A multi-parameter fusion asphalt mixture acoustic emission sensor layout optimization method

Abstract

The application provides a multi-parameter fusion asphalt mixture acoustic emission sensor layout optimization method, comprising: arranging a plurality of acoustic emission sensors at equal intervals along a straight line parallel to the longitudinal direction of the surface of the asphalt mixture to be measured, performing an acoustic emission pulse test, and measuring amplitude, ringing count and wave speed; determining an effective arrangement range in which acoustic emission signals can be effectively received according to the distance at which each index attenuates by 50%, and taking the minimum value as the initial interval of the acoustic emission sensor arrangement; according to the CRLB principle, an optimization function of the acoustic emission sensor network layout is constructed; and the optimization function in step S2 is solved by using an improved genetic algorithm. The application determines the physical space constraint of the effective arrangement of the sensor through standard attenuation testing and multi-parameter fusion, constructs a regional overall positioning accuracy optimization function that fuses the exponential wave speed equation and the spatial uncertainty of the sound source, and improves the physical accuracy and reliability of the layout optimization model from a theoretical source.

Description

Technical Field

[0001] This application relates to the field of non-destructive testing and acoustic emission technology for road materials, and in particular to a method for optimizing the layout of acoustic emission sensors for asphalt mixtures using multi-parameter fusion. Background Technology

[0002] Acoustic emission (AE) technology, as a non-destructive testing method, can capture elastic wave signals released during the initiation and propagation of microcracks within materials in real time. It is widely used for monitoring and locating fatigue damage evolution in asphalt mixtures. Sensor network layout is one of the core tasks of a pavement acoustic emission monitoring system for fatigue damage location, directly determining the accuracy of time of arrival (TOD) calculations and the precision of damage source localization, thus ensuring the validity of monitoring data.

[0003] Determining the optimal sensor network layout in road monitoring systems has always been a key challenge for researchers in the field of nondestructive testing of road materials. An unreasonable sensor network layout will cause the control equations for source localization to become approximately singular, leading to abnormal sensitivity of the localization system to input errors. Existing sensor network layout methods generally suffer from significant deviations in their optimization or evaluation functions, primarily in the following aspects: First, most methods fail to consider the significant signal attenuation characteristics of asphalt mixtures as viscoelastic materials, employing a constant wave velocity assumption, which contradicts the physical law of wave velocity variation with propagation distance. For example, patent CN116341726A discloses a sensor network layout optimization method for mine microseismic monitoring, which uses a constant wave velocity model (which contradicts the actual physical law of wave velocity variation with distance), quantifies positioning accuracy using the Cramer-Rao lower bound (CRLB), and employs a genetic algorithm to optimize the sensor network layout for mine microseismic monitoring. Second, there is a lack of comprehensive constraints on multiple physical parameters such as waveform, spectrum, and amplitude, failing to consider the effective signal reception range as a prerequisite for layout optimization.

[0004] Therefore, there is an urgent need for an acoustic emission sensor layout optimization method that can integrate material attenuation characteristics, is based on multi-parameter constraints, and has adaptive capabilities, in order to improve the accuracy and reliability of fatigue damage monitoring of asphalt mixtures. Summary of the Invention

[0005] The purpose of this application is to provide a multi-parameter fusion method for optimizing the layout of acoustic emission sensors in asphalt mixtures. This method aims to improve the accuracy of acoustic emission monitoring systems in identifying the fatigue state of asphalt mixtures, optimize the arrangement of acoustic emission sensors, and enhance the effectiveness and accuracy of data acquisition. This invention selects amplitude, ring count, and wave velocity as key acoustic emission parameters. By optimizing the spatial distribution of acoustic emission sensors, it improves the monitoring system's coverage of fatigue damage areas and enhances the representativeness of the data, thereby more accurately identifying the damage evolution process of asphalt mixtures and providing scientific guidance for road maintenance.

