A shield machine fault diagnosis method based on angular domain periodicity and multi-point spatial response

Abstract

This invention relates to the field of fault diagnosis, specifically a method for diagnosing tunnel boring machine (TBM) faults based on angular domain periodicity and multi-point spatial response. The method includes the following steps: S1, acquiring relevant signals from the TBM main bearing; S2, calculating cumulative rotation angle; S3, envelope and impact scoring; S4, identifying candidate impact event characteristics; S5, determining angular phase periodicity; S6, determining spatial response consistency; S7, fusing and identifying impact sources; S8, suppressing external impacts; and S9, performing joint diagnosis of envelope order spectrum and multiple order, outputting the diagnostic results. This invention achieves alignment of impact events and rotation phase under variable rotation speed conditions through angular domain resampling, eliminates external interference such as geological impacts by utilizing multi-point amplitude distribution, propagation delay, and waveform consistency, and enhances periodic fault characteristics through joint analysis of the fundamental and multiple orders, thereby improving the diagnosis of TBM main bearing faults under complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis in signal analysis, specifically a fault diagnosis method for tunnel boring machines based on angular domain periodicity and multi-point spatial response. Background Technology

[0002] Tunnel boring machines (TBMs) are widely used in tunnel, mine, and underground engineering construction. Their drive systems typically include rotating components such as bearings, gears, couplings, and rotors. Under long-term heavy loads, impact loads, and alternating loads, these components are prone to localized damage such as pitting, spalling, cracks, wear, or loosening. When the damaged area periodically enters the load-bearing or meshing zone, it usually generates periodic impacts related to the shaft rotation angle. Existing technologies generally use envelope analysis, spectrum analysis, or order analysis of vibration signals to extract the characteristics of such faults.

[0003] However, the working environment of tunnel boring machines is complex. Geological activity, rock fracturing, blasting operations, vehicle movement, cutter collisions, and foundation vibrations can all generate large-amplitude external transient impacts. Such external impacts may exceed fixed alarm thresholds and form spurious spectral peaks within a limited observation time. Therefore, judging solely based on vibration amplitude or a single spectral peak is insufficient to effectively distinguish between internal equipment fault impacts and external environmental impacts, easily leading to misdiagnosis or missed diagnosis.

[0004] Furthermore, the rotational speed of the tunnel boring machine varies with geological conditions, load status, and construction parameters, causing the time interval of fault impacts to change accordingly. When using conventional isochronous sampling spectrum analysis, fault characteristics are prone to spectral broadening, energy dispersion, or peak shift. Strong external impacts can also contaminate the envelope signal and envelope order spectrum, reducing the visibility of weaker periodic fault impacts.

[0005] Current diagnostic methods largely rely on single vibration measurement points, failing to fully utilize the response amplitude, propagation delay, and attenuation characteristics of shock waves at different spatial locations. Since the propagation sources and spatial responses of internal equipment fault impacts and external environmental impacts typically differ, combining the response characteristics of multiple measurement points helps eliminate external interference. Simultaneously, utilizing the periodic characteristics of multiple related fault orders can reduce the influence of random background peaks on diagnostic results. Therefore, a tunnel boring machine (TBM) fault diagnosis method that comprehensively utilizes angular domain periodicity and multi-point spatial response characteristics is needed to improve the accuracy and reliability of fault identification when speed fluctuations and strong external impacts coexist. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a fault diagnosis method for tunnel boring machines based on angular domain periodicity and multi-point spatial response, thereby reducing false alarms caused by external disturbances such as geological impacts and improving the detection rate of periodic impact faults under variable speed conditions.

[0007] To achieve the above-mentioned objectives, this invention provides a technical solution: a method for diagnosing tunnel boring machine faults based on angular domain periodicity and multi-point spatial response, comprising the following steps:

[0008] S1. Collect relevant signals of the main bearing of the tunnel boring machine: Arrange at least two vibration sensors at different spatial positions of the main bearing of the tunnel boring machine, and set up a speed sensor. Use a unified clock to synchronously collect multi-channel vibration signals and speed signals.

