Bolt pretightening force intelligent monitoring method and system based on acoustic characteristics

Abstract

The invention discloses a bolt pre-tightening force intelligent monitoring method and system based on acoustic characteristics, and relates to the technical field of structure monitoring, and the method comprises the steps: synchronously obtaining three types of acoustic characteristic data of resonance characteristics, sound velocity ratio and nonlinear characteristics, carrying out the interactive calibration, and generating a credibility evaluation result; according to a credibility evaluation result, dynamically adjusting weight coefficients of the three types of acoustic characteristic data in a pre-tightening force calculation weight distribution strategy; inputting the weighted three types of acoustic characteristic data into a staged fusion model by adopting a staged fusion strategy to carry out pretightening force joint calculation; and establishing a long-term monitoring database, and according to deviation analysis of historical monitoring data and a real-time resolving result, updating acquired acoustic characteristic data parameters and a weight distribution strategy to form a closed-loop calibration mechanism. The technical problems of poor dynamic adaptability of multi-source acoustic characteristic data and calculation error accumulation in bolt pre-tightening force monitoring under complex working conditions are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of structural monitoring, and particularly to an intelligent monitoring method and system for bolt pre-tightening force based on acoustic characteristics. Background Art

[0002] The bolt pre-tightening force is a core parameter for the reliability of mechanical connections, directly affecting the load-bearing capacity and fatigue life of structures. From the perspective of technical implementation, although the existing sound velocity ratio measurement method can reflect the macroscopic stress state, it lacks the ability to characterize the local plastic deformation of the bolt micro-contact interface. For example, when fretting wear occurs on the bolt flange surface, the sudden change in the propagation path impedance of longitudinal and transverse waves will cause the distortion of the sound velocity ratio characteristics, and the existing algorithm takes it as the only solution basis, ultimately resulting in the risk of overestimating the pre-tightening force. The resonance frequency method is restricted by the performance attenuation of the acoustic coupling agent, and is prone to frequency shift misjudgment due to the change in the equivalent stiffness of the resonance cavity in a long-term vibration environment, and it cannot distinguish the essential difference between abnormal interface contact and real pre-tightening force loss.

[0003] A more profound technical contradiction is reflected in the data fusion strategy. The traditional linear weighted fusion model ignores the dynamic correlation between different acoustic characteristics: the response characteristics of the sound velocity ratio and the non-linear harmonic amplitude to the pre-tightening force are significantly different (the former shows an approximate linear relationship, and the latter is strongly correlated with the non-linear stiffness of the contact interface), but under the coupling action of multi-physical fields, the credibility of the two will change dynamically with the service stage of the bolt. The existing technology uses fixed weight coefficients for fusion calculation, which not only leads to detection blind spots in local stress concentration areas, but also causes false alarms due to feature conflicts in the initial stage of bolt loosening. In addition, the lack of a closed-loop calibration mechanism makes the system unable to optimize the acquisition strategy based on historical data and difficult to break through the accuracy barrier brought by the performance attenuation of sensors, which is the key bottleneck for the current intelligent monitoring technology to be widely applied in large-scale engineering.

[0004] In view of the above problems, there is an urgent need for an intelligent monitoring method and system for bolt pre-tightening force based on acoustic characteristics that integrates multi-dimensional acoustic characteristics and has dynamic calibration capabilities to solve the technical bottleneck and provide a high-precision and strong-robust solution for bolt connection health management. Summary of the Invention

[0005] In view of the deficiencies of the existing technology, the present invention provides an intelligent monitoring method and system for bolt pre-tightening force based on acoustic characteristics.

[0006] To achieve the above object, the technical solution of the present invention is as follows:

[0007] In the first aspect, the present invention discloses an intelligent monitoring method for bolt pre-tightening force based on acoustic characteristics, including the following steps:

[0008] S1. Synchronously acquire three types of acoustic feature data, namely resonance characteristics, sound velocity ratio, and nonlinear characteristics, and perform interactive calibration to generate a credibility evaluation result:

[0009] Transmit a swept-frequency signal to the resonant cavity and record the frequency offset of the resonant cavity to generate resonance characteristics;

[0010] Alternately excite longitudinal waves and transverse waves, and use the shape factor method to measure the propagation time difference between the longitudinal wave and the transverse wave to obtain the sound velocity ratio;

[0011] Apply a fundamental wave excitation, capture the amplitude of the second harmonic, and obtain the nonlinear characteristics of the fundamental wave and the second harmonic;

[0012] S2. According to the credibility evaluation result, dynamically adjust the weight coefficients of the three types of acoustic feature data in the pre-tightening force calculation weight allocation strategy:

[0013] When the resonance characteristics indicate abnormal contact, reduce the weight of the sound velocity ratio;

[0014] When the nonlinear characteristics exceed the preset threshold, increase the sampling frequency of the sound velocity ratio;

[0015] Establish a feedback relationship between the weight adjustment coefficient and the historical monitoring error and feedback it to the weight allocation strategy;

[0016] S3. Adopt a phased fusion strategy, and input the weighted three types of acoustic feature data into a phased fusion model for joint calculation of the pre-tightening force:

[0017] S301. Perform primary calculation, and calculate the macroscopic pre-tightening force based on the sound velocity ratio;

[0018] S302. Perform microscopic correction, and use the nonlinear characteristics to correct the local stress distribution;

[0019] S303. Perform credibility verification, and compare the resonance characteristics with the material theoretical impedance value to generate a real-time calculation result;

[0020] S4. Establish a long-term monitoring database, and update the parameters of the acoustic feature data acquisition and the weight allocation strategy according to the deviation analysis of the historical monitoring data and the real-time calculation result to form a closed-loop calibration mechanism:

[0021] S401. Statistically calculate the resonance characteristic frequency drift rate every week and update the reference frequency;

[0022] S402. Analyze the coupling degree of the sound velocity ratio and the nonlinear characteristics every month and optimize the weight function;

[0023] S403. Calibrate the sensor installation torque every quarter to ensure stable acoustic coupling;

[0024] Establish historical pre-tightening force change trend data and predict the decay curve of the prediction weight adjustment parameter;

[0025] S5. When the deviation of the calculation results of the three types of acoustic feature data exceeds 5%, trigger exception handling and execute in sequence:

[0026] S501. Trigger the self-cleaning program of the composite acquisition sensor;

[0027] S502. Re-acquire three groups of the acoustic feature data and take the average value;

[0028] S503. If it is still out of tolerance, switch to the backup sensor channel.

[0029] In a second aspect, the present invention discloses an intelligent bolt pre-tightening force monitoring system based on acoustic features, which uses the above-mentioned intelligent bolt pre-tightening force monitoring method based on acoustic features, and includes:

[0030] A composite acquisition module, configured to synchronously acquire three types of acoustic feature data including resonance characteristics, sound velocity ratio, and non-linear characteristics, and perform interactive calibration to generate a credibility evaluation result; the composite acquisition module includes a composite acquisition sensor and a calibration module, and the composite acquisition sensor integrates a resonance cavity generator, a dual-wave transceiver, and a non-linear probe;

[0031] A dynamic weighting processor, which adjusts the weight coefficient based on the credibility evaluation result and executes step S2 of claim 1;

[0032] A phased calculation engine, configured to adopt a phased fusion strategy to input the weighted three types of acoustic feature data into a phased fusion model for joint pre-tightening force calculation; the phased calculation engine includes a macroscopic calculation unit and a microscopic correction unit;

[0033] A self-optimizing control module, configured to establish a long-term monitoring database, and update the acquisition parameters and weight allocation strategy according to the deviation analysis of historical monitoring data and real-time calculation results to form a closed-loop calibration mechanism; the self-optimizing control module includes a historical database and a parameter regulator;

[0034] An exception handling unit, configured to trigger exception handling when the deviation of the calculation results of the three types of acoustic feature data exceeds 5%; the exception handling unit includes a backup sensor and a channel switcher.

[0035] Compared with the prior art, the beneficial effects of the present invention are:

[0036] 1. The present invention mainly aims at the problem of pre-tightening force failure in bolt connection structures in industrial scenarios due to complex factors such as vibration, temperature change, and material aging. By synchronously collecting three types of acoustic feature data, namely the frequency shift of the resonant cavity, the longitudinal / transverse wave velocity ratio, and the non-linear harmonic amplitude ratio, a dynamic weight allocation and phased solution model is constructed to achieve continuous and accurate monitoring of the bolt stress state. Its core application scenarios include, but are not limited to, key parts such as bolts of wind turbine towers, bridge anchoring structures, and connectors of aerospace equipment, solving the problems of insufficient reliability and high false alarm rate of traditional single acoustic parameter monitoring methods in complex environments.

