Method and System for Monitoring Displacement During Installation of Self-Expanding Anchor Bolts

Abstract

This invention discloses a method and system for monitoring the installation displacement of self-expanding anchor bolts, belonging to the field of health monitoring and IoT sensing technology. The method includes: obtaining an event sequence through transient pulse detection technology and energy integration; obtaining a statistical feature vector through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis; obtaining a correlation attenuation scale and a dynamic correlation scale through energy density conversion, time correlation function, and percolation analysis; and obtaining a state health index and graded early warning signals through a fusion algorithm and trend analysis, and executing corresponding processing measures. This invention solves the problem of delayed early warning of sudden displacement failure of anchor bolts caused by initial interface defects due to thermal damage during installation and the inability to perceive the dynamic evolution of micro-damage in real time. It realizes online quantitative assessment and risk advance warning of the health status of the installation interface of self-expanding anchor bolts, improving the accuracy, real-time performance, and reliability of engineering decision-making in anchor bolt system fault prediction and health management.

Description

Technical Field

[0001] This invention relates to the field of health monitoring and Internet of Things sensing technology, and more specifically to a method and system for monitoring the installation displacement of a self-expanding anchor bolt. Background Technology

[0002] In fields such as civil engineering and mining that involve the safety of major structures, the reliability of mechanical anchoring connections is crucial. Traditional methods for assessing the installation quality and service status of anchor bolts mainly rely on post-hoc, local, or even destructive means such as installation torque control or pull-out tests, which make it difficult to perceive the initial defects of the installation interface and the dynamic evolution of micro-damage under long-term loads in real time.

[0003] With the development of the Internet of Things (IoT) and intelligent sensing technologies, vibration signal-based structural health monitoring, fault prediction, and health management offer new avenues for online assessment. However, existing IoT-based monitoring solutions primarily focus on macroscopic vibrations, lacking in-depth analytical capabilities for interface damage and making it difficult to construct accurate physical indicators that characterize the precursors of system instability. Furthermore, when utilizing IoT architectures for massive data aggregation and analysis, existing methods suffer from deficiencies in model adaptability and data security, leading to delays in early warning of microscopic damage evolution in advanced installation techniques such as self-expanding anchor bolts. Therefore, existing technologies have shortcomings. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method and system for monitoring the installation displacement of self-expanding anchor bolts. By acquiring raw vibration signal data, a statistical feature vector is obtained based on event sequences combined with power-law fitting technology. Based on the event sequences and statistical feature vectors, a correlation attenuation scale and a dynamic correlation scale are obtained. A state health index and graded early warning signals are derived based on the correlation attenuation scale and the dynamic correlation scale. Corresponding processing measures are then executed based on the graded early warning signals. This achieves quantitative perception, early warning, and proactive response to the dynamic evolution of microscopic damage at the anchor bolt installation interface. It solves the problem of delayed early warning for sudden failures caused by initial interface defects due to thermal damage during installation processes, which prevents the perception of the dynamic evolution of microscopic damage. This improves the monitoring system's predictive ability, response timeliness, and reliability for sudden displacement risks, providing precise closed-loop decision support for fault prediction and health management of self-expanding anchor bolts.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] This invention provides a method for monitoring the installation displacement of a self-expanding anchor bolt, applicable to self-expanding anchor bolts including diamond sintered cutting tools and cooling wax, comprising:

[0007] Acquire raw vibration signal data;

[0008] Based on the original vibration signal data, an event sequence is obtained through transient pulse detection technology and energy integration.

[0009] Based on the event sequence combined with power-law fitting technology, statistical feature vectors are obtained through statistical distribution analysis, energy distribution analysis and detrended fluctuation analysis.

[0010] Based on the event sequence and the statistical feature vector, the correlation decay scale and dynamic correlation scale are obtained through energy density conversion, time correlation function and percolation analysis.

[0011] Based on the aforementioned correlation decay scale and dynamic correlation scale, a state health index is calculated through a fusion algorithm. A graded early warning signal is obtained through a trend analysis algorithm and a multi-level threshold criterion logic. Corresponding processing measures are then executed based on the graded early warning signal.

[0012] As a further improvement of the present invention, the step of obtaining an event sequence based on the original vibration signal data through transient pulse detection technology and energy integration includes:

[0013] Based on the original vibration signal data, the denoised vibration signal data is obtained by digital bandpass filtering.

[0014] Based on the denoised vibration signal data, the pulse start point is obtained using transient pulse detection technology;

[0015] An event sequence is obtained by energy integration based on the pulse initiation point. The event sequence includes micro-rupture events and the relative energy of the events.

[0016] As a further improvement of the present invention, the step of obtaining a statistical feature vector based on the event sequence combined with power-law fitting technology through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis includes:

[0017] Based on the event sequence, the event clustering index is obtained through statistical distribution analysis and power-law fitting techniques.

[0018] Based on the event sequence, the energy power law exponent is obtained through energy distribution analysis and power law fitting techniques.

[0019] Based on the event sequence, the long-range correlation index of the event sequence is obtained through detrended fluctuation analysis;

[0020] Based on the event clustering index, energy power law index, and long-range correlation index, a statistical feature vector is obtained by integration.

[0021] As a further improvement of the present invention, the step of obtaining the event clustering index based on the event sequence through statistical distribution analysis and power-law fitting techniques includes:

[0022] Based on the event sequence, a set of event waiting time intervals is obtained by calculating the adjacent time differences;

[0023] Based on the set of event waiting time intervals, the waiting time distribution is obtained through statistical distribution analysis and probability density calculation.

[0024] Based on the aforementioned waiting time distribution, the event clustering index is obtained using a power-law fitting technique based on least squares.

[0025] As a further improvement of the present invention, the step of obtaining the long-range correlation index of the event sequence through detrended volatility analysis based on the event sequence includes:

[0026] Based on the event sequence, a one-dimensional time signal sequence is obtained through equal-interval counting and sequence reconstruction.

[0027] Based on the one-dimensional time signal sequence, a detrended fluctuation sequence is obtained through detrended fluctuation analysis;

[0028] Based on the detrended volatility sequence, a volatility function is calculated, and a long-range correlation index is obtained by fitting the volatility function using the least squares method.

[0029] As a further improvement of the present invention, the step of obtaining the correlation decay scale and dynamic correlation scale based on the event sequence and the statistical feature vector through energy density transformation, time correlation function and percolation analysis includes:

[0030] Based on the event sequence and the statistical feature vector, an energy density sequence is obtained through energy density transformation.

[0031] Based on the energy density sequence, the associated decay scale is obtained through time correlation function and curve fitting techniques;

[0032] Based on the aforementioned correlation decay scale and event sequence, a spatiotemporal correlation matrix is ​​obtained by constructing a dynamic directed network;

[0033] Based on the spatiotemporal correlation matrix, the largest correlation cluster is determined through percolation analysis and network connectivity identification techniques.

[0034] Based on the largest association cluster, the dynamic association scale is obtained by calculating the energy percentage.

[0035] As a further improvement of the present invention, the step of obtaining the spatiotemporal correlation matrix by constructing a dynamic directed network based on the correlation decay scale and event sequence includes:

[0036] Based on the event sequence, a set of dynamic directed network nodes is constructed using node initialization technology;

[0037] Based on the aforementioned correlation decay scale, the potential causal relationship between any two events is determined using dynamic time window judgment rules.

[0038] Based on the potential causal relationships and event sequences, the weights of directed edges are obtained through an exponentially decaying weight model.

[0039] Based on the set of nodes in the dynamic directed network and all weighted directed edges, a spatiotemporal correlation matrix is ​​obtained through matrix generation techniques.

[0040] As a further improvement of the present invention, the state health index is calculated by a fusion algorithm based on the correlation decay scale and the dynamic correlation scale, a graded early warning signal is obtained by a trend analysis algorithm and a multi-level threshold criterion logic, and corresponding processing measures are executed based on the graded early warning signal, including:

[0041] Based on the aforementioned correlation decay scale and dynamic correlation scale, the state health index is calculated using a fusion algorithm.

[0042] A normal baseline is established based on historical health indices using baseline learning and trend analysis algorithms.

[0043] Based on the deviation of the real-time calculated health status index from the normal benchmark and the changing trend of the health status index, risk assessment is performed through multi-level threshold judgment logic to obtain graded early warning signals.

[0044] Based on the tiered early warning signal, corresponding processing measures are executed according to preset response rules. The processing measures include at least one of increasing the data acquisition frequency, triggering system self-check, generating operation and maintenance prompts, and sending emergency shutdown commands.

[0045] This invention provides a self-expanding anchor bolt device, which applies the above-mentioned self-expanding anchor bolt installation displacement monitoring method, including:

[0046] Diamond sintered cutting tools are used to directly cut and enlarge holes in the substrate during anchor bolt installation.

[0047] Cooling wax is stored in an annular cooling wax storage cavity located inside the reaming head of the diamond sintering cutter, and is used to cool the diamond sintering cutter during anchor bolt installation.

[0048] This invention provides a self-expanding anchor installation displacement monitoring system, comprising the aforementioned self-expanding anchor device and a server, wherein the server includes:

[0049] A signal acquisition module is used to acquire raw vibration signal data, which is obtained through the self-expanding hole anchor bolt device.

[0050] The event extraction module is used to obtain an event sequence based on the original vibration signal data through transient pulse detection technology and energy integration;

[0051] The intelligent analysis module is used to obtain statistical feature vectors based on the event sequence and power-law fitting technology through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis; and to obtain correlation decay scales and dynamic correlation scales based on the event sequence and the statistical feature vectors through energy density conversion, time correlation functions, and percolation analysis.

[0052] The early warning output module is used to calculate the state health index based on the correlation attenuation scale and the dynamic correlation scale through a fusion algorithm, obtain a graded early warning signal through a trend analysis algorithm and a multi-level threshold criterion logic, and execute corresponding processing measures based on the graded early warning signal.

