Multi-dimensional sensing and self-adaptive tool changing method and system for shield tunneling machine under low-speed pressure maintaining

Abstract

The invention discloses a multi-dimensional sensing and self-adaptive tool changing method and system for a shield tunneling machine under low-speed pressure maintaining. The tool changing system comprises a sensing layer, a decision-making layer and a control layer. The tool changing method comprises the steps that three-dimensional comprehensive data are obtained through a sensor; the three-dimensional comprehensive data are fused through a sensor data fusion module and sent into a multi-modal space-time diagram neural network MST-GNN, and tool changing priorities, tool changing type labels, the tool changing number and tool changing numbers are output; instantaneous positioning is conducted on the target tool through the tool changing type label; and angle error calculation is carried out, a zero relative window is confirmed, and tool changing is achieved. In order to cope with a tedious process that a traditional method must completely stop and release pressure, the invention proposes that under the micro shutdown working condition of low speed of a cutter head and pressure maintaining of a soil bin, defect-service life-geological three-dimensional perception, millisecond decision, RFID instantaneous positioning and five spline closed-loop tool changing are fused, shutdown pressure release and same-speed ring rotation are not needed, and the tool changing duration is shortened from hour to minute.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of shield machine tunneling and cutter changing, in particular to a multi-dimensional perception and self-adaptive cutter changing method and system under low-speed pressure maintaining of a shield machine. BACKGROUND

[0002] Generally, shield tunnels need to pass through complex strata such as uneven soft and hard, high-erosion boulders, soft on top and hard on bottom, and developed karst caves.

[0003] Based on the complex strata in the tunnel, the shield machine needs to replace the cutter for tunneling. The existing shield machine cutter changing technology has the following problems:

[0004] 1) The traditional cutter changing technology relies on complete rotation of the cutter head and pressure relief of the soil chamber, resulting in interruption of tunneling for several hours, forced extension of the construction window, and doubling of labor and equipment costs, and the risk of pressure relief operation in a high-gas and water-rich environment is extremely high;

[0005] 2) After shutdown, manual visual inspection or disassembly inspection is generally used, which cannot capture multiple defects such as wear, cracks, corrosion, chipping and rock layer mutation in real time;

[0006] 3) There is a lack of real-time fusion model of three-dimensional data of defects-life-geology, and the timing of cutter changing is only based on experience, which causes waste of cutters too early or causes cutter chipping, cutter head jamming and even ground subsidence too late;

[0007] 4) The positioning link requires an additional same-speed rotating ring or a large mechanical compensation mechanism, which has complex structure, large weight, many fault points, time-consuming and expensive maintenance, and is harsh on cutter head space, which is not conducive to compact design. SUMMARY

[0008] The purpose of the present application is to provide a multi-dimensional perception and self-adaptive cutter changing method and system under low-speed pressure maintaining of a shield machine, which can provide a unified, accurate and efficient cutter changing method for the shield machine, in order to solve the problem of few cutter changing methods for shield machines under complex ground conditions and lack of pertinence.

[0009] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme: a multi-dimensional perception and self-adaptive cutter changing method under low-speed pressure maintaining of a shield machine, the cutter changing method comprising the following steps:

[0010] S1. In the perception layer, three-dimensional comprehensive data of the cutter are acquired in real time by the sensor;

[0011] S2. In the decision layer, the three-dimensional comprehensive data are fused by a sensor data fusion module in the perception layer and sent to a multi-modal spatio-temporal graph neural network MST-GNN, and the cutter changing priority, cutter changing type label, cutter changing number and cutter changing number are outputted;

[0012] S3. In the control layer, the target tool is instantaneously positioned by the tool changing type label and in cooperation with the low-speed rotation state of the cutter head;

[0013] S4. In the control layer, when the tool robot is synchronized with the target tool angular velocity, angular error calculation is performed to confirm the zero relative window, so that the tool robot completes the tool pulling, tool insertion and locking, and realizes the tool changing.

[0014] As a further scheme of the application: in the perception layer, the three-dimensional comprehensive data includes wear defect data A1, corrosion defect data A2 and fracture defect data A3 in the defect dimension A, rock hardness data B1 in the geological hardness dimension B, and service life data C1 in the service life dimension C;

[0015] Wherein, the wear defect data A1 is obtained by an electromagnetic mutual inductance wear sensor to obtain the radial wear amount ΔW of the cutter ring;

[0016] The corrosion defect data A2 is obtained by an electrochemical impedance spectroscopy corrosion sensor to obtain the corrosion area percentage ΔC of the bearing sealing area;

[0017] The fracture defect data A3 is obtained by an acoustic emission crack sensor to obtain the crack length ΔL of the cutter shaft or the acoustic emission energy cumulative value AE;

[0018] The rock hardness data B1 is obtained by an advanced geological radar antenna to obtain the uniaxial compressive strength UCS of the rock stratum;

[0019] The service life data C1 is obtained by a strain type load cycle counter to obtain the equivalent load cycle number Neq;

[0020] The three-dimensional comprehensive data is processed by a sensor data fusion module to output an N×5×T space-time tensor for decision-making by the decision layer.

[0021] As a further scheme of the application: in the perception layer, in the micro-stop working condition of the cutter head rotating at a low speed of ≤2 rpm and the soil bin being pressure-maintained, the necessity of tool changing is judged based on the defect dimension A, the geological hardness dimension B and the service life dimension C, which has:

[0022] In the defect dimension A, the tool defects are divided into:

[0023] The wear defect data A1 takes the radial wear amount ΔW of the cutter ring as an index, and when ΔW≥25mm, it is determined that A1 reaches the tool changing threshold;

[0024] The corrosion defect data A2 takes the corrosion area percentage ΔC of the bearing sealing area as an index, and when ΔC≥30%, it is determined that A2 reaches the tool changing threshold;

[0025] Fracture defect data A3, with the length of the cutter shaft crack ΔL or the acoustic emission energy cumulative value AE as the index, when ΔL≥15mm or AE≥100mV·ms, it is determined that A3 defect reaches the cutter replacement threshold;

[0026] Wherein, if any tool defect reaches its threshold, the cutter replacement instruction of the defect dimension A is triggered;

[0027] In the geological hardness dimension B, the geological hardness grade is defined:

[0028] Soft rock: UCS≤30Mpa, replaced with a scraper type cutter;

[0029] Medium-hard rock: 30MPa<UCS≤90MPa, replaced with a roller type cutter;

[0030] Extremely hard rock: UCS>90MPa, heavy roller type cutter;

[0031] When the UCS value obtained by the forward-looking geological radar antenna or the cutter head vibration inversion crosses from a low grade to a high grade, or the continuous excavation length in the extremely hard rock grade is greater than or equal to 10m, the cutter replacement instruction of the geological hardness dimension B is triggered, and the cutter type is replaced according to the geological hardness grade;

[0032] In the life dimension C, the service life of the cutter is defined as the equivalent load cycle number Neq, and when Neq≥design rated life Nr, the cutter replacement instruction of the life dimension C is triggered.