[0037] 2. The contact impedance of the sensor is calibrated in real time through the resonant cavity frequency data to eliminate abnormal acoustic coupling caused by bolt surface oxidation or oil pollution; the longitudinal / transverse wave velocity ratio is used to reflect the macroscopic stress change, and the non-linear harmonic characteristics are combined to capture the microscopic interface slip, forming a dual verification mechanism from the whole to the local. The dynamic weight allocation strategy adaptively adjusts the contribution degrees of the three types of data according to the environmental interference intensity, and preferentially uses more anti-interference characteristic parameters in case of sensor anomalies or sudden vibrations. The phased solution model introduces the constraint conditions of acoustoelastic theory, combines physical laws with data-driven methods, and avoids error accumulation caused by pure algorithm models deviating from mechanical principles. The closed-loop calibration mechanism reversely optimizes parameters such as the acquisition frequency and excitation power through historical data, enabling the system to have the ability of self-evolution in long-term monitoring and significantly reducing the need for manual maintenance.

[0038] 3. Through the complementary enhancement of multi-modal acoustic features and the closed-loop control logic, the present invention essentially improves the monitoring accuracy of pre-tightening force under complex working conditions. The integrated design of the resonant cavity structure and the acoustic wave transceiver reduces the difficulty of sensor deployment, enabling it to be directly integrated into the standard structure of the bolt head. The anomaly processing unit ensures the continuity of monitoring through self-cleaning and channel switching functions, avoiding system downtime caused by the failure of a single sensor. Compared with the disadvantages of traditional methods relying on high-frequency manual re-inspection, its self-optimization mechanism can extend the calibration period to the quarterly level, especially suitable for scenarios such as offshore platforms and deep mountain transmission towers that are difficult to maintain frequently. Through the design of the inherent physical correlation of acoustic features, the monitoring performance bottleneck is broken through without increasing the hardware cost, providing reliable technical support for the life prediction and active maintenance of bolt connection structures of major equipment. Description of the Drawings

[0039] The disclosure of the present invention will be described with reference to the accompanying drawings. It should be understood that the drawings are only for illustrative purposes and are not intended to limit the scope of protection of the present invention. In the drawings, the same reference numerals are used to refer to the same components. Among them:

[0040] Figure 1 is the step flow chart of the present invention;

[0041] Figure 2Flowchart of step S1 for synchronously acquiring three types of acoustic feature data of the present invention;

[0042] Figure 3 Flowchart of step S4 for the long-term monitoring database of the present invention;

[0043] Figure 4 Flowchart of step S5 for triggering abnormal handling of the present invention;

[0044] Figure 5 Composition diagram of the system modules of the present invention;

[0045] Figure 6 Functional schematic diagram of the composite acquisition module 101 of the present invention;

[0046] Figure 7 Functional schematic diagram of the dynamic weighting processor 120 of the present invention;

[0047] Figure 8 Functional schematic diagram of the phased resolution engine 130 of the present invention;

[0048] Figure 9 Functional schematic diagram of the self-optimizing control module 140 of the present invention;

[0049] Figure 10 Functional schematic diagram of the abnormal handling unit 150 of the present invention. Detailed implementation manners

[0050] It is easy to understand that according to the technical solution of the present invention, without changing the essential spirit of the present invention, those of ordinary skill in the art can propose various interchangeable structural ways and implementation manners. Therefore, the following detailed implementation manners and the accompanying drawings are only exemplary descriptions of the technical solution of the present invention, and should not be regarded as the whole of the present invention or as a limitation or restriction on the technical solution of the present invention.

[0051] Application overview:

[0052] In the prior art, the current mainstream technology mainly relies on acoustic detection means of a single physical quantity (such as the ultrasonic velocity method or the resonant frequency method). Its core defect lies in the insufficient adaptability to complex working conditions: on the one hand, the physical sensitivity of a single characteristic parameter is easily interfered by environmental temperature fluctuations, changes in interface contact states, and material nonlinear effects, resulting in a deviation of the measurement result from the actual stress state; on the other hand, the existing methods generally adopt a static weight distribution model, which is difficult to cope with the multi-factor coupling effects such as material creep and friction coefficient attenuation during the service process of bolts, and error drift is likely to occur during long-term monitoring. More severely, when there is fretting wear or corrosion at the sensor coupling interface, the traditional methods lack an effective self-calibration mechanism, and often need to maintain the accuracy through manual re-measurement or disassembly and calibration, which seriously restricts the continuous monitoring ability in industrial scenarios.

[0053] To solve the above problems, the inventors found that the resonance characteristics, sound speed ratio, and nonlinear acoustic characteristics are complementary in their responses to changes in pre-tightening force: the resonance frequency offset is sensitive to the interface contact state, the sound speed ratio can characterize the macroscopic stress distribution, and the nonlinear harmonic amplitude can reflect the microscopic plastic deformation. Based on the dynamic correlation law of the three, a multi-characteristic collaborative calibration mechanism is proposed, and a credibility evaluation model is constructed by establishing physical constraint relationships between acoustic parameters (such as the acoustoelastic theory equation of resonance frequency and sound speed ratio). Further research found that when micro-movement wear occurs at the sensor coupling interface, the deviation degree between the sound speed ratio and the resonance characteristics shows exponential growth, and the nonlinear response of the nonlinear harmonic amplitude to the contact pressure can effectively correct this error, thus forming a dynamic weight adjustment rule and a staged fusion solution strategy.

[0054] In specific implementation, the system synchronously acquires three types of data: resonance characteristics, longitudinal / transverse wave propagation sound speed ratio, and nonlinear characteristics, and performs time synchronization and physical consistency verification in the calibration module: when the resonance frequency offset exceeds the material theoretical impedance threshold, mark the credibility level of the current sound speed ratio data, trigger resampling of the sound speed ratio and reduce its weight coefficient; at the same time, based on the nonlinear mapping relationship between the nonlinear harmonic amplitude and the macroscopic pre-tightening force, construct a local stress correction model. In the staged solution process, first calculate the initial value of the macroscopic pre-tightening force using the sound speed ratio, then correct the local deviation through the stress gradient distribution of the nonlinear characteristics, and finally verify whether the solution result conforms to the material impedance characteristics in combination with the resonance characteristics, forming a three-level solution closed-loop of "global-local-verification". The dynamic weight adjustment mechanism establishes a feedback model based on historical error data. When it is detected that the coupling degree between the sound speed ratio and the nonlinear characteristics decreases, the weight of the resonance characteristics is automatically increased to enhance the robustness of the system.

[0055] Compared with the traditional single-characteristic detection method, this solution breaks through the adaptability limitation of the static model to complex working conditions: through the multi-characteristic dynamic calibration mechanism, it can effectively suppress the sound speed ratio drift caused by temperature fluctuations; the staged fusion model solves the problem of acoustic parameter decoupling caused by material creep, and controls the long-term monitoring error within ±2% during the service life of the bolt; the closed-loop self-optimization function further realizes the dynamic compensation of the hardware performance attenuation. When the sensor installation torque drops by 15% due to vibration, the system automatically adjusts the resonance cavity sweep frequency range and excitation power through quarterly calibration data to maintain the acoustic coupling stability. This intelligent monitoring system that combines software and hardware provides an all-weather high-precision solution for the bolt health management of key structures.

[0056] After introducing the basic concept of the present invention, the embodiments of the present invention will be specifically introduced below with reference to the accompanying drawings.

[0057] Embodiment 1:

[0058] As Figure 1As shown in the figure, an intelligent monitoring method for bolt pre-tightening force based on acoustic characteristics includes the following steps:

[0059] S1. Synchronously acquire three types of acoustic characteristic data, namely resonance characteristics, sound velocity ratio, and nonlinear characteristics, and perform interactive calibration to generate a credibility evaluation result;

[0060] As Figure 2 shown in the figure, the acquisition method for synchronous acquisition in S1 is as follows:

[0061] Emit a swept-frequency signal to the resonance cavity, and record the frequency offset of the resonance cavity to generate resonance characteristics;

[0062] Alternately excite longitudinal waves and transverse waves, and use the shape factor method to measure the propagation time difference between the longitudinal wave and the transverse wave to obtain the sound velocity ratio;

[0063] Apply a fundamental wave excitation, capture the amplitude of the second harmonic, and obtain the nonlinear characteristics of the fundamental wave and the second harmonic.

[0064] In step S1, the three types of acoustic characteristic data are interactively calibrated to generate a credibility evaluation result. Interactive calibration means that after obtaining the three types of acoustic characteristic data, since different types of data may have errors, noise, or inconsistent measurement standards, they need to be interactively calibrated. After obtaining the three types of acoustic characteristic data of synchronous acquisition of resonance characteristics, sound velocity ratio, and nonlinear characteristics, data preprocessing is performed. Cubic spline interpolation is performed on the asynchronously sampled data through time domain alignment to unify the timestamps; again, wavelet threshold denoising is used for the resonance characteristics through noise reduction processing, Kalman filtering is applied to the sound velocity ratio data, and phase-locked amplification technology is used for the nonlinear characteristics to suppress environmental noise.