[0006] To achieve the above objectives, this application provides the following technical solution: This application provides a method for optimizing the layout of acoustic emission sensors in asphalt mixtures using multi-parameter fusion, comprising the following steps: S1. Several acoustic emission sensors are evenly spaced along a straight line parallel to the longitudinal direction of the asphalt mixture surface to be tested. The acoustic emission sensor at one end is used as the excitation source, and the remaining acoustic emission sensors receive the signal. Multiple active acoustic emission pulse tests are conducted to measure the amplitude, ring count, and wave velocity, and an exponential fit is performed on the wave velocity equation. The effective arrangement range in which the acoustic emission signal can be effectively received is determined by the distance when the amplitude and ring count first decrease by 50% and the distance when the wave velocity attenuation rate first reaches 50%. The smallest distance among the three effective arrangement ranges is taken as the initial spacing of the acoustic emission sensors. S2. Based on the CRLB (Crame-Rao lower bound) principle and combined with the wave velocity equation, an optimization function for the acoustic emission sensor network layout is constructed. The optimization function is as follows: ; In the formula, Represents the CRLB matrix of the first... Column and number Elements in the row; Location of the acoustic emission sensor; for The abbreviation indicates the source of the sound. integral; A focal region comprised of all possible hypocenter locations. As the sound source In the monitoring area The probability density of occurrence; S3. The optimization function described in step S2 is solved using an improved genetic algorithm to minimize the integral value of the optimization function, thereby obtaining the optimal acoustic emission sensor network layout scheme.

[0007] Furthermore, in step S2, the wave velocity equation is: ; in, , For fitting parameters, , To the sound source Distance at location, It is a natural constant.

[0008] Furthermore, the formula for the wave velocity attenuation rate is: ; In the formula, The reference wave velocity is the measured wave velocity of the acoustic emission sensor used as the excitation source. The measured velocity of the sound wave is the actual velocity at the measuring point of the acoustic emission sensor.

[0009] Furthermore, in step S2, the steps for constructing the optimization function for the acoustic emission sensor network layout are as follows: S2.1 Calculate the selected acoustic emission sensor With reference sensor From the travel time equations between them, we get: ; In the formula, Acoustic emission sensor With reference sensor The theoretical time difference between them; The acoustic emission signal propagates from the sound source to the acoustic emission sensor. The time of departure; For acoustic emission signals to propagate from the sound source to the reference sensor The time of departure; From sound source to acoustic emission sensor The Euclidean distance; For varying distances of transmission The wave speed model of exponential change; , For fitting parameters, , To the sound source Distance at location, It is a natural constant; The integral symbol is used. The differential symbol; S2.2 Calculate the sound source In the monitoring area Probability density of occurrence : ; In the formula, M represents the number of acoustic emission sensors; It is the inverse of the covariance matrix; The normalization term determined by the determinant of the covariance matrix; It is a natural exponential function; This is the transpose of the time difference measurement error vector; S2.3, to Take the natural logarithm and compare it with the coordinates of the sound source. Find the partial derivative and combine it with the expected value. Construct the Fisher information matrix, invert it to obtain the CRLB, and combine it with the Jacobian matrix. CRLB can be simplified to: ; In the formula, For noise variance; It is the transpose of the Jacobian matrix; For fixed structure matrix ; It is an M×3 Jacobian matrix; S2.4 Integrate the microseismic events over the entire monitoring area to minimize the lower bound of the global positioning error, and obtain the optimization function.

[0010] Furthermore, in step S2.2, the time difference data is modeled as a time difference model including errors: ; in, It represents its theoretical truth value without error; This indicates the actual time difference collected. The noise is measured and follows a zero-mean Gaussian random distribution, independent of the time difference and the coordinates of the sound source.

[0011] Furthermore, The matrix expression is: ; In the formula, Theoretical to time difference The first-order partial derivatives with respect to the x, y, and z coordinates of the sound source; It is the symbol for partial differential.

