A method for inverting characteristic parameters of acoustic elastic force measurement of top cover bolts

By combining acoustoelastic theory with laboratory and field test data and genetic algorithms, the acoustoelastic parameters of the top cover bolts were inverted, solving the problem of the quantitative relationship between the acoustic time difference and preload of the top cover bolts, and realizing high-precision online monitoring and improved reliability.

CN123116776APending Publication Date: 2026-11-17CHINA YANGTZE POWER +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611111806.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-24
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies make it difficult to obtain a quantitative relationship between the acoustic time difference of the top cover bolts and the preload through theoretical manuals, resulting in inaccurate preload assessment and affecting the reliability of online monitoring of hydropower units.

Method used

A physical model of bolt stress and ultrasonic transit time was established using acoustoelastic theory. Combined with laboratory and field test data, key acoustoelastic parameters were inverted using genetic algorithms and various surrogate models to establish a reliable "acoustic transit time-preload" mapping model.

Benefits of technology

It achieves high-precision online monitoring of top cover bolt preload, reduces the difficulty and cost of parameter calibration, and improves the reliability and robustness of monitoring, making it suitable for rapid promotion and application in similar units.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116776A_ABST
    Figure CN123116776A_ABST
Patent Text Reader

Abstract

A method for inverting characteristic parameters of acoustoelastic force measurement on top cover bolts includes the following steps: establishing a physical model of bolt stress and ultrasonic transit time based on acoustoelastic theory, deriving theoretical formulas, and determining key parameters to be inverted; conducting two-level calibration tests in the laboratory and on-site to obtain high-precision stress and ultrasonic transit time benchmark data, providing a real verification basis for parameter inversion; generating parameter samples, training and comparing various surrogate models, and selecting the optimal fast calculation model that fits the "stress-ultrasonic transit time" mapping; transforming parameter identification into an error minimization optimization problem, using a genetic algorithm for global optimization, outputting the optimal parameters, and achieving reuse within the same model. Using the inversion method of this application, a universal and high-precision intelligent parameter inversion paradigm based on measured data and physical laws can be formed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the technical field of hydropower unit condition monitoring and fault diagnosis, and specifically relates to a method for inverting characteristic parameters of acoustic elastic force measurement of top cover bolts. Background Technology

[0002] Reliable pre-tightening of the bolts on the top cover of a pumped storage power station is crucial for ensuring the safe operation of the unit. Acoustic elastic force measurement technology assesses bolt pre-tightening force by measuring the change in ultrasonic time-of-flight caused by stress, making it an ideal means of achieving online condition monitoring of this critical component.

[0003] However, the quantitative relationship between acoustic time difference and preload is highly dependent on the material's elastic modulus, Poisson's ratio, third-order elastic constant, and structural equivalent parameters. These parameters are significantly affected by material processing, installation conditions, and environmental factors, and are difficult to obtain directly from theoretical manuals. Inaccurate parameter values ​​will directly lead to systematic deviations in preload assessment, seriously affecting monitoring reliability.

[0004] Therefore, the core prerequisite for breaking through the bottleneck of engineering application of this technology and realizing high-precision online monitoring is to accurately invert these key acoustic projectile parameters based on field and test data and establish a reliable "acoustic time difference-preload" mapping model. Summary of the Invention

[0005] This invention provides a method for inverting the characteristic parameters of acoustic elastic force measurement of top cover bolts to establish a reliable "acoustic time difference-preload" mapping model.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for inverting characteristic parameters of acoustic elastic force measurement of top cover bolts includes the following steps: Step 1: Principle Modeling: Based on the acoustoelastic theory, establish a physical model of bolt stress and ultrasonic transit time, complete the theoretical formula derivation, and determine the key parameters to be inverted; Step 2, Test Calibration: Conduct two-level calibration tests, one in the laboratory and one in the field, to obtain high-precision stress and acoustic time difference reference data, providing a real verification basis for parameter inversion; Step 3: Proxy Model Construction: Generate parameter samples, train and compare multiple proxy models, and select the best one to obtain a fast calculation model that fits the "stress-acoustic time difference" mapping; Step 4, Intelligent Parameter Inversion: The parameter identification is transformed into an error minimization optimization problem. A genetic algorithm is used to find the global optimal solution, output the optimal parameters, and enable reuse within the same model.

