Shield cutter friction coefficient real-time monitoring and wear trend prediction method and device

Abstract

The invention discloses a shield cutter friction coefficient real-time monitoring and wear trend prediction method and device, and the method comprises the steps: transmitting a high-frequency pulse wave through an ultrasonic sensor embedded in a cutter working surface, collecting a reflection signal of a cutter-rock soil interface, extracting the multi-dimensional acoustic characteristics of the reflection signal, and synchronously collecting the temperature of a friction region; compensating the multi-dimensional acoustic features through a nonlinear regression model, inputting the compensated features into a support vector regression machine, outputting a real-time friction coefficient, and predicting a wear trend through an autoregressive integral moving average model based on a historical sequence of mu; the device comprises a sound-temperature integrated probe, an embedded processing unit and an anti-vibration sealing structure, and provides core data support for shield cutter service life management, maintenance strategy optimization and digital construction.

Description

Technical Field

[0001] This invention relates to a method and device for real-time monitoring of the friction coefficient and prediction of wear trend of tunnel boring machine (TBM) cutters, specifically a method and device for real-time monitoring of the friction coefficient and prediction of wear trend of TBM cutters based on ultrasonic sensing and temperature compensation, belonging to the field of intelligent monitoring technology for tunnel boring equipment. Background Technology

[0002] The problem of cutter wear in tunnel boring machine (TBM) construction is becoming increasingly prominent, with downtime and cutter replacement costs exceeding one million yuan per instance, accounting for over 40% of equipment failure rates. Currently, the industry generally relies on manual inspection, requiring 8-24 hours of downtime every 50-100 meters of tunneling. This not only severely hinders construction efficiency but also poses significant safety risks such as rock instability and water inrush. Although some technologies attempt to indirectly estimate wear by collecting macroscopic parameters such as thrust and torque from the TBM, errors often exceed 25% due to geological interference and signal lag. Furthermore, these methods only identify wear when it reaches 30% or more, failing to detect early coating failures. More advanced offline monitoring methods, such as industrial endoscopes or 3D scanning, can obtain local cutter morphology, but require downtime and TBM opening, making dynamic monitoring impossible and exposing workers to high-risk environments.

[0003] While online sensing technology has been gradually applied in recent years, it still faces severe challenges: 1) Vibration monitoring suffers from a signal-to-noise ratio below 0.5 due to strong mechanical vibration of the cutterhead, completely submerging characteristic frequencies; 2) Acoustic emission technology is affected by high-density mud coating, resulting in signal attenuation exceeding 80% and significant loss of effective data; 3) Although resistive wear sensors can be embedded in the cutterhead, they fail to seal under mud pressures above 15 MPa, with an average lifespan of less than 200 hours. It is worth noting that while ultrasonic reflection is used to monitor wear, its sensors are externally mounted on the cutterhead and not integrated in situ, and the signal drift caused by frictional temperature rise is not considered, resulting in a measured error as high as 18.7%. It can only output the cumulative wear amount and cannot dynamically identify the coefficient of friction.

[0004] The core bottlenecks can be summarized in three aspects: First, there is a lack of in-situ integrated sensing solutions adapted to the extreme working conditions of tunnel boring machine (TBM) cutters, and existing external or contact sensors are difficult to operate stably for extended periods. Second, the severe temperature rise accompanying the friction process causes acoustic signal characteristics to drift by more than 30%, and existing technologies have not established effective compensation mechanisms. Third, monitoring functions are limited to post-event diagnosis and cannot provide early warnings of wear trend inflection points, resulting in delayed maintenance decisions. These deficiencies severely restrict the development of intelligent and unmanned TBM construction. Summary of the Invention

[0005] The technical solution of this invention addresses the technical problems existing in the prior art and provides a solution that is significantly different from the prior art. Specifically, the purpose of this invention is to solve the above-mentioned shortcomings of the prior art by proposing a method and device for real-time monitoring of the friction coefficient and prediction of wear trend of shield tunneling cutter based on ultrasonic sensing and temperature compensation. Through in-situ dynamic sensing of the friction state of the cutter-rock interface and prediction of wear trend, this invention aims to break through the three major technical barriers of in-situ sensing, temperature drift suppression and trend prediction.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for real-time monitoring of the friction coefficient and prediction of wear trends of tunnel boring machine cutters, the prediction method comprising:

[0007] S1: High-frequency pulse waves are emitted by an ultrasonic sensor embedded in the working surface of the tool to collect the reflected signals at the tool-soil interface;

[0008] S2: Extract the multi-dimensional acoustic features of the reflected signal, including:

[0009] a) Peak attenuation rate of envelope

[0010]

[0011] Among them, A t Represents the original amplitude attenuation rate, where A0 is the initial amplitude, and the unit is %;

[0012] b) Spectral centroid shift Δf c

[0013]

[0014] in, For real-time spectrum centroid; As the baseline value;

[0015] c) Subband energy entropy E s

[0016]

[0017] Where, p i The energy percentage of the i-th frequency band, N = 5;

[0018] S3: Synchronously acquire the temperature T of the friction zone, and compensate for multi-dimensional acoustic characteristics using a μ-T nonlinear regression model, including:

[0019]

[0020] in, α represents the amplitude attenuation rate after temperature compensation. A T is the amplitude temperature drift coefficient; T is the real-time friction zone temperature; Tref For reference temperature; To compensate for the centroid shift in the post-spectral composition; β f This is the frequency temperature drift coefficient; To compensate for the entropy of the subband energy; γ E This is the energy entropy temperature drift coefficient;

[0021] S4: Compensated features Input a support vector regression machine and output the real-time friction coefficient μ;

[0022] Wherein, the kernel function is a radial basis:

[0023]

[0024] in, Let be the Gaussian radial basis kernel function, representing the sample and Inner product in a high-dimensional feature space; and γ represents the sample vector in the input space; γ is the bandwidth parameter of the kernel function. The squared Euclidean distance between the sample vectors;

[0025] S5: Based on the historical sequence of μ, the wear trend is predicted using an autoregressive integral moving average model.

[0026]

[0027] Where, μ t Let be the target variable at time t; c is the constant term of the model. For the autoregressive part; ∈ t The white noise error term at time t; The moving average portion;

[0028] When the predicted value μ t +Δt exceeds the threshold μ crit An alert is triggered at any time.

[0029] As a further aspect of the present invention: in step S2, the extraction of multi-dimensional acoustic features of the reflected signal also needs to satisfy the following:

[0030] Calculate the short-time energy decay slope

[0031]

[0032] Where k is the slope of the linear regression; t i The midpoint of the i-th time window; E i The energy of the i-th time window; M = 10;

[0033] Extracting the energy variance of wavelet packet decomposition nodes

[0034]

[0035] Where, σ 2 E represents the variance of node energy. j The energy of the j-th wavelet packet node; This represents the average energy of the nodes.

[0036] As a further aspect of the present invention: in step S3, the temperature drift coefficient calibration method must satisfy the following:

[0037] In a constant temperature friction test bench, measurements were taken at different temperatures T. k Acoustic characteristics F raw,k F is fitted using the least squares method raw = f(T) function curve, solve for α A β f γ E .

[0038] As a further aspect of the present invention: the threshold μ in step S5 crit The following settings also need to be met:

[0039] μ crit =μ0+3σ μ

[0040] Where μ0 is the reference friction coefficient for the coating health state; σ μ This represents the standard deviation of historical fluctuations.

[0041] A monitoring device for implementing real-time monitoring of the friction coefficient and prediction of wear trends of tunnel boring machine cutters, the monitoring device comprising:

[0042] The acoustic-temperature integrated probe, encapsulated within a silicon nitride wear-resistant coupling layer, includes a piezoelectric ultrasonic transducer and a K-type thermocouple. The piezoelectric ultrasonic transducer has a center frequency of 2.5MHz, an acoustic wave incident angle θ = 15° ± 2°, and the axial spacing d of the K-type thermocouple is ≤ 2mm. The temperature measurement response time is < 50ms.

[0043] The embedded processing unit includes an adaptive bandpass filter, an FPGA-based real-time SVR calculation module, and an ARIMA prediction coprocessor. The adaptive bandpass filter has a passband of 1.8-5.2MHz, and the inference latency of the FPGA-based real-time SVR calculation module is <10ms.

[0044] The vibration-resistant sealing structure includes metal-armored optical fibers and a pressure-balanced sealing connector (402), with a pressure rating of ≥20MPa.

[0045] Preferably, the silicon nitride wear-resistant coupling layer (203) needs to meet the following requirements:

[0046] Acoustic impedance Z = 28 ± 2 × 10⁶ Rayl; thickness h = λ / 4; surface hardness ≥ 1500 HV.