[0053] This invention utilizes raw vibration signal data acquired through IoT intelligent sensing. It obtains event sequences using transient pulse detection technology and energy integration, generates statistical feature vectors using power-law fitting technology, and further derives correlation attenuation and dynamic correlation scales based on the event sequences and statistical feature vectors. Based on these correlation attenuation and dynamic correlation scales, a state health index and graded early warning signals are obtained through a fusion algorithm and trend analysis. Corresponding processing measures are automatically executed according to the graded early warning signals to form a complete health management closed loop. This achieves continuous quantitative sensing, advanced graded early warning, and proactive response to the dynamic evolution of microscopic damage at the installation interface of self-expanding anchors. It solves the problem of delayed early warning of sudden failures caused by initial interface defects due to thermal damage during installation processes, which prevents the perception of the dynamic evolution of microscopic damage. This improves the monitoring system's ability to predict displacement risks, the timeliness of early warnings, the accuracy of response, and the reliability of decision-making. It provides a precise, adaptive, and closed-loop control technology for fault prediction and health management of anchor systems. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating the steps of the self-expanding anchor installation displacement monitoring method of the present invention.

[0055] Figure 2 This is a structural diagram of the self-expanding anchor bolt device of the present invention;

[0056] Figure 3 This is a flowchart illustrating the steps of preprocessing the original vibration signal data and obtaining the event sequence in this invention.

[0057] Figure 4 This is a flowchart illustrating the steps involved in obtaining statistical feature vectors according to the present invention.

[0058] Figure 5 This is a schematic diagram of the self-expanding hole type anchor bolt installation displacement monitoring system of the present invention.

[0059] Reference numerals: 1. Wax injection hole; 2. Diamond sintered cutting tool; 3. Annular cooling wax storage chamber; 4. Locking key plate; 5. Hole reamer; 6. Support sleeve; 7. Anti-loosening nut; 8. Wax storage chamber. Detailed Implementation

[0060] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof.

[0061] Identical parts are indicated by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "up," and "down" used in the following description refer to directions in the accompanying drawings, while the terms "bottom surface," "top surface," "inner," and "outer" refer to directions toward or away from the geometric center of a specific part, respectively.

[0062] The term "and / or" in the following text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0063] like Figure 1 As shown, the present invention provides a method for monitoring the installation displacement of a self-expanding anchor bolt, comprising:

[0064] Acquire raw vibration signal data;

[0065] Based on the original vibration signal data, an event sequence is obtained through transient pulse detection technology and energy integration.

[0066] Based on event sequence combined with power law fitting technology, statistical feature vectors are obtained through statistical distribution analysis, energy distribution analysis and detrended fluctuation analysis.

[0067] Based on event sequences combined with statistical feature vectors, the correlation decay scale and dynamic correlation scale are obtained through energy density conversion, time correlation function and percolation analysis.

[0068] Based on the correlation decay scale and dynamic correlation scale, the state health index is calculated by a fusion algorithm. The graded early warning signal is obtained by a trend analysis algorithm and a multi-level threshold criterion logic. The corresponding processing measures are then executed based on the graded early warning signal.

[0069] The original vibration signal data is the axial vibration state signal of the self-expanding anchor system. It is continuously collected by the MEMS accelerometer built into the self-expanding anchor at a preset sampling rate, and then packaged by the embedded processor and uploaded to the cloud server through low power wide area network protocols such as NB-IoT / LoRa. The preset sampling rate is obtained by calibrating the time scale characteristics and passband frequency range of the micro-fracture event at the interface of the self-expanding anchor.

[0070] The transient pulse detection technology is a transient event detection technology using the STA / LTA algorithm. The STA / LTA algorithm is a classic algorithm that calculates the ratio of short-term energy / amplitude average (STA) to long-term background noise average (LTA) through a sliding window. When the ratio exceeds a preset ratio threshold, the event is determined to have occurred. The preset ratio threshold is determined by comprehensively considering the background noise statistics of the self-expanding anchor system under no-load conditions, the on-site environmental noise calibration, and the training of historical micro-fracture event samples. The STA / LTA algorithm is used for event detection in low signal-to-noise ratio scenarios. It is used to identify the pulse start point of energy abrupt changes caused by interface damage from the filtered original vibration signal. Since the cooling wax of the self-expanding anchor melts into a liquid state during the cutting and expanding process, it dissipates frictional heat through phase change heat absorption, effectively reducing tool thermal damage and substrate powder sintering. This solves the carbonization problem caused by high frictional temperature of diamond sintered cutting tools, optimizes the initial anchoring interface quality, and reduces irregular vibration interference during the cutting process. This makes the filtered signal more accurately reflect the micro-fracture state of the anchor and substrate interface.

[0071] Energy integration is a processing method that uses the identified pulse start point as the center, extracts a fixed time window waveform, and calculates its energy. It is used to quantify the relative energy of each discrete micro-event. The event sequence is a dataset describing the time-energy correlation of micro-fracture events. It is obtained by combining the energy integration after determining the event start time through transient pulse detection technology. Specifically, it is an array set including event time and event energy within a time window, representing the set of discrete micro-damages occurring within an analysis period. The diamond sintered cutting tool of the self-expanding anchor is made by high-temperature and high-pressure sintering process. It has extremely high hardness and wear resistance. It can rotate to form holes in high-strength substrates such as concrete and rock without pre-drilling. It can also continuously cut when encountering steel bars, ensuring the continuity of hole formation and the accuracy of hole diameter. This makes the anchor and substrate interface fit tightly, providing a stable physical basis for event sequence extraction.

[0072] Power-law fitting is a data analysis technique that obtains the power-law exponent by fitting the event interval distribution and energy distribution. It is used to quantify the statistical distribution law of waiting time and energy in an event sequence, where waiting time is the time interval between adjacent events in the event sequence. Statistical distribution analysis is a data analysis method for analyzing the event interval distribution, used to obtain the event clustering index. A decrease in the event clustering index indicates that events are shifting from random to clustered. Energy distribution analysis is a data analysis method for analyzing the energy magnitude distribution of all events in an event sequence. By fitting their complementary cumulative distribution, it obtains the energy power-law exponent. A decrease in the energy power-law exponent indicates an increase in the proportion of high-energy damage events. Detrended fluctuation analysis (DFA) is a scaling analysis method for quantifying the long-range power-law correlation and self-similarity of non-stationary time series. By analyzing the event sequence, it obtains the long-range correlation index, used to characterize the memory effect and persistence of microscopic damage events over time. An increase in the long-range correlation index indicates enhanced sequence persistence.

[0073] The statistical feature vector is a vector data describing the macroscopic statistical regularity of an event. It is obtained by processing the event sequence through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis. Specifically, it is a three-dimensional vector including the event clustering index, energy power law index, and long-range correlation index. Thanks to the precise hole-expanding capability and stable interface bonding effect of the self-expanding anchor bolt, the temporal and energy characteristics of micro-damage in the event sequence are more regular, making the calculation of the statistical feature vector more accurate and effectively characterizing the macroscopic evolution trend of interface micro-damage.

[0074] Energy density conversion is a process that transforms discrete event sequences into continuous energy density sequences. By using fixed small time intervals as windows, the sum of event energies within each window is calculated to obtain a continuous event energy density sequence, which provides a continuous data foundation for subsequent time correlation function calculations. Because the cooling wax of the self-expanding anchor bolt reduces energy loss and environmental interference during the cutting process, the original vibration signal has a higher signal-to-noise ratio, and the converted energy density sequence better reflects the true distribution of damage energy. The fixed small time interval is obtained by comprehensively calibrating the time scale characteristics of micro-fracture events at the interface of the self-expanding anchor bolt, the preset sampling rate, and the sliding window length of the STA / LTA algorithm. The time correlation function is a function that describes the correlation of the energy density sequence at different delay times, used to quantify the similarity or memory length of the sequence itself at different time points.

[0075] Percolation analysis is an analytical method that treats the dynamic spatiotemporal correlation matrix as a network, calculates its weakly connected components, and determines the largest connected subgraph. It is used to obtain the dynamic correlation scale, which is the ratio of the sum of event energies to the total energy in the largest connected subgraph. The correlation decay scale is a physical parameter characterizing the correlation decay characteristics of the energy density sequence. It is obtained by fitting the exponential decay curve of the time correlation function. Its physical meaning is the current dynamic relaxation time of the system. Its growth is a direct indicator that the system is undergoing critical slowdown and tending towards instability.

[0076] The dynamic correlation scale is a scalar parameter characterizing the degree of correlation of system damage. It is obtained by performing percolation analysis on the constructed spatiotemporal correlation network and calculating the ratio of the sum of the energy of all events contained in the largest connected subgraph to the total analysis energy. The value range is [0,1]. The larger the dynamic correlation scale, the more likely that most of the damage energy has been correlated into a whole. It is used to directly quantify the critical degree of micro-damage evolving from an isolated state to a highly correlated and coordinated state in spatiotemporal space. The structural stability of the self-expanding hole anchor avoids false signals caused by non-damaging loosening and ensures that the correlation attenuation scale and the dynamic correlation scale can accurately reflect the dynamic evolution process of micro-damage.

[0077] The State Health Index is a comprehensive evaluation indicator that integrates the correlation decay scale and the dynamic correlation scale. It is calculated using a weighted fusion algorithm, with the weights adaptively adjusted based on the anchor bolt installation conditions and baseline learning data. A larger value indicates a more stable anchor bolt installation state. The trend analysis algorithm analyzes the time evolution trend of the correlation decay scale to identify its growth trend and assist in early warning judgment. The multi-level threshold criterion logic is an early warning judgment rule constructed based on the baseline value and its fluctuation range learned in the stable phase. The stable phase is the initial service phase after the self-expanding anchor bolt is installed, the interface micro-fracture activity tends to converge, and the dynamic correlation scale and correlation decay scale remain stable. The baseline value includes the dynamic correlation scale baseline, the correlation decay scale baseline, and the State Health Index baseline, which are used to trigger different levels of early warning based on the deviation of the real-time value from the baseline value. The graded early warning signal is an early warning indication signal that distinguishes risk levels. It is obtained through the trend analysis algorithm and the multi-level threshold criterion logic, including a yellow warning for the attention level and a red alert for the action level.