[0012] Furthermore, in step S3, the improved genetic algorithm includes the following steps: S3.1 Randomly generate an initial population and use binary encoding to represent the acoustic emission sensor layout scheme; one chromosome represents a set of acoustic emission sensor layout schemes, and each gene in the chromosome corresponds to a candidate acoustic emission sensor position. A gene value of 1 indicates that an acoustic emission sensor is placed at that position, and a gene value of 0 indicates that no acoustic emission sensor is placed at that position. S3.2 Calculate the fitness value of the individual using the optimization function in step S2.4, wherein the gene values ​​at all positions on the chromosome satisfy the formula. , For chromosome number The gene value at each location, that is, the number of genes in the chromosome that indicate the presence of acoustic emission sensors, is equal to the actual number of acoustic emission sensors M to be deployed. S3.3 Termination Condition Judgment: Set an iteration termination condition. If the condition is met, output the current best individual and end the algorithm; otherwise, proceed to step S3.4. The iteration termination condition adopts one of the following two criteria: Criterion 1: The difference between the fitness value of the best individual in the current population and the average fitness value of the population is less than a preset threshold; Criterion 2: After a certain number of iterations, the optimal fitness value of the population remains unchanged; S3.4. Iteratively optimize the population using selection, crossover, and mutation operators. First, a truncation selection strategy is adopted, ranking individuals by fitness and retaining only the top 70% of superior individuals for the next generation, while the remaining 30% are eliminated. Then, crossover is performed on the selected individuals with a probability of 0.85 to generate new individuals. Finally, mutation is performed on the individual genes with a probability of 0.01 to generate a new population. Then, return to step S3.2 to evaluate the fitness of the next generation population. This process is repeated until the termination condition in step S3.3 is triggered.

[0013] The technical solution of this application has the following beneficial effects: This application determines the physical spatial constraints for effective sensor deployment through standard attenuation testing and multi-parameter fusion, and constructs a regional overall positioning accuracy optimization function that integrates the exponential wave velocity equation and the spatial uncertainty of the sound source. It breaks through the constant wave velocity assumption commonly used in past sensor layout optimizations, introducing the exponential wave velocity equation characterizing signal attenuation into the derivation of CRLB, thus improving the physical accuracy and reliability of the layout optimization model from a theoretical perspective.

[0014] To address the limitations of judging the range of a single parameter, this invention innovatively proposes to integrate the attenuation characteristics of three key acoustic emission parameters: amplitude, ring count, and wave velocity. By establishing attenuation selection criteria for each parameter, a "greatest common divisor" deployment range is comprehensively determined to ensure that all valid signals are received. This provides a precise and reliable pre-constraint for subsequent optimization of the search space, avoiding positioning failures caused by missed signal detection. Attached Figure Description

[0015] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. Wherein: Figure 1 This is a schematic diagram of the standard attenuation test setup according to an embodiment of the present invention.

[0016] Figure 2 This is a diagram showing the amplitude attenuation of the excitation source pulse signal in an embodiment of the present invention.

[0017] Figure 3 This is a ringing count attenuation diagram of the excitation source pulse signal in an embodiment of the present invention.

[0018] Figure 4 This is the attenuation curve of the excitation source pulse signal in an embodiment of the present invention.

[0019] Figure 5This is the result of the acoustic emission sensor layout optimization in an embodiment of the present invention.

[0020] Figure 6 This is a comparative example of the optimized layout of the acoustic emission sensor in this invention.

[0021] Explanation of reference numerals in the attached figures: 1-Acoustic emission sensor, 2-Signal amplifier, 3-Data acquisition instrument, 4-Computer processing unit. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] A method for optimizing the layout of acoustic emission sensors in asphalt mixtures through multi-parameter fusion includes the following steps: S1. Along a straight line parallel to the length of the asphalt mixture surface to be tested, a series of acoustic emission sensors are arranged at equal intervals of 0.3m-0.5m. The acoustic emission sensors are connected to a data acquisition instrument and a computer processing unit through signal amplifiers to convert the acquired acoustic signals into digital signals for analysis. The leftmost acoustic emission sensor is used as the excitation source to emit a pulse signal, and the other acoustic emission sensors receive the signal. Three active acoustic emission pulse tests are conducted, and the parameters are averaged. Amplitude, ring count, and wave velocity were selected as key parameters for exponential fitting of the wave velocity equation. , , For fitting parameters, , To the sound source Distance at location, It is a natural constant; The attenuation criteria for each parameter are: amplitude, distance at which the ring count first attenuates by 50%, and wave velocity attenuation rate. (Pick The distance when (as a reference wave speed) first reaches 50%; Establish attenuation selection target: Determine the effective arrangement range (attenuation less than 50%) where the acoustic emission signal can be effectively received under different parameters (amplitude, ring count, wave velocity). Take the smallest distance among the three effective arrangement ranges as the initial spacing of the acoustic emission sensor deployment to ensure that all parameters (amplitude, ring count, wave velocity) can be reliably collected. S2. Based on the CRLB (Crame-Rao lower bound) principle and combined with the wave velocity equation, an optimization function for the acoustic emission sensor network layout is constructed. The optimization function is as follows:

[0024] ;

[0025] In the formula, vector The coordinates of the sound source; Represents the CRLB matrix of the first... Column and number Elements in the row; For acoustic emission sensors (location); for The abbreviation indicates the source of the sound. integral; A focal region comprised of all possible hypocenter locations. As the sound source In the monitoring area The probability density of occurrence.

[0026] S3. The optimization function described in step S2 is solved using an improved genetic algorithm to minimize the integral value of the optimization function, thereby obtaining the optimal acoustic emission sensor network layout scheme.

[0027] Furthermore, in step S2, the steps for constructing the optimization function for the acoustic emission sensor network layout are as follows: S2.1, with The selected acoustic emission sensor is used as a reference sensor. With the sound source The Euclidean distance is Then, combining the wave speed equation Acoustic emission sensor and The travel time equation (theoretical to time difference) based on the hyperbolic positioning principle is as follows: ; S2.2 In real-world scenarios, due to measurement noise in the arrival time difference data, the arrival time difference containing errors is modeled as follows: ;in, It represents its theoretical truth value without error; This represents the actual time difference collected. The noise is measured to follow a zero-mean Gaussian random distribution and is independent of the time difference and the coordinates of the sound source; Let the standard deviation of the time difference measurement noise be . , If the covariance matrix is ​​M×M dimensional, then the specific expression is: Where M is the number of acoustic emission sensors, To obtain the standard deviation of noise in time zone measurement, For fixed structure matrix ; Therefore, the sound source In the monitoring area Probability density of occurrence for: ; In the formula, M represents the number of acoustic emission sensors; It is the inverse of the covariance matrix; It is a natural exponential function; This is the error vector for time difference measurement; S2.3, to Take the natural logarithm and compare it with the coordinates of the sound source. Find the partial derivative and combine it with the expected value. We obtain the Fisher information matrix, and then inverse it to obtain the CRLB (Crame-Rao lower bound): ; Regarding the above Take the natural logarithm and differentiate the formula, then substitute it into... and The specific expression simplifies CRLB to: ; In the formula, The matrix is ​​an M×3 Jacobian matrix, and its expression is: .

[0028] S2.4 Integrating the microseismic events across the entire monitoring area, with the objective of minimizing the lower bound of the global location error, yields the optimization function: ; In the formula, For acoustic emission sensors (position); vector The coordinates of the sound source; Represents the CRLB matrix of the first... Column and number Elements in the row; for The abbreviation indicates the source of the sound. integral; A focal region comprised of all possible hypocenter locations. As the sound source In the monitoring area The probability density of occurrence.

[0029] Furthermore, in step S3, the improved genetic algorithm includes the following steps: S3.1 Randomly generate an initial population and use binary encoding to represent the acoustic emission sensor layout scheme; one chromosome represents a set of acoustic emission sensor layout schemes, and each gene in the chromosome corresponds to a candidate acoustic emission sensor position. A gene value of 1 indicates that an acoustic emission sensor is placed at that position, and a gene value of 0 indicates that no acoustic emission sensor is placed at that position. S3.2 Calculate the fitness value of an individual using the fitness function, where the gene values ​​at all positions on the chromosome satisfy the formula. , For chromosome number The gene value at each location, i.e., the number of genes in the chromosome that represent the placement of acoustic emission sensors, is equal to the actual number M of acoustic emission sensors to be placed; the fitness function is selected according to the content of the object, and here we use the optimization function in step S2.4 as the fitness function.

