Health state diagnosis method and system for shock absorber device of rock bolting machine

Abstract

The application discloses an anchor rod machine damping device health state diagnosis method and system, relates to the technical field of anchor rod machine component testing, and comprises the following steps: after identifying the operation state of the anchor rod machine, a multi-source original data set is established; steady-state baseline features are extracted, multi-frequency disturbance is applied, and mapped forced response signals are collected; the forced response signals are demodulated in multiple scales to generate an initial feature set; feature reweighting under baseline correction is performed to establish degradation mode feature components; and the degradation mode feature components are input into a time series causal diagram network to output a current health state score and the contribution weight of each degradation mode. The application solves the technical problems in the prior art that health state diagnosis of the anchor rod machine damping device relies on a single signal, it is difficult to accurately identify degradation modes, and health management has hysteresis and insufficient accuracy, and achieves the technical effects of accurately diagnosing the health state of the damping device and improving the accuracy and timeliness of health management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of anchor rod machine component testing, in particular to an anchor rod machine damping device health state diagnosis method and system. BACKGROUND

[0002] In the operation process of the anchor rod machine, the damping device as the core component directly affects the equipment operation stability and operation safety. After long-term bearing of complex loads such as vibration impact, hydraulic fluctuation, etc., the damping device is prone to degradation problems such as damping failure, stiffness attenuation, hydraulic system aging, etc. The current health diagnosis technology for the damping device of the anchor rod machine mainly relies on single vibration signal monitoring or regular shutdown detection, which has two key limitations: first, single signal cannot comprehensively reflect the multi-dimensional degradation characteristics of the damping device, which is easy to cause misjudgment due to environmental interference or local signal abnormality, and cannot accurately locate the degradation mode of the core performance such as damping transmission, hysteresis stiffness, pressure recovery, etc.; second, regular detection has a lag, which is difficult to capture the sudden degradation of the damping device in dynamic operation in time, and lacks effective prediction and early warning of the degradation trend, which often leads to expansion of damping device failure due to failure to intervene in time, and further affects the overall operation efficiency of the anchor rod machine, and even causes safety hazards.

[0003] The prior art has the technical problems that the health state diagnosis of the damping device of the anchor rod machine relies on single signal, which is difficult to accurately identify the degradation mode, resulting in lag and insufficient accuracy of health management. SUMMARY

[0004] The present application provides an anchor rod machine damping device health state diagnosis method and system, which is used to solve the technical problems that the health state diagnosis of the damping device of the anchor rod machine relies on single signal in the prior art, which is difficult to accurately identify the degradation mode, resulting in lag and insufficient accuracy of health management.

[0005] In view of the above problems, the present application provides an anchor rod machine damping device health state diagnosis method and system.

[0006] In a first aspect of the present application, an anchor rod machine damping device health state diagnosis method is provided, which comprises:

[0007] After identifying the operating status of the anchor bolting machine, multi-source sensors are activated to synchronously collect vibration, hydraulic pressure, temperature, acoustic emission signals, displacement, and force signals to establish a multi-source raw dataset. Steady-state baseline features are extracted from the multi-source raw dataset based on the operating status. Active excitation is adaptively configured based on the steady-state baseline features and the operating status. Multi-frequency disturbances are applied to the anchor bolting machine's damping device based on the active excitation, and mapped forced response signals are collected. Natural vibration response signals are extracted from the steady-state baseline features, and the forced response signals are demodulated at multiple scales to establish explicit damping transfer spectra, hysteresis stiffness curves, pressure recovery curves, and impact attenuation curves, generating an initial feature set. Feature reweighting under baseline correction is performed on the initial feature set to establish degradation mode feature components. The degradation mode feature components are input into a time-series causal graph network, outputting the current health status score and the contribution weights of each degradation mode.

[0008] In one possible implementation, the adaptive configuration of active excitation based on the steady-state baseline characteristics and the operating state includes: establishing an ultimate load constraint based on the operating state and the steady-state baseline characteristics; under the ultimate load constraint, reading the historical health score of the anchor bolting machine, configuring the excitation amplitude based on the historical health score and the steady-state baseline characteristics; setting a random disturbance factor, and under the ultimate load constraint, performing a random probability disturbance of the excitation amplitude based on the random disturbance factor, and establishing active excitation using the random probability disturbance result.

[0009] In one possible implementation, the multi-scale demodulation of the forced response signal includes: using the natural vibration response signal as a zero-disturbance reference, performing a forced response signal fusion analysis to establish a fused signal, wherein the fused signal is a signal set that eliminates the influence of operating conditions and environmental disturbances; performing multi-scale demodulation on the fused signal, wherein the demodulation scale is adaptively set based on historical degradation weights; and establishing explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve, and impact attenuation curve based on the multi-scale demodulation results.

[0010] In one possible implementation, establishing explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve, and impact attenuation curve based on multi-scale demodulation results includes: performing multi-scale frequency domain analysis on the multi-scale demodulation results, calculating the ratio of response amplitude to excitation amplitude, and using the ratio of response amplitude to excitation amplitude as a damping transfer characteristic; after weighting and enhancing the damping transfer characteristic with degradation mode contribution weights, establishing a damping transfer spectrum; performing multi-scale segmentation on the displacement-force cycle in the multi-scale demodulation results, calculating the average slope of each cycle, and then, based on the degradation mode contribution weights, establishing a damping transfer spectrum. After the degradation mode contribution weights are weighted and made explicit by the average slope, hysteresis stiffness curves are established by arranging them in cyclic order. Multi-scale analysis is performed on the hydraulic pressure in the multi-scale demodulation results to calculate the time curve for pressure recovery to steady state after the disturbance ends. The time curve is then weighted and made explicit by the degradation mode contribution weights to establish a pressure recovery curve. Multi-scale envelope analysis is performed on the transient impact events in the multi-scale demodulation results to extract the amplitude decay curve over time. The decay curve is then weighted and made explicit by the degradation mode contribution weights to establish an impact decay curve.

[0011] In one possible implementation, the step of performing feature reweighting under baseline correction on the initial feature set to establish degradation mode feature components includes: standardizing the explicit features in the initial feature set based on steady-state baseline features, where the steady-state baseline features are features characterizing the health benchmark state of the damping device; inputting the standardization result into the adaptive feature degradation mode explicit model, performing feature reweighting, and establishing degradation mode feature components.

[0012] In one possible implementation, the step of inputting the feature components of the degradation mode into a temporal causal graph network and outputting the current health status score and the contribution weights of each degradation mode includes: establishing a degradation trend prediction result based on the current health status score and the contribution weights of each degradation mode; constructing a disturbance excitation using the degradation trend prediction result; conducting an active disturbance test on the vibration damping device based on the disturbance excitation and establishing a test feedback set; verifying the degradation trend prediction result using the test feedback set, and updating the current health status score and the contribution weights of each degradation mode using the test verification result.

[0013] In one possible implementation, the multi-source sensor includes an acceleration array for collecting vibrations, the acceleration array being respectively deployed on the frame side, the upper end of the shock absorber, and the lower end of the shock absorber, and the multi-source sensor is redundantly deployed and performs self-calibration processing before data acquisition.