[0007] Furthermore, in step one, the physical model is a nonlinear relationship model between the bolt axial stress and the ultrasonic longitudinal wave transit time, and the parameter to be inverted is the elastic modulus. Poisson's ratio of materials elastic constant and equivalent length coefficient .

[0008] Furthermore, the physical model established for bolt stress and ultrasonic transit time is as follows: ; In the formula, For the longitudinal wave acoustic time difference of ultrasound, It is the equivalent length coefficient. It is the length of the bolt before it is tightened. It is the density of the material. It is the elastic modulus. It is a second-order elastic constant. Let Lamé constant be . and It is the third elastic constant of the material. It is stress. It is the Poisson's ratio of the material.

[0009] Furthermore, in step two, the first-level test serves as a benchmark test, where standard specimens are precisely loaded in a controlled laboratory environment to obtain high signal-to-noise ratio stress-acoustic time difference benchmark data pairs. The second-level test is the in-situ test, which involves applying calibration loads to the in-service top cover bolts of the hydropower unit under real working conditions and collecting stress and acoustic time difference data in the actual operating environment.

[0010] Furthermore, in step three, the Latin hypercube sampling method is used to generate N sets of sampling points in the parameter space to verify the impact of different datasets on the subsequent inversion results.

[0011] Furthermore, in step three, multiple surrogate models are used, including parallel-constructed response surface models, Kriging models, and BP neural network models. The model with the highest prediction accuracy and the strongest generalization ability is selected as the final "stress-acoustic time difference" predictor.

[0012] Furthermore, in step three, the dataset generated by Latin hypercube sampling is divided into a training set and a test set. After training the parameters of the three models using the training set, the coefficient of determination and root mean square error are used to objectively evaluate and select the final "stress-sound time difference" predictor on the test set.

[0013] Furthermore, step four specifically includes the following steps: S41. Define the inversion optimization problem: For a given set of field measured data on the time difference of sound, find a set of parameters that minimize the error between the time difference of sound predicted by the surrogate model and the measured value. S42. Global Optimization Using Genetic Algorithm: The above optimization problem is solved using a genetic algorithm with real-number encoding. S43. Output the individual with the highest fitness across all generations; this is the optimal parameter estimate obtained through inversion [E]. , v , l , m , β ].

[0014] Furthermore, step S42 specifically includes: S421, Encoding: Real number encoding is used, and the five parameters to be inverted are directly used as chromosome genes; S422, Fitness Assessment: Input each individual in the population into the surrogate model selected in step three, and quickly calculate its corresponding predicted time difference; S423. Genetic operations, including selection, crossover, and mutation; S424. Iteration and Termination: Repeated selection, crossover, and mutation to generate a new generation of population until the maximum number of iterations is reached or fitness converges.

[0015] Furthermore, in step S423, a tournament selection strategy is adopted, in which several individuals are randomly selected from the current population for comparison, and the individual with the highest fitness is retained to enter the mating pool. Crossover uses simulated binary crossover or arithmetic crossover operators to perform linear or nonlinear combinations of the genes of parent individuals to generate offspring individuals containing new characteristics; The mutation introduces a multinomial mutation operator to randomly perturb certain gene loci of an individual with a certain low probability.

[0016] The present invention can achieve the following beneficial effects: 1. This application proposes a method that combines field testing in hydropower projects with data-driven machine learning and intelligent optimization algorithms. It not only solves the core pain point of parameter calibration in the sonic elastic force measurement method, but also forms a universal and high-precision intelligent parameter inversion paradigm based on measured data and physical laws, providing strong technical support for online health monitoring of key fasteners in hydropower and other fields.

[0017] 2. This study directly uses field and factory test data. The proxy model learns the real physical relationships and naturally includes all the complex factors that are difficult to model, such as micro-contact and material inhomogeneity. The inversion results are closer to engineering reality and have stronger robustness.

[0018] 3. By systematically comparing response surface models, Kriging models, and BP neural networks, and using data as a benchmark, the most suitable model for the nonlinear mapping of "stress-acoustic time difference" is selected, avoiding the limitations that may exist in a single model and ensuring the scientific nature and reliability of the core links of the inversion process.

[0019] 4. Genetic algorithms maintain a population and explore multiple regions of the parameter space simultaneously, resulting in higher parallel search efficiency. Furthermore, through crossover operations, they effectively combine excellent genes from different solutions, accelerating convergence towards the global optimum.