[0078] This embodiment achieves precise perception of the initial defects and dynamic evolution of micro-damping at the anchor bolt installation interface by deeply integrating a self-expanding anchor bolt structure, which includes precise hole enlargement using diamond sintered cutting tools, cooling wax for temperature reduction and disturbance reduction, and stable data acquisition using a built-in sensing cavity, with an IoT monitoring method. This solves the problem of delayed early warning of sudden failures caused by initial interface defects due to thermal damage during installation and the inability to perceive the dynamic evolution of micro-damping. It improves the monitoring system's ability and reliability in predicting sudden displacement risks and provides precise decision support for the operation and maintenance of self-expanding anchor bolts.

[0079] Furthermore, this embodiment provides a self-expanding anchor bolt device, comprising:

[0080] Diamond sintered cutting tools are used to directly cut and enlarge holes in the substrate during anchor bolt installation.

[0081] Cooling wax, stored in an annular cooling wax storage cavity inside the reaming head of the diamond sintering cutter, is used to cool the diamond sintering cutter during anchor bolt installation.

[0082] Among them, the diamond sintered cutting tool is made through a high-temperature and high-pressure sintering process. The composite structure formed by the diamond particles and the metal binder gives the tool extremely high hardness and wear resistance. It can directly rotate and cut into high-strength substrates such as concrete and rock to form anchor holes without pre-drilling, thus achieving self-reaming. When encountering reinforcing bars in the reaming path, the diamond sintered cutting tool can continue cutting through, thereby ensuring the continuity of the hole formation. The annular cooling wax storage cavity is a closed annular structure inside the reaming head, which stores solid cooling wax. The cooling wax is a solid wax material, which is distributed through the annular distribution of the annular cooling wax storage cavity. The system achieves uniform storage and coverage of the high-heat area around the cutting tool, enabling the cooling wax in the annular cooling wax storage cavity to rapidly and uniformly melt into a liquid state during the cutting and reaming process of the diamond sintered cutting tool. This phase transition from solid to liquid absorbs and carries away a large amount of frictional heat, thereby effectively reducing the temperature of the area in contact with the hole wall and playing a cooling role. This aims to reduce tool thermal damage, inhibit substrate powder sintering, suppress the carbonization problem caused by high frictional temperature of the diamond sintered cutting tool, ensure the smooth progress of the reaming process, and optimize the quality of the initial anchoring interface.

[0083] Specifically, such as Figure 2As shown, the self-expanding anchor used in this embodiment includes the following structure: wax injection hole 1, diamond sintered blade 2, annular cooling wax storage cavity 3, locking key 4, reaming head 5, support sleeve 6, anti-loosening nut 7, and wax storage chamber 8. From the perspective of anchor application principle, the high hardness cutting characteristics of the diamond sintered blade 2 solve the problems of easy obstruction by reinforcing bars and insufficient hole diameter accuracy during traditional anchor hole expansion, ensuring that the anchor can be accurately installed in the preset position, making the anchor and the substrate interface fit tightly, and laying a structural foundation for stable acquisition of vibration signals. Cooling wax is injected through the wax injection hole 1 and stored in the wax storage chamber 8 and the annular cooling wax storage cavity 3. The annular cooling wax storage cavity 3 is located inside the reaming head 5 of the diamond sintering tool, ensuring that the cooling wax stored inside the annular cooling wax storage cavity can directly cover and act on the main high-heat area generated by cutting, thereby giving full play to the thermal melting and cooling function of the cooling wax, reducing energy loss and irregular vibration during the cutting process, reducing the influence of environmental interference on the vibration signal, and making the acquired original vibration signal more accurately reflect the micro-fracture state of the anchor bolt and substrate interface, thus improving the effectiveness and reliability of the signal. The locking key plate 4 and the support sleeve 6 are used for the mechanical locking and support of the anchor bolt, and the anti-loosening nut 7 is used for tightening and preventing loosening, together ensuring the structural stability of the anchor bolt during long-term use.

[0084] This embodiment of the IoT-based intelligent sensing displacement monitoring method is based on vibration signals collected from self-expanding anchors installed and fastened to a specific structure within the substrate. The sensing cavity built into the anchor provides installation space for the MEMS accelerometer, ensuring that the sensor can capture vibration signals in real time and accurately, caused by micro-crack propagation and frictional slippage at the anchor-substrate interface due to stress changes. These signals serve as the core data source for the monitoring method, and their stability and accuracy directly determine the reliability of subsequent event sequence extraction, statistical feature analysis, and early warning judgment. Therefore, the self-expanding anchor in this embodiment is not only the object of the monitoring method, but its structural characteristics, including precise hole-expanding capability, stable signal acquisition environment, and anti-interference design, also provide crucial physical support for displacement monitoring during the installation of the self-expanding anchor. This achieves a deep integration of anchor installation function and displacement monitoring function, ensuring that the monitoring method can effectively identify displacement risks during the installation and use of the self-expanding anchor.

[0085] Furthermore, this embodiment provides a step for obtaining an event sequence based on raw vibration signal data through transient pulse detection technology and energy integration, including:

[0086] Based on the original vibration signal data, the noise-reduced vibration signal data is obtained by digital bandpass filtering;

[0087] Based on the denoised vibration signal data, the pulse start point is obtained through transient pulse detection technology;

[0088] The event sequence is obtained by energy integration based on the pulse initiation point. The event sequence includes micro-fracture events and the relative energy of the events.

[0089] Digital bandpass filtering is a signal processing technique that allows signals within a specific frequency range to pass through while suppressing frequency components outside that range. It is used to filter out low-frequency structural vibration interference and high-frequency electronic measurement noise from the original vibration signal data, while retaining the mid-to-high frequency vibration components related to the micro-fractures at the interface of the self-expanding anchor bolt. Because the cooling wax of the self-expanding anchor bolt reduces irregular vibrations during the cutting process, there are fewer interference components in the original signal, and the filtered signal can more accurately retain the damage characteristic frequency components. The denoised vibration signal data is a more regular time-domain waveform of the vibration acceleration sequence after digital bandpass filtering of the original vibration signal data, which retains the vibration characteristics related to the interface of the self-expanding anchor bolt. The characteristic frequency components related to damage are selected by eliminating irrelevant interference signals. The pulse start point refers to the starting point in the denoised vibration signal data where the signal energy or amplitude rises sharply in a short period of time, triggered by a micro-fracture event. A micro-fracture event is a damage process such as the propagation of micro-cracks or frictional slippage at the interface between the self-expanding anchor and the substrate due to stress changes. It is manifested as a transient pulse waveform in the vibration signal and is the core event reflecting the interface state change of the self-expanding anchor. The event relative energy is a physical scalar value used to quantify the intensity of a single micro-fracture event. It is calculated by summing the squares of the vibration signal amplitude within a fixed duration window centered on the pulse start point.

[0090] Specifically, such as Figure 3 As shown, the received raw vibration signal data is first... Perform digital bandpass filtering, with the preset passband frequency range of the digital bandpass filter being [specified value]. ,in This is the lower passband frequency, used to suppress low-frequency environmental and structural vibrations. The upper passband frequency is used to suppress high-frequency noise. This filtering process suppresses low-frequency structural vibration interference and high-frequency noise interference in the original vibration signal data, resulting in denoised vibration signal data that retains the frequency components characteristic of the self-expanding anchor interface damage. In other words, the frequency components are concentrated in... Denoising vibration signal data within the range Then, the denoised vibration signal data... Applying STA / LTA transient pulse detection technology, the sliding window length of the STA algorithm is set to... The sliding window length of the LTA algorithm is ,in The ratio of short-time average energy to long-time average background energy of the denoised vibration signal data is calculated in real time using a sliding window. ,when First time exceeding the preset ratio threshold At that time, the moment corresponding to the current sliding window is determined as the pulse start point. Then, starting from each pulse... Centered on, from Extracting time intervals The signal segment within, in which The event time window length is determined by comprehensively calibrating the event based on the time-scale characteristics of micro-fracture events at the interface of the self-expanding anchor bolt, the sliding window length of the STA / LTA algorithm, and the preset sampling rate. The amplitudes of all data points within this segment are then summed by squares, i.e., the calculation... ,in It belongs to this time interval, thus obtaining the pulse start point. The corresponding event relative energy Finally, all pulse start points and their corresponding event relative energies identified within an analysis time window are combined into a binary list in chronological order. This constitutes an event sequence characterizing the microscopic damage activity at the interface of the self-expanding anchor bolt within a time window, wherein... This represents the total number of micro-fracture events within the time window.

[0091] For example, this embodiment assumes that the vibration signal of a certain anchor bolt is analyzed, and the passband of the digital bandpass filter is set. The lower cutoff frequency Upper limit cutoff frequency The STA / LTA algorithm parameters are set to a short time window. Long window Preset ratio threshold Preset event time window length The cloud server received a message with a duration of [duration missing]. Sampling frequency is raw vibration signal data Then, the denoised vibration signal data is first obtained by passing it through the aforementioned bandpass filter. ; for the denoised vibration signal data The ratio of short-time average energy to long-time average background energy of the denoised vibration signal data was obtained by using a sliding window calculation based on the STA / LTA algorithm. ,exist First time to meet The first pulse start point is obtained at this time. Similarly, we can identify them sequentially. and Then, taking the starting point of each pulse as the center, the denoised vibration signal data is extracted. exist For signals within a time interval, calculate the amplitude of all data points within that segment and sum the squares to obtain the relative energy of the event. Calculated in this way, , ; Ultimately A total of [number] were identified within the analysis window. A series of micro-fracture events were obtained, resulting in an event sequence. It should be noted that the preset filter passband, STA / LTA algorithm window length and threshold, energy integration window length, and other parameter values ​​in this embodiment are merely examples. Those skilled in the art can make adaptive adjustments based on actual signal characteristics, noise levels, and the physical scale of the damage event. This embodiment does not impose any limitations on these adjustments.

[0092] This embodiment achieves precise removal of interference signals through digital bandpass filtering, reliable identification of the pulse initiation point of micro-fracture events through the STA / LTA algorithm, quantitative calculation of the relative energy of events through fixed time window energy integration, and formation of a standardized event sequence through time-series arrangement. This enables the accurate and efficient extraction of micro-fracture event sequences reflecting the interface damage state of self-expanding anchor bolts from complex vibration signals, improving the anti-interference capability and accuracy of event identification, and providing high-quality basic data support for subsequent statistical feature analysis and critical correlation scale calculation.