[0014] In one possible implementation, the output of the current health status score and the contribution weights of each degradation mode further includes: establishing a first health warning signal based on the current health status score, wherein the first health warning signal is a real-time warning signal; obtaining historical health status scores, performing time-series smoothing analysis using the historical health status scores and the current health status score, and establishing a second health warning signal, wherein the second health warning signal is a smoothed anomaly signal; performing warning fusion on the first health warning signal and the second health warning signal, outputting a fused warning signal, and performing warning management based on the fused warning signal.

[0015] In one possible implementation, the early warning management based on the fused early warning signal includes: after issuing an early warning based on the fused early warning signal, obtaining an early warning response action; reconstructing the monitoring cycle based on the early warning response action and the fused early warning signal, and performing health status monitoring and management under the monitoring cycle.

[0016] A second aspect of this application provides a health status diagnosis system for a bolting machine vibration damping device, the system being used to implement the health status diagnosis method for the bolting machine vibration damping device described in this application, the system comprising:

[0017] The module for establishing the original dataset is used to activate multi-source sensors to synchronously collect vibration, hydraulic pressure, temperature, acoustic emission signals, displacement, and force signals after identifying the working state of the anchor bolting machine, thus establishing a multi-source original dataset. The module for acquiring the forced response signal is used to extract steady-state baseline features from the multi-source original dataset based on the working state, and adaptively configure active excitation based on the steady-state baseline features and the working state. Based on the active excitation, multi-frequency disturbances are applied to the vibration damping device of the anchor bolting machine, and the mapped forced response signal is acquired. The module for generating the initial feature set is used to extract the natural vibration response signal from the steady-state baseline features, perform multi-scale demodulation on the forced response signal, and establish explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve, and impact attenuation curve to generate an initial feature set. The module for reweighting the features under baseline correction is used to reweight the initial feature set to establish degradation mode feature components. The module for obtaining the health status score is used to input the degradation mode feature components into a time-series causal graph network and output the current health status score and the contribution weight of each degradation mode.

[0018] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0019] After identifying the operational status of the anchor bolting machine, a multi-source raw dataset is established. Steady-state baseline features are extracted from the multi-source raw dataset based on the operational status. Active excitation is adaptively configured based on the steady-state baseline features and the operational status. Multi-frequency disturbances are applied to the vibration damping device of the anchor bolting machine based on the active excitation, and the mapped forced response signal is collected. After extracting the natural vibration response signal from the steady-state baseline features, the forced response signal is demodulated at multiple scales to generate an initial feature set. Feature reweighting under baseline correction is performed on the initial feature set to establish degradation mode feature components. The degradation mode feature components are input into a time-series causal graph network, which outputs the current health status score and the contribution weight of each degradation mode. This achieves the technical effect of accurately diagnosing the health status of the vibration damping device, improving the accuracy and timeliness of health management. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A schematic flowchart of the health status diagnosis method for the anchor bolt machine vibration damping device provided in the embodiments of this application;

[0022] Figure 2 A schematic diagram of the health status diagnosis system for the anchor bolt machine vibration damping device provided in this application embodiment.

[0023] Figure labeling: Module 10 for establishing the original dataset, Module 20 for acquiring the forced response signal, Module 30 for generating the initial feature set, Module 40 for reweighting the features, and Module 50 for obtaining the health status score. Detailed Implementation

[0024] This application provides a method and system for diagnosing the health status of anchor bolt machine vibration damping devices, which addresses the technical problems in the prior art where the diagnosis of the health status of anchor bolt machine vibration damping devices relies on a single signal and is difficult to accurately identify degradation patterns, resulting in lagging and inaccurate health management.

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0026] Example 1, asFigure 1 As shown, this application provides a method for diagnosing the health status of an anchor bolting machine's vibration damping device, the method comprising:

[0027] Step S100: After identifying the working status of the anchor bolting machine, activate the multi-source sensors to synchronously collect vibration, hydraulic pressure, temperature, acoustic emission signals, displacement and force signals, and establish a multi-source raw dataset.

[0028] Specifically, the first step is to identify the operating status of the bolting machine and determine its current operating condition. After the operating status is identified, pre-deployed multi-source sensors are activated. These sensors include an acceleration array for vibration data acquisition, sensors for hydraulic pressure, temperature, acoustic emission signals, displacement, and force. The multi-source sensors are deployed in a redundant manner and have undergone self-calibration before data acquisition. All sensors synchronously begin data acquisition, capturing vibration, hydraulic pressure, temperature, acoustic emission signals, displacement, and force signals in real time during the bolting machine's operation. This raw data from different dimensions is then integrated and stored, ultimately establishing a multi-source raw dataset covering all parameters required for diagnosis.

[0029] Step S200: Extract steady-state baseline features from the multi-source raw dataset according to the operation status, and adaptively configure active excitation according to the steady-state baseline features and the operation status. Apply multi-frequency disturbance to the damping device of the anchor bolt machine based on the active excitation, and collect the mapped forced response signal.

[0030] Specifically, based on the identified operating status of the anchor bolting machine, data from the stable operation phase of the equipment is first selected from the multi-source raw dataset. Steady-state baseline features characterizing the health baseline state of the damping device are then extracted using feature extraction techniques. Next, combining these steady-state baseline features with the current operating status, an ultimate load constraint that the damping device can withstand is established. Within this constraint range, the historical health score of the anchor bolting machine is read, and the initial excitation amplitude is determined based on the historical health score and the steady-state baseline features. Subsequently, a random disturbance factor is set, and under the ultimate load constraint, the initial excitation amplitude is randomly perturbed based on the random disturbance factor. The perturbation results are used to construct an active excitation containing multi-frequency components. Finally, multi-frequency perturbations are applied to the damping device of the anchor bolting machine based on this active excitation. Simultaneously, the mapping signal generated by the damping device under the multi-frequency perturbation is collected in real time by multi-source sensors, i.e., the forced response signal.

[0031] Step S300: After extracting the natural vibration response signal from the steady-state baseline features, the forced response signal is demodulated at multiple scales to establish explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve and impact attenuation curve, and generate an initial feature set.

[0032] Specifically, the natural vibration response signal of the anchor bolt machine's damping device under no external active excitation is separated from the extracted steady-state baseline features, and this signal is used as the zero-disturbance reference for subsequent analysis. Next, using the natural vibration response signal as a reference, the collected forced response signal is fused to eliminate the influence of operational fluctuations and environmental interference, resulting in a fused signal that reflects only the damping device's own response characteristics. Subsequently, the scale parameters for multi-scale demodulation are adaptively set according to the historical degradation weight of the anchor bolt machine's damping device, and multi-scale demodulation is performed on the fused signal. Based on the demodulation results, the ratio of response amplitude to excitation amplitude was calculated through multi-scale frequency domain analysis. After weighted explicit enhancement by degradation mode contribution weights, a damping transfer spectrum was established. The displacement-force cycle was segmented into multiple scales, and the average slope of each cycle was calculated. After weighted explicit enhancement, the curves were arranged in cycle order to establish hysteresis stiffness curves. Multi-scale analysis of hydraulic pressure was performed to obtain the time curve for pressure recovery to steady state. After weighted explicit enhancement, a pressure recovery curve was established. Multi-scale envelope analysis was performed on transient impact events to extract amplitude decay curves. After weighted explicit enhancement, an impact decay curve was established. Finally, the four explicit features—damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve, and impact decay curve—were integrated to generate an initial feature set for subsequent health status diagnosis.

[0033] Step S400: Perform baseline correction on the initial feature set to reweight the features and establish the degradation mode feature components.