[0009] S2. Cumulative rotation angle calculation: The collected rotation speed signal is preprocessed, and the instantaneous rotation frequency, cumulative rotation angle and cumulative rotation number of the corresponding shaft of the main bearing are calculated based on the preprocessed rotation speed signal. An equal angle sampling grid is established, and the vibration signal of each channel is resampled into angular domain vibration signal according to the equal angle sampling grid.

[0010] S3. Envelope and Impact Score: The vibration signals of each channel are detrended, bandpass filtered, or spectral kurtosis selected. The envelope signal of each channel is obtained through Hilbert transform. The robust normalized impact score of each channel is calculated based on the median and median absolute deviation of the envelope signal. The robust normalized impact scores of each channel are then fused to obtain the global impact score. The impact peak value is detected based on an adaptive threshold, and adjacent impact peak values ​​are merged into candidate impact events.

[0011] S4. Characteristics of candidate impact events: Determine the occurrence time and corresponding cumulative rotation angle of each candidate impact event, and extract the response peak value or response energy, cross-channel normalized amplitude vector, inter-channel propagation delay, cross-channel waveform correlation coefficient, impact duration, kurtosis, impulse factor, frequency band energy, and instantaneous rotation speed before and after the impact event in different vibration channels.

[0012] S5. Angular phase periodicity judgment: For the candidate fault order to be detected, the cumulative rotation angle corresponding to each candidate impact event is folded into a candidate fault cycle to obtain the folded angular phase; the angular phase concentration is calculated based on the folded angular phase to evaluate the periodic synchronization between the candidate impact event and the spindle rotation angle.

[0013] S6. Spatial response consistency judgment: The amplitude spatial mode stability is calculated based on the distance or correlation coefficient between the cross-channel normalized amplitude vectors corresponding to different candidate impact events, the propagation delay stability is calculated based on the dispersion of the propagation delay between channels corresponding to multiple candidate impact events, and the cross-channel waveform consistency is calculated based on the correlation coefficient of the impact response waveform in different vibration channels.

[0014] S7. Impact Source Fusion and Discrimination: The angular phase concentration, amplitude spatial mode stability, propagation delay stability, and cross-channel waveform consistency are fused to construct the fault confidence of the candidate fault order; based on the fault confidence, the candidate impact events are divided into internal tunnel boring machine fault impacts and external environmental impacts.

[0015] S8. External shock suppression: Identify the signal segment that is determined to be an external environmental shock, and perform processes such as elimination, interpolation replacement, prediction replacement, windowing smoothing or weight reduction on the signal segment to obtain the angular domain vibration signal after external shock suppression.

[0016] S9. Envelope order spectrum and multiple order combined diagnosis: Calculate the envelope order spectrum based on the angular domain vibration signal after external impact suppression, extract the candidate fault order and the spectral peaks near at least one integer multiple order, and calculate the multiple order combined evidence; fuse the multiple order combined evidence with the fault confidence, and output the shield machine main bearing fault diagnosis result based on the judgment results in multiple consecutive rotation windows.

[0017] Preferably, in step S1, the vibration sensors are disposed at different spatial locations on the main bearing housing, support, or foundation of the tunnel boring machine; the number of vibration sensors is not less than two, preferably three, and the three vibration sensors are respectively disposed at the bottom, left side, and right side of the main bearing of the tunnel boring machine; the vibration sensors are one or more of the following: acceleration sensor, velocity sensor, displacement sensor, acoustic emission sensor, or dynamic pressure sensor; the rotational speed sensor is used to collect instantaneous rotational speed, rotational frequency, or key phase signal, and each vibration channel and the rotational speed channel are synchronously collected using a unified clock.

[0018] The present invention has the following beneficial effects:

[0019] 1. This invention restores the stable period of fault impacts that vary with rotational speed in the angular domain through rotational speed integration and equal-angle resampling. Simultaneously, it utilizes angular phase periodicity and multi-point spatial response characteristics to avoid directly classifying high-amplitude geological impacts as equipment failures. External impacts are removed or downweighted before calculating the order spectrum, improving the visibility of weakly periodic fault components.