[0020] 5. Intelligent inversion from observation data to parameters yields invariant material properties of bolt components and invariant characteristic parameters of structural connection dimensions, enabling rapid application on similar units and reducing calibration difficulty and cost. Attached Figure Description

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of a method for inverting characteristic parameters of acoustic elastic force measurement of top cover bolts according to the present invention; Figure 2 This is a schematic diagram of the bolt connection structure of the present invention. Detailed Implementation

[0022] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings, which illustrate embodiments of the present application. However, the present application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that the disclosure of this application will be thorough and complete.

[0023] A method for inverting characteristic parameters of acoustic elastic force measurement of top cover bolts, such as Figure 1 As shown, it includes the following steps: Step 1: Principle Modeling: Based on the acoustoelastic theory, establish a physical model of bolt stress and ultrasonic transit time, complete the theoretical formula derivation, and determine the key parameters to be inverted.

[0024] First, based on the theory of acoustoelasticity, a parameterized physical model is established between the axial preload of the bolt and the longitudinal wave time difference of the ultrasonic wave. Through parameter sensitivity analysis, it is determined which parameters have an important impact on the results.

[0025] Specifically, when a bolt is pre-tightened, its length elongates, and the propagation speed of ultrasonic waves in the bolt material changes due to the alteration of internal stress. Therefore, the propagation time of ultrasonic waves in the bolt differs significantly before and after pre-tightening. Based on this mechanism, ultrasonic longitudinal or transverse wave technology can be directly used to detect the pre-tightening force of large bolts.

[0026] Under uniaxial stress conditions, the relationship between the longitudinal wave velocity propagating along the stress direction and the stress can be described by the following equation: (1) in, It is the density of the material. It is the longitudinal wave velocity under stress. It is a second-order elastic constant. Let Lamé constant be . and It is the third elastic constant of the material. It is stress.

[0027] For a stress-free bolt, the longitudinal wave velocity under stress-free conditions is: (2) Substituting equation (2) into equation (1), we get: (3) in, The acoustic coefficient can be written as: (4) Take the square root of formula (3) as... Performing a Taylor expansion on the independent variable, ignoring higher-order terms, yields: (5) The above equation shows that the longitudinal wave velocity is proportional to the stress, and the acoustic elastic coefficient is... The value is usually negative, meaning that the speed of sound decreases as stress increases. Because it is difficult to measure the speed of sound directly, the time of sound is often used to calculate the preload.

[0028] Under normal circumstances, the stress on a bolt is not uniformly distributed along the axial direction. For example... Figure 2 As shown, the portion of the bolt end extending beyond the nut is often in a stress-free state, or experiences a significantly lower level of axial stress than the shaft itself. This indicates the length of the unloaded portion at the end of the bolt. For shorter bolts, this area represents a large proportion of the overall length, and its impact cannot be ignored.

[0029] Furthermore, in the vicinity of the bolt head, there are also low-stress or even stress-free zones. The length of the unloaded section of the bolt head also needs to be considered in the analysis. It is the length of the bolt before it is tightened. This is the length of the bolt's stressed section, also known as the effective length. (Equivalent length coefficient) It is the effective length and original length The ratio of .

[0030] like Figure 2 As shown, effective length It can be expressed as the length between the bolt head and the nut plus half the thickness of the bolt head plus half the thickness of the nut.

[0031] Through derivation and calculation, the elastic modulus is adopted. Poisson's ratio of materials elastic constant and equivalent length coefficient To represent the acoustic transit time, the relationship between stress and acoustic transit time can be obtained as follows: ; Based on the theory of acoustoelasticity, the axial stress of the bolt was established. With the longitudinal wave time difference of ultrasound The model describes a nonlinear relationship between the input and output parameters. The core of this model lies in the fact that its input-output relationship is determined by five key material and structural parameters, including the elastic modulus. Poisson's ratio of materials elastic constant .

[0032] By using parameter sensitivity analysis, it is determined that these five parameters jointly determine the input-output relationship of the model, thereby defining the core objective of parameter inversion: to solve for these parameters that cannot be directly measured but directly determine the measurement accuracy through observable acoustic time difference. At the same time, through sensitivity analysis of the mechanistic model parameters, the main factors to be inverted are identified, and a quantitative analysis of the uncertainty of the system results is carried out accordingly.

[0033] Step 2, Test Calibration: Conduct two-level calibration tests, one in the laboratory and one in the field, to obtain high-precision stress and acoustic time difference reference data, providing a real verification basis for parameter inversion.