[0093] Furthermore, this embodiment provides a step-by-step approach to obtain statistical feature vectors based on event sequence combined with power-law fitting techniques, through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis, including:

[0094] Based on event sequences, an event clustering index is obtained through statistical distribution analysis and power-law fitting techniques.

[0095] Based on event sequences, the energy power law exponent is obtained through energy distribution analysis and power law fitting techniques.

[0096] Based on the event sequence, the long-range correlation index of the event sequence is obtained through detrended fluctuation analysis;

[0097] Based on the event clustering index, energy power law index, and long-range correlation index, a statistical feature vector is obtained by integration.

[0098] Among them, the event clustering index is a power-law index characterizing the clustering characteristics of the event occurrence time distribution in an event sequence. It is obtained by statistically analyzing the waiting time intervals of adjacent events in the event sequence and fitting the probability density distribution of the waiting time using power-law fitting techniques. A decrease in its value indicates that the events are shifting from a random distribution to a clustered distribution. The energy power-law index is a power-law index characterizing the energy distribution characteristics of events in an event sequence. It is obtained by sorting the energy of all events in the event sequence, calculating the cumulative probability, and fitting the complementary cumulative distribution of energy using power-law fitting techniques. A decrease in its value indicates an increase in the proportion of high-energy damage events. The long-range correlation index is a Hearst-like index characterizing the long-range power-law correlation and self-similarity of an event sequence. It is obtained by processing the reconstructed one-dimensional time signal sequence through detrending fluctuation analysis, and by detrending, fluctuation function calculation and fitting. An increase in its value indicates an enhancement of the sequence's persistence and memory effect.

[0099] Specifically, such as Figure 4 As shown, firstly based on the event sequence By calculating the time difference between adjacent events ,in , to obtain A set of event waiting time intervals with time differences Then, for the event wait time interval set... Statistical distribution analysis was performed, and the probability density of time differences occurring within each waiting time interval was calculated using kernel density estimation to obtain the waiting time distribution. .

[0100] Specifically, kernel density estimation uses a set of event wait time intervals. As input, this set is directly derived from the time-series records of damage processes such as microcrack propagation and frictional slip at the anchor-substrate interface caused by stress changes. It reflects the clustering or dispersion characteristics of damage events along the time axis. The discrete data is then analyzed using kernel density estimation. Perform smooth probability density estimation and output a continuous probability density function. This establishes a bridge between discrete event time intervals and continuous statistical distributions. When damage occurs in a random phase... It exhibits an approximately exponential distribution characteristic; as damage gradually clusters and the system approaches critical instability, small intervals... The probability of its occurrence increased significantly, leading to Presenting the power-law tail in a double logarithmic coordinate system, this method avoids the sensitivity of the histogram method to interval division, more accurately captures the overall shape and tail characteristics of the waiting time distribution, and provides stable and continuous probability density estimates for subsequent power-law fitting based on least squares, ensuring the event clustering index... Physical reliability.

[0101] Based on waiting time distribution For data points in a log-log coordinate system, a power-law fitting technique based on least squares is used to find a straight line. The best fit yields the event clustering index. The proportional sign This represents the linear proportional relationship under double logarithmic coordinates in power-law fitting, i.e. and The relationship between them is approximately linear.

[0102] Then the event sequence All events relative energy Sort the energy sets from largest to smallest to obtain an ordered energy set. ; Calculate each energy value in an ordered energy set The cumulative probability of that energy value is calculated as the proportion of events with a value greater than or equal to that value out of the total number of events. The distribution comprising all data points including the relative energy and cumulative probability of events is called the complementary cumulative distribution. Based on the linear segment of the complementary cumulative distribution in log-log coordinates, a power-law fitting technique based on least squares is used to find the straight line. The best fit yields the energy power law exponent. .

[0103] Next, the event sequence Total time length Divided into By dividing the data into several equal short-range time intervals, the number of micro-fracture events occurring within each short-range time interval is counted, resulting in a length of [missing value]. One-dimensional time signal sequence ,in For one-dimensional time signal sequences Detrended volatility analysis involves eliminating local trends through integration, segmentation, and polynomial fitting within each segment, ultimately yielding a detrended volatility sequence. At different time scales Functional relationship on; through fitting and The linear relationship between them, the slope of which is the long-range correlation index. .

[0104] Event clustering index The value range is between 1.0 and 3.0. A value close to 2.0 indicates that the events occur according to a random Poisson distribution. When the value is less than 2.0 and decreases, it indicates that the event gradually shifts from a random Poisson distribution to a clustered distribution over time, and the clustering of micro-damage increases; energy power-law exponent The value ranges from 1.0 to 2.5. A decreasing value indicates an increased proportion of high-energy damage events in the total events, suggesting that the energy release structure within the system is becoming unstable; long-range correlation index The value range is between 0 and 1.5. Time indicates that the sequence is uncorrelated white noise. This indicates that the sequence has a positive long-range correlation, meaning that a period of high activity is more likely to be followed by another period of high activity. The increased numerical value indicates an enhanced memory effect and persistence of microscopic damage events, suggesting the system is approaching a critical slowing state. Finally, the three scalar parameters, the event clustering index, will be calculated. Energy power law index Long-range correlation index By combining them in sequence, a three-dimensional statistical feature vector is formed. This is to comprehensively describe the macroscopic statistical laws governing the microscopic damage activity at the interface of self-expanding anchor bolts within the current time window.

[0105] For example, this embodiment assumes that an analysis window contains... Event sequence of micro-fracture events First, calculate the time interval between adjacent events to obtain the included... A set of event waiting time intervals with time differences ,in , , indicating the first An index of micro-fracture events was generated; then, the waiting time distribution was power-law fitted using kernel density estimation, when the fitting interval was... The event clustering index is obtained by using the least squares method. Then The energy values ​​of each event are sorted from largest to smallest, and their complementary cumulative distribution is calculated within the logarithmic coordinate range of the energy values. Power-law fitting is performed within the inner region to obtain the energy power-law exponent. Next, the total duration will be... by Divide into intervals By dividing the time intervals and counting the number of events in each interval, a one-dimensional time signal sequence is obtained. Detrending fluctuation analysis was performed on the one-dimensional time signal sequence, and fitting was performed within the time scale range. and The linear relationship is used to obtain its slope, which is the long-range correlation index. The final integrated statistical feature vector of this analysis window is: Among them, when the event clustering index When the event clustering index is less than the typical Poisson process value of 2.0, it indicates that the events have a certain degree of clustering; the energy power law index Within the common range Within this range, it is indicated that the energy distribution conforms to a power-law characteristic; long-range correlation index This indicates that the event sequence has persistence, meaning that one period of high activity is more likely to be followed by another period of high activity. It should be noted that the preset fitting interval, time division interval, and detrending fluctuation analysis scale range in this embodiment are merely examples. Those skilled in the art can make adaptive adjustments based on the density, energy range, and temporal correlation characteristics of the actual event sequence; this embodiment does not impose any limitations on this.

[0106] This embodiment quantifies the cluster characteristics of events through statistical distribution analysis and power-law fitting techniques, characterizes the energy distribution pattern through energy distribution analysis and power-law fitting techniques, mines long-range correlations of sequences through detrended fluctuation analysis, and forms standardized statistical feature vectors through feature integration. This achieves a comprehensive and accurate characterization of the macroscopic evolution trend of event sequences of microscopic damage activities at the anchor bolt interface, improves the scientificity and effectiveness of feature characterization and the comprehensiveness and robustness of state description, and provides a reliable statistical basis for subsequent critical correlation scale calculations.

[0107] Furthermore, this embodiment provides a step for obtaining an event clustering index based on an event sequence through statistical distribution analysis and power-law fitting techniques, including:

[0108] Based on the event sequence, the set of event waiting time intervals is obtained by calculating the adjacent time differences;

[0109] Based on the set of event waiting time intervals, the waiting time distribution is obtained through statistical distribution analysis and probability density calculation.

[0110] Based on the waiting time distribution, the event clustering index is obtained by power-law fitting technique based on least squares.

[0111] Here, the adjacent time difference is the difference between the occurrence times of two micro-rupture events that are sequentially adjacent in the event sequence; the event waiting time interval set is a mathematical set consisting of all adjacent time differences within an analysis window; the probability density is a function describing the likelihood of a random variable taking a value near a certain point, i.e., the probability of a waiting time occurring within a unit time interval; the least squares-based power law fitting technique is a mathematical optimization method that determines the optimal power law relationship parameters by minimizing the sum of squared vertical distances between the fitted line and the data points in the double logarithmic coordinate system, and is used to extract the event clustering index from the waiting time distribution.

[0112] Specifically, first obtain the event sequence. ,in For the first The timing of micro-fracture events and Calculate in sequence ,in Thus, we obtain from A set of event wait time intervals consisting of 10 elements Then, for the event wait time interval set... Perform statistical distribution analysis, and set the range of waiting time values ​​as follows: ,in It is the minimum value in the set of event wait time intervals. The maximum value in the set of event wait time intervals is used to evenly divide this range into... The group interval is The interval is defined, and the number of adjacent time differences contained in each interval is counted, i.e., the frequency. ,in Calculate the probability density of each interval. This yields the waiting time distribution with the midpoint of the interval as the x-axis and the probability density as the y-axis. ,in For the first The midpoint value of each interval; finally, in a double logarithmic coordinate system, the waiting time distribution is fitted to a power-law distribution model using a power-law fitting technique based on least squares. ,in The normalization constant is The event clustering index is obtained by taking the logarithm of the model. ,make Transformed into a linear model By minimizing ,in , The event clustering index is obtained by solving the problem. .

[0113] Based on the temporal distribution characteristics of event sequences, this embodiment transforms discrete event occurrence time sequences into a stable scalar index that reflects the degree of clustering of event occurrence patterns through adjacent time difference calculation, waiting time distribution construction, and least squares power law fitting. This improves the accuracy of identifying the evolution trend of damage event occurrence patterns and provides core parameter support for subsequent statistical feature vector construction.