[0020] 2. This invention improves the reliability of fault characteristic order identification by jointly judging the basic order and multiple multiple orders. The threshold is determined based on robust statistics such as the median and median absolute deviation, which can adapt to different equipment, loads, and noise levels. By utilizing the amplitude ratio and arrival delay of different measurement points, it is possible to further determine whether the impact is closer to the bottom, left, or right side of the equipment. Attached Figure Description

[0021] Figure 1 This is a flowchart of the present invention;

[0022] Figure 2 This is a schematic diagram of variable speed;

[0023] Figure 3 This is a comparison chart of global impact scores under normal and fault conditions;

[0024] Figure 4 It is a comparison diagram of the envelope order spectrum between healthy and faulty states. Detailed Implementation

[0025] 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.

[0026] The specific embodiments of the present invention will now be described with reference to the accompanying drawings to enable those skilled in the art to better understand the invention. It should be particularly noted that in the following description, detailed descriptions of known functions and designs that might obscure the main content of the invention will be omitted here.

[0027] Please see Figure 1 This invention provides a technical solution: a method for diagnosing tunnel boring machine faults based on angular domain periodicity and multi-point spatial response, comprising the following steps:

[0028] S1. Acquiring signals related to the main bearing of the tunnel boring machine: At least two vibration sensors are arranged on the main bearing of the tunnel boring machine. In this embodiment, three vibration measuring points are set, located at the bottom, left, and right sides of the equipment, respectively. A speed sensor is also set to acquire the instantaneous rotational speed, rotational frequency, or key phase signal of the shaft. All vibration channels and the speed channel are synchronized using a unified clock. The vibration sensors can be acceleration sensors, velocity sensors, displacement sensors, acoustic emission sensors, or dynamic pressure sensors.

[0029] S2, Cumulative Rotation Angle Calculation: If the speed signal is expressed in revolutions per minute... In other words, the instantaneous frequency conversion is:

[0030]

[0031] The cumulative turning angle is:

[0032]

[0033] The cumulative number of revolutions is:

[0034]

[0035] When the rotational speed signal contains glitches or short-term gaps, median filtering, Kalman filtering, spline interpolation, or amplitude limiting can be used first, followed by integration. An isoangular mesh is then established based on the accumulated rotation angle.

[0036]

[0037] in, The angular sampling interval is defined. The angular domain signal of each channel is obtained through linear interpolation, spline interpolation, or anti-aliasing resampling. .

[0038] S3. Envelope and Impact Scoring: Detrending, bandpass filtering, or spectral kurtosis selection is applied to each vibration channel to preserve the structural resonance components excited by the impact. Analytical signals are obtained through Hilbert transform.

[0039]

[0040] The envelope signal is:

[0041]

[0042] A sliding window method was used to calculate the background median and median absolute deviation, and a robust standardized score was obtained. The multi-channel global impact score can be expressed as:

[0043]

[0044] in, Used to indicate the degree of impact occurring simultaneously in multiple channels within a neighborhood at the same time. When When the adaptive threshold is exceeded, the corresponding location is marked as a candidate impact. For multiple peaks with an interval shorter than the impact response duration, they are merged into a single impact event.

[0045] S4. Characteristics of candidate impact events: For the first One candidate impact event, extract: occurrence time. Cumulative turning angles Peak values ​​or energy of each channel Cross-channel normalized magnitude vector Propagation delay between channels The parameters include the waveform correlation coefficient between channels, impact duration, kurtosis, impulse factor, and bandgap energy, as well as the instantaneous rotational speed before and after the impact event. Normalization of the multi-channel amplitude vectors can reduce the influence of the absolute impact intensity of different events on spatial pattern determination.