[0034] Specifically, for the explicit features in the initial feature set, such as the damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve, and impact attenuation curve, standardization is performed based on the steady-state baseline features. The steady-state baseline features characterize the health baseline state of the damping device. Standardization eliminates interference from different operating conditions and sensor data acquisition differences, making the values ​​of each explicit feature more comparable. After standardization, the results are input into an adaptive feature degradation mode explicit model. This model performs feature reweighting, assigning higher weights to features strongly correlated with the degradation state of the damping device to strengthen their representation of degradation information, while reducing the weights of features with weak correlation or prone to interference to weaken their impact. Through baseline correction and feature reweighting, a degradation mode feature component that accurately reflects the key information of each degradation mode of the damping device is finally established.

[0035] Step S500: Input the feature components of the degradation mode into the temporal causal graph network, and output the current health status score and the contribution weight of each degradation mode.

[0036] Specifically, the established degradation mode feature components are first input into a pre-constructed temporal causal graph network. This network combines temporal analysis and causal reasoning capabilities, learning the correspondence between historical degradation data and health status of the anchor bolt machine damper, as well as the intrinsic relationships between different degradation modes, to perform calculations and analysis on the input degradation mode feature components. During the calculation process, the network first quantitatively assesses the overall health status of the damper, outputting a current health status score presented in specific numerical form, intuitively reflecting the health level of the damper. Simultaneously, it identifies and calculates the contribution weights of various degradation modes affecting the current health status, such as damping degradation, stiffness degradation, and hydraulic system degradation, clarifying the degree of influence of different degradation modes on the health status of the damper. Finally, it synchronously outputs the current health status score and the contribution weights of each degradation mode, providing data support for subsequent health warnings and maintenance decisions of the damper.

[0037] In one possible implementation, step S200 further includes:

[0038] Step S210: Establish ultimate load constraints based on the operating state and the steady-state baseline characteristics.

[0039] Step S220: Under the ultimate load constraint, read the historical health score of the anchor bolting machine, and configure the excitation amplitude based on the historical health score and the steady-state baseline characteristics.

[0040] Step S230: Set a random disturbance factor, and under the limit load constraint, perform random probability disturbance of the excitation amplitude based on the random disturbance factor, and establish active excitation using the random probability disturbance result.

[0041] Specifically, the anchor bolting machine control system calls upon the operating condition data output by the operation status identification module, such as drilling speed, thrust level, and current operation cycle stage, to clarify the actual load requirements of the vibration damping device under the current operation. Simultaneously, from the steady-state baseline features of the multi-source raw dataset, key performance parameter thresholds for the vibration damping device under healthy conditions are extracted, including maximum allowable vibration acceleration, hydraulic system safe pressure range, displacement stroke limit, and upper limit of force bearing capacity. Then, a load coupling analysis algorithm is used to correlate the actual load requirements corresponding to the operating state with the performance thresholds in the steady-state baseline features, eliminating instantaneous overload interference caused by operating condition fluctuations, and determining the load boundaries that the vibration damping device can safely withstand under the current operating scenario. The upper load limit must be lower than 90% of the performance thresholds in the steady-state baseline features to reserve safety redundancy, while the lower load limit must meet the minimum excitation requirements for normal response of the vibration damping device under the operating state. Finally, the calculated load boundary parameters, such as vibration amplitude of 0.1~0.8g, hydraulic pressure of 5~18MPa, and force of 10~50kN, are written into the system constraint configuration file to form the ultimate load constraint, ensuring that the subsequent active excitation will not exceed the safe operating range of the vibration damping device.

[0042] Based on the established ultimate load constraints, historical health score data of the anchor bolt machine damping device is retrieved. These scores, derived from quantitative health status results output during historical diagnostic processes, reflect the performance degradation of the damping device at different times. Subsequently, the historical health scores are correlated with extracted steady-state baseline characteristics: if the historical health score is high, close to the health benchmark level, it indicates good overall performance of the damping device. In this case, based on the standard load parameters corresponding to the steady-state baseline characteristics, an excitation amplitude slightly higher than the benchmark is configured within the ultimate load constraint range to accurately detect potential minor degradation. If the historical health score is low, indicating some degree of performance degradation, conservative performance parameters from the steady-state baseline characteristics are referenced, and a relatively mild excitation amplitude is configured under the ultimate load constraint to avoid excessive excitation that could damage the damping device. Through this method of combining historical health status with benchmark characteristics, an excitation amplitude suitable for the current state of the damping device and meeting safety constraints is ultimately determined.

[0043] Based on historical disturbance test data and the interference characteristics of the operating environment of the anchor bolt machine vibration damping device, the parameter range of the random disturbance factor was determined through statistical analysis. This range included amplitude fluctuation coefficient, frequency disturbance step size, and probability distribution type. A normal distribution was selected to simulate random fluctuations under actual working conditions, thus completing the setting of the random disturbance factor. Subsequently, using the configured reference excitation amplitude as input, within the established ultimate load constraint range, a Monte Carlo sampling algorithm was used to randomly perturb the reference excitation amplitude: based on a preset normal distribution probability density function, multiple sets of random disturbance values ​​conforming to the probability distribution were generated and superimposed with the reference excitation amplitude to obtain multiple sets of candidate excitation amplitudes. Then, a load compliance verification algorithm was used to screen the candidate excitation amplitudes, eliminating outliers exceeding the upper limit of the ultimate load constraint or falling below the lower limit, retaining valid excitation amplitudes that meet safety requirements. Finally, the valid excitation amplitudes were distributed to multiple preset disturbance frequencies, such as 5Hz, 10Hz, and 15Hz, generating a set of excitation signals containing multiple frequency components and multiple amplitude combinations, thereby establishing an active excitation capable of applying multi-frequency disturbances to the vibration damping device.

[0044] In one possible implementation, step S300 further includes:

[0045] Step S310: Using the natural vibration response signal as a zero-disturbance reference, perform the forced response signal fusion analysis to establish a fused signal, which is a signal set that eliminates the influence of operating conditions and environmental disturbances.

[0046] Step S320: Perform multi-scale demodulation on the fused signal, wherein the demodulation scale is adaptively set based on the historical degradation weight.

[0047] Step S330: Based on the multi-scale demodulation results, establish explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve and impact attenuation curve.

[0048] Specifically, the natural vibration response signal extracted from the steady-state baseline characteristics is first determined as the zero-disturbance benchmark. This benchmark accurately reflects the inherent vibration characteristics of the anchor bolting machine's vibration damping device when there is no external active excitation and it is unaffected by fluctuations in operating conditions and environmental interference, serving as the reference standard for subsequent signal analysis. Subsequently, for the collected forced response signal, a signal fusion analysis algorithm is used to correlate it with the zero-disturbance benchmark: by calculating the difference between the forced response signal and the zero-disturbance benchmark in the time domain, signal interference caused by changes in operating conditions, such as additional vibrations caused by increases or decreases in the anchor bolting machine's drilling load, is eliminated; simultaneously, the spectral distributions of the two are compared in the frequency domain to filter redundant frequency components caused by environmental factors, such as vibration transmission from surrounding equipment and the impact of temperature fluctuations on the signal. After dual fusion processing in the time and frequency domains, the final result is a signal set that retains only the true response information of the vibration damping device to active excitation, i.e., a fused signal that eliminates the influence of operating conditions and environmental disturbances.