[0114] Furthermore, this embodiment provides a step for obtaining the energy power law exponent based on an event sequence through energy distribution analysis and power law fitting techniques, including:

[0115] Based on the event sequence, a complementary cumulative distribution is obtained through energy distribution analysis, including sorting and cumulative probability calculation.

[0116] Based on complementary cumulative distribution, the energy power law exponent is obtained by power law fitting technique based on least squares.

[0117] The cumulative probability is the proportion of micro-fracture events with energy greater than or equal to a preset energy threshold in the event sequence to the total number of events. The preset energy threshold is determined by comprehensively considering the background noise calibration of the self-expanding anchor system under no-load conditions, the statistical analysis of on-site environmental noise, and the baseline data from the learning of the stable phase. The complementary cumulative distribution is a distribution curve and functional relationship constructed with event energy as the abscissa and the corresponding cumulative probability as the ordinate. It is obtained by calculating the cumulative probability point by point after sorting the event energies and is used to intuitively reflect the distribution pattern of events with different energy levels.

[0118] Specifically, firstly, based on the event sequence Extract the relative energy of all micro-fracture events, and arrange the energy values ​​in the set in descending order to construct an ordered energy set. ,in Then, the sorted ordered energy set... Perform cumulative probability calculations for each sorted energy value. ,in Statistically ordered energy set greater than or equal to Number of events The corresponding cumulative probability is calculated. ,in, , And thus obtain from The dataset consists of complementary cumulative distributions; finally, using a power-law fitting technique based on least squares, the complementary cumulative distributions are fitted into a power-law distribution model. ,in The normalization constant is The energy power-law exponent is obtained by taking the logarithm of the model. ,make Transformed into a linear model The complementary cumulative distribution dataset Transform into ,in By minimizing Solving for the results The value of is obtained, which gives the energy power law exponent. .

[0119] Based on the energy scaling characteristics of event sequences, this embodiment achieves the quantitative extraction of the energy distribution pattern of micro-fracture events through energy sorting, complementary cumulative distribution analysis, cumulative probability calculation, and least squares power law fitting. This enhances the sensitivity to changes in the proportion of high-energy damage events, improves the quantification and scientific level of judging the evolution trend of damage intensity distribution, and provides key energy distribution parameters for statistical feature vectors.

[0120] Furthermore, this embodiment provides a step for obtaining the long-range correlation index of an event sequence based on detrended volatility analysis, including:

[0121] Based on the event sequence, a one-dimensional time signal sequence is obtained through equal-interval counting and sequence reconstruction.

[0122] Based on a one-dimensional time signal sequence, a detrended fluctuation sequence is obtained through detrended fluctuation analysis.

[0123] Based on the detrended volatility sequence, the long-range correlation index is obtained by calculating the volatility function and fitting it with the least squares method.

[0124] Interval counting involves dividing the event sequence into time spans at preset fixed time intervals and counting the number of micro-fracture events occurring within each time interval. This method converts discrete event sequences into continuous time signals. The preset fixed time intervals are determined by comprehensively considering the time-scale characteristics of micro-fracture events at the self-expanding anchor interface, the preset sampling rate, and the time resolution requirements for detrended fluctuation analysis. Sequence reconstruction involves arranging the event counts obtained through interval counting in chronological order according to the time intervals to form a standardized one-dimensional time series, which meets the input data format requirements for detrended fluctuation analysis. The one-dimensional time signal sequence is based on a fixed time interval as the time axis and corresponds to... The number of events within an interval is a continuous sequence of amplitudes, obtained through equal-interval counting and sequence reconstruction; the detrended fluctuation sequence is obtained by subtracting the corresponding trend value from the original sequence after fitting a local trend to a segmented one-dimensional time signal sequence, used to eliminate the interference of local trends on the long-range correlation analysis of the sequence; the fluctuation function is a statistic characterizing the degree of fluctuation of a one-dimensional time signal sequence at different time scales, obtained by calculating the standard deviation of the detrended fluctuation sequence, used to quantify the fluctuation characteristics of the sequence; the least squares fitting method is a fitting method that solves the power-law relationship parameters between the fluctuation function and the time scale by minimizing the sum of squares between the fitted line and the fluctuation function data points, used to obtain the long-range correlation index.

[0125] Specifically, first obtain the event sequence obtained earlier. The time span of the event sequence is determined to be The preset fixed time interval is ,in To ensure a reasonable proportion of non-zero windows, The time span is the minimum value in the set of event wait time intervals. At fixed time intervals Divided into The nth consecutive and equal-length time interval, the The time range of each time interval is ,in , This is the floor function. For the last time interval, its right boundary... It may exceed the final moment of the event sequence. Within the time window corresponding to the portion exceeding the time limit, no events occur, and the event count is 0. The number of micro-fracture events contained in each time interval is counted. Arranged in order of time intervals, a one-dimensional time signal sequence is obtained. .

[0126] Then, the one-dimensional time signal sequence Perform detrending volatility analysis and set the time scale. The range of values ​​is For each time scale One-dimensional time signal sequence Divided into non-overlapping subsequences, where For each subsequence, the function is used to round down. ,in Fitting local trend lines using linear regression. Calculate the difference between each data point in the subsequence and the corresponding trend line to obtain the detrended volatility subsequence for each subsequence. By concatenating all the detrended volatility subsequences in order, the complete detrended volatility sequence is obtained. Finally, calculate each time scale. Corresponding fluctuation function That is, the volatile subsequence after trend detrending. The standard deviation of the mean, i.e. ,in For the first The first detrended oscillation subsequence From 10 data points, we obtain a set of fluctuation functions. The relationship between the wave function and the time scale Take the logarithm and transform it into ,in The proportionality constant is a leading coefficient characterizing the amplitude of time series fluctuations in detrended volatility analysis, obtained adaptively through a least-squares fitting process. The solution is obtained by fitting this linear relationship using the least-squares method. The value of is used to obtain the long-range correlation index. .

[0127] Based on the temporal correlation characteristics of event sequences, this embodiment achieves accurate quantification of the long-range memory or persistence of non-stationary, micro-rupture event sequences through equal-interval counting and sequence reconstruction, segmented detrending processing, fluctuation function calculation and least-squares fitting. This improves the depth and reliability of mining the deep dynamic characteristics of temporal correlation patterns of damage events, enhances the analysis accuracy of sequence persistence and memory effects, and supplements the statistical feature vector with key temporal correlation parameters.

[0128] Furthermore, this embodiment provides a step-by-step approach to obtain the correlation decay scale and dynamic correlation scale based on event sequences combined with statistical feature vectors, through energy density transformation, time correlation functions, and percolation analysis, including:

[0129] Based on the event sequence combined with statistical feature vectors, an energy density sequence is obtained through energy density transformation.

[0130] Based on the energy density sequence, the associated decay scale is obtained through time correlation function and curve fitting techniques.

[0131] Based on the correlation decay scale and event sequence, a spatiotemporal correlation matrix is ​​obtained by constructing a dynamic directed network;

[0132] Based on the spatiotemporal correlation matrix, the largest correlation cluster is determined through percolation analysis and network connectivity identification techniques.

[0133] Based on the maximum correlation cluster, the dynamic correlation scale is obtained by calculating the energy proportion.

[0134] Among these methods, energy density transformation is a process that converts discrete event sequences into continuous energy density sequences. It calculates the sum of event energies within a fixed time window to obtain the energy density sequence, providing a continuous data foundation for subsequent time correlation function calculations. The window size can be adaptively adjusted based on the long-range correlation index in the statistical feature vector to better reflect the system's inherent time correlation structure. Curve fitting is a strategy that selects the fitting function model and fitting interval based on the event clustering index in the statistical feature vector, used to extract the correlation decay scale under non-exponential decay modes. Dynamic directed networks are network models that, during network construction, not only consider time correlation but also incorporate statistical feature vectors to correct edge weights, providing a more comprehensive... The statistical characteristics and dynamic correlations of geofusion events are analyzed. The spatiotemporal correlation matrix is ​​a matrix representation of a dynamic directed network. It is a weighted directed matrix with the same dimension as the number of events in the event sequence. The matrix element values ​​are the directed edge weights after correction by combining statistical feature vectors. If there is no correlation, the value is 0. It is used to quantify the coupling relationship between the dynamic correlation strength and statistical features between events. The network connectivity identification technology is a graph theory method based on depth-first search or disjoint-set data structure algorithm to identify network connected components. It is used to determine the largest connected subgraph from the spatiotemporal correlation matrix. The largest correlation cluster is the connected subgraph with the most event nodes or the largest total weight identified from the spatiotemporal correlation matrix through percolation analysis. Percolation analysis is a graph theory analysis method based on network connectivity theory. It is used to identify the correlation structure evolution characteristics of damage events from the spatiotemporal correlation matrix.

[0135] Specifically, firstly based on event sequences and statistical eigenvectors Energy density conversion is performed based on the long-range correlation index. Dynamically set the basic time window length for energy density conversion The adjustment rules are as follows: To achieve long-term correlation index When the value is greater than the preset sequence persistence threshold, a larger window is used to smooth noise; when persistence is weak, a smaller window is used to preserve details. The dynamic time window length for energy density conversion is determined by a preset sequence persistence threshold based on the long-range correlation index. The theoretical critical value is determined by combining the physical characteristics of the microscopic damage at the interface of the self-expanding anchor bolt when it approaches a critical state. In detrended fluctuation analysis, This indicates that the sequence is uncorrelated white noise. This indicates that the sequence has positive long-range correlation; when the system tends to become unstable, The persistence of a sequence typically shows an increasing trend, indicating that the memory effect of the injury event is enhanced. The preset persistence threshold is usually set to 1.0, which serves as the dividing point between strong persistence and weak persistence.

[0136] Next, the total time span Divided into An energy density sequence is obtained by calculating the sum of the event energies within each consecutive window. Then, based on the energy density sequence Calculate the time correlation function The discrete correlation function values ​​are obtained, where Energy density sequence Length, Indicates a time delay.