[0033] As a further scheme of the application: in the perception layer, a sensor system is arranged on the cutter head and cutter of the shield tunneling machine for real-time acquisition of three-dimensional comprehensive data of the defect dimension A, the geological hardness dimension B and the life dimension C, and is arranged and connected as follows:

[0034] In the defect dimension A:

[0035] The wear monitoring module is embedded with two groups of electromagnetic mutual inductance wear sensors in the circumferential direction, same radius and equal angle interval inside the cutter ring base body of each replaceable roller on the front surface of the cutter head, for real-time output of the cutter ring radial wear amount ΔW;

[0036] The corrosion monitoring module: a group of electrochemical impedance spectroscopy corrosion sensors are attached to the inner surface of the bearing sealing end cover of each roller, for real-time output of the bearing sealing area corrosion area percentage ΔC;

[0037] The fracture monitoring module: a group of acoustic emission crack sensors are pasted on the transition fillet area of the cutter shaft of each roller, for real-time capture of the acoustic emission energy cumulative value AE of crack initiation and expansion;

[0038] In the geological hardness dimension B: at least three groups of ground penetrating radar antennas are arranged equidistantly along the circumferential direction of the outer periphery of the cutter head, which are used to emit electromagnetic waves in front of the tunnel face before the shield tunneling machine is excavated and receive reflected signals to obtain the uniaxial compressive strength UCS of the rock stratum;

[0039] In the service life dimension C: at least one corresponding strain load cycle counter is symmetrically pasted on the surface of the cutter shaft of each roller cutter along the axial centerline, which is used to collect the torque and thrust load spectrum of the roller cutter in the excavation process in real time and calculate the equivalent load cycle number Neq;

[0040] The sensor data fusion module is arranged in the explosion-proof electric control box at the cutter head central rotary joint, and includes a multi-channel synchronous sampling unit, a time synchronization unit and an edge computing unit;

[0041] The multi-channel synchronous sampling unit is connected with A1, A2, A3, B1 and C1 through shielded cables, and the sampling frequency is greater than or equal to 1 kHz;

[0042] The time synchronization unit receives a 1PPS second pulse signal from the main control PLC of the shield tunneling machine, which is used to timestamp the three-dimensional comprehensive data, and the time synchronization error is less than or equal to 1 ms.

[0043] As a further scheme of the application: in the perception layer, a multi-dimensional state matrix is constructed based on the three-dimensional comprehensive data, and the "isolated state of a single cutter" is upgraded to a global coupling relationship network of "cutter head-cutter-geology-time" through graph structure modeling, which specifically includes:

[0044] Taking N replaceable cutters on the cutter head as nodes, and generating a five-dimensional state vector for each cutter i in real time:

[0045] S it =[wear corrosion fracture geological hardness service life] T

[0046] =[ΔW it ΔC it ΔL it / AE it UCS it Neq it ] T

[0047] Wherein, ΔW it , ΔC it , ΔL it / AE it , UCS it and Neq it are directly provided by A1, A2, A3, B1 and C1, i represents the serial number of the cutter, and t represents a certain moment of monitoring;

[0048] 2) Stack the five-dimensional state vectors of the N tools in order of tool number to form an N×5-dimensional global state matrix:

[0049]

[0050]

[0051] 3) Using the hobs as nodes, with node features being the aforementioned five-dimensional state vector, and the mechanical coupling relationship between the hobs and the rotational dynamics of the cutter head as edges, construct an undirected graph G = (V, E), where the edge weight formula in the undirected graph G is as follows:

[0052]

[0053] Edge weight Q ij The physical distance d between hobs i and j on the cutter head ij The speed of the cutter head, ω, and σ and α are determined together, and are preset hyperparameters.

[0054] 4) Introduce a time dimension into the node features of the undirected graph G, and transform the state matrix M over T consecutive sampling periods. t Stacked as N×5×T spacetime tensors.

[0055] As a further aspect of the present invention: in the decision-making layer, the N×5×T spatiotemporal tensor is processed by a multimodal spatiotemporal graph neural network (MST-GNN) to obtain the tool-changing decision result, specifically including:

[0056] The N×5×T spatiotemporal tensor is modally decomposed into five-channel tensors: wear channel, corrosion channel, fracture channel, geological hardness channel, and lifetime channel. LayerNorm is applied independently to each channel to obtain the normalized five-channel feature tensor.

[0057] X c ∈R N×T c∈{1,…,5};

[0058] Spatial Graph Convolutional Network (S-TCN) uses the adjacency matrix of an undirected graph G as a mask to perform graph convolution at each time point τ∈{1,…,T}:

[0059]

[0060] Where l represents the layer index and D is the degree matrix. These are trainable spatial weights; after passing through the Ls layer, the spatial feature tensor is obtained. ds represents the spatial embedding dimension;

[0061] The temporal causal convolutional layer T-TCN performs dilated causal convolution along the temporal dimension for each node i:

[0062] Z i =CausalConv1D(H s [i,:,:];W t ,k t ,d t )

[0063] Where, k t d is the kernel size. t As the expansion factor, via L s The time feature tensor is obtained after the layer. T′ is the time step after downsampling;

[0064] Multimodal attention fusion: After linearly mapping the five-channel feature tensor, calculate the cross-modal attention weights.

[0065]

[0066] Among them, Q c K c V c They are query, key-value mapping, and output fused features respectively.

[0067] The results are output through the task output header. The output results include:

[0068] Tool change priority P∈[0,1] N ;

[0069] Tool replacement type label T∈{wear type, corrosion type, fracture type, geological adaptability type, end-of-life type} N ;

[0070] Tool change count K = Round! (∑) i σ(P i ≥θ)), the threshold θ is adjusted in real time by engineering constraints;

[0071] Tool change number set ID = {i | P i ≥θ}.

[0072] As a further aspect of the present invention: in the control layer, instantaneous positioning of the target tool includes the following steps:

[0073] An absolute encoder is coaxially mounted on the end face of the tool turret spindle, and the output signal is transmitted to the tool robot controller in real time via an EtherCAT bus.

[0074] Each replaceable hob has a passive UHF-RFID tag embedded inside its base. The passive UHF-RFID tag contains a UID and a fixed offset Δθ. i , where Δθ iThe mechanical angle difference between the tool center line and the encoder zero position;

[0075] The tool robot controller performs a one-time scan after power-on: through slow rotation of the tool disc, the RFID read-write antenna array reads each tool UID and Δθ in turn i An index array Index[N] = {Δθ1, Δθ2, …, ΔθN} is established in the tool robot controller memory, which cannot be powered off N};

[0076] At any t0 moment, the real-time angle of the tool disc is read out by the absolute value encoder The instantaneous target angle coordinate of the Nth tool is:

[0077]

[0078] Dynamic compensation and closed loop, specifically including: the tool robot controller inputs θ target (t0) to the quintic spline trajectory planner, generating an angular displacement-time curve of the end effector.