[0065] After generating the credibility evaluation result, after completing the interactive calibration of the three types of acoustic characteristic data, according to the calibrated three types of acoustic characteristic data, the credibility of the data is evaluated using a preset evaluation method and standard. The credibility evaluation result can reflect the reliability, accuracy of the data, and whether it can be used for subsequent analysis and decision-making.

[0066] The steps are as follows: Based on the acoustoelastic theory formula of the sound velocity ratio and the resonance frequency, verify the data consistency:

[0067]

[0068] k is the calibration coefficient, and R0 is the sound velocity ratio in the stress-free state. If the deviation between the measured resonance frequency offset Δf and the theoretical value exceeds 5%, the sound velocity ratio data is marked as suspicious.

[0069] Perform a threshold judgment on the nonlinear harmonic amplitude. When When it is determined that there is a significant non - linear effect, U1 represents the voltage amplitude of the fundamental wave excitation signal (unit: V), that is, the response amplitude of the single - frequency sine wave emitted by the sensor after propagating in the medium, and U2 represents the voltage amplitude of the second - harmonic signal (unit: V).

[0070] Input the verification result and real - time working condition parameters (temperature, vibration intensity, bolt service life) into the fuzzy inference system, and output the comprehensive credibility result.

[0071] Compared with traditional technologies, existing acoustic detection methods mostly rely on isolated acquisition of single characteristic parameters such as resonant frequency or sound - velocity ratio. Their physical characterization dimension is single and lacks a cross - verification mechanism. In traditional solutions, the resonant characteristics are prone to drift and distortion due to micro - motion wear at the sensor coupling interface. The sound - velocity ratio is significantly interfered by ambient temperature fluctuations, and the non - linear characteristics are difficult to independently support the calculation due to weak signals. Such isolated detection modes often lead to the accumulation of calculation errors due to the lack of data credibility evaluation when facing the dynamic changes of bolt pre - tightening force, material creep, and complex working condition interferences, and it is difficult to distinguish abnormal states such as real pre - tightening force loss and sensor coupling failure. This solution breaks through the one - sidedness dilemma of single - physical - quantity detection through a multi - characteristic collaborative perception and interactive calibration mechanism, and suppresses systematic errors caused by characteristic conflicts or data islands from the root cause.

[0072] Through the above technical solutions, this application realizes the full - dimension fusion calculation of multi - source acoustic characteristics, and constructs an adaptive dynamic perception system for bolt health state assessment. The resonant characteristics monitor the interface contact state in real - time, the sound - velocity ratio accurately captures macroscopic stress changes, the non - linear characteristics synchronously analyze the microscopic damage evolution, and the three types of data achieve cross - verification and dynamic weight adaptation through a physical correlation model. Each module elevates the discrete characteristics to a joint perception field with spatio - temporal consistency through time - series synchronization and logical coupling, and can still maintain the stability of pre - tightening force calculation in a complex interference environment, providing cross - scale technical support for the reliable monitoring of the bolt connection state of industrial equipment.

[0073] S2. According to the credibility evaluation result, dynamically adjust the weight coefficients of the three types of acoustic characteristic data in the weight distribution strategy for pre - tightening force calculation:

[0074] When the resonant characteristics show abnormal contact, reduce the weight of the sound - velocity ratio;

[0075] When the non - linear characteristics exceed the preset threshold, increase the sampling frequency of the sound - velocity ratio;

[0076] Establish a feedback relationship between the weight adjustment coefficient and the historical monitoring error and feedback it to the weight distribution strategy;

[0077] In step S2, the weight coefficients of the three types of acoustic feature data in the pre-tightening force calculation weight distribution strategy are dynamically adjusted mainly according to the described credibility evaluation results. The core purpose of this adjustment is to adaptively optimize the weight distribution by real-time monitoring the changes in various acoustic features, so that the subsequent pre-tightening force calculation based on these weights can be more accurate and reliable.

[0078] When the resonance feature indicates abnormal contact, it means that the current acoustic state may deviate from the normal expectation. Such abnormal contact may be caused by reasons such as uneven material surface or loose connecting components. At this time, the abnormal contact will have a significant impact on the sound speed, resulting in a decrease in the credibility of the information reflected by the sound speed. Therefore, it is necessary to reduce the weight of the sound speed. The weight adjustment method can assume that the original weight of the sound speed is relatively high in the overall weight distribution, such as 0.5. However, when it is detected that the resonance feature indicates abnormal contact, according to the set adjustment rule, its weight may be reduced to 0.3, etc. The specific reduction amplitude can be determined according to the actual degree of abnormality and the system's preset adjustment strategy.

[0079] When the non-linear feature exceeds the preset threshold, it indicates that the non-linear characteristics of the current material or system are relatively significant and may be in a certain special state (such as close to the yield limit of the material, there are local defects, etc.). At this time, it is necessary to collect sound speed samples more frequently to capture its dynamic changes more accurately. The weight adjustment method correspondingly increases the sampling frequency of the sound speed. For example, the original sound speed data is collected once per second, and it is increased to twice per second or more. At the same time, in terms of weight distribution, since more attention is paid to the sound speed information at this time, the weight of the sound speed may be increased from 0.3 to 0.4, etc., to give it a more important position in the subsequent calculation.

[0080] The feedback mechanism establishes the feedback relationship between the weight adjustment coefficient and the historical monitoring error, that is, through the analysis of historical data, the error situation generated by calculating the pre-tightening force under different weight distributions is statistically analyzed. For example, if it is found that at a certain moment, the pre-tightening force calculation error is large due to the unreasonable weight setting of a certain type of acoustic feature, then in the subsequent weight adjustment, reverse correction will be made according to this historical error information. The feedback information based on the historical monitoring error is fed back into the weight distribution strategy to finely adjust the weight coefficients of various acoustic features. If it is found that a certain type of acoustic feature has contributed a large amount to the calculation error in the past period (that is, the error is caused by too high or too low weight of this feature), then its weight is correspondingly reduced or increased to optimize the overall calculation accuracy.

[0081] The dynamic adjustment in S2 includes:

[0082] When the resonance cavity frequency offset exceeds the material's theoretical impedance value, mark that the credibility of the sound speed ratio drops by one level;

[0083] When the harmonic amplitude ratio of the non-linear feature exceeds a preset threshold, trigger the data recalibration acquisition of the sound velocity ratio.

[0084] Specific description of the dynamic adjustment in step S2: When the resonance cavity frequency offset exceeds the material's theoretical impedance value, this indicates that a significant change has occurred in the current material's internal acoustic state, which may be caused by external stress changes, internal material structure damage, etc. This large frequency offset will reduce the credibility of the results calculated based on the current sound velocity. Mark the credibility of the sound velocity ratio as decreased by one level, for example, from a high credibility level to a medium credibility level, and accordingly, appropriate adjustments will be made in the weight allocation to reduce the sound velocity weight.

[0085] Adjustment related to the harmonic amplitude of the non-linear feature: When the harmonic amplitude of the non-linear feature exceeds the preset threshold, it indicates that the non-linear effect of the material is relatively strong. At this time, the original calculation based on conventional sound velocity data may no longer be accurate. Trigger the data recalibration acquisition of the sound velocity, that is, re-acquire the sound velocity according to the new sampling strategy and parameters to obtain sound velocity data that can better reflect the current non-linear state. At the same time, the weight allocation will also be dynamically adjusted according to this change, and the sound velocity weight may be increased to highlight the importance of the newly acquired data in subsequent calculations.

[0086] The method for dynamically adjusting the weight coefficient is as follows:

[0087] Define the initial weight W0 as an empirical value, such as W0 = [0.5, 0.3, 0.2] (corresponding to the resonance feature, the sound velocity ratio, and the non-linear feature);

[0088] Real-time weight W t = W0 · [1 - α · Δf, 1 + β · ΔR, 1 + γ · ΔU], where α, β, and γ are attenuation factors calibrated based on material properties and environmental temperature, and Δf, ΔR, and ΔU are real-time change amounts.

[0089] Specifically, the setting of the initial weight is determined based on the comprehensive consideration of the physical sensitivity and reliability priority of the acoustic features in the bolt pre-tightening force calculation. These initial weight values are determined based on a large number of previous experiments and engineering practice experiences. When the system is initially running and there is not enough real-time adjustment basis, these are used as the initial importance measurement of various acoustic features in the pre-tightening force calculation. The specific logic of W0 = [0.5, 0.3, 0.2] (corresponding to the resonance, the sound velocity ratio, and the non-linear feature) is as follows:

[0090] (1) Resonance feature (0.5):

[0091] The highest weight is assigned because the resonance frequency offset is directly sensitive to the interface contact state (such as bolt loosening and fretting wear of the flange surface). Experiments show that in the initial stage of bolt pre-tightening force loss (such as within a 10% decrease in pre-tightening force), the rate of change of the resonance frequency offset is 3-5 times that of the sound velocity ratio, which can quickly respond to contact anomalies, so it is used as the core monitoring index.