[0030] S3.3 Termination Condition Judgment: Set an iteration termination condition. If the condition is met, output the current best individual and end the algorithm; otherwise, proceed to step S3.4. The iteration termination condition adopts one of the following two criteria: Criterion 1: The difference between the fitness value of the best individual in the current population and the average fitness value of the population is less than a preset threshold; Criterion 2: After a certain number of iterations, the optimal fitness value of the population remains unchanged; S3.4. Iteratively optimize the population using selection, crossover, and mutation operators. First, a truncation selection strategy is adopted, ranking individuals by fitness and retaining only the top 70% of superior individuals for the next generation, while the remaining 30% are eliminated. Then, crossover is performed on the selected individuals with a probability of 0.85 to generate new individuals. Finally, mutation is performed on the individual genes with a probability of 0.01 to generate a new population. Then, return to step S3.2 to evaluate the fitness of the next generation population. This process is repeated until the termination condition in step S3.3 is triggered.

[0031] Example This embodiment was conducted indoors, using a standard asphalt mixture four-point bending beam specimen as the asphalt mixture to be tested as an example. The standard asphalt mixture four-point bending beam specimen provides a three-dimensional monitoring area (equivalent to a pavement asphalt mixture layer), with dimensions of 380 mm × 63.5 mm × 50 mm (length × width × height). The number of acoustic emission sensors to be arranged is 8, as shown in Table 1. First, the coordinates of the acoustic emission sensors are randomly generated on the surface of the specimen, with the origin (0,0,0) defined as the vertex of the lower left corner of the specimen.

[0032] Table 1 Initial layout of the acoustic emission sensor provided in this embodiment

[0033] A method for optimizing the layout of acoustic emission sensors in asphalt mixtures through multi-parameter fusion includes the following steps: S1. Conduct standard attenuation tests to determine the effective deployment range of multi-parameter fusion; S1.1 Place a standard asphalt mixture four-point bending beam specimen with dimensions of 380mm × 63.5mm × 50mm (length × width × height) on the test bench. For example... Figure 1 As shown, acoustic emission sensor 1 is connected to data acquisition instrument 3 and computer processing unit 4 via signal amplifier 2 to convert the acquired acoustic signals into digital signals for analysis. A series of acoustic emission sensors are arranged at equal intervals of 30 mm along a straight line parallel to the length of the specimen surface. The leftmost acoustic emission sensor is used as the excitation source to emit a pulse signal, and the remaining acoustic emission sensors receive the signal. Three active acoustic emission pulse tests are performed, and the parameters are averaged. Amplitude, ring count, and wave velocity were selected as key parameters. Attenuation selection criteria were established: the distance at which the amplitude and ring count first decreased by 50%; and the wave velocity attenuation rate. (Pick The distance when the baseline wave speed first reaches 50%, such as Figure 2-4 As shown, based on the above standards, the effective arrangement range (effective receiving distance) for the acoustic emission signals corresponding to amplitude, ring count, and wave velocity that can be effectively received are calculated to be 330mm, 180mm, and 330mm, respectively.

[0034] Based on the principle of "taking the smallest distance among the three effective arrangement ranges," the intersection of the three is taken, and the farthest effective distance between the acoustic emission sensor and the sound source in the monitoring area is determined to be 180mm. This distance will serve as the spatial constraint for subsequent optimization of the acoustic emission sensor layout. This is the basis for ensuring that the valid signals of all key parameters can be reliably received.

[0035] S1.2. Based on the wave velocity data measured in step S1.1, calculate the wave velocity as a function of propagation distance. The relationship between the changes is fitted with an exponential equation to obtain the fitting parameters. , and wave speed equation , =4994.71, =−0.00159.