[0079] As a further scheme of the application: in the control layer, the instantaneous target angle coordinate θ target (t) is obtained by instantaneous positioning of the target tool, and the tool robot is pre-planned, speed-chased, and error-closed loop controlled under the micro-stop tool changing working condition, which includes:

[0080] In the tool robot joint space, with the current joint angle vector q curr as the starting point and the target joint angle vector q target as the terminal point, a quintic polynomial trajectory q(t) is generated for quintic spline trajectory pre-planning, and its expression is:

[0081] q(t) = a0 + a1t + a2t 2 +a3t 3 +a4t 4 +a5t 5 ,(0≤t≤Δt pred )

[0082] The boundary conditions that the quintic polynomial trajectory q(t) must satisfy at the starting point and the terminal point include:

[0083] Starting from the current actual joint angle of the robot, position jump is avoided:

[0084] q(0) = q curr

[0085] Accurately reach the target tool angle within the specified time Δt pred , achieving "zero relative" alignment:

[0086] q(Δt pred )=q target

[0087] Start velocity, acceleration is zero, ensure smooth start:

[0088]

[0089] End velocity must match with tool angular velocity ω, make robot end and tool relative static:

[0090]

[0091] End acceleration is zero, avoid mechanical impact or residual vibration:

[0092]

[0093] Where ω is tool instantaneous angular velocity, calculated by encoder differential in real time:

[0094]

[0095] Spline solver automatically give theoretical time Δt pred to reach target according to joint maximum velocity v max and maximum acceleration a max ;

[0096] Do velocity feedforward and angular synchronization, controller take -ω as end angular velocity feedforward command, ensure relative angular velocity Δω≈0 between robot end and target tool at t=Δt pred , after reaching target angle, system switch to angular synchronization mode;

[0097] 4) Complete velocity feedforward and angular synchronization, do error closed loop and reentry mechanism, real-time angular error is:

[0098] Δθ(t)=θ target (t)-θ end-effector (t)

[0099] When |Δθ(t)|≤0.5° and holding time >50ms, judge zero relative window established, allow tool changer action;

[0100] If |Δθ(t)|>0.5° or joint torque mutation >15N·m, immediately trigger reentry: tool robot decelerate to zero, record q reentry ; recalculate θ target (t).

[0101] As a further scheme of the application: in the control layer, once the zero relative window is established and confirmed, the tool robot controller immediately starts the following tool changing sequence, and the process specifically includes:

[0102] After the zero relative window is confirmed, the locking state detection is performed, the normal force Fn of the tool locking surface is monitored in real time through the six-dimensional force sensor, and only when Fn≤2N and lasts for more than 50 ms, it is determined that the locking force has been released, and the tool is allowed to be pulled out;

[0103] After the electromagnetic / pneumatic lock is loosened, the Z-axis electric sliding table is pushed out by 50 mm at a speed of 100 mm / s to complete the tool pulling out;

[0104] The end-of-travel photoelectric switch is detected, and if it is not in place within 3 seconds, it is interrupted and an alarm is given;

[0105] The tool robot grabs a new tool, and the Z-axis sliding table is inserted into the tool seat in reverse at a speed of 80 mm / s;

[0106] After insertion, a locking torque of 180 N·m is applied, and the torque closed loop tolerance is ±5 N·m;

[0107] After locking is completed, the six-dimensional force sensor confirms that the normal force Fn' is greater than or equal to 800 N, ensuring locking;

[0108] The end UHF-RFID reader reads the new tool UID and compares it with the tool robot controller array, and after verification, the life counter is cleared; if UID verification fails twice, it is marked as an exception.

[0109] A tool changing system for multi-dimensional perception and adaptive tool changing of a shield machine at low speed and pressure maintaining, the tool changing system comprising:

[0110] The perception layer perceives the tool state and collects data, including an electromagnetic mutual inductance wear sensor, an electrochemical impedance spectroscopy corrosion sensor, an acoustic emission crack sensor, a ground penetrating radar antenna, a strain load cycle counter, and a sensor data fusion module;

[0111] The decision layer is used for intelligent judgment of the three-dimensional comprehensive data obtained by the perception layer, and generates instructions through a multi-modal spatio-temporal graph neural network MST-GNN, outputs the tool changing priority, the tool changing type label, the tool changing number and the tool changing number, and sends them to the control layer through the EtherCAT bus;

[0112] The control layer performs instantaneous positioning of the tool to be replaced, and performs trajectory closed loop on the tool robot according to the position, and completes the tool changing action, including the tool robot, the absolute value encoder, the passive UHF-RFID tag, the tool robot controller and the six-dimensional force sensor.

[0113] The beneficial effects of the present application are:

[0114] 1) Through the micro-stop design of low-speed rotation of the cutter head and pressure maintenance of the soil bin, the traditional lengthy process of complete stop and pressure relief is completely abandoned, the interruption time of single tool replacement is compressed from hours to minutes, and the tunnel excavation period is significantly shortened;

[0115] 2) The three-dimensional perception system of defect-life-geology replaces manual visual inspection, and real-time fusion of electromagnetic mutual inductance, electrochemical impedance, acoustic emission, geological radar and strain gauge data makes the wear, corrosion, cracks and changes in rock hardness obvious, the accuracy of tool replacement decision is improved by more than one time, and the economic loss and safety risk caused by premature scrap or sudden blade collapse are avoided;

[0116] 3) The absolute value rotary encoder cooperates with the passive UHF-RFID tag to realize millisecond-level relative static window, and the robot can accurately pull and insert the tool without rotating ring at the same speed, the whole machine structure is lighter, the reliability is higher, and the maintenance amount is smaller;

[0117] 4) Through the dynamic adjustment of threshold value by multi-modal spatio-temporal graph neural network MST-GNN, the scraper, cutter or heavy cutter can be automatically matched according to different strata, the tool utilization rate is improved by more than 30%, and the spare parts inventory and transportation cost are reduced, which provides an integrated solution of high efficiency, safety and economy for complex geological long-distance tunnel construction;

[0118] 5) The present application proposes to fuse the three-dimensional perception of defect-life-geology, millisecond decision, RFID instantaneous positioning and five times spline closed loop tool replacement under the micro-stop working condition of low-speed cutter head and pressure maintenance of soil bin, without stopping and pressure relief, without rotating ring at the same speed, and the tool replacement time is reduced from hours to minutes. BRIEF DESCRIPTION OF DRAWINGS

[0119] Figure 1 It is a schematic diagram of the overall process structure of the present application;

[0120] Figure 2 It is a signal transmission flowchart of the present application;

[0121] Figure 3 It is a tool replacement type, sensor selection and position distribution analysis diagram of the present application;

[0122] Figure 4 It is a decision layer overall system flowchart and framework diagram of the present application;

[0123] Figure 5 It is a control layer overall system flowchart and framework diagram of the present application. DETAILED DESCRIPTION

[0124] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be clearly and completely described below, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work belong to the scope of protection of the present application.

[0125] Embodiment one, as shown in Figure 1 and Figure 2 , the embodiment provides a multi-dimensional perception and adaptive tool changing method of shield machine under low-speed pressure maintaining, which is based on the low-speed rotation of the cutter head (rotation), the micro-stop of the soil bin pressure maintaining, and divides the whole process into perception layer, decision layer and control layer, and specifically includes the following steps:

[0126] Firstly, in the perception layer, under the micro-stop working condition of the cutter head with ≤2rpm low-speed rotation and the soil bin pressure maintaining, the necessity of tool changing is judged based on the defect dimension A, the geological hardness dimension B and the service life dimension C.