[0046] S5. Angular phase periodicity determination: For the order to be detected Calculate the folding angle phase of each candidate impact event:

[0047]

[0048] If the impact is caused by a failure in the stationary rotating component, then The angular phases are concentrated near one or more stable phases; if the impact originates from the geological or construction environment, its angular phase relative to the axis of rotation is usually dispersed. The angular phase concentration is calculated from this. Periodicity can also be evaluated using the Rayleigh test, circular variance, phase histogram entropy, or event sequence order spectrum.

[0049] S6. Spatial Response Consistency Judgment: When the same fault source inside rotating equipment propagates to each sensor via a fixed mechanical transmission path, its normalized amplitude distribution and inter-channel propagation delay usually exhibit high stability. External impacts from multiple different locations, with changes in propagation direction, incident angle, and coupling path, typically result in significant dispersion in the cross-channel spatial response mode. Spatial response stability can be expressed as:

[0050]

[0051] The propagation delay stability can be expressed as:

[0052]

[0053] in, For the number of channels, This is the time delay normalization coefficient.

[0054] S7. Impact Source Fusion Judgment: Targeting Candidate Orders Constructing fault confidence:

[0055]

[0056] in, For angular phase concentration, For propagation delay stability, As evidence of multiple order combinations, to For weights.

[0057] S8. External shock suppression: For identified external environmental shocks, the shock segment is replaced with a linear interpolation of the signals on both sides of the segment.

[0058] S9. Envelope Order Spectrum and Multiple Order Joint Diagnosis: A Discrete Fourier Transform is performed on the processed angular domain envelope signal to obtain the envelope order spectrum. For candidate fault orders, not only the fundamental order but also integer multiple orders are detected. In this embodiment, for a 12th-order periodic impact fault, the spectral peaks at orders 12, 24, 36, and 48 can be jointly calculated. Even if a healthy background peak exists near order 12, as long as orders 24, 36, and 48 show synchronous enhancement, high-level multiple order joint evidence can still be formed, thereby reducing misjudgment based on a single spectral peak. The final alarm preferably adopts a continuous window confirmation mechanism: a fault alarm is only output when the fault confidence exceeds a threshold within multiple consecutive revolution windows.

[0059] To verify the effectiveness of this invention, an experiment was conducted: an accelerometer was installed at the bottom, left, and right sides of the main bearing of the tunnel boring machine, and a rotational speed sensor was also installed. Vibration and rotational speed signals were collected synchronously for 300 seconds. The equipment rotational speed fluctuated continuously during operation. The instantaneous rotational speed was integrated to obtain a cumulative rotational speed curve; subsequently, the vibration signals from the three channels were resampled at equal angles with a fixed number of sampling points per rotation.

[0060] The three vibration channels were subjected to resonant bandpass filtering and Hilbert envelope processing. Robust impact scores were calculated based on the sliding median and median absolute deviation of the envelopes of each channel, and then fused to obtain the global impact score.

[0061] Figure 2 This is a comparison chart of the global impact scores under healthy and faulty conditions. Under healthy conditions, there are a small number of random impacts with high amplitudes in the global impact score. Although these impacts exceed the event threshold, their angular phase relative to the axis of rotation is not concentrated, and the normalized amplitude distribution of different events varies significantly across the three measuring points. Therefore, they are classified as external environmental impacts.

[0062] Under fault conditions, a large number of periodic impacts occur in addition to random environmental impacts. After mapping the impact locations to the cumulative rotation angle, the 12-fold phase of the impacts shows a clear concentration trend, and the amplitude ratio and propagation delay of multiple impacts at the bottom, left, and right measuring points remain relatively stable. Therefore, it is determined to be a 12th-order fault impact inside the equipment.

[0063] Figure 4 This is a comparison chart of the envelope order spectrum for healthy and faulty states. The envelope order spectrum is calculated after weight reduction for the impact zone of the external environment, and spectral values ​​are extracted near orders 12, 24, 36, and 48. A 12th-order fault alarm is output based on the combined enhancement degree of multiple orders.