[0034] To address the issue of unknown model parameters, a two-stage calibration experiment was designed and executed. The first stage was a baseline experiment, in which standard specimens were precisely loaded in a controlled laboratory environment to obtain high signal-to-noise ratio stress-transit time of day baseline data pairs. The coefficient values ​​in the stress-transit time of day correspondence were calibrated using the experimental data, providing a basis for subsequent model validation.

[0035] The second stage involves in-situ testing at the hydroelectric generator site. Calibration loading is applied to the in-service top cover bolts under real-world operating conditions, and stress and acoustic time-of-flight data are collected in the actual operating environment. Data collected on a real generator unit, encompassing environmental disturbances and structural complexity, is used to ultimately verify the engineering robustness of the model.

[0036] The data obtained from the two-stage experiments together constitute the factual basis and verification standard of the entire technical solution, providing a true and reliable benchmark for subsequent parameter inversion.

[0037] Step 3: Proxy Model Construction: Parameter samples are generated using Latin hypercube sampling, and various proxy models are trained and compared to select the best one for a fast calculation model that adapts to stress and acoustic time difference mapping.

[0038] After analyzing the influence weights of each parameter in the relationship between acoustic time difference and stress, this study performs accurate inversion of the five input parameters based on the acoustic time difference-stress calibration results measured in the field.

[0039] First, the Latin hypercube sampling (LHS) method is used to generate N sets of sampling points in the parameter space. N is set to 2000, 5000, and 10000 to verify the impact of different datasets on the subsequent inversion results. The input is the acoustic time difference. The output is the bolt stress. The data is randomly divided into a training set for model learning and a test set for final performance evaluation.

[0040] To overcome the potential fitting bias or overfitting risks of a single model and ensure the highest accuracy and robustness of the stress-acoustic time difference mapping relationship, this embodiment constructs three representative surrogate models in parallel and selects the best one through comparison using a test set. The specific implementation process is as follows: The first proxy model: response surface model A response surface model is constructed using polynomial regression. Given the nonlinear relationship between the stress-acoustic time difference and the various parameters, a second-order polynomial is chosen as the basis function in this embodiment. Let the input parameter vector be: ; Response variables in response surface model for: ; in, , These are undetermined regression coefficients. The random error is represented by the coefficient matrix. The sample data generated by Latin hypercube sampling is fitted using the least squares method to solve for the coefficient matrix.

[0041] The model has a simple structure, fast computation speed, and good physical interpretability, making it suitable for initially capturing global trends among parameters.

[0042] The second proxy model: the Kriging model. A Kriging model is constructed using Gaussian process regression. This model assumes that the output response is a weighted sum of the global multinomial model and local biases, expressed as: ; in the formula It is a global trend item. The Gaussian random process with a mean of 0 and a variance of σ² is used, and its covariance function is the Gaussian correlation function: ; Solving hyperparameters using the maximum likelihood estimation method .

[0043] Kriging models not only provide predicted values, but also the variance of the predicted results, i.e., uncertainty estimates, which have unique advantages in dealing with strong nonlinear local fluctuations.

[0044] The third proxy model: BP neural network: A multi-layer feedforward neural network was constructed to fit a highly complex nonlinear mapping. The network structure consists of an input layer with 5 nodes corresponding to 5 inversion parameters, a hidden layer, and an output layer with 1 node corresponding to the acoustic time difference coefficients.

[0045] The hidden layer activation function is a hyperbolic tangent sigmoid function, and the output layer uses a linear function. The Levenberg-Marquardt backpropagation algorithm is used for training to minimize the mean squared error between the network output and the true analytical solution. An L2 regularization term is introduced to prevent overfitting and ensure the model's generalization ability on unseen data.

[0046] The dataset generated by Latin hypercube sampling was divided into training and test sets in an 8:2 ratio. After training the parameters of the three models using the training set, the model with the highest prediction accuracy and strongest generalization ability was objectively evaluated on the test set using metrics such as the coefficient of determination (R²) and root mean square error (RMSE) as the final stress-acoustic time difference predictor. Finally, the trained surrogate model was selected to replace the tedious analytical formula calculations for efficient inversion in the subsequent genetic algorithm.