[0036] S2. Construct an optimization function for sensor network layout based on the CRLB principle and wave velocity equation; S2.1 Assume that all possible hypocenter locations together constitute a hypocenter region. ,vector yes A random vector with prior probability density is The sensor network consists of 9 sensors, whose locations are denoted as... ,by The selected acoustic emission sensor is used as a reference sensor. With the sound source The Euclidean distance is ; Without loss of generality, select the sensor As a reference sensor; then the sensor is obtained. and The travel time equations between them are: , ; S2.2 In real-world scenarios, arrival time difference data inevitably contains measurement noise; therefore, the arrival time difference with errors is modeled as follows: ;in, It represents its theoretical truth value without error; This represents the actual time difference collected; the measurement noise is a zero-mean Gaussian random distribution, independent of the time difference and the coordinates of the sound source, and is represented by... express.

[0037] Let the standard deviation of the time difference measurement noise be . , The covariance matrix is ​​8×8 in dimension, and its specific expression is as follows: , To obtain the standard deviation of noise in time zone measurement, .

[0038] thus, As the sound source In the monitoring area The probability density of occurrence is: ; In the formula, M represents the number of acoustic emission sensors; It is the inverse of the covariance matrix; It is a natural exponential function; This is the error vector for time difference measurement; S2.3 For any given source coordinates and sensor network layout, CRLB can provide the minimum bound of the standard deviation of the source coordinate parameter estimates; the CRLB of the parameters to be estimated can be determined by taking the inverse of the Fisher matrix. ; By analyzing the above Taking the natural logarithm and differentiating the formula, CRLB can be simplified to... . The matrix is ​​an M×3 Jacobian matrix, and its expression is: .

[0039] S2.4 Finally, by integrating the microseismic events over the entire monitoring area, the optimized function is obtained as follows: .

[0040] S3. Solve the optimization function in step S1 using an improved genetic algorithm to obtain the optimal sensor network layout scheme, specifically including the following steps: S3.1. Randomly generate an initial population and use binary encoding to represent the sensor layout scheme. One chromosome represents a set of acoustic emission sensor layout schemes. Each gene in the chromosome corresponds to a candidate sensor position. A gene value of 1 indicates that a sensor is placed at that position, and a gene value of 0 indicates that no sensor is placed at that position.

[0041] S3.2 Calculate the individual fitness value using the optimization function from step S2.4. The gene values ​​at all positions on the chromosome satisfy the formula... , For individual Gene values ​​at each location; S3.3 Determine if the termination condition is met. The termination condition is that after a certain number of consecutive iterations, the optimal fitness of the population no longer changes, or the difference between the optimal fitness and the average fitness is less than a threshold. If the condition is met, the iteration terminates and the optimal individual is output; where the number of consecutive iterations is set to 100, and the threshold is... Set as If the termination condition is not met, proceed to step S3.4.

[0042] S3.4. Iteratively optimize the population using selection, crossover, and mutation operators. First, adopt a truncation selection strategy, sort individuals by fitness, retain only the top 70% of superior individuals to enter the next generation, and eliminate the remaining 30% of individuals. Then, perform crossover operation on the selected individuals with a probability of 0.85 to generate new individuals. Finally, perform mutation operation on the individual genes with a probability of 0.01 to generate a new population (next generation population).

[0043] After completing the above operations, return to step S3.2 to evaluate the fitness of the next generation population.

[0044] This embodiment uses the sensor network layout optimization method based on CRLB and improved genetic algorithm of the present invention to calculate and obtain the optimal acoustic emission sensor layout diagram with 8 acoustic emission sensors, as shown in the figure. Figure 5 As shown in Table 2, the specific coordinates of each acoustic emission sensor are shown. The origin (0,0,0) is defined as the vertex of the lower left corner of the specimen. Figure 5 This is the result of the layout optimization of the acoustic emission sensor in this embodiment.