Claims

1. A method for fault diagnosis of tunnel boring machines based on angular domain periodicity and multi-point spatial response, characterized in that, Includes the following steps: S1. Collect relevant signals of the main bearing of the tunnel boring machine: Arrange at least two vibration sensors at different spatial positions of the main bearing of the tunnel boring machine, and set up a speed sensor. Use a unified clock to synchronously collect multi-channel vibration signals and speed signals. S2. Cumulative rotation angle calculation: The collected rotation speed signal is preprocessed, and the instantaneous rotation frequency, cumulative rotation angle and cumulative rotation number of the corresponding shaft of the main bearing are calculated based on the preprocessed rotation speed signal. An equal angle sampling grid is established, and the vibration signal of each channel is resampled into angular domain vibration signal according to the equal angle sampling grid. S3. Envelope and Impact Score: The vibration signals of each channel are detrended, bandpass filtered, or spectral kurtosis selected. The envelope signal of each channel is obtained through Hilbert transform. The robust normalized impact score of each channel is calculated based on the median and median absolute deviation of the envelope signal. The robust normalized impact scores of each channel are then fused to obtain the global impact score. The impact peak value is detected based on an adaptive threshold, and adjacent impact peak values ​​are merged into candidate impact events. S4. Characteristics of candidate impact events: Determine the occurrence time and corresponding cumulative rotation angle of each candidate impact event, and extract the response peak value or response energy, cross-channel normalized amplitude vector, inter-channel propagation delay, cross-channel waveform correlation coefficient, impact duration, kurtosis, impulse factor, frequency band energy, and instantaneous rotation speed before and after the impact event in different vibration channels. S5. Angular phase periodicity judgment: For the candidate fault order to be detected, the cumulative rotation angle corresponding to each candidate impact event is folded into a candidate fault cycle to obtain the folded angular phase; the angular phase concentration is calculated based on the folded angular phase to evaluate the periodic synchronization between the candidate impact event and the spindle rotation angle. S6. Spatial response consistency judgment: The amplitude spatial mode stability is calculated based on the distance or correlation coefficient between the cross-channel normalized amplitude vectors corresponding to different candidate impact events, the propagation delay stability is calculated based on the dispersion of the propagation delay between channels corresponding to multiple candidate impact events, and the cross-channel waveform consistency is calculated based on the correlation coefficient of the impact response waveform in different vibration channels. S7. Impact Source Fusion and Discrimination: The angular phase concentration, amplitude spatial mode stability, propagation delay stability, and cross-channel waveform consistency are fused to construct the fault confidence of the candidate fault order; based on the fault confidence, the candidate impact events are divided into internal tunnel boring machine fault impacts and external environmental impacts. S8. External shock suppression: Identify the signal segment that is determined to be an external environmental shock, and perform processes such as elimination, interpolation replacement, prediction replacement, windowing smoothing or weight reduction on the signal segment to obtain the angular domain vibration signal after external shock suppression. S9. Envelope order spectrum and multiple order combined diagnosis: Calculate the envelope order spectrum based on the angular domain vibration signal after external impact suppression, extract the candidate fault order and the spectral peaks near at least one integer multiple order, and calculate the multiple order combined evidence; fuse the multiple order combined evidence with the fault confidence, and output the shield machine main bearing fault diagnosis result based on the judgment results in multiple consecutive rotation windows.

2. The tunnel boring machine fault diagnosis method based on angular domain periodicity and multi-point spatial response as described in claim 1, characterized in that: In step S1, the vibration sensors are installed at different spatial locations on the main bearing housing, support, or foundation of the tunnel boring machine; the number of vibration sensors is not less than two, preferably three, and the three vibration sensors are respectively installed at the bottom, left side, and right side of the main bearing of the tunnel boring machine; the vibration sensors are one or more of the following: acceleration sensor, velocity sensor, displacement sensor, acoustic emission sensor, or dynamic pressure sensor; the rotational speed sensor is used to collect instantaneous rotational speed, rotational frequency, or key phase signal, and each vibration channel and the rotational speed channel are synchronously collected using a unified clock.

Images