[0047] This step employs a data-driven machine learning approach, constructing a parameter set of five parameters based on the Latin hypercube sampling method. Using the theoretical analytical expression from step one, multiple surrogate models are trained to establish a high-speed, high-precision approximate mapping relationship from parameter combinations to predicted stress-acoustic transit time. By using a neural network surrogate model, the extensive computation of complex analytical expressions is avoided, efficiently obtaining the stress-acoustic transit time correspondence under different parameter combinations.

[0048] Step 4, Intelligent Parameter Inversion: The parameter identification is transformed into an error minimization optimization problem. A genetic algorithm is used to find the global optimal solution, output the optimal parameters, and enable reuse within the same model.

[0049] This stage utilizes the surrogate model constructed in step three to generate a large dataset of stress-acoustic time difference correspondences. Based on the correspondence coefficients obtained from experimental calibration, a genetic algorithm is used to globally optimize five parameters in a multi-dimensional parameter space to achieve rapid and accurate inversion of key material parameters. Specifically, the following steps are included: S41. Define the inversion optimization problem. The parameter identification problem is transformed into an optimization problem, with five material parameters as optimization variables. , , For a given set of on-site measured data on the time difference of sound, we seek a set of parameters that minimize the error between the time difference of sound predicted by the surrogate model and the measured value.

[0050] S42, Global Optimization Using Genetic Algorithm The above optimization problem is solved using a genetic algorithm with real-number encoding, and the steps are as follows: S421, Encoding: Real number encoding is used, directly treating the five parameters to be inverted as chromosome genes. Within the physical boundaries defined by the parameters, Latin hypercube sampling is used to generate M initial individuals to form the initial population P0. This initialization method ensures a uniform distribution of the initial solution in the search space, avoiding the algorithm from starting the search from a single initial value.

[0051] S422, Fitness Assessment: Each individual in the population (i.e., a set of parameter combinations) is input into the surrogate model selected in the previous stage, and its corresponding predicted time difference is quickly calculated. The fitness value of each individual is calculated according to the objective function. Here, the surrogate model is used to replace complex physical formulas or finite element calculations, enabling thousands of iterative evaluations to be completed in milliseconds.

[0052] S423. Genetic operations, specifically including selection, crossover, and mutation. Selection employs a tournament selection strategy, randomly selecting several individuals from the current population for comparison, retaining the individual with the highest fitness (i.e., the smallest prediction error) into the mating pool to ensure the inheritance of superior genes.

[0053] Crossover employs simulated binary crossover or arithmetic crossover operators to linearly or nonlinearly combine the genes of parent individuals to generate offspring individuals with new characteristics, thereby exploring new regions in the parameter space.

[0054] Mutation introduces a multinomial mutation operator to randomly perturb certain gene loci of an individual with a low probability. This step is crucial for maintaining population diversity and escaping local optima.

[0055] S424. Iteration and Termination: Repeated selection, crossover, and mutation to generate a new generation of population until the maximum number of iterations is reached or fitness converges.

[0056] The individual with the highest fitness across all generations is the optimal parameter estimate obtained through inversion [E]. , v , l , m , β ].

[0057] This step transforms the parameter identification problem into an optimization problem aimed at minimizing the error between the inversion results and the experimental calibration results. A genetic algorithm, a global optimizer, is employed to perform a parallel and adaptive search within the possible parameter space. In each iteration, the algorithm rapidly evaluates the merits of candidate parameter sets by invoking the pre-trained surrogate model, thereby efficiently approximating the global optimum.

[0058] The process is fully automated and ultimately outputs a set of high-confidence parameter estimates that best match the inversion model with the actual experimental calibration. It completes the intelligent inversion from observation data to parameters, and obtains the material invariants of bolt components and the characteristic parameter invariants of structural connection dimensions. It can be quickly promoted and applied on similar units, reducing calibration difficulty and cost.

[0059] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for inverting characteristic parameters of acoustic elastic force measurement of top cover bolts, characterized in that, Includes the following steps: Step 1: Principle Modeling: Based on the acoustoelastic theory, establish a physical model of bolt stress and ultrasonic transit time, complete the theoretical formula derivation, and determine the key parameters to be inverted; Step 2, Test Calibration: Conduct two-level calibration tests, one in the laboratory and one in the field, to obtain high-precision stress and acoustic time difference reference data, providing a real verification basis for parameter inversion; Step 3: Proxy Model Construction: Generate parameter samples, train and compare multiple proxy models, and select the best one to obtain a fast calculation model that fits the "stress-acoustic time difference" mapping; Step 4, Intelligent Parameter Inversion: The parameter identification is transformed into an error minimization optimization problem. A genetic algorithm is used to find the global optimal solution, output the optimal parameters, and enable reuse within the same model.