[0127] The necessity of tool changing is judged through the above three dimensions, and is executed in the following manner, as shown in Figure 3 :

[0128] 1) In the defect dimension A, the tool defects are divided into:

[0129] 11) wear defect data A1, taking the cutter ring radial wear amount ΔW as the index, when ΔW≥25mm, it is judged that A1 reaches the tool changing threshold;

[0130] 12) corrosion defect data A2, taking the bearing sealing area corrosion area percentage ΔC as the index, when ΔC≥30%, it is judged that A2 reaches the tool changing threshold;

[0131] 13) fracture defect data A3, taking the tool shaft crack length ΔL or the acoustic emission energy cumulative value AE as the index, when ΔL≥15mm or AE≥100mV·ms, it is judged that A3 defect reaches the tool changing threshold;

[0132] Wherein, if any tool defect reaches its threshold, the tool changing instruction of the defect dimension A is triggered.

[0133] 2) In the geological hardness dimension B, the geological hardness grade is defined:

[0134] 21) soft rock: UCS≤30Mpa, replace with scraper type cutter;

[0135] 22) medium-hard rock: 30MPa<UCS≤90MPa, replace with roller type cutter;

[0136] 23) extremely hard rock: UCS>90MPa, heavy-duty roller type cutter;

[0137] When the UCS value obtained by the advanced geological radar antenna (or cutterhead vibration inversion) in real time or the cutterhead vibration inversion crosses from a lower grade to a higher grade, or the continuous excavation length of the extreme hard rock grade is ≥10 m, the cutter replacement instruction of the geological hardness dimension B is triggered, and the cutter type is replaced according to the geological hardness grade.

[0138] 3) In the service life dimension C, the service life of the cutter is defined as the equivalent load cycle number Neq, and when Neq≥ the design rated life Nr, the cutter replacement instruction of the service life dimension C is triggered.

[0139] It should be noted that ΔW, ΔC, ΔL, AE, UCS and Neq are all collected in real time by the sensor system arranged on the cutterhead and the cutter and processed synchronously by the sensor data fusion module.

[0140] Secondly, in the perception layer, in order to obtain three-dimensional comprehensive data of the defect dimension, the geological hardness dimension and the service life dimension in real time, a sensor system is arranged on the cutterhead and the cutter of the shield tunneling machine, and the sensor system comprises:

[0141] 1) Obtain wear defect data, corrosion defect data and crack defect data in the defect dimension A

[0142] 11) Wear monitoring module: at least two groups of electromagnetic mutual inductance wear sensors are embedded in the cutter ring base body of each replaceable hob on the front surface of the cutterhead along the circumferential direction, the same radius and the same angle interval, the at least two groups of electromagnetic mutual inductance wear sensors are arranged in an upper and lower staggered manner in the thickness direction of the cutter ring, the staggered distance is 30%-40% of the thickness of the cutter ring, to form a differential measurement structure, for real-time output of the radial wear amount ΔW of the cutter ring;

[0143] 12) Corrosion monitoring module: at least one group of electrochemical impedance spectroscopy corrosion sensors are attached to the inner surface of the bearing sealing end cover of each hob, wherein the probe of the electrochemical impedance spectroscopy corrosion sensor and the metal surface of the end cover maintain an electrolyte gap of 0.1mm-0.3mm, and are isolated from the external mud by a sealing rubber ring, for real-time output of the corrosion area percentage ΔC of the bearing sealing area;

[0144] 13) Fracture monitoring module: at least one group of acoustic emission crack sensors are pasted on the transition fillet area of the cutter shaft of each hob; the center frequency of the acoustic emission crack sensor is 150kHz±10%, and the acoustic emission crack sensor is tightly attached to the surface of the cutter shaft through a high-temperature resistant coupling agent, for real-time capture of the acoustic emission energy cumulative value AE of crack initiation and propagation.

[0145] 2) Obtain rock hardness data in the geological hardness dimension B

[0146] Three groups of advanced geological radar antennas are arranged at equal intervals along the circumferential direction of the cutter head. The advanced geological radar antennas are shielded dipole antennas, the radiation surfaces of which are flush with the front surface of the cutter head, and the antennas are isolated from the external rock and soil through wear-resistant ceramic shells. The three groups of antennas work in time division, forming a 120° sector scan. The antennas are used to emit electromagnetic waves with a center frequency of 100 MHz in the range of 0-30 m in front of the tunnel face before the shield tunneling machine starts tunneling, and receive reflected signals, so as to obtain the uniaxial compressive strength UCS of the rock in front.

[0147] 3) Obtain life data in life dimension C

[0148] On the surface of the cutter shaft of each roller cutter, at least one corresponding strain load cycle counter is symmetrically pasted along the axial centerline. The strain load cycle counter is a full-bridge strain gauge group, the measurement direction of which is consistent with the direction of the maximum principal stress of the cutter shaft, and is protected through a waterproof glue layer and a metal foil shielding layer. The strain load cycle counter is used to collect the torque and thrust load spectrum of the roller cutter in the tunneling process in real time, and calculate the equivalent load cycle number Neq.

[0149] 4) Sensor data fusion module: arranged in the explosion-proof electric control box at the cutter head central rotary joint, including a multi-channel synchronous sampling unit, a time synchronization unit and an edge computing unit.

[0150] The multi-channel synchronous sampling unit is connected with A1, A2, A3, B1 and C1 through shielded cables, and the sampling frequency is ≥1 kHz;

[0151] The time synchronization unit receives a 1PPS second pulse signal from the main control PLC of the shield tunneling machine, and is used to timestamp all three-dimensional integrated data, and the time synchronization error is ≤1 ms;

[0152] The edge computing unit performs real-time calculation and caching on ΔW, ΔC, AE, UCS and Neq, and uploads the three-dimensional integrated data to the cutter robot controller through an EtherCAT bus.

[0153] Thirdly, in the perception layer, a multi-dimensional state matrix is constructed based on the real-time collected three-dimensional integrated data, and the “isolated state of a single cutter” is upgraded to a global coupling relationship network of “cutter head-cutter-geology-time” through graph structure modeling. This step specifically includes:

[0154] 1) Taking N replaceable cutters on the cutter head as nodes, and generating a five-dimensional state vector for each cutter i in real time:

[0155] S it =[wear corrosion fracture geological hardness life] T

[0156] =[ΔW it ΔC it ΔLit / AE it UCS it New it ] T

[0157] where, ΔW it , ΔC it , ΔL it / AE it , UCS it and Neq it are directly provided by A1, A2, A3, B1 and C1 respectively, i represents the serial number of the cutter, and t represents a certain moment of monitoring;

[0158] 2) Stack the five-dimensional state vectors of N cutters in the order of cutter number to form an N*5-dimensional global state matrix:

[0159]

[0160] 3) Take the hob as the node, the node feature as the above-mentioned five-dimensional state vector, the mechanical coupling relationship between the hobs and the rotating dynamics of the cutter head as the edge, and construct an undirected graph G=(V, E), wherein the edge weight formula in the undirected graph G is as follows:

[0161]

[0162] The edge weight Q ij is determined by the physical distance d ij between the hob i and the hob j on the cutter head and the rotating speed ω of the cutter head, and σ and α are preset hyperparameters;

[0163] 4) Introduce the time dimension into the node feature of the undirected graph G, and stack the state matrices M t of the continuous T sampling periods into an N*5*T spatio-temporal tensor to provide "zero delay" raw data for subsequent decision-making.