[0092] (2) Sound velocity ratio (0.3):

[0093] The sound velocity ratio (the ratio of the longitudinal wave to the transverse wave propagation velocity) is strongly correlated with the acoustoelastic effect of the macroscopic pre-tightening force, but its response to local plastic deformation (such as stress concentration at the thread root) is weak. The weight is slightly lower than that of the resonance characteristics because it is easily disturbed by temperature drift (the temperature drift coefficient is about 0.05% / °C), and a dynamic calibration mechanism is required to suppress errors.

[0094] (3) Nonlinear characteristics (0.2):

[0095] The amplitude of the second harmonic is highly sensitive to the microscopic nonlinear stiffness of the contact interface (such as microcracks and plastic deformation), but its signal amplitude is usually 1-2 orders of magnitude lower than that of the fundamental wave, and the signal-to-noise ratio (SNR) is relatively low, so the initial weight is the smallest. The weight increase depends on a threshold triggering mechanism (such as the harmonic amplitude ratio exceeding a preset threshold).

[0096] Through the real-time weight calculation method, it is possible to dynamically adjust the weight of various acoustic characteristics in the pre-tightening force calculation according to the real-time changes of these characteristics to meet the calculation requirements under different working conditions and material states.

[0097] Compared with traditional technologies, traditional bolt pre-tightening force monitoring methods mostly rely on the threshold judgment mechanism of a single acoustic characteristic parameter (such as the resonance frequency with a fixed weight or the static sound velocity ratio). Their linear calculation models are difficult to adapt to the nonlinear interference caused by the multi-physical field coupling effect under complex working conditions. Such methods have three inherent defects: First, the fixed weight allocation strategy cannot respond to the credibility offset of acoustic parameters caused by the dynamic changes of material properties (such as temperature drift and stress relaxation); second, there is a lack of a feedback optimization mechanism for the collaborative relationship of multi-characteristic data, and misjudgment is likely to occur when local anomalies (such as thread plastic deformation and microcracks on the contact surface) cause characteristic contradictions; third, the traditional resampling mechanism is decoupled from the calculation model, making it difficult to timely correct the data distortion caused by environmental interference or sensor drift. This solution fundamentally solves the problem of adaptive fusion of multi-source heterogeneous acoustic characteristics under complex working conditions by constructing a dynamic mapping relationship between the acoustic characteristic credibility and the weight coefficient.

[0098] Through the above technical solutions, the present application realizes a closed-loop optimization system for multi-dimensional acoustic monitoring data, deeply integrating the physical sensitivity of acoustic features, environmental anti-interference ability and algorithm self-learning ability. The system effectively suppresses the distortion diffusion of acoustic parameters caused by microscopic deformation of the contact interface by quantifying in real time the inhibitory effect of abnormal resonance characteristics on the credibility of the sound velocity ratio; at the same time, based on the threshold triggering mechanism of non-linear characteristics, a dynamic balance between the data acquisition accuracy and the sensitivity of the solution model is constructed, enabling the sound velocity recalibration process to be synchronized with the evolution of microscopic material damage. More importantly, through the synergistic effect of historical error feedback and weight decay factor, the system can autonomously identify the true contribution degree of various acoustic features under different working conditions, significantly improving the early warning ability of sudden pre-tightening force failure while suppressing slow-changing interferences such as temperature drift, and forming an intelligent solution system with both environmental robustness and working condition adaptability.

[0099] S3. Adopt a staged fusion strategy, and input the weighted three types of acoustic feature data into a staged fusion model for joint pre-tightening force calculation:

[0100] S301. Perform primary calculation to calculate the macroscopic pre-tightening force based on the sound velocity ratio;

[0101] S302. Perform microscopic correction to correct the local stress distribution using the non-linear characteristics;

[0102] S303. Perform credibility verification, and compare the resonance characteristics with the theoretical impedance value of the material to generate a real-time calculation result;

[0103] The staged fusion strategy establishes a staged fusion model as follows:

[0104] The first stage calculates the macroscopic pre-tightening force where K1 is the acoustoelastic coefficient, v p 、v s are the longitudinal wave and transverse wave sound velocities respectively;

[0105] The second stage is microscopic correction, and the correction value where F0 is the initial pre-tightening force converted based on the bolt installation torque, F0 = τ·k·d, τ is the installation torque value, k is the thread coefficient, and d is the nominal diameter of the bolt;

[0106] Determine the final pre-tightening force F = F1 ± ΔF.

[0107] The sign of the correction value ΔF is determined by the correction rule:

[0108] When the frequency offset Δf of the resonance characteristic and the change amount ΔR of the sound velocity ratio satisfy:

[0109] take the negative sign;

[0110] When it is positive, where η is the material interface impedance coefficient.

[0111] Specifically, based on the sound velocity ratio weight value dynamically adjusted in step S2, step S301 preferentially uses the sound velocity ratio data to calculate the macroscopic pre-tightening force. The sound velocity ratio is an important parameter that reflects the propagation characteristics of longitudinal waves and transverse waves under specific conditions. According to the principles of physical acoustics, the propagation velocities of longitudinal waves and transverse waves in a medium are related to factors such as the stress state of the medium. By measuring and utilizing this sound velocity ratio information, a relationship model with the macroscopic pre-tightening force is established:

[0112] Formula:

[0113] Where K1 is the acoustoelastic coefficient (calibrated by material properties); v p , v s The weighted longitudinal wave / transverse wave sound velocity (the weight is inherited from the sound velocity ratio weight in step S2); the acoustoelastic coefficient K1 is a constant related to the acoustic characteristics of the material determined through a large number of experiments and theoretical studies, which relates the square of the sound velocity ratio to the macroscopic pre-tightening force. If the sound velocity weight is reduced due to abnormal contact in S201, then the confidence levels of v p , v s decrease, and the error needs to be compensated by increasing the redundancy coefficient of K1.

[0114] Step S302, based on the non-linear characteristic weight adjusted in the step, focuses on using the high-frequency harmonic amplitude to correct the local stress distribution. The microscopic correction mainly takes into account that in actual situations, the local stress distribution of the bolt will be affected by some non-linear characteristics, and these effects may not have been fully considered in the preliminary solution. Therefore, the non-linear characteristics are used to correct the local stress distribution, so as to calculate the pre-tightening force more accurately.

[0115] Second-stage correction value Where F0 is the calculated initial pre-tightening force converted based on the bolt installation torque, K2 is the non-linear correction factor, U1 and U2 are the reference harmonic amplitude and the real-time harmonic amplitude (the dynamic change of the non-linear characteristics directly affects the correction amplitude; it is also a coefficient related to material properties, etc., and it plays an important role in the microscopic correction calculation. By introducing the initial pre-tightening force F0 based on the installation torque and related velocity ratios and logarithmic functions, etc., various factors are comprehensively considered to correct the pre-tightening force.

[0116] After obtaining the pre-tightening force results after preliminary calculation and microscopic correction, it is necessary to verify their credibility. This is because in practical engineering applications, we need to ensure the accuracy and reliability of the calculation results, making them consistent with the material's theoretical resistance value, etc. Compare the harmonic resonance characteristics with the real-time calculation results of the material's theoretical resistance value. By comparing the harmonic resonance characteristics related to the pre-tightening force obtained by calculation with the known material's theoretical resistance value, if the two are in agreement within a certain error range, the calculation result is considered to have a high credibility; otherwise, it is necessary to further check and adjust the calculation model or parameters, etc.

[0117] By establishing a phased fusion model through a phased fusion strategy, overall, the calculation of the pre-tightening force is gradually improved through the calculations of the above three sub-steps. Starting from the macroscopic pre-tightening force calculation based on the sound velocity ratio, to the microscopic correction considering non-linear characteristics, and then to the credibility verification through comparison with the material's theoretical resistance value, a complete calculation system is formed. The sign of the correction value ΔF is determined by the correction rule, and this rule for determining the sign of the correction value is based on the research and summary of the material interface characteristics and the relationship between acoustic characteristic changes. By reasonably judging the sign of the correction value, the accuracy of the pre-tightening force calculation is further improved.

[0118] Example scenario simulation of the dynamic interaction between the phased fusion model and S2:

[0119] S2 detects that the non-linear harmonic amplitude exceeds the threshold, and increases the sound velocity weight from 0.3 to 0.4;

[0120] The non-linear correction factor in S302 is synchronously increased to strengthen the non-linear correction term;

[0121] If the verification in S303 fails (Δf exceeds the limit), the S203 feedback mechanism is triggered to reduce the resonance characteristic weight and re-calculate.