[0137] Based on the event clustering index in the statistical eigenvector Guided adaptive curve fitting, if the event clustering index If the value is less than the preset event clustering index threshold, it indicates strong event clustering, and the power-law decay model should be preferred. Fitting the data, clustered events may lead to slower association decay; if the event clustering index... If the value is close to the typical value of a random distribution, then the exponential decay model is selected. Perform a fitting; the fitting interval is also based on the event clustering index. Adjustments are made to exclude extremely short-duration noise in the event sequence where the time difference is much smaller than the minimum value of the event waiting interval set, and extremely long-duration unreliable data where the time difference is much larger than the maximum value of the event waiting interval set. The preset event clustering index threshold is determined by the event clustering index. The theoretical critical value is determined by combining the physical characteristics of the micro-damage at the interface of the self-expanding anchor bolt, which shifts from a random distribution to a clustered distribution. Approaching 2.0, the event behaves as an approximate Poisson random process; when Furthermore, as the value decreases, it indicates that the events are gradually shifting from a random distribution to a clustered distribution, and the system is approaching a critical state. The preset event clustering index threshold is set at 1.8, which serves as the dividing point between strong clustering and random distribution.

[0138] The power law exponent was obtained by least squares fitting. Then through The relationship is transformed into equivalent relaxation time, resulting in the correlation decay scale. Subsequently, based on the correlation decay scale Event Sequence and statistical eigenvectors Construct a dynamic directed network, treating each event as a node to form a node set. For any two events and If satisfied Then establish from point to Directed edges; weights of directed edges Calculated using a weighted model that integrates exponential decay and statistical characteristics, i.e. Among them, the energy power law exponent As an index of energy, it makes in When the proportion of high-energy events increases, the weight of the influence of high-energy events on their subsequent events is non-linearly amplified, which better reflects the physical correlation characteristics under the evolution of energy distribution; based on all nodes and weighted directed edges, the spatiotemporal correlation matrix is ​​obtained. If an edge exists, then Otherwise .

[0139] Next, the spatiotemporal correlation matrix By using percolation analysis and network connectivity identification techniques, ignoring edge directions, the spatiotemporal correlation matrix is... Treating it as an undirected weighted adjacency matrix, a depth-first search algorithm is used to traverse all nodes, identify all connected components, and select the connected component with the largest sum of node weights (i.e., event energies) as the largest association cluster. Ultimately, based on the maximum association cluster Calculate the dynamic correlation scale First, extract the largest association cluster. Energy values ​​corresponding to all event nodes Calculate its total energy. ,in The total number of event nodes in the largest associative cluster, where the event indices in the event sequences corresponding to these nodes are: The corresponding energy values ​​are respectively , , To sum the variables, iterate through all variables within the cluster. 1 event; then obtain the event sequence within the current analysis time window. All Total energy of an event ,in Given the event sequence number, iterate through all events within the window. One event; based on Obtain dynamic correlation scale The dynamic correlation scale is a dimensionless scalar with a value range of [value range missing]. The larger the value, the higher the proportion of damage energy gathered in the maximally connected subgraph, reflecting the greater the degree to which micro-damage events evolve from an isolated state to a highly correlated and cooperative state in space and time, and the closer the system is to critical instability.

[0140] For example, this embodiment assumes that an analysis window contains... An event sequence of events and their corresponding statistical feature vectors According to the long-range correlation index The basic time window length for energy density conversion is set as follows: The adjusted window obtained according to the adjustment rules is: Next, the total time span Divided into An energy density sequence is obtained by calculating the sum of the event energies within each consecutive window. Then based on the energy density sequence Calculate the time correlation function of this event sequence Due to the event clustering index Less than the preset event clustering index threshold This indicates that the event sequence has strong clustering, therefore a power-law decay model is chosen. for The power law exponent is obtained by fitting within an interval of time steps. This leads to the calculation of the correlation attenuation scale. Each time step corresponds to an actual time of [number]. Subsequently, based on the correlation decay scale Event Sequence and statistical eigenvectors Construct a dynamic directed network for any two events. and If satisfied Then a directed edge is constructed with a weight of Then, based on all nodes and edges, the spatiotemporal correlation matrix is ​​obtained. Percolation analysis was performed on the spatiotemporal correlation matrix to identify the largest correlation cluster as the largest connected subgraph. The sum of the energies of all events in the largest cluster is calculated to obtain... The sum of the energy of all events within the total analysis window is Then the dynamic correlation scale Among them, when the associated attenuation scale The corresponding actual time is When, it indicates that the current dynamic relaxation time of the system is 44 ms; when the dynamic correlation scale When this value is reached, it indicates that more than 71% of the damage energy has been correlated into a whole, and the system damage correlation degree is relatively high. It should be noted that the preset base time window, event clustering index threshold, correlation function fitting interval, and other parameter values ​​in this embodiment are only examples. Those skilled in the art can make adaptive adjustments according to the actual signal sampling rate, event density, and correlation characteristics. This embodiment does not impose any limitations on this.

[0141] This embodiment achieves deep coupling between statistical features and dynamic correlation scales by combining statistical feature vectors to adaptively adjust the energy density conversion interval, optimizing the fitting weights of the time correlation function by fusing energy power law exponents, correcting the dynamic network construction rules based on the event clustering index, and quantifying the degree of damage correlation through percolation analysis and energy proportion calculation. This makes the calculation of correlation decay scale and dynamic correlation scale not only based on instantaneous signals, but also on the macroscopic statistical evolution law of event sequences, improving the accuracy, dynamic adaptability and physical consistency of correlation scale calculation, enhancing the physical consistency of critical state criteria and the robustness of the early warning system, and providing core physical indicators that integrate macroscopic statistical laws and microscopic correlation characteristics for subsequent graded early warning.

[0142] Furthermore, this embodiment provides a step for obtaining a spatiotemporal correlation matrix by constructing a dynamic directed network based on the correlation decay scale and event sequence, including:

[0143] Based on event sequences, a set of dynamic directed network nodes is constructed through node initialization techniques;

[0144] Based on the correlation decay scale, the potential causal relationship between any two events is determined by a dynamic time window judgment rule.

[0145] Based on potential causal relationships and event sequences, the weights of directed edges are obtained through an exponentially decaying weight model.

[0146] Based on the set of nodes in a dynamic directed network and all weighted directed edges, a spatiotemporal correlation matrix is ​​obtained through matrix generation techniques.

[0147] Among them, node initialization technology is a processing method that uniquely maps each micro-rupture event in the event sequence to a network node, used to establish the basic element set of the network; the set of dynamic directed network nodes is a set composed of all event nodes; the dynamic time window determination rule is a criterion based on the correlation decay scale and combined with the event clustering index to determine the maximum time interval between events to determine whether there may be a causal relationship; potential causal relationship is the dynamic influence relationship between the preceding event and the subsequent event within the dynamic time window, and its determination result is consistent with the evolution trend represented by the event clustering index; the exponential decay weight model is a method based on the classical exponential decay... Based on the model, a weight calculation formula is used to power-correct the event energy term by combining the energy power-law exponent in the statistical feature vector. This is used to more accurately quantify the transmission strength of the event's influence under a specific energy distribution pattern. The weight of the directed edge is a non-negative scalar that characterizes the coupling effect between the potential causal relationship strength and the statistical feature. It is determined by the energy of the first event, the time difference between the two events, and the energy power-law exponent. The larger the value, the stronger the relationship and the more it conforms to the macroscopic energy distribution law. The matrix generation technique is a technique that transforms the node and directed edge weights of the dynamic directed network into a matrix. This technique is used to transform the network structure that integrates statistical features into a matrix form that is easy to calculate.

[0148] Specifically, firstly based on event sequences Each event is initialized using node initialization technology. Create a corresponding node This constitutes a set of nodes in a dynamic directed network. Then based on the correlation attenuation scale Apply the dynamic time window judgment rule, that is, for any two different events and and its corresponding nodes and If the time difference satisfies Then determine the event. Regarding the event There is a potential causal relationship, denoted as ; then based on potential causal relationships Relative energy of events in an event sequence and statistical feature vectors Energy power law index The weights of directed edges are calculated using a statistically weighted exponential decay weighting model, resulting in... The exponential decay weighted model introduces As an exponent of the energy term, the weighting calculation can respond to the overall change in the system's energy distribution, i.e. Increasing the value implies a stronger relative influence of high-energy events, thus integrating macroscopic statistical regularities into the measurement of microscopic correlation strength; ultimately based on node sets. and the set of directed edges with all calculated weights Construct a matrix generation technique spatiotemporal correlation matrix If If it exists, then the element ,otherwise And regulations .

[0149] This embodiment, based on the temporal and energy characteristics of event sequences, correlation decay scales, and statistical feature vectors, achieves adaptive adjustment of network weights to the macroscopic statistical state of the system through node initialization, dynamic time window determination of fused clustering index, weight calculation of fused energy power law index, and matrix generation. It realizes the coupled representation of dynamic correlation between events and macroscopic statistical laws, improves the accuracy, adaptability, and physical logic of correlation representation, and improves the accuracy of the representation of the real damage correlation and evolution stage of the system by constructing a spatiotemporal correlation matrix. It provides a high-quality network model foundation that integrates multi-dimensional features for subsequent percolation analysis.

[0150] Furthermore, this embodiment provides a step-by-step approach: calculating a state health index based on a correlation decay scale and a dynamic correlation scale using a fusion algorithm; obtaining a graded early warning signal using a trend analysis algorithm and multi-level threshold criterion logic; and executing corresponding processing measures based on the graded early warning signal. The steps include:

[0151] Based on the correlation decay scale and the dynamic correlation scale, the state health index is calculated by a fusion algorithm.

[0152] A normal baseline is established based on historical health indices using baseline learning and trend analysis algorithms.

[0153] Based on the deviation of the real-time calculated health status index from the normal benchmark and the changing trend of the health status index, risk assessment is performed through multi-level threshold judgment logic to obtain graded early warning signals.

[0154] Based on the graded early warning signals, corresponding processing measures are executed according to the preset response rules. The processing measures include at least one of the following: increasing the data collection frequency, triggering system self-check, generating operation and maintenance prompts, and sending emergency shutdown commands.