[0164] Fourthly, in the decision-making layer, the N*5*T spatio-temporal tensor is processed by a multi-modal spatio-temporal graph neural network MST-GNN to obtain a tool changing decision result. The mechanical coupling between the cutters is captured by a spatial graph convolution S-TCN, and the evolution trend of the defects-geology-life is captured by a time causal convolution T-TCN, as shown in Figure 4 , which specifically includes:

[0165] 1) In the input adaptation, the N*5*T spatio-temporal tensor is split into five channel tensors according to the modalities, which are the wear channel, the corrosion channel, the fracture channel, the geological hardness channel and the life channel, and LayerNorm is independently performed on each channel to obtain a normalized five-channel feature tensor:

[0166] Xc ∈ R N×T ,c∈{1,…,5};

[0167] 2) Spatial graph convolution S-TCN performs graph convolution on the adjacency matrix of the undirected graph G for each time point τ∈{1,…,T}:

[0168]

[0169] where l denotes the layer index, D is the degree matrix, is the trainable spatial weight; after Ls layers, the spatial feature tensor dsis the spatial embedding dimension;

[0170] 3) Temporal causal convolution layer T-TCN performs dilated causal convolution along the time dimension for each node i:

[0171] Z i = CausalConv1D(H s [i,:,:]; W t ,k t ,d t )

[0172] where k t is the kernel size, d t is the dilation factor, after L s layers, the temporal feature tensor T′is the down-sampled time step;

[0173] 4) Multimodal attention fusion: after linear mapping of the five-channel feature tensor, cross-modal attention weights are calculated:

[0174]

[0175] where Q c , K c , V c are query, key, and value mappings, respectively, and the output fusion feature

[0176] 5) The output result is output through the task output head, including:

[0177] Tool change priority P∈[0,1] N ;

[0178] Tool change type label T∈{wear type, corrosion type, fracture type, geological adaptation type, life expiration type} N ;

[0179] Tool change quantity K=Round!(∑ i σ(P i≥ θ), the threshold value θ is adjusted in real time by engineering constraints;

[0180] The tool change number set ID = {i | P i ≥ θ}.

[0181] The training optimal weight model of the multi-modal spatio-temporal graph neural network MST-GNN is deployed in the local edge computing unit of the shield machine, and the single forward inference delay is ≤ 50 ms; the inference result is sent to the control layer through the EtherCAT bus with an update period of 1 kHz.

[0182] Fifth: in the control layer, according to the tool change type label output by the multi-modal spatio-temporal graph neural network MST-GNN, and cooperating with the low-speed rotation state of the cutter head, the required tool change tool (target tool) is instantaneously positioned. As shown in the figure, the process specifically includes: Figure 5

[0183] 1) The cutter head main shaft end surface is coaxially installed with a 25-bit absolute value encoder, the resolution is 0.0055° / tick, and the output signal is transmitted in real time to the tool robot controller through the EtherCAT bus with a frequency of 4 kHz;

[0184] 2) A passive UHF-RFID tag is embedded in the base of each replaceable hob, and a 64-bit UID and a fixed offset Δθ i , are fixed in the passive UHF-RFID tag, wherein Δθ i is the mechanical angle difference between the tool center line and the encoder zero position, with a precision of 0.001°;

[0185] 3) The tool robot controller performs a one-time scan after power-on: through the slow rotation of the cutter head for one revolution, the RFID read-write antenna array reads the UID and Δθ i of each tool in turn, and establishes an index array Index[N] = {Δθ1, Δθ2, …, Δθ N} in the tool robot controller memory, which cannot be powered off;

[0186] 4) At any t0 moment, the real-time angle of the cutter head is read out by the absolute value encoder Then the instantaneous target angle coordinate of the Nth tool is:

[0187]

[0188] The formula is refreshed once every 0.25 ms by the tool robot controller, ensuring that the positioning error is ≤ 0.01°;

[0189] 5) Dynamic compensation and closed loop, specifically including: the tool robot controller inputs θ target (t0) to the quintic spline trajectory planner to generate an angular displacement-time curve of the end effector.​

[0190] Note that during the movement of the tool holder, the 4 kHz EtherCAT feedback is used to correct the trajectory in real time, achieve angular synchronization closed loop, and ensure that the relative angular velocity of the end and the target tool is ≤0.02° / s; if the RFID reading fails, the absolute value encoder and the mechanical zero point double channel are used for verification, and the positioning error is still ≤0.02°.

[0191] Sixth: In the control layer, the instantaneous target angle coordinate θ target (t) is received, and the tool robot is trajectory pre-planned, speed chased, and error closed loop controlled under the micro-stop tool changing working condition. As shown in Figure 5 , this step has the following steps:

[0192] 1) In the joint space of the tool robot, a quintic polynomial trajectory q(t) is generated with the current joint angle vector q curr as the starting point and the target joint angle vector q target as the end point, and quintic spline trajectory pre-planning is performed, and its expression is:

[0193] q(t) = a0 + a1t + a2t 2 +a3t 3 +a4t 4 +a5t 5 ,(0≤t≤Δt pred )

[0194] 2) The boundary conditions that the quintic polynomial trajectory q(t) must satisfy at the starting point and the end point include:

[0195] 21) Starting from the current actual joint angle of the robot, avoid position jump:

[0196] q(0) = q curr

[0197] 22) Accurately reach the target tool angle within the specified time Δt pred , and achieve "zero relative" alignment:

[0198] q(Δt pred ) = q target

[0199] The starting point speed and acceleration are zero, ensuring smooth start (no jitter):

[0200]

[0201] The end point speed must match the tool holder angular velocity ω, so that the robot end and the tool are relatively stationary:

[0202]

[0203] The final acceleration should be zero to avoid mechanical impact or residual vibration.

[0204]

[0205] Where ω is the instantaneous angular velocity of the cutter head, which is calculated in real time by the differential of the encoder:

[0206]

[0207] The spline solver is based on the maximum joint velocity v max Maximum acceleration a max The theoretical arrival time Δt is automatically given. pred Typical range: 0.15–0.30 s;

[0208] 3) Perform velocity feedforward and angular synchronization. The controller uses -ω as the end-point angular velocity feedforward command to ensure that at t = Δt pred When the relative angular velocity between the robot end effector and the target tool is Δω≈0, after reaching the target angle, the system switches to angular synchronization mode with a control cycle of 250μs, position loop gain Kp≥800Hz, and velocity feedforward coefficient Kv=1.

