steps: Based on the geological exploration data and real-time tunneling parameters of the current tunneling section, identify the current geological type; Based on the identified geological type, the target range for stable operation of the total thrust and the cutterhead drive torque is matched and calculated from a preset database; Control the propulsion system and cutterhead drive system of the tunnel boring machine to gradually adjust the total thrust and cutterhead drive torque to the target range for machine stability; While adjusting the tunneling parameters, a preparatory command is sent to the tail grouting system. The preparatory command is used to switch the grouting system to pressure stabilization mode in preparation for responding to subsequent linkage control. The pressure fluctuations in the soil chamber and the attitude of the tunnel boring machine are monitored. When both stabilize within the permissible threshold within a set time, the tunnel boring machine is determined to have entered a stable operating state.

6. The method according to claim 1, characterized in that, Step S3, controlling the tool turret to perform indexing positioning and lock according to the tool change decision, includes the following steps: Based on the target tool position determined by the tool change decision, calculate the step-by-step rotation sequence that the tool head needs to execute; Based on the step-by-step rotation sequence, the tool head drive system is controlled to perform step-by-step rotation; After the cutter head rotates to a predetermined angle, the locking mechanism is activated to fix the cutter head in place.

7. The method according to claim 1, characterized in that, Prior to step S4, a visual-assisted correction step is also included, comprising: The actual position and orientation of the target tool mount on the locked tool turret are obtained through a vision sensor. The actual pose is compared with the theoretical workstation derived from the tool change decision to generate a pose deviation. The robotic arm is controlled to compensate for the motion trajectory based on the posture deviation.

8. The method according to claim 1, characterized in that, Step S4 includes the following steps: Based on the tool change decision, the tool disassembly and installation operation instructions of the robotic arm are generated; The robotic arm is controlled to perform a tool changing operation according to the work instructions. During the execution of the operation instructions by the robotic arm, a grouting pressure adjustment instruction is generated in real time based on the action status and sent to the tail grouting system to maintain stable grouting pressure.

9. The method according to claim 8, characterized in that, The real-time generation of grouting pressure adjustment commands based on the action status specifically includes: Monitor the action phase and load status of the robotic arm's end effector; Based on the pre-established mapping relationship, the disturbance trend of the cutterhead replacement operation on the soil chamber and segment lining is predicted according to the action stage and load state. Based on the disturbance trend, a time-sequential feedforward control command is generated to drive the tail grouting system to adjust the grouting pressure in advance.

10. A collaborative control system for online cutter replacement in a tunnel boring machine for implementing the method of any one of claims 1-9, characterized in that, include: The intelligent decision-making module is used to determine the wear status of the cutters and generate cutter replacement decisions based on multi-source sensor data from the operation of the tunnel boring machine through an artificial intelligence model. The collaborative control center is used to respond to the tool change decision and coordinate the control of the following modules; The tunneling and attitude stabilization module is used to automatically adjust the tunneling parameters according to the instructions of the collaborative control center to enable the tunnel boring machine to enter a stable state and to establish a pressure buffer for the soil chamber. The tool head precision positioning module is used to control the tool head to perform indexing positioning and lock it according to the instructions and tool changing decisions of the collaborative control center. The automatic tool changer and linkage execution module is used to control the robotic arm to perform tool changing operations after the cutter head is locked, and to control the robotic arm to link with the shield tail grouting system to maintain stable grouting pressure during tool changing. The quality verification module is used to verify the tightening torque and perform visual verification on newly installed cutting tools; The data management and process control module is used to record cutter change data and guide subsequent operation processes, and to instruct the tunnel boring machine to resume normal tunneling after all operations are completed.

Images