[0049] The system data interface retrieves the historical health diagnosis database of the anchor bolt machine vibration damping device, extracting the degradation mode contribution weight data from each diagnosis record, i.e., historical degradation weights, such as the past weight proportions of modes like damping degradation, stiffness attenuation, and hydraulic system degradation. A weighted average algorithm is then used to calculate the historical average weight value for each degradation mode. Subsequently, based on a preset degradation weight-demodulation scale mapping rule (e.g., degradation modes with a historical average weight ≥ 50% correspond to a fine scale of 1–3 kHz, those with a weight between 30% and 50% correspond to a medium scale of 3–8 kHz, and those with a weight < 30% correspond to a coarse scale of 8–15 kHz), the frequency band parameters for multi-scale demodulation are adaptively allocated, combining the historical average weights of each degradation mode. Next, a wavelet packet multi-scale decomposition algorithm is used, based on the configured frequency band parameters, to perform layered demodulation of the fused signal: a high sampling rate, such as 10 kHz, is used for fine-scale decomposition to capture signal features corresponding to minor degradations; corresponding sampling rates, such as 5 kHz and 2 kHz, are used for medium and coarse-scale decomposition to balance feature extraction accuracy and computational efficiency. During demodulation, the matching degree between the decomposition results at each scale and the historical degradation weights is checked in real time. If the signal-to-noise ratio of the feature signal at a certain scale is lower than the threshold, the frequency band range of that scale is dynamically adjusted to ensure that the demodulation results can accurately cover the feature information of each degradation mode, and finally complete the multi-scale demodulation processing of the fused signal.

[0050] Based on the multi-scale demodulation results, explicit characteristic curves are constructed in four categories: First, multi-scale frequency domain analysis is performed on the demodulation results to calculate the ratio of the response amplitude of the damping device to the active excitation amplitude at different frequencies. This ratio is used as the damping transfer characteristic, and it is weighted and explicitly enhanced by combining the contribution weight of historical degradation modes. The data is then arranged in frequency order to generate a damping transfer spectrum. Next, the displacement-force signal in the demodulation results is segmented into multiple complete displacement-force cycles. The average slope of each cycle is calculated, and after weighting and enhancement by the contribution weight of degradation modes, the hysteresis stiffness is plotted in the order of the cycles. The demodulated hydraulic pressure signal is analyzed to determine the pressure change process after the active excitation disturbance ends at multiple scales. The time required for the pressure to recover to the pressure value corresponding to the steady-state baseline characteristic is calculated, forming a pressure recovery time curve. After weighting and enhancing by the degradation mode contribution weight, the pressure recovery curve is obtained. Finally, multi-scale envelope analysis is performed on the transient impact event signal in the demodulated results to extract the attenuation data of the impact amplitude over time. After weighting and enhancing by the degradation mode contribution weight, the impact attenuation curve is plotted along the time axis. Finally, the establishment of four types of explicit curves is completed, providing core feature data for the subsequent generation of the initial feature set.

[0051] In one possible implementation, step S330 further includes:

[0052] Step S331: Perform multi-scale frequency domain analysis on the multi-scale demodulation results, calculate the ratio of response amplitude to excitation amplitude, and use the ratio of response amplitude to excitation amplitude as a damping transfer characteristic.

[0053] Step S332: After enhancing the damping transfer characteristics by weighting the degradation mode contribution weights to make them explicit, a damping transfer spectrum is established.

[0054] Step S333: Perform multi-scale segmentation on the displacement-force cycle in the multi-scale demodulation result, calculate the average slope of each cycle, and then perform weighted explicit enhancement of the average slope according to the contribution weight of the degradation mode. Finally, arrange the cycles in order to establish the hysteresis stiffness curve.

[0055] Step S334: Perform multi-scale analysis on the hydraulic pressure in the multi-scale demodulation results, calculate the time curve for the pressure to recover to steady state after the disturbance ends, and enhance the explicitness of the time curve by weighting the contribution weight of the degradation mode to establish the pressure recovery curve.

[0056] Step S335: Perform multi-scale envelope analysis on the transient impact event in the multi-scale demodulation results, extract the amplitude decay curve over time, and enhance the weighted explicitness of the degradation mode contribution weight on the decay curve to establish the impact decay curve.

[0057] Specifically, taking the obtained multi-scale demodulation results as the processing object, a frequency domain transformation algorithm, such as Fast Fourier Transform, is first used to perform multi-scale frequency domain analysis on signals at different demodulation scales. This converts the demodulated fused signal in the time domain into a frequency domain signal, clearly presenting the response characteristics of the damping device at different frequency bands. Next, the excitation amplitude data corresponding to the established active excitation is retrieved. For the frequency domain signal at each demodulation scale, the forced response signal amplitude corresponding to each frequency point is extracted. The ratio of the response amplitude to the excitation amplitude at each frequency point is calculated through division. This ratio directly reflects the energy transfer efficiency and damping dissipation capacity of the damping device under different frequency excitations, and is a parameter characterizing damping performance; therefore, it is determined as the damping transfer characteristic.

[0058] A data retrieval algorithm was used to extract the contribution weight data of each degradation mode of the anchor bolt machine damping device from the system's historical diagnostic database. This data was based on the statistical results of the impact of each degradation mode on the device's performance degradation in past diagnostic records. Next, a feature weighting algorithm was used to multiply the obtained damping transfer characteristics (the ratio of response amplitude to excitation amplitude at each frequency point) with the contribution weight of the corresponding degradation mode point by point. This operation amplified the feature values ​​of the frequency points associated with high contribution weight degradation modes while weakening the feature values ​​of the frequency points corresponding to low contribution weight modes, thus achieving an explicit enhancement of the damping transfer characteristics. Finally, a curve fitting algorithm was used, with frequency as the horizontal axis and the weighted enhanced damping transfer characteristics as the vertical axis, to smooth the feature data of all frequency points, generating a continuous curve that intuitively reflects the damping transfer performance of the damping device at different frequencies, thus completing the establishment of the damping transfer spectrum.

[0059] For the displacement-force signals included in the multi-scale demodulation results, multiple complete displacement-force cycles are identified based on the periodic changes in the signal over time. Each displacement-force cycle refers to the process where the displacement starts from an initial value, changes, and then returns to the initial value, while the force also completes a complete change. Next, a multi-scale segmentation algorithm is used to segment each identified displacement-force cycle, incorporating different demodulation scales during segmentation to ensure accurate capture of the details of displacement and force changes within each cycle. After segmentation, linear fitting is performed on the displacement and force data within each displacement-force cycle, and the average slope of each cycle is obtained by calculating the slope of the fitted line. This average slope reflects the average stiffness level of the damping device within that cycle; a larger slope indicates stronger stiffness of the damping device in the corresponding cycle. Subsequently, the contribution weights of each degradation mode are retrieved. These contribution weights are determined based on past health diagnostic data and can reflect the degree of influence of different degradation modes on the overall performance degradation of the damping device. The average slope of each cycle is multiplied by the corresponding degradation mode contribution weight. This operation amplifies the changes in the average slope associated with high-contribution-weight degradation modes, making these stiffness characteristics, which are more crucial for health status diagnosis, more prominent. Simultaneously, it weakens the interference of low-contribution-weight degradation modes on the average slope, achieving a weighted explicit enhancement of the average slope. Finally, according to the temporal order of the displacement-force cycles, the average slopes after weighted explicit enhancement are arranged sequentially. With the cycle number as the horizontal axis and the weighted average slope as the vertical axis, these data points are connected to form a curve, which is the hysteretic stiffness curve that intuitively reflects the stiffness variation law of the damping device.