[0155] Among them, the fusion algorithm is a mathematical operation method that weights and fuses the correlation attenuation scale and the dynamic correlation scale to obtain a comprehensive evaluation index, which is used to comprehensively reflect the current dynamic relaxation characteristics and damage correlation degree of the anchor system; the state health index is a scalar index calculated by the fusion algorithm. The higher the value, the more stable the installation interface of the self-expanding anchor and the better the health status; the historical state health index is a state health index that is continuously calculated and stored in the stable stage after the installation of the self-expanding anchor, and is used to characterize the typical behavior of the system under normal working conditions.

[0156] Baseline learning is a data learning method based on historical state health indices, which uses statistical modeling to determine the central trend and normal fluctuation range of the indices, and is used to establish a reference benchmark for the health status of the system. Trend analysis algorithm is an algorithm that uses moving average and linear fitting to identify the long-term direction of change of state health indices over time, and is used to determine whether the system state is trending towards stability or instability.

[0157] The normal baseline is a set of parameters obtained through baseline learning that describes the typical values ​​and reasonable fluctuation range of the health index under normal system conditions. It includes the baseline value of the health index and its standard deviation, and is used as a reference for real-time status evaluation. The real-time calculated health index is the current health index calculated based on the event sequence and statistical feature vector within the latest time window. The degree of deviation and trend of the real-time calculated health index relative to the normal baseline is a risk assessment basis obtained by calculating the relative deviation between the real-time value and the baseline value and combining the trend analysis results. It is used to quantify the degree and direction of evolution of the current state from the normal range.

[0158] The multi-level threshold criterion logic consists of threshold conditions and judgment rules corresponding to different risk levels preset based on a normal baseline. The preset different risk levels are quantified by the degree of deviation of the health status index from the normal baseline and the changing trend identified by the trend analysis algorithm. These include attention-level thresholds and action-level thresholds, which are used to trigger corresponding level warnings based on the real-time health status index and its changing trend. The graded warning signals are risk indication signals output by the multi-level threshold criterion logic, including attention-level yellow warnings and action-level red alerts, which are used to provide intuitive and graded risk prompts for operation and maintenance decisions. The preset response rules are a set of measures pre-configured according to different graded warning signals, which are implemented by calling corresponding control commands. The handling measures are specific operations automatically triggered according to the graded warning signals, including at least one of the following: increasing the data collection frequency, triggering system self-checks, generating operation and maintenance prompts, and sending emergency shutdown commands, which are used to achieve closed-loop health management from warning to response.

[0159] Specifically, the first step is to obtain the associated attenuation scale sequence. and dynamic correlation scale sequence ,in Given the total number of historical analysis windows, a fusion algorithm is used to calculate the historical health index. The specific implementation process of the fusion algorithm is as follows: First, the correlation decay scale is normalized to obtain the normalized correlation decay scale. ,in For the baseline value of the associated attenuation scale, For the first The correlation decay scale of each analysis window and Then, baseline normalization is performed on the dynamic correlation scale to obtain the normalized dynamic correlation scale. ,in For dynamically correlated scale baseline values, For the first The dynamic correlation scale of each analysis window, and when hour, .

[0160] The contribution weights were then determined based on the instability simulation data during the baseline learning period. and ,in and Through formula Calculate the first Historical health index in each analysis window ,all Constructing a historical health index series .

[0161] Then, baseline learning is performed based on the historical state health index sequence to calculate the baseline value of the state health index. The standard deviation of the health status index is Baseline value of health status index Standard deviation of the health status index Together they form a normal baseline; then, during long-term monitoring, each new analysis time window is used. Real-time calculation of the current correlation decay scale and dynamic correlation scale The real-time health index is calculated using the same steps as described in the fusion algorithm. At the same time, set the length of the sliding window to ,in Given a preset positive integer, extract the current time and the time before the current time. A sliding window sequence is composed of real-time health indices. The slope of the trend line was obtained by fitting the sequence using linear regression. Calculate the deviation of the real-time health index from the normal baseline. Sum of deviations .

[0162] Finally, risk assessment is performed based on multi-level threshold criterion logic. The specific rules of the multi-level threshold criterion logic are as follows: when the condition is met... and When a yellow alert for alert is triggered; when the conditions are met... or At that time, an action-level red alert was triggered, in which These are the minimum and maximum values ​​of the deviation rate threshold. If the slope threshold of the trend line is negative, the system will output a corresponding level of early warning signal; based on the level of early warning signal, the corresponding processing measures will be automatically executed according to the preset response rules.

[0163] The preset response rules are as follows: When the graded warning signal is a yellow warning of concern level, the first set of processing measures is executed, including: increasing the data acquisition frequency from the basic frequency to the encrypted frequency, triggering the system self-check program, and sending a concern level prompt message to the operation and maintenance platform, suggesting that inspections be arranged or monitoring be strengthened. The basic frequency is a preset sampling rate, which is obtained by calibrating the time scale characteristics and passband frequency range of micro-fracture events at the interface of self-expanding anchor bolts, and is used for the regular continuous acquisition of the original vibration signal. The encrypted frequency is the data acquisition frequency increased from the basic frequency to the encrypted frequency after the yellow warning of concern level is triggered. Its specific value is 2 to 5 times the basic frequency. It is comprehensively calibrated by the time scale characteristics of micro-fracture events at the interface of self-expanding anchor bolts, the sliding window length of the STA / LTA algorithm, and the preset sampling rate to ensure that more dense micro-damage events can be captured during the risk escalation phase.

[0164] When the tiered early warning signal is an action-level red alert, the second set of processing measures is executed, including: sending an emergency shutdown command to the field controller via industrial bus or wireless network, triggering the anchor bolt's built-in or external mechanical reinforcement device, automatically generating a maintenance work order and pushing it to the relevant responsible person's terminal, activating the audible and visual alarm device, and further increasing the data acquisition frequency to the highest monitoring frequency to closely monitor subsequent state changes. The highest monitoring frequency, determined after triggering the action-level red alert, is 5 to 10 times the base frequency. It is determined through comprehensive analysis of the self-expanding anchor bolt system's unloaded background noise statistics, on-site environmental noise calibration, and training with historical micro-fracture event samples. This is used to closely monitor subsequent state changes with the highest time resolution during the critical instability stage. The execution of these processing measures is achieved by the cloud server calling the corresponding control commands based on the tiered early warning signal, ensuring a timely and reliable response.

[0165] For example, this embodiment assumes that a self-expanding anchor system, during a baseline learning period of 72 hours of stable operation after installation, has analyzed and obtained the following data: A historical health index, whose dynamic correlation scale baseline value is obtained through statistical learning, is used to derive a baseline value for each historical state health index. The baseline value of the associated attenuation scale is The baseline value of the health status index is Standard deviation is The contribution weights determined based on the instability simulation data are respectively the weights of the associated decay scale. Dynamic correlation scale weight The length of the sliding window is This involves examining monitoring data within the most recent 10 analysis windows, assuming the preset deviation rate thresholds are as follows: and The trend line slope threshold is set to At a specific point in time during long-term monitoring, the current correlation decay scale is calculated in real time based on the latest time window. Current dynamic correlation scale .

[0166] Real-time health index is calculated using a fusion algorithm. Extract the current time before The real-time health indices of each window constitute a sliding window sequence, and the slope of the trend line is obtained through linear regression fitting. Then the deviation value is calculated. The deviation rate is That is, the absolute value of the deviation rate. .

[0167] Then, a judgment is made based on the multi-level threshold criterion logic, due to the absolute value of the deviation rate. The threshold is greater than the preset maximum deviation rate. And the slope of the trend line Therefore, the triggering conditions for an action-level red alert are met, and the system outputs an action-level red alert, indicating a significant decrease in the stability of the anchor bolt installation interface and a high risk. Then, based on the action-level red alert, the system immediately executes the second set of processing measures according to the preset response rules: the cloud server sends an emergency stop command to the field programmable logic controller through the industrial IoT gateway, cutting off the power source of the equipment where the anchor bolt is located, and simultaneously triggering the hydraulic locking mechanism built into the anchor bolt to prevent further displacement. Then, the system automatically generates a maintenance work order containing information such as time, location, and health index, and sends it to the maintenance engineer via SMS and application push. The on-site audible and visual alarm is activated to warn surrounding personnel. Subsequently, the system increases the data acquisition frequency from the basic frequency of once per hour to the maximum monitoring frequency of once per minute to closely monitor subsequent status changes. This response process is completed within milliseconds to seconds, effectively avoiding the occurrence of sudden displacement failure accidents while meeting the real-time requirements of the project. It should be noted that the preset baseline learning period, contribution weight, sliding window length, deviation rate thresholds at each level, and trend slope thresholds in this embodiment are merely examples. Those skilled in the art can make adaptive adjustments based on actual engineering safety margin requirements, data stability, and early warning sensitivity requirements. This embodiment does not impose any restrictions on these adjustments.

[0168] This embodiment achieves comprehensive quantification of correlation attenuation scale and dynamic correlation scale through an adaptive weighted fusion algorithm, establishes personalized normal benchmarks through baseline learning, captures the evolution trend of health status through a sliding window linear regression algorithm, and realizes risk classification through multi-level threshold judgment logic that integrates deviation degree and change trend. Based on the classified early warning signal, corresponding processing measures are automatically triggered, realizing accurate and layered early warning of installation displacement risk of self-expanding hole anchor bolts. This improves the engineering practicality and decision guidance of the early warning signal, reduces the false alarm rate and false alarm rate, and provides clear and operable decision basis and closed-loop management capability for anchor bolt operation and maintenance.

[0169] Furthermore, such as Figure 5 As shown in the figure, this application provides a self-expanding anchor installation displacement monitoring system, which includes a signal acquisition module, an event extraction module, an intelligent analysis module, and an early warning output module.

[0170] Signal acquisition module: Acquires raw vibration signal data;

[0171] Event extraction module: Based on the original vibration signal data, the event sequence is obtained through transient pulse detection technology and energy integration;

[0172] Intelligent Analysis Module: Based on event sequences combined with power-law fitting technology, statistical feature vectors are obtained through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis; based on event sequences combined with statistical feature vectors, correlation decay scale and dynamic correlation scale are obtained through energy density transformation, time correlation function, and percolation analysis.