[0122] Step S3 integrates the acoustical characteristic weights dynamically adjusted in step S2 into the pre-tightening force calculation through a three-level process of "macroscopic calculation → microscopic correction → credibility verification". The input data of S3 are the weighted acoustical characteristics (sound velocity ratio, non-linear characteristics, resonance characteristics) output by S2, and the weight allocation directly affects the calculation priority at each stage. The credibility verification result of S303 can trigger the secondary adjustment of the weights in S2 in the reverse direction (for example, when the verification fails, re-calibrate the weights through error analysis). The positive or negative sign of the correction value ΔF is determined by the dynamic relationship between the resonance characteristic and the sound velocity ratio, ensuring that the correction direction is consistent with the physical state.

[0123] Compared with traditional technologies, traditional methods for detecting bolt pre-tightening force usually rely on static calculation models of single parameters, such as solely relying on installation torque conversion or simple linear regression of single acoustic characteristics. These methods have significant defects: torque conversion is easily interfered by non-linear factors such as friction coefficient drift and thread wear, resulting in deviation in pre-tightening force calculation; while for the analysis of sound velocity ratio or resonance characteristics based on fixed weights, in the face of complex working conditions such as sudden changes in material interface impedance and abnormal contact surfaces, the coupling relationship between acoustic characteristics cannot be dynamically corrected. Especially when there is local stress concentration or non-linear attenuation in the acoustic wave propagation path, traditional models are difficult to distinguish between macroscopic pre-tightening force and microscopic stress distortion, resulting in unstable calculation results. This solution overcomes three major technical problems: dynamic evaluation of the confidence level of acoustic characteristics under complex working conditions, coupled analysis of macroscopic and microscopic stress fields, and real-time credibility verification, by constructing a multi-dimensional collaborative calculation system of sound velocity ratio, non-linear harmonics and resonance characteristics.

[0124] Through the above technical solution, this application realizes a precise pre-tightening force calculation system based on acoustic characteristics adaptation. The system innovatively introduces a phased fusion strategy within the framework of acoustoelastic theory, enabling the formation of a closed-loop optimization between macroscopic sound velocity ratio calculation and microscopic non-linear correction: the dynamic weight mechanism can autonomously adjust the contribution degree of acoustic characteristics and trigger feedback recalculation according to real-time working conditions such as abnormal contact and harmonic distortion; the sign determination rule of the correction value effectively avoids the systematic deviation caused by artificial experience intervention in traditional methods through the dynamic matching of material interface impedance and the change amount of acoustic characteristics; the two-way interaction between the credibility verification module and the feature weight adjustment constructs an all-process self-correction ability from data acquisition, model calculation to result verification, significantly improving the robustness and environmental adaptability of pre-tightening force monitoring in complex engineering scenarios.

[0125] S4. Establish a long-term monitoring database, and update the parameters of the acoustic characteristic data and the weight distribution strategy according to the deviation analysis of historical monitoring data and the real-time calculation results, forming a closed-loop calibration mechanism.

[0126] As Figure 3 shown, the long-term monitoring database in step S4 is automatically executed:

[0127] S401. Statistically analyze the resonance characteristic frequency drift rate every week and update the reference frequency;

[0128] S402. Analyze the coupling degree of the sound velocity ratio and the non-linear characteristics every month and optimize the weight function;

[0129] S403. Calibrate the sensor installation torque every quarter to ensure stable acoustic coupling;

[0130] Establish historical pre-tightening force change trend data and predict the attenuation curve of the weight adjustment parameters.

[0131] The closed-loop calibration mechanism in step S4 includes:

[0132] Statistically calculate the standard deviation σ of historical monitoring data monthly, preset a standard deviation threshold σ0. If σ > σ0, then execute:

[0133] Expand the sweep frequency range of the resonant cavity to ±20%;

[0134] Shorten the excitation interval between longitudinal waves and transverse waves by 20%;

[0135] Increase the output power of the excitation probe for non-linear characteristics by 10%.

[0136] Specifically, in the automatic execution steps of the long-term monitoring database, in step S401 during the operation of the long-term monitoring database, the drift rate of the harmonic vibration characteristic frequency is statistically calculated weekly. This is because in the actual operating environment, various factors may cause changes in the harmonic vibration characteristic frequency, and through weekly statistics, this change trend can be captured in a timely manner. Then, based on the statistically obtained drift rate, the reference frequency is updated. Ensure that subsequent calculations and analyses based on frequency are all established on an accurate reference, providing an accurate frequency reference for the entire monitoring and calibration system.

[0137] Step S402 performs a sound speed analysis once a month, specifically analyzing the coupling degree between the sound speed and non-linear characteristics. Since the coupling relationship between the sound speed and non-linear characteristics will have an important impact on the overall monitoring results, through monthly analysis, the change situation of this coupling degree can be dynamically understood. According to the analysis results, the weight function is optimized, so that the weight function can more accurately reflect the actual situation, and thus in subsequent various calculations and decisions, the weights can be reasonably allocated to improve the overall accuracy.

[0138] Step S403 calibrates the installation torque of the sensor once a quarter. During the long-term operation of the sensor, its installation torque may change due to various mechanical stresses, environmental vibrations, etc., and the accuracy of the installation torque is directly related to the stability of acoustic coupling. Through quarterly calibration, ensure that the sensor installation torque is in an accurate state, thereby ensuring stable acoustic coupling, enabling the sensor to accurately collect relevant acoustic data, and providing a reliable data basis for the entire monitoring system.

[0139] Step S404 establishes historical prestress change trend data. During the operation of the entire system, the change of prestress will have an impact on many related parameters, especially on the weight adjustment parameters. By analyzing and summarizing the historical prestress change trend, the attenuation curve of the weight adjustment parameters can be predicted. This helps to plan and adjust relevant parameter settings in advance to adapt to possible future changes and ensure the long-term stable operation of the system.

[0140] The standard deviation σ of historical monitoring data is statistically calculated monthly, and a preset standard deviation threshold σ0 is used as the judgment basis. When the statistically obtained standard deviation σ is greater than the preset standard deviation threshold σ0, corresponding adjustment measures in the closed-loop calibration mechanism need to be executed.

[0141] The specific adjustment measures are as follows:

[0142] (1) The frequency sweep range of the harmonic vibration cavity will be expanded to ±20%. When the standard deviation of historical monitoring data becomes larger, it indicates that the current working state of the harmonic vibration cavity may not be able to well adapt to the actual situation. By expanding the frequency sweep range, relevant acoustic characteristics can be captured more comprehensively, improving the accuracy and adaptability of monitoring.

[0143] (2) The interval between the excitation of longitudinal waves and transverse waves will be shortened by 20%. Such an adjustment is considered in the case of large current data fluctuations (large standard deviation). By shortening the interval between the excitation of longitudinal waves and transverse waves, relevant wave characteristic data can be obtained more densely, which helps to analyze and process possible abnormal situations more timely and accurately.

[0144] (3) The output power of the excitation probe for non-linear characteristics will be increased by 10%. When the data fluctuates greatly, appropriately increasing the output power of the excitation probe can enhance the excitation effect of non-linear characteristics, so as to obtain data related to non-linear characteristics more clearly, which helps to analyze and calibrate the entire system more accurately.

[0145] Through the above automatic execution steps of the long-term monitoring database and the operation of the closed-loop calibration mechanism, it can effectively ensure that the entire system continuously adjusts and optimizes its own parameters according to the actual situation during the long-term operation process, maintaining a stable and accurate operating state.

[0146] Compared with traditional technologies, traditional acoustic monitoring systems mostly rely on fixed parameter thresholds and periodic manual calibration. Their reference frequency settings remain unchanged for a long time, making it difficult to promptly capture sensor drifts caused by factors such as environmental temperature and humidity changes and mechanical stress relaxation. The weight distribution strategy in traditional methods is fixed and cannot dynamically adapt to the evolution of the coupling relationship between sound velocity and non-linear characteristics, easily causing continuous accumulation of characteristic calculation errors. At the same time, there is a lack of a quantitative monitoring mechanism for the stability of sensor installation torque, and the gradual deterioration of the acoustic coupling state is often discovered only after a failure occurs. Historical data is only used for trend backtracking, and a parameter decay prediction model has not been constructed, resulting in lag in system maintenance. These defects make traditional systems face multiple risks such as reference inaccuracy, weight mismatch, and coupling failure during long-term monitoring, seriously affecting the reliability of the health assessment of prestressed structures.

[0147] Through the above technical solutions, the present application realizes the adaptive closed-loop optimization of the acoustic monitoring system. The system dynamically corrects the reference frequency through weekly frequency drift analysis, effectively eliminating the continuous interference of environmental factors on the resonance characteristics. The monthly coupling degree calculation enables the weight function to continuously fit the dynamic relationship between the sound velocity and the nonlinear characteristics, significantly improving the accuracy of multi-dimensional data fusion. The quarterly torque calibration mechanism incorporates the acoustic coupling stability into the system self-check scope, fundamentally ensuring the effectiveness of the data collected by the sensor. The historical pre-tightening force trend modeling constructs the prediction ability of parameter attenuation, making the weight adjustment forward-looking. When the monitoring data fluctuation exceeds the limit, the linkage response of sweep range expansion, excitation interval compression, and excitation power increase forms a feature capture enhancement mechanism under abnormal conditions. The synergistic effect of these technologies enables the system to continuously maintain high-precision monitoring during long-term operation, establishing a full-cycle calibration closed-loop of "real-time perception - deviation analysis - parameter optimization - trend prediction".