[0045] Table 2 Optimal acoustic emission sensor layout obtained in this embodiment

[0046] Comparative Example To verify the superiority of this invention over traditional methods, a comparative example was set up for comparison. The comparative example used the exact same four-point bending fatigue specimen, monitoring area, prior sound source distribution p(u), initial local acoustic emission sensor (Table 1), and number of target sensors (M=8) as the aforementioned embodiments. It also employed the same improved genetic algorithm process and parameters (population size, selection, crossover, mutation probability, and termination condition) for solving the problem. The only difference was that the comparative example did not use the exponential wave velocity model fitted in step S1.2, but instead used a constant wave velocity (3500 m / s) based on the conventional simplified assumptions of traditional acoustic emission localization to construct and solve the optimization function; and it no longer used amplitude, ring count, or wave velocity attenuation rate for distance limitation.

[0047] The steps of the acoustic emission sensor layout optimization method in this comparative example are as follows: S1. Conduct standard attenuation tests to determine the effective deployment range of multi-parameter fusion; The multi-parameter fusion attenuation test in step S1.1 and the wave velocity equation fitting in step S1.2 are not performed; the average wave velocity v = 3500 m / s is directly selected based on the acoustic wave propagation characteristics of the specimen.

[0048] Step S2: Construct an optimization function for sensor network layout based on the CRLB principle; S2.1 is the same as S2.1 in Example 1; S2.2 In real-world scenarios, arrival time difference data inevitably contains measurement noise; arrival time difference with errors can be modeled as... , This represents its error-free theoretical true value. The measurement noise is a zero-mean Gaussian random process and is independent of the time difference and source coordinates. express; If the standard deviation of noise in time difference measurement is , The covariance matrix is ​​8×8 in dimension, and its specific expression is as follows: , To obtain the standard deviation of noise in time zone measurement, ; thus, For the sound source u in the monitoring area Probability density of occurrence ,in , It is the three-dimensional coordinates of the sound source. It's speed; S2.3. For any given source coordinates and sensor network layout, CRLB can provide a minimum bound on the standard deviation of the source coordinate parameter estimates. Generally, the CRLB of the parameters to be estimated can be determined by taking the inverse of the Fisher matrix. The result of taking the inverse of the Fisher matrix (the definition of CRLB) is: ; By taking the natural logarithm and differentiating, CRLB can be simplified to... . Let be a Jacobian matrix, and its matrix expression is: ; S2.5 Finally, the optimization function is obtained by integrating the microseismic events over the entire monitoring area: .

[0049] S3 is exactly the same as step S3 in the embodiment; In summary, in this comparative example, the average wave speed is a constant wave speed calculated by dividing distance by velocity; furthermore, the Jacobian matrix... It is also different from that in the embodiments.

[0050] The optimal layout of the eight acoustic emission sensors obtained in this comparative example is shown in [reference needed]. Figure 6 The specific coordinates are shown in Table 3.

[0051] Table 3. Layout results of the eight acoustic emission sensors in the comparative example.

[0052] Using the average Euclidean distance between sensors The formula reflecting the "dispersion" of the sensor network is as follows: ; In the formula, M represents the number of acoustic emission sensors. For the first The three-dimensional coordinates of an acoustic emission sensor For the first The three-dimensional coordinates of an acoustic emission sensor.

[0053] Within the effective monitoring range, the average Euclidean distance A larger value indicates a more dispersed distribution of acoustic emission sensors and better geometric intersection conditions, which can more accurately pinpoint the sound source location, reduce errors, and improve positioning accuracy.

[0054] With the initial layout of the acoustic emission sensor (Table 1), the CRLB value is 73.25. The CRLB value is 85.64; in the optimal acoustic emission sensor layout (Table 2) of the embodiments, the CRLB value is 41.42. The CRLB value is 155.72; in the comparative acoustic emission sensor layout (Table 3), the CRLB value is 58.6. It is 195.52.

[0055] The ultimate goal of this application is to achieve higher theoretical positioning accuracy. Comparing the initial layout, the embodiment, and the comparative example, it is clear that the initial layout has the worst positioning accuracy and the sensor distribution is too concentrated; although the comparative example... Compared to the embodiment, the larger size and more dispersed distribution of acoustic emission sensors result in lower CRLB value positioning accuracy. The embodiment achieves the lowest CRLB while maintaining reasonable dispersion, and theoretically achieves optimal positioning accuracy, thus realizing the best combination of positioning accuracy and layout rationality, which is superior to the traditional layout method.