[0060] For the hydraulic pressure data recorded in the multi-scale demodulation results, a multi-scale analysis algorithm is used to analyze the pressure changes across different time dimensions. Multi-scale analysis progressively decomposes the hydraulic pressure signal from fine to coarse scales. The fine scale captures the details of instantaneous pressure fluctuations, while the coarse scale identifies the overall trend of pressure changes. This multi-dimensional analysis comprehensively grasps the changes in hydraulic pressure during and after active excitation disturbances. Next, the end time of the active excitation disturbance is determined. Starting from this point, the hydraulic pressure changes over time are continuously tracked, and the time required for the pressure to decrease or rise from the value at the end of the disturbance to the steady-state pressure value is calculated. The steady-state pressure value refers to the standard hydraulic pressure value of the damping device under healthy baseline conditions without active excitation disturbances. By comparing the real-time pressure with the steady-state pressure value, the time points when the pressure approaches the steady-state value are recorded. These time points are integrated with the corresponding pressure values ​​to form a time curve showing the pressure recovering to steady state after the disturbance ends. Next, the contribution weights of each degradation mode are retrieved. These weights are derived from past health diagnostic data of the anchor bolt machine's vibration damping device and reflect the impact of different degradation modes, such as aging of hydraulic system seals and hydraulic oil leakage, on the overall performance degradation of the damping device. The pressure value corresponding to each time point on the time curve is multiplied by the corresponding degradation mode contribution weight. This weighted processing amplifies the pressure recovery characteristics associated with high-contribution-weight degradation modes, making this more crucial information for health status assessment more prominent, while reducing the interference of low-contribution-weight degradation modes on the pressure recovery curve, thus achieving weighted explicit enhancement of the time curve. Finally, with time as the horizontal axis and the weighted explicit enhancement pressure value as the vertical axis, the processed pressure-time data points are connected sequentially in chronological order to form a curve that intuitively reflects the hydraulic pressure recovery process and recovery efficiency after the disturbance ends. This curve is the pressure recovery curve.

[0061] For the transient impact event-related signals recorded in the multi-scale demodulation results, a multi-scale envelope analysis algorithm is used to process the signals. Multi-scale envelope analysis analyzes the transient impact signal at different time and frequency scales. First, signal decomposition techniques are used to break down the complex impact signal into multiple sub-signals of different scales. Then, the envelope of each sub-signal is calculated separately to capture the amplitude variation characteristics of the transient impact event at different stages, avoiding the omission of key impact details by single-scale analysis. Next, based on the envelope data obtained from the multi-scale envelope analysis, complete information on the amplitude variation of the transient impact event over time is extracted. Using time as the axis, the impact amplitude data corresponding to different moments are integrated to form a curve that clearly shows the gradual decrease in amplitude from the peak after the impact. This curve is the amplitude decay curve over time, which intuitively reflects the damping capability of the shock absorption device for transient impacts. Then, the contribution weights of each degradation mode are retrieved from historical diagnostic data. These contribution weights are determined based on past health diagnostic results of the anchor bolt machine's damping device and represent the degree of impact of different degradation modes, such as elastic failure of damping components and aging of damping materials, on the overall performance degradation of the device. The amplitude corresponding to each time point on the attenuation curve is multiplied by the corresponding degradation mode contribution weight. Through this weighted processing, the amplitude attenuation characteristics associated with high contribution weight degradation modes are amplified, making this information, which is more important for judging the health status of the damping device, more prominent, while reducing the interference of low contribution weight degradation modes on the attenuation curve, thus achieving weighted explicit enhancement of the attenuation curve. Finally, with time as the horizontal axis and the weighted explicit enhancement amplitude as the vertical axis, the processed amplitude-time data points are connected sequentially in chronological order to form a curve that accurately reflects the amplitude attenuation law of transient impact events and the impact attenuation performance of the damping device. This curve is the impact attenuation curve.

[0062] In one possible implementation, step S400 further includes:

[0063] Step S410: The explicit features in the initial feature set are standardized based on the steady-state baseline features, which are features that characterize the health reference state of the shock absorption device.

[0064] Step S420: Input the standardization result into the adaptive feature degradation mode explicit model, perform feature reweighting, and establish degradation mode feature components.

[0065] Specifically, the explicit features within the initial feature set are defined. These features include the core parameters corresponding to the established damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve, and impact decay curve, such as the ratio of response amplitude to excitation amplitude in the damping transfer spectrum, the average slope in the hysteresis stiffness curve, the pressure recovery time in the pressure recovery curve, and the amplitude decay rate in the impact decay curve. Next, the steady-state baseline features extracted from the multi-source raw dataset are retrieved. These features are standard parameters obtained by analyzing signals such as vibration, hydraulic pressure, and displacement under healthy baseline conditions, accurately characterizing the normal performance level of the damping device without degradation. Subsequently, a feature standardization algorithm is used to compare the actual values ​​of each explicit feature in the initial feature set with the standard values ​​corresponding to the steady-state baseline features. By calculating the ratio of (actual feature value - baseline feature value) / baseline feature value, the dimensional differences between different feature dimensions, such as vibration amplitude, pressure value, and time, are eliminated, and all explicit features are uniformly converted into dimensionless standardized data, ensuring the consistency and accuracy of subsequent feature processing.

[0066] An adaptive feature degradation pattern explicitness model was constructed using the random forest algorithm. During construction, historical degradation data of the anchor bolt machine damping device was used as the training set. Input features included standardized parameters such as damping transfer features, stiffness features, pressure recovery features, and impact attenuation features obtained from previous diagnoses. Output labels were the actual contribution weights of each degradation pattern, such as damping degradation, stiffness attenuation, and hydraulic system failure. Next, multiple subsamples were extracted from the training set using Bootstrap sampling. A decision tree was constructed for each subsample. At each decision tree node split, information gain was calculated by randomly selecting some features to determine the optimal split node, thereby improving the model's generalization ability and robustness. After model training, the standardized results were input into the model. Based on the voting results of multiple decision trees, the model analyzed the correlation strength between each standardized feature and different degradation patterns, and then performed feature reweighting, assigning higher weights to features closely associated with high-contribution degradation patterns and lower weights to features with weaker associations. Finally, all the weighted feature data are integrated, and feature information that can accurately characterize each degradation mode is selected to establish degradation mode feature components.

[0067] In one possible implementation, step S500 further includes:

[0068] Step S510: Establish a degradation trend prediction result based on the current health status score and the contribution weight of each degradation mode.

[0069] Step S520: Construct a perturbation stimulus using the degradation trend prediction results.

[0070] Step S530: Conduct an active disturbance test on the vibration damping device based on the first disturbance excitation, and establish a test feedback set.

[0071] Step S540: Use the test feedback set to test and verify the degradation trend prediction results, and use the test verification results to update the current health status score and the contribution weight of each degradation mode.