[0148] As Figure 4 shown, the intelligent bolt pre-tightening force monitoring method based on acoustic characteristics further includes:

[0149] S5. When the deviation of the calculation results of the three types of acoustic characteristic data exceeds 5%, trigger exception handling and execute in sequence:

[0150] S501. Trigger the self-cleaning program of the composite acquisition sensor;

[0151] S502. Re-collect three groups of the acoustic characteristic data and take the average value;

[0152] S503. If it is still out of tolerance, switch to the backup sensor channel.

[0153] As Figure 4 shown, when the deviation of the calculation results of the three types of acoustic characteristic data exceeds 5%, the exception handling process will be triggered. Under normal circumstances, the bolt pre-tightening force result obtained through reasonable calculation of the acoustic characteristic data should be within a certain accuracy range. Once the deviation exceeds 5%, it is very likely that there are some problems in the current monitoring system, such as the sensor being contaminated, the data acquisition process being interfered with, or the sensor itself malfunctioning, etc. Therefore, a series of exception handling steps need to be executed to ensure the accuracy of the monitoring and the stability of the system.

[0154] A 5% deviation threshold. Referring to the stress-strain non-linear interval in material mechanics, for the monitoring of bolt pre-tightening force, when the deviation of the acoustic characteristic data calculation result reaches 5%, this is likely to mean that the stress state of the bolt has approached or started to enter the area that may cause obvious changes in material mechanical properties, similar to the stress-strain curve starting to deviate from the linear interval. In this case, even if it has not really entered the obvious non-linear interval, there is a high possibility that the pre-tightening force state of the bolt may have abnormal changes or there may be problems with the monitoring system itself (such as sensor performance changes, external interference, etc.). Therefore, setting a 5% deviation threshold can be used as an early warning signal to prompt the system to check and make possible adjustments.

[0155] If the deviation threshold is set too low, such as 1% or 2%, it may cause the system to be too sensitive, frequently trigger the abnormal handling process, and increase unnecessary operations such as self-cleaning, re-acquiring data, and even switching sensors. This will not only increase the energy consumption and operating cost of the system, but also may introduce new errors due to frequent operations. If the deviation threshold is set too high, such as 10% or higher, it may cause the system to fail to detect the abnormal changes in the bolt pre-tightening force in time, miss the early warning and handling opportunities for problems such as bolt loosening and insufficient pre-tightening force, thus affecting the safety and reliability of the entire structure. Considering comprehensively, a 5% deviation threshold can ensure timely detection of potential problems without making the system too sensitive, and it is a relatively reasonable balance between monitoring accuracy and system operation stability.

[0156] In step S501, when the deviation exceeds 5% and triggers abnormal handling, the first thing to execute is to trigger the self-cleaning program of the composite acquisition sensor. In many cases, pollutants such as dust and oil may adhere to the surface of the sensor, which will affect the acquisition accuracy of the acoustic signal by the sensor, resulting in a large deviation in the calculation result. By triggering the self-cleaning program, the built-in cleaning mechanism of the sensor (such as by specific air flow purging, slight vibration, etc.) can be used to clean the surface of the sensor, remove impurities that may affect the acquisition accuracy, and prepare for subsequent accurate data re-acquisition.

[0157] After the sensor self-cleaning program is completed in step S502, the operation of step S1 will be re-executed to re-acquire three groups of the acoustic characteristic data. This is because there may be accidental errors in a single data acquisition. By collecting multiple groups of data and taking the average value, the influence of such accidental errors on the final result can be reduced to a certain extent. And theoretically, the data re-acquired after the sensor is cleaned should be closer to the true value. After collecting three groups of data, the system will automatically calculate the auto-calibration of these three groups of data, generate a credibility evaluation result as the new acoustic characteristic data for subsequent calculations and other operations.

[0158] In step S503, if after the sensor self-cleaning and re-acquisition of data and taking the average value, the new calculated result of the acoustic feature data still has a deviation exceeding 5%, it is necessary to consider that the currently used sensor may have relatively serious faults or other irreparable problems. At this time, the system will automatically switch to the backup sensor channel and continue to collect and calculate the acoustic feature data using the backup sensor. The backup sensor is in a standby state under normal circumstances, but when the main sensor has problems and cannot be solved after preliminary processing, it can take over the work of the main sensor in a timely manner to ensure that the intelligent monitoring of the bolt pre-tightening force can continue to operate normally without monitoring interruption and other situations.

[0159] Compared with the traditional technology, the traditional bolt pre-tightening force monitoring method mainly relies on manual regular inspections or single-sensor static detection, and its technical defects are mainly reflected in two aspects: response lag and system vulnerability. Traditional manual detection requires shutdown and disassembly for verification, which not only affects the operation efficiency of the equipment, but also is difficult to capture the real-time change of the pre-tightening force under dynamic working conditions; while conventional automated monitoring systems mostly adopt a fixed-threshold alarm mechanism, lacking the ability to classify and process composite factors such as sensor performance degradation and environmental interference, and are prone to misjudgment or missed detection due to single data anomaly. This solution aims at the pain points of poor monitoring continuity and weak anti-interference ability existing in the traditional technology, and effectively solves the problem of monitoring inaccuracy caused by sensor contamination, instantaneous interference or hardware failure in industrial scenarios by constructing a dynamic response system for the deviation of acoustic feature calculation, realizing the technical leap from passive alarm to active diagnosis.

[0160] Through the above technical solutions, this application realizes the multi-dimensional reliability improvement of the bolt pre-tightening force monitoring system. Based on the intelligent grading response mechanism with a 5% deviation threshold, it not only avoids the excessive sensitivity of the traditional system to minor deviations, but also can start the predictive processing process before the pre-tightening force enters the critical state. The self-cleaning function of the composite sensor combined with the mean processing of three groups of data significantly improves the anti-instantaneous interference ability of acoustic feature acquisition, while the dual-channel redundancy design fundamentally guarantees the continuity of the monitoring process. This closed-loop abnormal processing system can not only actively eliminate common interference sources such as sensor surface contamination and abnormal data acquisition, but also seamlessly switch to the backup channel in case of hardware failure, enabling the monitoring system to maintain stable operation under complex working conditions and providing continuous and reliable data support for the mechanical state assessment of key structures.

[0161] Embodiment 2:

[0162] As Figure 5 shown, the intelligent bolt pre-tightening force monitoring system based on acoustic features uses the above-mentioned intelligent bolt pre-tightening force monitoring method based on acoustic features, including:

[0163] The composite acquisition module 101 is used to synchronously acquire three types of acoustic feature data, namely resonance characteristics, sound velocity ratio, and nonlinear characteristics, perform interactive calibration, and generate a credibility evaluation result. The composite acquisition module 101 includes a composite acquisition sensor 110 and a calibration module 102. The composite acquisition sensor integrates a resonance cavity generator 111, a dual-wave transceiver 112, and a nonlinear probe 113.

[0164] The dynamic weighting processor 120 is configured to dynamically adjust the weight coefficients of the three types of acoustic feature data in the pre-tightening force calculation weight distribution strategy based on the credibility evaluation result.

[0165] The staged calculation engine 130 is used to adopt a staged fusion strategy to input the weighted three types of acoustic feature data into a staged fusion model for joint pre-tightening force calculation. The staged calculation engine 130 includes a macro calculation unit 131 and a micro correction unit 132.

[0166] The self-optimizing control module 140 is used to establish a long-term monitoring database, update the acquisition parameters and weight distribution strategy according to the deviation analysis of historical monitoring data and real-time calculation results, and form a closed-loop calibration mechanism. The self-optimizing control module 140 includes a historical database 141 and a parameter regulator 142.

[0167] The anomaly handling unit 150 is used to trigger anomaly handling when the deviation of the calculation results of the three types of acoustic feature data exceeds 5%. The anomaly handling unit 150 includes a backup sensor 151 and a channel switch 152.

[0168] As Figure 6 shown, the composite acquisition module 101 is the core data acquisition unit of the system, responsible for synchronously acquiring three types of key acoustic feature data (resonance characteristics, sound velocity ratio, nonlinear characteristics), and ensuring the reliability of the data through interactive calibration. The resonance cavity generator 111 detects the resonance characteristics (such as frequency offset, amplitude attenuation) of the bolt connection interface through acoustic wave excitation at a specific frequency, and generates resonance feature data. The dual-wave transceiver 112 emits longitudinal waves (v p ) and transverse waves (v s ), and calculates the sound velocity ratio by measuring the propagation speed difference between the two, which reflects the macroscopic stress state of the bolt. The nonlinear probe 113 captures the nonlinear response of the material after being stressed (such as harmonic amplitude, waveform distortion), generates nonlinear feature data, and is used to correct the local stress distribution. The calibration module 102 synchronously calibrates the original data, eliminates environmental interference (such as temperature drift, electromagnetic noise), and generates a credibility evaluation result (such as data confidence level, error range).