[0209] 4) Complete velocity feedforward and angle synchronization, implement error closure and reentry mechanisms, and the real-time angle error is:

[0210] Δθ(t)=θ target (t)-θ end-effector (t)

[0211] When |Δθ(t)|≤0.5° and the holding time>50ms, the zero relative window is determined to be established, and the tool changing mechanism is allowed to operate;

[0212] If |Δθ(t)|>0.5° or the joint torque mutation>15N·m, immediately trigger reentry: the robot decelerates to zero and records q. reentry Recalculate θ target (t); with q reentry Five splines are generated again starting from the current point, with a maximum of three re-entries allowed, and the total time is less than 1.5 seconds. If three consecutive re-entries fail, the tool change is abandoned and the next opportunity is waited for.

[0213] It is important to note that once the zero relative window is established and confirmed, the tool change mechanism is allowed to operate, and the tool robot controller immediately initiates the following tool change sequence, the process of which includes:

[0214] After the zero relative window is confirmed, the locked state detection is performed, the normal force Fn of the tool locking surface is monitored in real time through the six-dimensional force sensor, and only when Fn≤2N and lasts for more than 50 ms, it is determined that the locking force has been released, and the tool pulling step is allowed to enter;

[0215] After the electromagnetic / pneumatic lock is loosened, the Z-axis electric sliding table is pushed out by 50 mm at a speed of 100 mm / s to complete the tool pulling;

[0216] The end-of-travel photoelectric switch is detected, and if it is not in place within 3 seconds, it is interrupted and an alarm is given;

[0217] The tool robot grabs a new tool, and the Z-axis sliding table is inserted into the tool seat in reverse at a speed of 80 mm / s;

[0218] After insertion, a locking torque of 180 N·m is applied, and the torque closed loop tolerance is ±5 N·m;

[0219] After locking is completed, the six-dimensional force sensor confirms that the normal force Fn' is greater than or equal to 800 N, ensuring reliable locking;

[0220] The end UHF-RFID reader reads the new tool UID within 0.2 seconds and compares it with the tool robot controller array, and after verification, the life counter is cleared; if UID verification fails, two re-readings still fail, and it is marked as abnormal; the entire pulling-inserting-locking-verification process is less than or equal to 1.2 seconds, and if it exceeds the time limit or is out of tolerance, it immediately returns to the safety position, retries once, and if it fails twice, it gives up this time tool changing.

[0221] In embodiment two, a multi-dimensional perception and self-adaptive tool changing system for a shield machine at low speed is provided, which can realize the tool changing method in embodiment one, and the tool changing system comprises:

[0222] 1) Perception layer, perception of tool state and data collection, which includes:

[0223] 11) Electromagnetic mutual inductance wear sensor: embedded in each hob ring, real-time measurement of radial wear ΔW of the hob ring, used to judge the "wear type" tool changing demand;

[0224] 12) Electrochemical impedance spectroscopy corrosion sensor: attached to the inner surface of the bearing sealing end cover, real-time output of the bearing sealing area corrosion area percentage ΔC, used to judge the "corrosion type" tool changing demand;

[0225] 13) Acoustic emission crack sensor: fixed on the transition fillet of the tool shaft, real-time capture of crack length ΔL or acoustic emission energy cumulative value AE of the tool shaft, used to judge the "fracture type" tool changing demand;

[0226] 14) Ground Penetrating Radar Antenna: Three sets of evenly distributed cutting heads on the periphery, emitting 100 MHz electromagnetic waves forward 0-30 m, inverting the uniaxial compressive strength UCS of the rock stratum, to determine the "geological adaptation type" tool replacement requirement;

[0227] 15) Strain-based Load Cycle Counter: Attached to the surface of the tool shaft, it records the torque-thrust load spectrum in real time and calculates the equivalent load cycle number Neq, to determine the "end-of-life type" tool replacement requirement;

[0228] 16) Sensor Data Fusion Module: Located in the explosion-proof control box at the center of the cutting head, it synchronously samples (≥1 kHz) and timestamps the data from the above five types of sensors, and outputs an N×5×T spatiotemporal tensor for the decision layer to make decisions;

[0229] 2) Decision Layer, for intelligent judgment of tool state based on three-dimensional comprehensive data from the perception layer, and instruction generation through a multi-modal spatiotemporal graph neural network MST-GNN, including a shield machine local edge computing unit: running a multi-modal spatiotemporal graph neural network MST-GNN, completing a forward inference within 50 ms, outputting tool replacement priority, tool replacement type label, tool replacement number, and tool replacement number, and issuing them to the control layer at a frequency of 1 kHz through an EtherCAT bus;

[0230] 3) Control Layer, for instantaneous positioning of tools that need to be replaced, and trajectory closed-loop control of the tool robot based on its position to complete the tool replacement action, which includes:

[0231] 31) Tool Robot: Under the micro-stop working condition of low-speed rotation of the cutting head, it automatically completes the whole set of actions of "positioning - trajectory planning - angle synchronization - tool extraction - tool insertion - locking - identity confirmation" according to the "tool replacement number + target angular coordinate" instruction issued by the decision layer;

[0232] 32) 25-bit Absolute Value Encoder: Installed on the end face of the cutting head spindle, with a resolution of 0.0055° / tick, it measures the cutting head rotation angle φ(t) in real time, providing a reference for "millisecond-level instantaneous positioning";

[0233] 33) Passive UHF-RFID Tag: Embedded in the base of each tool, storing a 64-bit UID and a fixed angular offset Δθ i from the encoder zero position, used to calculate the instantaneous target angular coordinate θ target (t) of the tool;

[0234] 34) Tool Robot Controller: Receiving encoder and RFID data, it generates a quintic spline trajectory in real time, realizes joint space trajectory planning and angular synchronization closed-loop control of the robot, and has a control period of 250 μs;

[0235] 35) Six-axis force sensor: mounted on the robot end, real-time detection of the normal force and torque during the process of pulling / plugging the tool, used for locking / unlocking state judgment and safety protection.

[0236] It is apparent for a person skilled in the art that the present application is not limited to the details of the above-described exemplary embodiments, but that it can be implemented in other concrete forms without departing from the spirit or essential characteristics of the application. Therefore, the embodiments should be considered in all respects as illustrative and not restrictive, the scope of the application being defined by the appended claims rather than that of the above description, and it is therefore intended that all changes and modifications that come within the meaning and range of equivalency of the claims are to be embraced by the application. Any reference signs in the claims should not be construed as limiting the claims concerned.

[0237] Furthermore, it should be understood that although the present specification describes only a single independent technical solution for each embodiment, the specification is described in this way only for the sake of clarity, and a person skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by a person skilled in the art.

Claims

1. A multi-dimensional sensing and adaptive cutterhead changing method for tunnel boring machines under low-speed pressure maintenance, characterized in that, The tool changing method includes the following steps: S1. In the sensing layer, three-dimensional comprehensive data of the tool is obtained in real time through sensors; S2. In the decision-making layer, the three-dimensional comprehensive data is fused by the sensor data fusion module in the sensing layer and sent into the multi-modal spatio-temporal graph neural network MST-GNN, and the tool changing priority, tool changing type label, tool changing quantity and tool changing number are output; S3. In the control layer, through the tool changing type label, and in cooperation with the low-speed rotation state of the cutter head, instantaneous positioning of the target tool is performed; S4. In the control layer, when the angular velocity of the tool robot is synchronized with the target tool, angular error calculation is performed to confirm the zero relative window, so that the tool robot completes tool extraction, tool insertion and locking, and realizes tool changing.