[0072] Specifically, a Long Short-Term Memory (LSTM) network was selected as the machine learning algorithm for predicting degradation trends. This algorithm can effectively capture long-term dependencies in time-series data and is suitable for processing sequence data where the contribution weights of health status scores and degradation patterns change over time. When building the model, historical diagnostic data was used as the training set. Input features included health status scores at each past diagnosis time and the contribution weights of various degradation patterns, such as damping degradation, stiffness decay, and hydraulic system failure. The output labels were set as the health status scores and contribution weights of each degradation pattern at subsequent time points. During model training, hyperparameters such as the number of hidden layer neurons and the learning rate were adjusted, and the backpropagation algorithm was used to continuously optimize the model parameters, enabling the model to accurately learn the changing patterns of health status and degradation patterns in historical data. After training, the current health status score and the contribution weights of each degradation mode are input into the LSTM model. Based on the learned temporal patterns, the model will output the predicted health status score for a future period (such as the next week or month), as well as the trajectory of the change in the contribution weights of each degradation mode. For example, it is predicted that the contribution weight of damping degradation will increase from the current 40% to 55% in the next month, and the health status score will decrease from the current 80 points to 72 points. These predicted data together constitute the degradation trend prediction results that can clearly show the future direction of the performance of the shock absorber.

[0073] Based on the high-risk degradation modes predicted in the degradation trend forecast, such as the predicted significant increase in the contribution weight of stiffness decay in the short to medium term, the core objective of the primary perturbation excitation is determined. Perturbation parameters that can accurately excite the characteristics of high-risk degradation modes are designed. Referring to the logic of adaptive active excitation configuration, the excitation amplitude range is set in conjunction with the health status change trend in the forecast results to avoid exceeding the ultimate load constraint. The perturbation frequency is determined based on the sensitive frequencies of the high-risk degradation modes; for example, for stiffness decay, a frequency band near its resonant frequency can be selected as the perturbation frequency. Simultaneously, the perturbation duration is controlled to ensure that a complete response signal can be acquired, ultimately constructing a highly targeted primary perturbation excitation.

[0074] According to the constructed disturbance excitation parameters, an active disturbance is applied to the anchor bolt machine's vibration damping device. During this process, multiple source sensors are activated to synchronously collect signals such as vibration, hydraulic pressure, displacement, and force after the disturbance. These signals reflect the actual response of the vibration damping device under specific disturbances. The collected raw signals are preprocessed, including noise removal and filtering of valid data segments. The processed signals are then correlated and integrated with the corresponding disturbance parameters, such as excitation amplitude, frequency, and duration, to form a test feedback set containing the correspondence between "disturbance parameters and response signals."

[0075] The test feedback set undergoes data preprocessing to filter out valid response signals collected during active disturbance testing, such as vibration, hydraulic pressure, and displacement signals. Noise interference and outlier data points are removed to ensure data accuracy and reliability. Next, the preprocessed actual response signals are compared and analyzed with the corresponding predicted response data in the degradation trend prediction results. The difference between the actual vibration amplitude and the predicted amplitude at a specific frequency in the test feedback set is calculated. Simultaneously, the actual and predicted values ​​of key parameters such as pressure recovery time and impact decay rate are compared to quantify the degree of deviation. Subsequently, testing and verification are performed based on the deviation analysis results: if the deviation value is within the preset reasonable range, such as not exceeding 10%, it indicates that the degradation trend prediction result is consistent with the actual situation. At this time, the current health status score is slightly corrected based on the test feedback set, such as fine-tuning by 1 to 3 points according to the actual response signal, and the contribution weight of each degradation mode remains basically stable; if the deviation value exceeds the reasonable range, the cause of the deviation is analyzed in depth, such as the prediction model not fully considering a certain potential degradation factor. Combining the new feature information in the test feedback set, the actual impact of each degradation mode on the device performance is reassessed, and the contribution weight of the corresponding degradation mode is adjusted, such as adjusting the weight of degradation modes with high deviation correlation by ±5% to 10%, and the current health status score is updated simultaneously, so that the updated score and weight are more consistent with the actual health status of the vibration damping device.

[0076] In one possible implementation, step S500 further includes:

[0077] The multi-source sensor includes an acceleration array for collecting vibrations. The acceleration array is respectively deployed on the frame side, the upper end of the shock absorber, and the lower end of the shock absorber. The multi-source sensor is redundantly deployed and performs self-calibration processing before data acquisition.

[0078] Specifically, the multi-source sensors include an acceleration array specifically designed for acquiring vibration signals. To comprehensively and accurately obtain vibration data from different parts of the shock absorber, this acceleration array is installed in three key locations: first, on the frame side of the anchor bolt machine, to capture the vibration transmitted from the frame to the shock absorber; second, at the top of the shock absorber, directly monitoring the vibration response of the upper part; and third, at the bottom of the shock absorber, to collect vibration information from the connection points between the lower part of the shock absorber and other components in real time. Simultaneously, to ensure the reliability and security of data acquisition and avoid data loss or distortion due to the failure of a single sensor, these multi-source sensors employ a multi-source redundant deployment, meaning multiple sensors are deployed to collect the same monitoring parameter. Furthermore, before officially executing the data acquisition task, all multi-source sensors undergo self-calibration processing, adjusting their accuracy through internal calibration procedures to ensure the accuracy of subsequently acquired signals such as vibration, hydraulic pressure, and temperature, providing a high-quality data foundation for establishing a multi-source raw dataset and conducting health status diagnosis.

[0079] In one possible implementation, step S500 further includes:

[0080] Step S550: Establish a first health warning signal based on the current health status score, wherein the first health warning signal is a real-time warning signal.

[0081] Step S560: Obtain historical health status scores, perform time-series smoothing analysis using the historical health status scores and current health status scores, and establish a second health warning signal, which is a smoothed abnormal signal.

[0082] Step S570: Perform early warning fusion on the first health early warning signal and the second health early warning signal, output a fused early warning signal, and perform early warning management based on the fused early warning signal.

[0083] Specifically, the warning threshold range corresponding to the current health status score is determined. This range is based on the health baseline state of the anchor bolt machine vibration damping device, i.e., the health score corresponding to the steady-state baseline characteristics, and the industry's equipment failure risk level classification standards. For example, a health score of 90-100 is set as the no-warning range, 70-89 as the mild warning range, 50-69 as the moderate warning range, and below 50 as the severe warning range. Next, the real-time output of the current health status score is compared with the preset threshold range. If the score is in the mild warning range, a first health warning signal is generated, indicating "a slight decline in equipment health status, requiring monitoring of subsequent changes"; if it is in the moderate warning range, a first health warning signal is generated, indicating "significant degradation in equipment health status, recommending a shortened monitoring cycle"; and if it is in the severe warning range, a first health warning signal is generated, indicating "severe deterioration in equipment health status, requiring immediate shutdown and inspection." This warning signal is directly generated based on the real-time updated health status score and can immediately reflect the current health status of the device.

[0084] The system retrieves all historical health status scores of the anchor bolt machine's vibration damping device from its historical database over a period of time (e.g., the past month or quarter). These scores record the device's health level at different points in time. Then, using a time-series smoothing analysis algorithm, such as the moving average method, the system selects the N most recent historical health scores and the current score to form a data sequence. The mean and standard deviation of this sequence are calculated. Based on the smoothing analysis results, if a continuous downward trend in the health score is observed (e.g., a continuous decline of more than 5% over multiple periods), or if the smoothed score suddenly exhibits abnormal changes exceeding the normal fluctuation range (e.g., a single smoothed score decline of more than 10 points), a second health warning signal reflecting a long-term trend anomaly is generated. This signal avoids false alarms caused by short-term random disturbances and is considered a smoothing anomaly signal.