2. The method for inverting the characteristic parameters of acoustic elastic force measurement of top cover bolts according to claim 1, characterized in that: In step one, the physical model is a nonlinear relationship model between the bolt axial stress and the ultrasonic longitudinal wave transit time, and the parameter to be inverted is the elastic modulus. Poisson's ratio of materials elastic constant and equivalent length coefficient .

3. The method for inverting the characteristic parameters of acoustic elastic force measurement of top cover bolts according to claim 2, characterized in that: The established physical model of bolt stress and ultrasonic transit time is as follows: ; In the formula, For the longitudinal wave acoustic time difference of ultrasound, It is the equivalent length coefficient. It is the length of the bolt before it is tightened. It is the density of the material. It is the elastic modulus. It is a second-order elastic constant. Let Lamé constant be . and It is the third elastic constant of the material. It is stress. It is the Poisson's ratio of the material.

4. The method for inverting the characteristic parameters of acoustic elastic force measurement of top cover bolts according to claim 1, characterized in that: In step two, the first-level test serves as a benchmark test. In a controlled laboratory environment, standard specimens are precisely loaded to obtain a high signal-to-noise ratio "stress-acoustic time difference" benchmark data pair. The second-level test is the in-situ test, which involves applying calibration loads to the in-service top cover bolts of the hydropower unit under real working conditions and collecting stress and acoustic time difference data in the actual operating environment.

5. The method for inverting the characteristic parameters of acoustic elastic force measurement of top cover bolts according to claim 1, characterized in that: In step three, the Latin hypercube sampling method is used to generate N sets of sampling points in the parameter space to verify the impact of different datasets on the subsequent inversion results.

6. The method for inverting the characteristic parameters of acoustic elastic force measurement of a top cover bolt according to claim 1 or 5, characterized in that: In step three, multiple surrogate models are used, including response surface model, kriging model and BP neural network model constructed in parallel, and the model with the highest prediction accuracy and the strongest generalization ability is selected as the final "stress-acoustic time difference" predictor.

7. The method for inverting the characteristic parameters of acoustic elastic force measurement of a top cover bolt according to claim 5, characterized in that: In step three, the dataset generated by Latin hypercube sampling is divided into a training set and a test set. After training the parameters of the three models using the training set, the coefficient of determination and root mean square error are used to objectively evaluate and select the final "stress-acoustic time difference" predictor on the test set.

8. The method for inverting the characteristic parameters of acoustic elastic force measurement of top cover bolts according to claim 1, characterized in that: Step four specifically includes the following steps: S41. Define the inversion optimization problem: For a given set of field measured data on the time difference of sound, find a set of parameters that minimize the error between the time difference of sound predicted by the surrogate model and the measured value. S42. Global Optimization Using Genetic Algorithm: The above optimization problem is solved using a genetic algorithm with real-number encoding. S43. Output the individual with the highest fitness across all generations; this is the optimal parameter estimate obtained through inversion [E]. , v , l , m , β ].

9. The method for inverting the characteristic parameters of acoustic elastic force measurement of top cover bolts according to claim 8, characterized in that: Step S42 specifically includes: S421, Encoding: Real number encoding is used, and the five parameters to be inverted are directly used as chromosome genes; S422, Fitness Assessment: Input each individual in the population into the surrogate model selected in step three, and quickly calculate its corresponding predicted time difference; S423. Genetic operations, including selection, crossover, and mutation; S424. Iteration and Termination: Repeated selection, crossover, and mutation to generate a new generation of population until the maximum number of iterations is reached or fitness converges.

10. The method for inverting the characteristic parameters of acoustic elastic force measurement of a top cover bolt according to claim 9, characterized in that: In step S423, a tournament selection strategy is adopted, in which several individuals are randomly selected from the current population for comparison, and the individual with the highest fitness is retained to enter the mating pool. Crossover uses simulated binary crossover or arithmetic crossover operators to perform linear or nonlinear combinations of the genes of parent individuals to generate offspring individuals containing new characteristics; The mutation introduces a multinomial mutation operator to randomly perturb certain gene loci of an individual with a certain low probability.