[0047] Preferably, the embedded processing unit (300) further includes a multi-sensor data fusion module (304), which associates the friction coefficient μ with the tunnel boring machine thrust F and torque M, as follows:

[0048]

[0049] Where HI is the health index and η is the aging coefficient. When HI < 0.7, a maintenance command is triggered.

[0050] The beneficial effects of this invention are as follows: High-frequency sound waves are emitted by an ultrasonic sensor embedded in the cutter surface, collecting the amplitude, spectrum, and envelope characteristics of the interface reflection signal; a temperature sensor is simultaneously integrated to construct a μ-T nonlinear regression, eliminating the interference of temperature drift on acoustic characteristics; based on the compensated features, a support vector regression machine is used to map the data to a real-time friction coefficient, and historical data is combined with time-series analysis to predict the wear inflection point. The corresponding device includes an ultrasonic module, a temperature module, an embedded processing unit, and a host computer. It employs a wear-resistant coupling layer and a shock-resistant sealing structure to adapt to the shield cutterhead environment, providing core data support for shield cutter life management, maintenance strategy optimization, and digital construction. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the structure of an embodiment of the present invention;

[0052] Figure 2 This is a flowchart of a method according to an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of the temperature compensation correction model according to an embodiment of the present invention;

[0054] In the picture:

[0055] 200. Acoustic-thermal integrated probe; 201. Piezoelectric ultrasonic transducer; 202. Type K thermocouple; 203. Silicon nitride wear-resistant coupling layer.

[0056] 300. Embedded processing unit; 301. Adaptive bandpass filter; 302. FPGA-based real-time SVR calculation module; 303. ARIMA prediction coprocessor.

[0057] 400. Vibration-resistant sealing structure; 401. Metal-armored optical fiber; 402. Pressure-balanced sealing joint. Detailed Implementation

[0058] 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 embodiments of the present invention, and not all embodiments. 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.

[0059] Example 1, as Figures 2 to 3 As shown in the figure, this embodiment provides a method for real-time monitoring of the friction coefficient and prediction of wear trends of tunnel boring machine (TBM) cutters. The prediction method includes:

[0060] First: A 1-5MHz high-frequency pulse wave is emitted by an ultrasonic sensor embedded in the working surface of the tool to collect the reflected signal at the tool-soil interface.

[0061] Second: Extract the multi-dimensional acoustic features of the reflected signal, including:

[0062] 1) Envelope peak attenuation rate

[0063]

[0064] Among them, A t Represents the original amplitude attenuation rate, where A0 is the initial amplitude, and the unit is %;

[0065] 2) Spectral centroid shift Δf c

[0066]

[0067] in, For real-time spectrum centroid; As the baseline value;

[0068] c) Subband energy entropy E s

[0069]

[0070] Where, p i The energy percentage of the i-th frequency band, N = 5;

[0071] In addition, the extraction of multi-dimensional acoustic features from reflected signals must also meet the following requirements:

[0072] Calculate the short-time energy decay slope

[0073]

[0074] Where k is the slope of the linear regression; t i The midpoint of the i-th time window; E i The energy of the i-th time window; M = 10;

[0075] Extracting the energy variance of wavelet packet decomposition nodes

[0076]

[0077] Where, σ 2 E represents the variance of node energy. j The energy of the j-th wavelet packet node; This represents the average energy of the nodes.

[0078] Third: The temperature T of the friction zone is collected synchronously, and the multi-dimensional acoustic characteristics are compensated by the μ-T nonlinear regression model.

[0079] Multi-dimensional acoustic feature compensation specifically includes:

[0080]

[0081] in, α represents the amplitude attenuation rate after temperature compensation. A T is the amplitude temperature drift coefficient; T is the real-time friction zone temperature; T ref For reference temperature; To compensate for the centroid shift in the post-spectral composition; β f This is the frequency temperature drift coefficient; To compensate for the entropy of the subband energy; γ E This is the energy entropy temperature drift coefficient;

[0082] In addition, the calibration method for temperature drift coefficients (including amplitude temperature drift coefficient, frequency temperature drift coefficient, and energy entropy temperature drift coefficient) must meet the following requirements:

[0083] In a constant temperature friction test bench, measurements were taken at different temperatures T. k Acoustic characteristics F raw,k F is fitted using the least squares method raw = f(T) function curve, solve for α A β f γ E .