[0085] The first health warning signal reflects the current abnormal health status of the vibration damping device in real time, while the second health warning signal reflects the long-term trend of abnormal health status. A comprehensive judgment is made on the warning level, abnormality type, and impact range of the two types of signals. If the first health warning signal shows a slight abnormality and the second health warning signal shows no trend of abnormality, a low-risk fused warning signal is generated after fusion. If the first health warning signal shows a moderate abnormality and the second health warning signal shows a continuous degradation trend, a medium-risk fused warning signal is generated after fusion. If the first health warning signal shows a severe abnormality, regardless of the trend of the second health warning signal, a high-risk fused warning signal is generated. After outputting the fused warning signal, corresponding warning management measures are initiated according to its risk level. At low risk, the regular health monitoring frequency is maintained; at medium risk, the monitoring cycle is shortened and personnel are arranged for inspections; at high risk, the equipment shutdown procedure is immediately triggered and a maintenance team is organized to conduct fault diagnosis and repair work, ensuring timely control of the health risks of the vibration damping device.

[0086] In one possible implementation, step S570 further includes:

[0087] Step S571: After issuing an early warning based on the fused early warning signal, obtain the early warning response action.

[0088] Step S572: Based on the early warning response action and the fused early warning signal, reconstruct the monitoring cycle and perform health status monitoring and management under the monitoring cycle.

[0089] Specifically, according to the preset early warning mechanism, the integrated early warning signal will be pushed to the equipment management terminal or relevant personnel in a specified form, such as a system pop-up, SMS notification, or audible and visual alarm, to ensure that the responsible personnel can receive the early warning information in a timely manner. After an early warning is issued, the specific response measures taken by relevant personnel or the system in response to the early warning signal will be continuously tracked and recorded, i.e., the early warning response actions. These actions include arranging for technicians to inspect the vibration damping device on-site, starting backup equipment to replace the current equipment, adjusting the operating parameters of the anchor bolt machine to reduce the load on the vibration damping device, or temporarily shutting down the machine to await further troubleshooting. The specific content and execution time of these response actions must be accurately obtained and stored.

[0090] Analyze the risk level of the fused early warning signal and the execution of the early warning response actions: If the fused early warning signal is low-risk and the early warning response action has completed preliminary checks and confirmed no serious problems, reconstruct the monitoring cycle to a regular cycle, such as daily health status monitoring; if the fused early warning signal is medium-risk, or the early warning response action indicates the need for continuous monitoring of the device status, such as minor degradation that does not require immediate shutdown for maintenance, shorten the monitoring cycle, such as monitoring every half day, to more frequently monitor health changes; if the fused early warning signal is high-risk, and the early warning response action involves shutdown for maintenance, after maintenance is completed, a short-term high-frequency monitoring cycle should be set initially, such as once per hour, and gradually adjusted to a regular cycle after confirming the device has stabilized. After reconstructing the monitoring cycle, start the data acquisition and health status diagnosis process of the multi-source sensors according to the new cycle to continuously track the health status of the vibration damping device, ensuring that new early warnings can be triggered in a timely manner when abnormal conditions recur, achieving dynamic and precise management of the device's health status.

[0091] Example 2, based on the same inventive concept as the method for diagnosing the health status of the anchor bolt machine vibration damping device in the foregoing examples, such as... Figure 2 As shown, this application provides a health status diagnosis system for anchor bolt machine vibration damping devices. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0092] The original dataset creation module 10 is used to activate multi-source sensors to synchronously collect vibration, hydraulic pressure, temperature, acoustic emission signals, displacement and force signals after identifying the working status of the anchor bolting machine, and to create a multi-source original dataset.

[0093] The forced response signal acquisition module 20 is used to extract steady-state baseline features from the multi-source raw dataset according to the operation status, and adaptively configure active excitation according to the steady-state baseline features and the operation status, apply multi-frequency disturbance to the shock absorption device of the anchor bolt machine based on the active excitation, and acquire the mapped forced response signal.

[0094] The initial feature set generation module 30 is used to extract the natural vibration response signal from the steady-state baseline features, perform multi-scale demodulation on the forced response signal, establish explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve and impact attenuation curve, and generate an initial feature set.

[0095] The feature reweighting module 40 is used to perform feature reweighting under baseline correction on the initial feature set to establish degradation mode feature components.

[0096] The health status score acquisition module 50 is used to input the feature components of the degradation mode into the temporal causal graph network and output the current health status score and the contribution weight of each degradation mode.

[0097] Furthermore, the system is also used to implement the following functions:

[0098] Based on the operating status and the steady-state baseline characteristics, an ultimate load constraint is established; under the ultimate load constraint, the historical health score of the anchor bolting machine is read, and the excitation amplitude is configured based on the historical health score and the steady-state baseline characteristics; a random disturbance factor is set, and under the ultimate load constraint, the excitation amplitude is randomly disturbed based on the random disturbance factor, and the active excitation is established using the random probability disturbance result.

[0099] Furthermore, the system is also used to implement the following functions:

[0100] Using the natural vibration response signal as a zero-disturbance reference, the forced response signal fusion analysis is performed to establish a fused signal, which is a signal set that eliminates the influence of operating conditions and environmental disturbances. Multi-scale demodulation is performed on the fused signal, and the demodulation scale is adaptively set based on historical degradation weights. Based on the multi-scale demodulation results, explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve, and impact attenuation curve are established.

[0101] Furthermore, the system is also used to implement the following functions:

[0102] Multi-scale frequency domain analysis is performed on the multi-scale demodulation results to calculate the ratio of response amplitude to excitation amplitude, which is then used as a damping transfer characteristic. After weighted explicit enhancement of the damping transfer characteristic by degradation mode contribution weights, a damping transfer spectrum is established. The displacement-force cycle in the multi-scale demodulation results is segmented into multiple scales, and the average slope of each cycle is calculated. Then, the average slope is weighted explicit enhancement based on degradation mode contribution weights, and the curves are arranged in cycle order to establish hysteresis stiffness curves. Multi-scale analysis is performed on the hydraulic pressure in the multi-scale demodulation results to calculate the time curve for pressure recovery to steady state after the disturbance ends. The time curve is weighted explicit enhancement based on degradation mode contribution weights to establish a pressure recovery curve. Multi-scale envelope analysis is performed on the transient impact event in the multi-scale demodulation results to extract the amplitude decay curve over time. The decay curve is weighted explicit enhancement based on degradation mode contribution weights to establish an impact decay curve.

[0103] Furthermore, the system is also used to implement the following functions:

[0104] The explicit features in the initial feature set are standardized based on the steady-state baseline features, which are features that characterize the health benchmark state of the damping device. The standardized features are then input into the adaptive feature degradation mode explicit model, and feature reweighting is performed to establish degradation mode feature components.

[0105] Furthermore, the system is also used to implement the following functions:

[0106] A degradation trend prediction result is established based on the current health status score and the contribution weight of each degradation mode; a primary disturbance excitation is constructed using the degradation trend prediction result; an active disturbance test of the vibration damping device is conducted based on the primary disturbance excitation, and a test feedback set is established; the degradation trend prediction result is tested and verified using the test feedback set, and the current health status score and the contribution weight of each degradation mode are updated using the test verification result.