[0169] Among them, for initial data acquisition, the resonator generator 111 is started, which emits sound waves at a fixed frequency and records the resonance response to generate resonance characteristic data (such as the frequency offset Δf). The dual-wave transceiver 112 alternately emits longitudinal waves and transverse waves, calculates the propagation time difference between the two, and outputs the sound velocity ratio. The nonlinear probe 113 captures weak harmonic signals through a high-sensitivity sensor and extracts nonlinear characteristics (such as the amplitude of the second harmonic).

[0170] The three types of data are aggregated to the original data collection point, and the calibration module 102 performs the following operations: Time synchronization calibration: Align the acquisition timestamps of the three groups of data to eliminate timing deviations. Environmental compensation: Correct the temperature drift error of the sound velocity ratio according to the data of the temperature and humidity sensors. Confidence evaluation: Compare with the reference values in the historical database and mark abnormal data (such as a sudden increase in the resonance frequency exceeding 10%). Output the calibrated characteristic data and confidence labels (such as "high confidence" and "needs review").

[0171] As Figure 7 As shown, the dynamic weighting processor 120 inputs three types of acoustic characteristic data (resonance characteristics, sound velocity ratio, and nonlinear characteristics) from the composite acquisition module 101 and their confidence evaluation results, and outputs a weighted acoustic characteristic data stream for the phased resolution engine 130 to calculate the pre-tightening force.

[0172] Among them, for initial weight assignment: Assign weights according to preset empirical values. For example: Resonance characteristic weight: 0.4, sound velocity ratio weight: 0.4, nonlinear characteristic weight: 0.2.

[0173] Weight dynamic adjustment logic: Adjustment basis: Real-time confidence label: If a certain type of data is marked as "low confidence", the weight is reduced proportionally. For example, when the confidence of the sound velocity ratio is downgraded, the weight is reduced from 0.4 to 0.3. Adjustment rules: Conditions for weight increase: High data consistency, low environmental interference, and low historical error. Conditions for weight reduction: Data mutation, large environmental interference, and high historical error.

[0174] Apply the adjusted weights to the original data to generate a weighted data stream. The weighted data stream is transmitted to the phased resolution engine 130 in real time to trigger the subsequent resolution process.

[0175] As Figure 8As shown in the figure, the macro-solver unit 131 prepares to receive the weighted data stream output by the dynamic weighting processor 120. The input data includes: sound velocity ratio, which reflects the overall stress state of the bolt; non-linear characteristics, which are used for subsequent local correction; and resonance characteristics, which assist in verifying the credibility of the data. The macro pre-tightening force is calculated based on the sound velocity ratio data, and an empirical model is used to calculate the macro pre-tightening force of the bolt (for example, the initial estimated value is 1200 N), which serves as the reference value for micro correction. At the same time, a data confidence label (such as "high credibility" or "needs review") is output to guide the subsequent correction direction. If the sound velocity ratio indicates that the material stress is close to the critical value of the linear range (such as 80% of the material yield strength), the macro-solver result will be marked as "high risk", indicating that subsequent micro correction needs to be focused on.

[0176] The micro correction unit 132 integrates the input data, receives the macro-solver result and the weighted non-linear characteristic data, and performs local stress correction. If the weight of the non-linear characteristics is increased to 0.4, the influence weight of the change in harmonic amplitude will increase synchronously. The stress concentration in the local area of the bolt is detected (such as abnormal stress at the thread root), and based on the non-linear characteristics (such as a sudden increase in harmonic amplitude by 15%), the local stress correction value is calculated. If the deviation between the macro-solver result and the theoretical value is large, the micro correction unit identifies local loosening through high-frequency harmonic data, and the correction value ΔF is taken as a positive value to compensate for the pre-tightening force.

[0177] The corrected result is output to generate the corrected pre-tightening force, and the reason for correction (such as "local stress concentration") is marked for reference in credibility verification. If the ratio of the resonance characteristic frequency offset Δf to the sound velocity ratio change ΔR exceeds the material interface impedance coefficient η, it is determined as a "suspicious result".

[0178] As Figure 9 As shown in the figure, the self-optimizing control module 140 dynamically updates the acquisition parameters and weight allocation strategy based on the deviation between the long-term monitoring data and the real-time solver result. It forms a two-way data interaction with the composite acquisition module 101 and the dynamic weighting processor 120 to achieve system self-calibration. By learning from historical data, the impact of abnormal events on the system is reduced.

[0179] Historical database 141: Stores three types of acoustic characteristic data (resonance characteristics, sound velocity ratio, non-linear characteristics) in the long term; records the pre-tightening force solver result each time (including the macro-solver value, micro correction value, and final result); saves the abnormal event log (such as sensor self-cleaning records and the number of backup sensor switches).

[0180] Parameter regulator 142: Solves the real-time pre-tightening force solver result output by the phased solver engine 130; abnormal event signals triggered by the abnormal handling unit 150 (such as a deviation exceeding 5%). Adjusts the sensor sampling frequency, trigger threshold, etc. of the composite acquisition module 101. Sends a weight coefficient adjustment instruction to the dynamic weighting processor 120 (such as reducing the weight of acoustic characteristics with low credibility).

[0181] As shown Figure 10 in the figure, the exception handling unit 150 is responsible for restoring system stability through multi-level fault tolerance measures when the data calculation deviation exceeds 5%. It monitors the pre-tightening force calculation results output by the phased calculation engine 130 in real time and compares them with the reference values in the historical database 141. Operations such as self-cleaning, data re-acquisition, and sensor switching are gradually triggered according to the severity of the deviation. Hardware-level fault tolerance is achieved through the backup sensor 151 and the channel switch 152.

[0182] When the deviation between the pre-tightening force calculation result and the historical reference value exceeds 5%, an exception signal is transmitted. The phased calculation engine 130 sends the deviation signal to the exception handling unit 150. The self-optimizing control module 140 synchronously records this exception event in the historical database 141 and marks it as "pending processing".

[0183] The exception handling unit 150 sends a "sensor self-cleaning instruction" to the composite acquisition module 101. After the cleaning is completed, the composite acquisition module 101 returns a "cleaning completed signal" to eliminate the attenuation of the acoustic signal caused by sensor contamination (such as a 30% reduction in the resonant characteristic frequency offset).

[0184] The composite acquisition module 101 re-acquires three groups of acoustic feature data, with each group separated by 1 second to avoid the influence of instantaneous interference. If the mean deviation still exceeds 5%, it is determined as a "hardware-level fault", triggering the sensor switching process and entering the next processing stage. The backup sensor 151 is enabled to re-acquire three groups of acoustic feature data. The channel switch 152 intervenes to switch the data transmission path from the main channel (such as CAN bus) to the backup channel (such as RS-485) to avoid communication interference.

[0185] Example of a typical operating scenario:

[0186] Scenario: Handling of sudden interference in bolt pre-tightening force monitoring

[0187] Exception trigger: The phased calculation engine 130 outputs a pre-tightening force deviation of 8% (F 实测 = 1080N), and the exception handling unit 150 is activated.

[0188] Self-cleaning execution: The resonant cavity generator 111 executes the self-cleaning program to remove the attached oil stain.

[0189] Data review: Three groups of data are re-acquired and averaged, and the deviation is reduced to 3%, and the system returns to normal.

[0190] Parameter optimization: The self-optimizing control module 140 records this event and optimizes the anti-interference algorithm of the dual-wave transceiver 112 to reduce the probability of future similar exceptions.

[0191] This application constructs a multi-dimensional acoustic feature collaborative perception system, breaks through the limitations of traditional single detection parameters, and significantly improves the accuracy and reliability of bolt pre-tightening force monitoring. Through the dynamic interaction calibration and weight allocation of resonance characteristics, sound velocity ratio, and non-linear characteristics, the system can real-time sense interference factors such as environmental temperature fluctuations, interface contact deterioration, and material non-linear effects, and automatically correct the calculation deviation based on the credibility evaluation results. For example, when the resonance cavity frequency offset is abnormal, the system actively reduces the weight ratio of the sound velocity ratio data to avoid misjudgment caused by sudden changes in the impedance of the acoustic wave propagation path; at the same time, the real-time monitoring of the non-linear harmonic amplitude can accurately capture the plastic deformation of the bolt micro-contact interface, making up for the blind area of the traditional sound velocity method in characterizing the local stress distribution, so as to achieve full-scale coverage from macroscopic stress to microscopic damage under complex working conditions.