[0173] Early warning output module: Based on the correlation decay scale and dynamic correlation scale, the health index is calculated by the fusion algorithm, and the graded early warning signal is obtained by the trend analysis algorithm and multi-level threshold criterion logic. The corresponding processing measures are then executed based on the graded early warning signal.

[0174] The signal acquisition module is used to sense the axial vibration status signal of the anchor bolt and substrate interface in real time. The embedded processing unit performs analog-to-digital conversion and data packaging at a preset sampling rate, and uploads the data to the cloud server via low-power wide-area network protocols such as NB-IoT / LoRa to obtain the raw vibration signal data. The unique diamond sintered cutting tool and cooling wax structure of the self-reaming anchor bolt ensures a smooth reaming process and a tight fit between the anchor bolt and substrate interface, reducing non-destructive vibration interference at the source and providing the signal acquisition module with a high signal-to-noise ratio and high stability raw vibration signal. The event extraction module, deployed on the server, processes the received raw vibration signal data. It filters out environmental and noise interference using digital bandpass filtering technology and identifies the pulse start point generated by interface micro-fractures using STA / LTA transient pulse detection technology, combined with a fixed time window energy... The quantity integral technique quantifies the relative energy of each event, ultimately outputting an event sequence composed of event time and event relative energy pairs. The functionality of the event extraction module directly relies on the high-quality raw data provided by the signal acquisition module. The intelligent analysis module is also deployed on the server to perform in-depth mining and feature extraction on the event sequence output by the event extraction module. First, based on the event sequence and power-law fitting technique, through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis, the event clustering index, energy power-law index, and long-range correlation index, which characterize the macroscopic statistical regularity of events, are extracted and integrated into a statistical feature vector. Subsequently, based on the event sequence and the statistical feature vector, through adaptive energy density transformation, time correlation function calculation, and percolation analysis, the correlation decay scale, which characterizes the dynamic relaxation characteristics of the system, and the dynamic correlation scale, which characterizes the degree of damage correlation, are further extracted. The analysis in this module is based on a series of regular and reliable event sequences guaranteed by the precise structure of the anchor bolts. The early warning output module, deployed on the server, is the decision-making terminal of the monitoring system. It receives the correlation attenuation scale and dynamic correlation scale output by the intelligent analysis module, and calculates the state health index, which represents the overall health status, through a fusion algorithm. Then, based on historical data, a normal benchmark is established through baseline learning, and the evolution trend of the state health index is judged by combining trend analysis algorithms. Finally, by integrating the deviation of the state health index from the normal benchmark and the changing trend of the state health index, a multi-level threshold judgment logic is constructed to conduct risk assessment and generate graded early warning signals such as yellow warning at the attention level or red alert at the action level. The early warning signals output by the early warning output module provide a direct and quantifiable basis for assessing the stability of the self-expanding hole anchor bolt installation interface and making operation and maintenance decisions.The self-expanding anchor installation displacement monitoring system provided in this embodiment is closely connected through data flow, forming a closed-loop monitoring logic. The signal acquisition module provides physical sensing data to the system, the event extraction module transforms the raw data into discrete events characterizing microscopic damage, the intelligent analysis module extracts multi-level and multi-scale dynamic features from the events, and the early warning output module performs comprehensive evaluation and risk warning based on these features. The entire system, through the combination of hardware and software, realizes intelligent monitoring of the self-expanding anchor from installation status perception to risk warning throughout the entire process.

[0175] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0176] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0177] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0178] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for monitoring displacement of a self-reaming anchor installation, applied to a self-reaming anchor comprising a diamond sintered blade and a cooling wax, characterized in that, include: Acquire raw vibration signal data; Based on the original vibration signal data, an event sequence is obtained through transient pulse detection technology and energy integration. Based on the event sequence combined with power-law fitting technology, statistical feature vectors are obtained through statistical distribution analysis, energy distribution analysis and detrended fluctuation analysis. Based on the event sequence and the statistical feature vector, the correlation decay scale and dynamic correlation scale are obtained through energy density conversion, time correlation function and percolation analysis. Based on the aforementioned correlation decay scale and dynamic correlation scale, a state health index is calculated through a fusion algorithm. A graded early warning signal is obtained through a trend analysis algorithm and a multi-level threshold criterion logic. Corresponding processing measures are then executed based on the graded early warning signal.

2. The method of monitoring displacement of a self-reaming anchor installation according to claim 1, wherein, The process of obtaining an event sequence based on the original vibration signal data using transient pulse detection technology and energy integration includes: Based on the original vibration signal data, the denoised vibration signal data is obtained by digital bandpass filtering; Based on the denoised vibration signal data, the pulse start point is obtained using transient pulse detection technology; An event sequence is obtained by energy integration based on the pulse initiation point. The event sequence includes micro-rupture events and the relative energy of the events.

3. The method of monitoring displacement of a self-reaming anchor installation according to claim 1, wherein, The statistical feature vector obtained based on the event sequence combined with power-law fitting technology, through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis, includes: Based on the event sequence, the event clustering index is obtained through statistical distribution analysis and power-law fitting techniques. Based on the event sequence, the energy power law exponent is obtained through energy distribution analysis and power law fitting techniques. Based on the event sequence, the long-range correlation index of the event sequence is obtained through detrended fluctuation analysis; Based on the event clustering index, energy power law index, and long-range correlation index, a statistical feature vector is obtained by integration.

4. The method for monitoring the installation displacement of self-expanding anchor bolts according to claim 3, characterized in that, The process of obtaining the event clustering index based on the event sequence through statistical distribution analysis and power-law fitting techniques includes: Based on the event sequence, a set of event waiting time intervals is obtained by calculating the adjacent time differences; Based on the set of event waiting time intervals, the waiting time distribution is obtained through statistical distribution analysis and probability density calculation. Based on the aforementioned waiting time distribution, the event clustering index is obtained using a power-law fitting technique based on least squares.

5. The method for monitoring the installation displacement of self-expanding anchor bolts according to claim 3, characterized in that, The process of obtaining the long-range correlation index of the event sequence through detrended volatility analysis based on the event sequence includes: Based on the event sequence, a one-dimensional time signal sequence is obtained through equal-interval counting and sequence reconstruction. Based on the one-dimensional time signal sequence, a detrended fluctuation sequence is obtained through detrended fluctuation analysis; Based on the detrended volatility sequence, a volatility function is calculated, and a long-range correlation index is obtained by fitting the volatility function using the least squares method.

6. The method for monitoring the installation displacement of self-expanding anchor bolts according to claim 1, characterized in that, The process of obtaining the correlation decay scale and dynamic correlation scale based on the event sequence and the statistical feature vector through energy density transformation, time correlation function, and percolation analysis includes: Based on the event sequence and the statistical feature vector, an energy density sequence is obtained through energy density transformation. Based on the energy density sequence, the associated decay scale is obtained through time correlation function and curve fitting techniques; Based on the aforementioned correlation decay scale and event sequence, a spatiotemporal correlation matrix is ​​obtained by constructing a dynamic directed network; Based on the spatiotemporal correlation matrix, the largest correlation cluster is determined through percolation analysis and network connectivity identification techniques. Based on the largest association cluster, the dynamic association scale is obtained by calculating the energy percentage.

7. The method for monitoring the installation displacement of self-expanding anchor bolts according to claim 6, characterized in that, The process of obtaining the spatiotemporal correlation matrix by constructing a dynamic directed network based on the correlation decay scale and event sequence includes: Based on the event sequence, a set of dynamic directed network nodes is constructed using node initialization technology; Based on the aforementioned correlation decay scale, the potential causal relationship between any two events is determined using dynamic time window judgment rules. Based on the potential causal relationships and event sequences, the weights of directed edges are obtained through an exponentially decaying weight model. Based on the set of nodes in the dynamic directed network and all weighted directed edges, a spatiotemporal correlation matrix is ​​obtained through matrix generation techniques.

8. The method for monitoring the installation displacement of self-expanding anchor bolts according to claim 1, characterized in that, The state health index is calculated using a fusion algorithm based on the correlation decay scale and dynamic correlation scale. A graded early warning signal is then obtained using a trend analysis algorithm and multi-level threshold criterion logic. Corresponding processing measures are then executed based on the graded early warning signal, including: Based on the aforementioned correlation decay scale and dynamic correlation scale, the state health index is calculated using a fusion algorithm. A normal baseline is established based on historical health indices using baseline learning and trend analysis algorithms. Based on the deviation of the real-time calculated health status index from the normal benchmark and the changing trend of the health status index, risk assessment is performed through multi-level threshold judgment logic to obtain graded early warning signals. Based on the tiered early warning signal, corresponding processing measures are executed according to preset response rules. The processing measures include at least one of increasing the data acquisition frequency, triggering system self-check, generating operation and maintenance prompts, and sending emergency shutdown commands.

9. A self-expanding anchor bolt device, employing the self-expanding anchor bolt installation displacement monitoring method as described in any one of claims 1-8, characterized in that, include: Diamond sintered cutting tools are used to directly cut and enlarge holes in the substrate during anchor bolt installation. Cooling wax is stored in an annular cooling wax storage cavity located inside the reaming head of the diamond sintering cutter, and is used to cool the diamond sintering cutter during anchor bolt installation.

10. A self-expanding anchor bolt installation displacement monitoring system, comprising the self-expanding anchor bolt device and server as described in claim 9, characterized in that, The server includes: A signal acquisition module is used to acquire raw vibration signal data, which is obtained through the self-expanding hole anchor bolt device. The event extraction module is used to obtain an event sequence based on the original vibration signal data through transient pulse detection technology and energy integration; The intelligent analysis module is used to obtain statistical feature vectors based on the event sequence and power-law fitting technology through statistical distribution analysis, energy distribution analysis, and detrended fluctuation analysis; and to obtain correlation decay scales and dynamic correlation scales based on the event sequence and the statistical feature vectors through energy density conversion, time correlation functions, and percolation analysis. The early warning output module is used to calculate the state health index based on the correlation attenuation scale and the dynamic correlation scale through a fusion algorithm, obtain a graded early warning signal through a trend analysis algorithm and a multi-level threshold criterion logic, and execute corresponding processing measures based on the graded early warning signal.

Images