[0107] Furthermore, the system is also used to implement the following functions:

[0108] The multi-source sensor includes an acceleration array for collecting vibrations. The acceleration array is respectively deployed on the frame side, the upper end of the shock absorber, and the lower end of the shock absorber. The multi-source sensor is redundantly deployed and performs self-calibration processing before data acquisition.

[0109] Furthermore, the system is also used to implement the following functions:

[0110] A first health warning signal is established based on the current health status score, wherein the first health warning signal is a real-time warning signal; historical health status scores are obtained, and time-series smoothing analysis is performed using the historical health status scores and the current health status scores to establish a second health warning signal, wherein the second health warning signal is a smoothed anomaly signal; the first health warning signal and the second health warning signal are fused together to output a fused warning signal, and warning management is performed based on the fused warning signal.

[0111] Furthermore, the system is also used to implement the following functions:

[0112] After issuing an early warning based on the fused early warning signal, an early warning response action is obtained; based on the early warning response action and the fused early warning signal, the monitoring cycle is reconstructed, and health status monitoring and management are performed under the monitoring cycle.

[0113] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0114] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0115] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for diagnosing the health state of a rock bolter damping device, characterized in that, The method comprises: After identifying the working state of the anchor rod machine, activate the multi-source sensor to synchronously collect vibration, hydraulic pressure, temperature, acoustic emission signal, displacement and force signal, and establish a multi-source original data set; According to the working state, extract the steady-state baseline feature from the multi-source original data set, and adaptively configure the active excitation according to the steady-state baseline feature and the working state, apply multi-frequency disturbance to the damping device of the anchor rod machine based on the active excitation, and collect the mapped forced response signal; After extracting the natural vibration response signal from the steady-state baseline feature, multi-scale demodulation is performed on the forced response signal to establish explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve and impact attenuation curve, and generate an initial feature set; Perform feature reweighting under baseline correction on the initial feature set to establish degradation mode feature components; Input the degradation mode feature components into the time series causal diagram network to output the current health state score and the contribution weight of each degradation mode; Take the natural vibration response signal as the zero disturbance reference, perform fusion analysis on the forced response signal to establish a fusion signal, which is a signal set eliminating the influence of working state and environmental disturbance; Perform multi-scale demodulation on the fusion signal, and adaptively set the demodulation scale based on the historical degradation weight; According to the multi-scale demodulation result, establish explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve and impact attenuation curve; Perform multi-scale frequency domain analysis on the multi-scale demodulation result, calculate the response amplitude to excitation amplitude ratio, and take the response amplitude to excitation amplitude ratio as the damping transfer feature; After the damping transfer feature is weighted and explicitly enhanced according to the degradation mode contribution weight, the damping transfer spectrum is established; After multi-scale segmentation of the displacement-force cycle in the multi-scale demodulation result, the average slope of each cycle is calculated, and then the average slope is weighted and explicitly enhanced according to the degradation mode contribution weight to establish the hysteresis stiffness curve in sequence; Perform multi-scale analysis on the hydraulic pressure in the multi-scale demodulation result, calculate the time curve of the pressure recovery to the steady state after the disturbance ends, and then perform time curve weighting and explicit enhancement through the degradation mode contribution weight to establish the pressure recovery curve; Perform multi-scale envelope analysis on the transient impact event in the multi-scale demodulation result, extract the amplitude decay curve with time, and then perform weighting and explicit enhancement on the decay curve according to the degradation mode contribution weight to establish the impact attenuation curve; Standardize the explicit features in the initial feature set based on the steady-state baseline feature, which is a feature representing the health reference state of the damping device; Input the standardization result into the adaptive feature degradation mode explicit model to perform feature reweighting processing and establish the degradation mode feature components.

2. The method according to claim 1, c h a r a c t e r i z e d b y, The adaptive configuration of the active excitation according to the steady-state baseline feature and the working state comprises: Establish the limit load constraint according to the working state and the steady-state baseline feature; Under the limit load constraint, read the historical health score of the anchor rod machine, and configure the excitation amplitude based on the historical health score and the steady-state baseline feature; A random disturbance factor is set, a random probability disturbance of an excitation amplitude is performed based on the random disturbance factor under the limit load constraint, and a random probability disturbance result is used to establish the active excitation.

3. The method according to claim 1, wherein the method comprises the steps of: determining the health state of the shock absorber of the roof bolter based on the comparison result of the first and second comparison results. After the degradation mode characteristic components are input into the time-series causal diagram network to output the current health state score and the contribution weight of each degradation mode, the method comprises: A degradation trend prediction result is established according to the current health state score and the contribution weight of each degradation mode; A primary disturbance excitation is constructed using the degradation trend prediction result; A test feedback set is established based on the primary disturbance excitation for active disturbance testing of the shock absorbing device; The degradation trend prediction result is tested and verified using the test feedback set, and the current health state score and the contribution weight of each degradation mode are updated using the test verification result.

4. The method according to claim 1, wherein the method is characterized by: The multi-source sensor comprises an acceleration array for collecting vibration, the acceleration array is respectively arranged on the rack side, the upper end of the shock absorber and the lower end of the shock absorber, and the multi-source sensor is redundantly arranged in a multi-source manner, and a self-correction process is performed before data collection is performed.

5. The method according to claim 1, wherein the method is characterized by: The output of the current health state score and the contribution weight of each degradation mode further comprises: A first health warning signal is established according to the current health state score, wherein the first health warning signal is a real-time warning signal; A historical health state score is obtained, time-series smoothing analysis is performed using the historical health state score and the current health state score, a second health warning signal is established, and the second health warning signal is a smoothed abnormal signal; The first health warning signal and the second health warning signal are warning fused to output a fused warning signal, and warning management is performed according to the fused warning signal.

6. The method according to claim 5, c h a r a c t e r i z e d b y, The warning management according to the fused warning signal comprises: After the warning is reported according to the fused warning signal, a warning response action is obtained; The monitoring cycle is reconstructed based on the warning response action and the fused warning signal, and health state monitoring management is performed in the monitoring cycle.

7. A health condition diagnosis system for a rock drilling rig damping device, characterized in that The system is used to implement the health state diagnosis method of the shock absorbing device of the anchor rod machine according to any one of claims 1-6, and the system comprises: An original data set establishment module is configured to activate the multi-source sensor to synchronously collect vibration, hydraulic pressure, temperature, acoustic emission signal, displacement and force signal after identifying the working state of the anchor rod machine, and establish a multi-source original data set; A forced response signal acquisition module is configured to extract a steady-state baseline feature from the multi-source original data set according to the working state, and adaptively configure an active excitation according to the steady-state baseline feature and the working state, apply a multi-frequency disturbance to the shock absorbing device of the anchor rod machine based on the active excitation, and acquire a mapped forced response signal; An initial feature set generation module is configured to extract a natural vibration response signal from the steady-state baseline feature, demodulate the forced response signal in multiple scales, establish explicit damping transfer spectrum, hysteresis stiffness curve, pressure recovery curve and impact attenuation curve, and generate an initial feature set; A feature reweighting module is configured to perform feature reweighting under baseline correction on the initial feature set to establish degradation mode characteristic components. The health state score acquisition module is configured to input the degradation mode feature components into a time-series causal diagram network, and output a current health state score and a contribution weight of each degradation mode.

Images