[0192] The innovative phased fusion model solves the problem of dynamic decoupling of multi-physical field coupling data, and for the first time deeply fuses the macroscopic pre-tightening force calculation of the sound velocity ratio with the microscopic correction of non-linear characteristics. By establishing a two-stage joint calculation mechanism, the system first calculates the macroscopic pre-tightening force based on the acoustoelastic effect, then combines the second harmonic amplitude to perform non-linear correction on the local stress concentration area, and introduces the comparison and verification of resonance characteristics and material theoretical impedance to effectively eliminate systematic errors caused by changes in sensor coupling state or material creep. Especially through the intelligent determination rule of the correction value sign (based on the dynamic ratio of frequency offset and sound velocity ratio change), the system can adaptively distinguish interface contact anomalies from real pre-tightening force losses, greatly reducing the false alarm rate, and the deviation between the calculation result and the real pre-tightening force can be controlled, with higher accuracy than traditional methods.

[0193] The closed-loop self-optimization mechanism constructed by the system realizes the dynamic evolution of the monitoring strategy and solves the stability problem of long-term monitoring in industrial scenarios. Through the correlation analysis of historical data and real-time calculation results in the long-term monitoring database, the system can automatically update key parameters such as the resampling frequency of the sound velocity ratio, non-linear excitation power, and resonance cavity sweep range, and predict the attenuation trend of the weight adjustment coefficient. For example, the coupling degree data of the sound velocity ratio and non-linear characteristics statistically calculated monthly can optimize the non-linear compensation coefficient of the weight function, while the quarterly sensor torque calibration ensures the stability of the acoustic coupling state at the hardware level. This software and hardware collaborative adaptive ability enables the system to continuously maintain high accuracy without manual intervention, especially suitable for the full life cycle management of bolt health status in harsh environments such as nuclear power and wind power.

[0194] The technical scope of the present invention is not limited to the content described above. Those skilled in the art can make various deformations and modifications to the above embodiments without departing from the technical idea of the present invention, and these deformations and modifications should all fall within the protection scope of the present invention.

Claims

1. An intelligent monitoring method for bolt pre-tightening force based on acoustic characteristics, characterized in that: It includes the following steps: S1. Synchronously acquire three types of acoustic feature data, namely resonance characteristics, sound velocity ratio, and nonlinear characteristics, and perform interactive calibration to generate a credibility evaluation result; S2. According to the credibility evaluation result, dynamically adjust the weight coefficients of the three types of acoustic feature data in the pre-tightening force calculation weight distribution strategy: When the resonance characteristics show abnormal contact, reduce the weight of the sound velocity ratio; When the nonlinear characteristics exceed the preset threshold, increase the sampling frequency of the sound velocity ratio; Establish a feedback relationship between the weight adjustment coefficient and the historical monitoring error and feedback it to the weight distribution strategy; S3. Adopt a phased fusion strategy, and input the weighted three types of acoustic feature data into a phased fusion model for joint calculation of the pre-tightening force: S301. Perform primary calculation to calculate the macroscopic pre-tightening force based on the sound velocity ratio; S302. Perform microscopic correction to correct the local stress distribution using the nonlinear characteristics; S303. Perform credibility verification to generate a real-time calculation result by comparing the resonance characteristics with the material theoretical impedance value; S4. Establish a long-term monitoring database, and update the parameters for acquiring the acoustic feature data and the weight distribution strategy according to the deviation analysis of the historical monitoring data and the real-time calculation result to form a closed-loop calibration mechanism.

2. The intelligent monitoring method for bolt pre-tightening force based on acoustic features according to claim 1, characterized in that: The acquisition method for synchronous acquisition in S1 is as follows: Emit a swept-frequency signal to the resonant cavity, and record the frequency offset of the resonant cavity to generate resonance characteristics; Alternately excite longitudinal waves and transverse waves, and use the shape factor method to measure the propagation time difference between the longitudinal wave and the transverse wave to obtain the sound velocity ratio; Apply a fundamental wave excitation and capture the second harmonic amplitude to obtain the nonlinear characteristics of the fundamental wave and the second harmonic.

3. The intelligent monitoring method for bolt pre-tightening force based on acoustic features according to claim 2, wherein: The dynamic adjustment in S2 includes: When the frequency offset of the resonant cavity exceeds the material theoretical impedance value, mark that the credibility of the sound velocity ratio drops by one level; When the harmonic amplitude ratio of the nonlinear characteristics exceeds the preset threshold, trigger data recalibration acquisition of the sound velocity ratio.

4. The intelligent monitoring method for bolt pre-tightening force based on acoustic characteristics according to claim 1, characterized in that: The phased fusion strategy establishes a phased fusion model as follows: The first stage calculates the macroscopic pre-tightening force where K1 is the acoustoelastic coefficient, v p and v s are the longitudinal wave and transverse wave sound velocities respectively; Second-stage correction value Where F0 is the initial pre-tightening force converted based on the bolt installation torque, F0 = τ·k·d, τ is the installation torque value, k is the thread coefficient, and d is the nominal diameter of the bolt; Determine the final pre-tightening force F = F1 ± ΔF.

5. The intelligent monitoring method for bolt pre-tightening force based on acoustic features according to claim 4, characterized in that: The sign of the correction value ΔF is determined by the correction rule: When the frequency offset Δf of the resonance characteristics and the change amount ΔR of the sound velocity ratio satisfy: When taking the negative sign; When it is, take the positive sign, where η is the material interface impedance coefficient.

6. The intelligent monitoring method for bolt pre-tightening force based on acoustic characteristics according to claim 1, characterized in that: The long-term monitoring database in step S4 is automatically executed: S401. Statistically calculate the frequency drift rate of the resonance characteristics every week and update the reference frequency; S402. Analyze the coupling degree of the sound velocity ratio and the nonlinear characteristics every month and optimize the weight function; S403. Calibrate the sensor installation torque every quarter to ensure stable acoustic coupling; Establish historical pre-tightening force change trend data and predict the attenuation curve of the weight adjustment parameters.

7. The intelligent monitoring method for bolt pre-tightening force based on acoustic features according to claim 6, characterized in that: The closed-loop calibration mechanism in step S4 includes: Statistically calculate the standard deviation σ of the historical monitoring data every month, preset a standard deviation threshold σ0, if σ > σ0, then execute: Expand the swept-frequency range of the resonant cavity to ±20%; Shorten the excitation interval between the longitudinal wave and the transverse wave by 20%; Increase the output power of the excitation probe for the nonlinear characteristics by 10%.

8. The intelligent monitoring method for bolt pre-tightening force based on acoustic characteristics according to claim 1, characterized in that: It also includes: S5. When the deviation of the calculation results of the three types of acoustic feature data exceeds 5%, trigger anomaly handling and execute in sequence: S501. Trigger the self-cleaning program of the composite acquisition sensor; S502. Re - collect three groups of the acoustic feature data and take the average value; S503. If it is still out of tolerance, switch to the backup sensor channel.

9. The intelligent monitoring method for bolt pre-tightening force based on acoustic characteristics according to claim 1, characterized in that: The method for dynamically adjusting the weight coefficient is as follows: Define the initial weight W0 as an empirical value, W0 = [0.5, 0.3, 0.2] (corresponding to the resonance feature, the sound - velocity ratio, and the non - linear feature); Real-time weight W t = W0·[1 - α·Δf, 1 + β·ΔR, 1 + γ·ΔU], where α, β, and γ are attenuation factors calibrated based on material properties and ambient temperature, and Δf, ΔR, and ΔU are real-time change amounts.

10. An intelligent monitoring system for bolt pre-tightening force based on acoustic features, characterized in that: Use the intelligent monitoring method for bolt pre - tightening force based on acoustic features as described in any one of claims 1 - 9, including: A composite acquisition module (101) for synchronously acquiring three types of acoustic feature data, namely resonance feature, sound - velocity ratio, and non - linear feature, and performing interactive calibration to generate a credibility evaluation result; the composite acquisition module (101) includes a composite acquisition sensor (110) and a calibration module (102), and the composite acquisition sensor integrates a resonance - cavity generator (111), a dual - wave transceiver (112), and a non - linear probe (113); A dynamic weighting processor (120) configured to dynamically adjust the weight coefficients of the three types of acoustic feature data in the weight - distribution strategy for pre - tightening force calculation based on the credibility evaluation result; A phased calculation engine (130) for using a phased fusion strategy to input the weighted three types of acoustic feature data into a phased fusion model for joint pre - tightening force calculation; the phased calculation engine (130) includes a macroscopic calculation unit (131) and a microscopic correction unit (132); A self - optimizing control module (140) for establishing a long - term monitoring database, and updating the acquisition parameters and the weight - distribution strategy according to the deviation analysis of historical monitoring data and real - time calculation results to form a closed - loop calibration mechanism; the self - optimizing control module (140) includes a historical database (141) and a parameter regulator (142); An exception - handling unit (150) for triggering exception handling when the deviation of the calculation results of the three types of acoustic feature data exceeds 5%; the exception - handling unit (150) includes a backup sensor (151) and a channel switcher (152).

Images