2. The tool changing method according to claim 1, characterized in that: In the sensing layer, the three-dimensional comprehensive data includes wear defect data A1, corrosion defect data A2 and fracture defect data A3 in the defect dimension A, rock layer hardness data B1 in the geological hardness dimension B, and life data C1 in the life dimension C; Among them, the wear defect data A1, the radial wear amount ΔW of the cutter ring is obtained through an electromagnetic mutual inductance type wear sensor; The corrosion defect data A2, the corrosion area percentage ΔC of the bearing seal area is obtained through an electrochemical impedance spectroscopy corrosion sensor; The fracture defect data A3, the crack length ΔL of the tool shaft or the accumulated acoustic emission energy value AE is obtained through an acoustic emission crack sensor; The rock layer hardness data B1, the uniaxial compressive strength UCS of the rock layer is obtained through an advanced geological radar antenna; The life data C1, the equivalent load cycle number Neq is obtained through a strain type load cycle counter; The three-dimensional comprehensive data is processed by the sensor data fusion module and outputs an N×5×T spatio-temporal tensor for decision-making by the decision-making layer.

3. The tool changing method according to claim 1, characterized in that: In the sensing layer, in the micro-stop working condition where the cutter head rotates at a low speed of ≤2 rpm and the soil chamber is kept pressurized, the necessity of tool changing is judged based on the defect dimension A, the geological hardness dimension B and the life dimension C, including: In the defect dimension A, the tool defects are classified as: For the wear defect data A1, with the radial wear amount ΔW of the cutter ring as the index, when ΔW≥25 mm, it is determined that A1 reaches the tool changing threshold; For the corrosion defect data A2, with the corrosion area percentage ΔC of the bearing seal area as the index, when ΔC≥30%, it is determined that A2 reaches the tool changing threshold; For the fracture defect data A3, with the crack length ΔL of the tool shaft or the accumulated acoustic emission energy value AE as the index, when ΔL≥15 mm or AE≥100 mV·ms, it is determined that the A3 defect reaches the tool changing threshold; Among them, if any tool defect reaches its threshold, the tool changing instruction for the defect dimension A is triggered; In the geological hardness dimension B, the geological hardness grade is defined: Soft rock: UCS≤30 Mpa, replace with a scraper type tool; Medium hard rock: 30 MPa < UCS≤90 MPa, replace with a hob type tool; Extremely hard rock: UCS>90 MPa, heavy hob type tool; When the real-time UCS value obtained by the advanced geological radar antenna or the tool disk vibration inversion jumps from a low grade to a high grade, or when the continuous tunneling length in the extremely hard rock grade is ≥10 m, the tool changing instruction for the geological hardness dimension B is triggered, and the tool type is replaced according to the geological hardness grade; In the life dimension C, the tool's service life is defined as the equivalent load cycle number Neq. When Neq ≥ the design rated life Nr, the tool change command in the life dimension C is triggered.

4. The tool changing method according to claim 2, characterized in that: In the perception layer, to acquire real-time comprehensive three-dimensional data of defect dimension A, geological hardness dimension B, and lifespan dimension C, a sensor system is deployed on the tunnel boring machine cutterhead and cutters, and arranged and connected as follows: In defect dimension A: The wear monitoring module has two sets of electromagnetic mutual inductance wear sensors embedded in the cutter ring base of each replaceable hob on the front of the cutter head, with the same radius and equal angle intervals along the circumference, to output the radial wear amount ΔW of the cutter ring in real time. Corrosion monitoring module: A set of electrochemical impedance spectroscopy corrosion sensors are attached to the inner surface of the bearing sealing end cap of each hob to output the percentage of corrosion area ΔC in the bearing sealing area in real time; Fracture monitoring module: A set of acoustic emission crack sensors are attached to the transition fillet area of ​​the cutter shaft of each hob to capture the cumulative acoustic emission energy AE of crack initiation and propagation in real time; In geological hardness dimension B: At least three sets of ground-penetrating radar antennas are equally spaced along the circumferential direction on the outer periphery of the cutterhead. These antennas are used to transmit electromagnetic waves in front of the tunnel face before the tunnel boring machine advances and to receive reflected signals in order to obtain the uniaxial compressive strength (UCS) of the rock strata. In the life dimension C: On the cutter shaft surface of each cutter, at least one pair of strain-type load cycle counters are symmetrically attached along the axial centerline to collect the torque and thrust load spectrum of the cutter in real time during the tunneling process and calculate the equivalent load cycle number Neq. Sensor data fusion module: Located in the explosion-proof electrical control box at the center rotary joint of the cutter head, it includes a multi-channel synchronous sampling unit, a time synchronization unit, and an edge computing unit; The multi-channel synchronous sampling unit is connected to A1, A2, A3, B1 and C1 respectively via shielded cables, with a sampling frequency ≥1kHz; The time synchronization unit receives a 1PPS pulse signal from the main control PLC of the tunnel boring machine, which is used to timestamp the three-dimensional integrated data, with a time synchronization error of ≤1ms.

5. The tool changing method according to claim 2, characterized in that: In the perception layer, a multi-dimensional state matrix is ​​constructed based on the aforementioned three-dimensional integrated data, and the "isolated state of a single tool" is upgraded to a globally coupled relationship network of "tool head-tool-geology-time" through graph structure modeling. This step specifically includes: Using N replaceable cutting tools on the tool turret as nodes, a five-dimensional state vector is generated in real time for each tool i: S it =[Wear, corrosion, fracture, geological hardness, life] T =[ΔW it ΔC it ΔL it / AE it UCS it Neq it ] T Wherein, ΔW it ΔC it ΔL it / AE it UCS it and Neq it Provided directly by A1, A2, A3, B1 and C1 respectively, where i represents the tool number and t represents a specific moment of monitoring; 2) Stack the five-dimensional state vectors of the N tools in order of tool number to form an N×5-dimensional global state matrix: 3) Using the hobs as nodes, with node features being the aforementioned five-dimensional state vector, and the mechanical coupling relationship between the hobs and the rotational dynamics of the cutter head as edges, construct an undirected graph G = (V, E), where the edge weight formula in the undirected graph G is as follows: Edge weight Q ij The physical distance d between hobs i and j on the cutter head ij The speed of the cutter head, ω, and σ and α are determined together, and are preset hyperparameters. 4) Introduce a time dimension into the node features of the undirected graph G, and transform the state matrix M over T consecutive sampling periods. t Stacked as N×5×T spacetime tensors.