[0084] Fourth: The compensated features Input a Support Vector Regression (SVR) machine and output the real-time friction coefficient μ.

[0085] Wherein, the kernel function is a radial basis:

[0086]

[0087] in, Let be the Gaussian radial basis kernel function, representing the sample and Inner product in a high-dimensional feature space; and γ represents the sample vector in the input space; γ is the bandwidth parameter of the kernel function. The squared Euclidean distance between the sample vectors;

[0088] In predicting the friction coefficient of tunnel boring machine cutters, the eigenvector after temperature compensation... It may have a nonlinear mapping relationship with the friction coefficient. By using the RBF kernel function, low-dimensional feature vectors can be mapped to a high-dimensional space, making the originally linearly inseparable problem linearly separable in the high-dimensional space, thereby improving the prediction accuracy of support vector regression (SVR) for the friction coefficient.

[0089] Fifth: Based on the historical sequence of μ, the wear trend is predicted using the Autoregressive Integral Moving Average (ARIMA) model.

[0090]

[0091] Where, μ t Let be the target variable at time t; c is the constant term of the model. For the autoregressive part; ∈ t The white noise error term at time t; This refers to the moving average (MA) portion;

[0092] When the predicted value μ t +Δt exceeds the threshold μ crit An alert is triggered at any time.

[0093] threshold μ crit The settings must meet the following requirements:

[0094] μ crit =μ0+3σ μ

[0095] Where μ0 is the reference friction coefficient for the coating health state; σ μ This represents the standard deviation of historical fluctuations.

[0096] Example 2, as Figure 1 As shown, this embodiment provides a monitoring device for implementing the prediction method in Embodiment 1. The monitoring device includes an integrated acoustic-temperature probe 200, an embedded processing unit 300, and a vibration-resistant sealing structure 400.

[0097] The acoustic-temperature integrated probe 200, encapsulated within a silicon nitride wear-resistant coupling layer 203, includes a piezoelectric ultrasonic transducer 201 and a K-type thermocouple 202. The center frequency of the piezoelectric ultrasonic transducer 201 is 2.5MHz, the acoustic wave incident angle θ = 15°±2°, the axial spacing d of the K-type thermocouple 202 is ≤2mm, and the temperature measurement response time is <50ms. The silicon nitride wear-resistant coupling layer 203 must meet the following requirements: acoustic impedance Z = 28±2×106Rayl; thickness h = λ / 4; and surface hardness ≥1500HV.

[0098] The embedded processing unit 300 includes an adaptive bandpass filter 301, an FPGA-based real-time SVR calculation module 302, and an ARIMA prediction coprocessor 303. The adaptive bandpass filter 301 has a passband of 1.8-5.2MHz, and the inference latency of the FPGA-based real-time SVR calculation module 302 is <10ms.

[0099] In addition, the embedded processing unit 300 also includes a multi-sensor data fusion module 304, which associates the friction coefficient μ with the tunnel boring machine thrust F and torque M, as follows:

[0100]

[0101] Where HI is the health index and η is the aging coefficient. When HI < 0.7, a maintenance command is triggered.

[0102] The vibration-resistant sealing structure 400 includes a metal-armored optical fiber 401 and a pressure-balanced sealing joint 402, with a pressure resistance rating of ≥20MPa.

[0103] Based on Embodiments 1 and 2, the problem of lack of in-situ sensing of the friction state at the tool-soil interface under complex working conditions can be solved. Embodiment 1 uses an ultrasonic sensor embedded in the tool surface to emit high-frequency sound waves and collect the amplitude, spectrum, and envelope characteristics of the interface reflection signal; simultaneously, a temperature sensor is integrated to construct a μ-T nonlinear regression to eliminate the interference of temperature drift on the acoustic characteristics; based on the compensated features, a support vector regression machine is used to map to the real-time friction coefficient, and the wear inflection point is predicted through time series analysis combined with historical data; Embodiment 2's corresponding device includes an ultrasonic module, a temperature module, an embedded processing unit, and a host computer, and adopts a wear-resistant coupling layer and a shock-resistant sealing structure to adapt to the shield cutterhead environment.

[0104] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0105] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those 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 those skilled in the art.