6. The tool changing method according to claim 5, characterized in that: In the decision-making layer, the N×5×T spatiotemporal tensor is processed by a multimodal spatiotemporal graph neural network (MST-GNN) to obtain the tool-changing decision result, specifically including: The N×5×T spatiotemporal tensor is modally decomposed into five-channel tensors: wear channel, corrosion channel, fracture channel, geological hardness channel, and lifetime channel. LayerNorm is applied independently to each channel to obtain the normalized five-channel feature tensor. X c ∈R N×T ,c∈{1,…,5}; Spatial Graph Convolutional Network (S-TCN) uses the adjacency matrix of an undirected graph G as a mask to perform graph convolution at each time point τ∈{1,…,T}: Where l represents the layer index and D is the degree matrix. These are trainable spatial weights; after passing through the Ls layer, the spatial feature tensor is obtained. ds represents the spatial embedding dimension; The temporal causal convolutional layer T-TCN performs dilated causal convolution along the temporal dimension for each node i: Z i =CausalConv1D(H s [i,:,:];W t ,k t ,d t ) Where, k t d is the kernel size. t As the expansion factor, via L s The time feature tensor is obtained after the layer. T′ is the time step after downsampling; Multimodal attention fusion: After linearly mapping the five-channel feature tensor, calculate the cross-modal attention weights. Among them, Q c K c V c They are query, key-value mapping, and output fused features respectively. The results are output through the task output header. The output results include: Tool change priority P∈[0,1] N ; Tool replacement type label T∈{wear type, corrosion type, fracture type, geological adaptability type, end-of-life type} N ; Tool change count K = Round! (∑) i σ(P i ≥θ)), the threshold θ is adjusted in real time by engineering constraints; Tool change number set ID = {i | P i ≥θ}.

7. The tool changing method according to claim 6, characterized in that: In the control layer, instantaneous positioning of the target tool includes the following steps: An absolute encoder is coaxially mounted on the end face of the tool turret spindle, and the output signal is transmitted to the tool robot controller in real time via an EtherCAT bus. Each replaceable hob has a passive UHF-RFID tag embedded inside its base. The passive UHF-RFID tag contains a UID and a fixed offset Δθ. i , where Δθ i This is the mechanical angle difference between the tool centerline and the encoder zero position; After the tool robot controller is powered on, it performs a one-time scan: by slowly rotating the tool head, the RFID read / write antenna array sequentially reads the UID and Δθ of each tool. i In the memory of the tool robot controller, a non-power-loss index array Index[N] = {Δθ1, Δθ2, ..., Δθ} is created. N }; At any time t0, the real-time angle of the tool head is read by the absolute encoder. Then the instantaneous target angular coordinates of the Nth tool are: Dynamic compensation and closed-loop control specifically include: the tool robot controller will adjust θ... target (t0) is input into the quintic spline trajectory planner to generate the angular displacement-time curve of the end effector.

8. The switching method according to claim 7, characterized in that: In the control layer, the instantaneous target angular coordinates θ are obtained by instantaneously positioning the target tool. target (t), under the micro-stop tool change condition, the tool robot performs trajectory pre-planning, speed chasing, and error closed-loop control. This step includes: Within the joint space of the cutting tool robot, with the current joint angle vector q curr Let q be the starting and target joint angle vectors. target Assuming the endpoint is t, a fifth-order polynomial trajectory q(t) is generated. A fifth-order spline trajectory pre-planning is then performed, and its expression is: q(t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 5 ,(0≤t≤Δt pred ) The boundary conditions that the fifth-order polynomial trajectory q(t) must satisfy at the starting and ending points include: Starting from the robot's "current actual joint angles," avoid position jumps: q(0)=q curr At the specified time Δt pred The inner diameter accurately reaches the target tool angle, achieving "zero relative" alignment: q(Δt pred )=q target Starting speed and acceleration are zero to ensure a smooth start. The end-effector velocity must be matched with the angular velocity ω of the cutter head to keep the robot end-effector relatively stationary with respect to the cutter. The final acceleration should be zero to avoid mechanical impact or residual vibration. Where ω is the instantaneous angular velocity of the cutter head, which is calculated in real time by the differential of the encoder: The spline solver is based on the maximum joint velocity v max Maximum acceleration a max The theoretical arrival time Δt is automatically given. pred ; To perform velocity feedforward and angular synchronization, the controller uses -ω as the end-effector angular velocity feedforward command to ensure that at t = Δt pred When the relative angular velocity between the robot end effector and the target tool is Δω≈0, the system switches to angular synchronization mode after reaching the target angle. 4) Complete velocity feedforward and angle synchronization, implement error closure and reentry mechanisms, and the real-time angle error is: Δθ(t)=θ target (t)-θ end-effector (t) When |Δθ(t)|≤0.5° and the holding time>50ms, the zero relative window is determined to be established, and the tool changing mechanism is allowed to operate; If |Δθ(t)|>0.5° or the joint torque mutation>15N·m, immediately trigger reentry: the tool robot decelerates to zero and records q. reentry Recalculate θ target (t).

9. The tool changing method according to claim 8, characterized in that, In the control layer, once the zero relative window is established and confirmed, the tool changer is allowed to move. The tool robot controller immediately initiates the following tool change sequence, the process of which includes: After the zero relative window is confirmed, the locking state is detected. The normal force Fn of the tool locking surface is monitored in real time by a six-dimensional force sensor. If Fn≤2N and lasts for more than 50ms, the locking force is determined to be released and the tool can be pulled out. After the electromagnetic / pneumatic lock is released, the Z-axis electric slide table pushes outward by 50mm at a speed of 100mm / s to complete the tool removal. The photoelectric switch detects the position at the end of the travel; if the position is not reached within 3 seconds, the system will interrupt and trigger an alarm. The tool robot picks up a new tool, and the Z-axis slide inserts it into the tool holder in the opposite direction at 80mm / s. After insertion, apply a locking torque of 180 N·m, with a torque closed-loop tolerance of ±5 N·m; After locking is completed, the six-dimensional force sensor confirms that the normal force Fn' ≥ 800N, ensuring locking; The end-effector UHF-RFID reader reads the new tool's UID and compares it with the tool robot controller's array. If the verification is successful, the lifespan counter is reset to zero. If the UID verification fails, and two rereads still fail, an anomaly is marked.

10. A cutter changing system for implementing the multi-dimensional sensing and adaptive cutter changing method for low-speed pressure maintenance of a tunnel boring machine as described in any one of claims 1 to 9, characterized in that, The tool changing system includes: The sensing layer senses and collects data on the tool's condition, including electromagnetic inductance wear sensors, electrochemical impedance spectroscopy corrosion sensors, acoustic emission crack sensors, advanced ground-penetrating radar antennas, strain gauge load cycle counters, and sensor data fusion modules. The decision layer is used to intelligently judge the tool status based on the three-dimensional comprehensive data obtained from the perception layer, and generates instructions through the multimodal spatiotemporal graph neural network MST-GNN, outputting tool change priority, tool change type label, tool change quantity and tool change number, and sending them to the control layer through the EtherCAT bus. The control layer performs instantaneous positioning of the tool to be replaced and performs trajectory closed-loop control of the tool robot based on its position to complete the tool changing action. It includes the tool robot, absolute encoder, passive UHF-RFID tag, tool robot controller and six-dimensional force sensor.

Images