Claims

1. A method for real-time monitoring of the friction coefficient and prediction of wear trend of tunnel boring machine (TBM) cutters, characterized in that, The prediction method includes: S1: High-frequency pulse waves are emitted by the ultrasonic sensor (201) embedded in the working surface (101) of the tool to collect the reflected signal of the tool-soil interface; S2: Extract the multi-dimensional acoustic features of the reflected signal, including: a) Peak attenuation rate of envelope Among them, A t Represents the original amplitude attenuation rate, where A0 is the initial amplitude, and the unit is %; b) Spectral centroid shift Δf c in, For real-time spectrum centroid; As the baseline value; c) Subband energy entropy E s Where, p i The energy percentage of the i-th frequency band, N = 5; S3: Synchronously acquire the temperature T of the friction zone, and compensate for multi-dimensional acoustic characteristics using a μ-T nonlinear regression model, including: in, α represents the amplitude attenuation rate after temperature compensation. A T is the amplitude temperature drift coefficient; T is the real-time friction zone temperature; T ref For reference temperature; To compensate for the centroid shift in the post-spectral composition; β f This is the frequency temperature drift coefficient; To compensate for the entropy of the subband energy; γ E This is the energy entropy temperature drift coefficient; S4: Compensated features Input a support vector regression machine and output the real-time friction coefficient μ; Wherein, the kernel function is a radial basis: in, Let be the Gaussian radial basis kernel function, representing the sample and Inner product in a high-dimensional feature space; and γ represents the sample vector in the input space; γ is the bandwidth parameter of the kernel function. The squared Euclidean distance between the sample vectors; S5: Based on the historical sequence of μ, the wear trend is predicted using an autoregressive integral moving average model. Where, μ t Let be the target variable at time t; c is the constant term of the model. For the autoregressive part; ∈ t The white noise error term at time t; This is the moving average portion; When the predicted value μ t +Δt exceeds the threshold μ crit An alert is triggered at any time.

2. The prediction method according to claim 1, characterized in that, In step S2, the extraction of multi-dimensional acoustic features of the reflected signal also needs to satisfy the following: Calculate the short-time energy decay slope Where k is the slope of the linear regression; t i The midpoint of the i-th time window; E i The energy of the i-th time window; M = 10; Extracting the energy variance of wavelet packet decomposition nodes Where, σ 2 E represents the variance of node energy. j The energy of the j-th wavelet packet node; This represents the average energy of the nodes.

3. The prediction method according to claim 1, characterized in that, In step S3, the temperature drift coefficient calibration method must meet the following requirements: In a constant temperature friction test bench, measurements were taken at different temperatures T. k Acoustic characteristics F raw,k F is fitted using the least squares method raw = f(T) function curve, solve for α A β f γ E .

4. The prediction method according to claim 1, characterized in that, The threshold μ in step S5 crit The following settings also need to be met: m crit =μ0+3σ μ Where μ0 is the reference friction coefficient for the coating health state; σ μ This represents the standard deviation of historical fluctuations.

5. A monitoring device for implementing the prediction method according to any one of claims 1 to 4, characterized in that, include: An integrated acoustic-temperature probe (200) is encapsulated within a silicon nitride wear-resistant coupling layer (203), comprising a piezoelectric ultrasonic transducer (201) and a K-type thermocouple (202). The piezoelectric ultrasonic transducer (201) has a center frequency of 2.5MHz and an acoustic wave incident angle θ = 15° ± 2°. The axial spacing d of the K-type thermocouple (202) is ≤ 2mm, and the temperature measurement response time is < 50ms. The embedded processing unit (300) includes an adaptive bandpass filter (301), an FPGA-based real-time SVR calculation module (302), and an ARIMA prediction coprocessor (303), wherein the passband of the adaptive bandpass filter (301) is 1.8-5.2MHz, and the inference latency of the FPGA-based real-time SVR calculation module (302) is <10ms. The vibration-resistant sealing structure (400) includes a metal-armored optical fiber (401) and a pressure-balanced sealing connector (402), with a pressure resistance rating of ≥20MPa.

6. The monitoring device according to claim 5, characterized in that: The silicon nitride wear-resistant coupling layer (203) must meet the following requirements: Acoustic impedance Z = 28 ± 2 × 10⁶ Rayl; thickness h = λ / 4; surface hardness ≥ 1500 HV.

7. The monitoring device according to claim 5, characterized in that: The embedded processing unit (300) further includes a multi-sensor data fusion module (304), which associates the friction coefficient μ with the tunnel boring machine thrust F and torque M, as follows: Where HI is the health index and η is the aging coefficient. When HI < 0.7, a maintenance command is triggered.

Images