[0056] The comparison results show that the CRLB value in the embodiment is smaller than that in the comparative example, a reduction of approximately 29.3%, indicating that theoretically, the acoustic emission sensor layout of the embodiment can achieve higher positioning accuracy. Figure 5 , Figure 6 From the layout perspective, the acoustic emission sensors in this embodiment are distributed on the top, bottom, and sides of the specimen, providing comprehensive three-dimensional coverage without any blind spots. In contrast, the acoustic emission sensors in the comparative embodiment are only located on the top (two at both ends) and bottom of the specimen, resulting in uneven three-dimensional coverage and potential blind spots. This is because attenuation is not considered for the parameters, and the algorithm pushes the acoustic emission sensors to the boundaries without restraint in pursuit of minimizing the CRLB value, leading to significant dispersion. The optimized acoustic emission sensor layout in this embodiment incorporates a realistic attenuation model, allowing for a more scientific balance between spatial coverage and effective signal reception distance.

[0057] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. 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 optimizing the layout of acoustic emission sensors in asphalt mixtures through multi-parameter fusion, characterized in that: Includes the following steps: S1. Several acoustic emission sensors are evenly spaced along a straight line parallel to the longitudinal direction of the asphalt mixture surface to be tested. The acoustic emission sensor at one end is used as the excitation source, and the remaining acoustic emission sensors receive the signal. Multiple active acoustic emission pulse tests are conducted to measure the amplitude, ring count, and wave velocity, and an exponential fit is performed on the wave velocity equation. The effective arrangement range in which the acoustic emission signal can be effectively received is determined by the distance when the amplitude and ring count first decrease by 50% and the distance when the wave velocity attenuation rate first reaches 50%. The smallest distance among the three effective arrangement ranges is taken as the initial spacing of the acoustic emission sensors. S2. Based on the CRLB (Crame-Rao lower bound) principle and combined with the wave velocity equation, an optimization function for the acoustic emission sensor network layout is constructed. The optimization function is as follows: ； In the formula, Represents the CRLB matrix of the first... Column and number Elements in the row; Location of the acoustic emission sensor; for The abbreviation indicates the source of the sound. integral; A focal region comprised of all possible hypocenter locations. As the sound source In the monitoring area The probability density of occurrence; S3. The optimization function described in step S2 is solved using an improved genetic algorithm to minimize the integral value of the optimization function, thereby obtaining the optimal acoustic emission sensor network layout scheme.

2. The method for optimizing the layout of acoustic emission sensors for asphalt mixtures through multi-parameter fusion according to claim 1, characterized in that: In step S2, the wave velocity equation is: ； in, , For fitting parameters, ; To the sound source Distance at location; It is a natural constant.

3. The method for optimizing the layout of acoustic emission sensors for asphalt mixtures through multi-parameter fusion according to claim 1, characterized in that: The formula for wave velocity attenuation rate is: ； In the formula, The reference wave velocity is the measured wave velocity of the acoustic emission sensor used as the excitation source. The measured velocity of the sound wave is the actual velocity at the measuring point of the acoustic emission sensor.

4. The method for optimizing the layout of acoustic emission sensors for asphalt mixtures through multi-parameter fusion according to claim 1, characterized in that: In step S2, the steps for constructing the optimization function for the acoustic emission sensor network layout are as follows: S2.1 Calculate the selected acoustic emission sensor With reference sensor From the travel time equations between them, we get: ； In the formula, Acoustic emission sensor With reference sensor The theoretical time difference between them; The acoustic emission signal propagates from the sound source to the acoustic emission sensor. The time of departure; For acoustic emission signals to propagate from the sound source to the reference sensor The time of departure; From sound source to acoustic emission sensor The Euclidean distance; For varying distances of transmission The wave speed model of exponential change; , For fitting parameters, ; To the sound source Distance at location, It is a natural constant; The integral symbol is used. The differential symbol; S2.2 Calculate the sound source In the monitoring area Probability density of occurrence : ； In the formula, M represents the number of acoustic emission sensors; It is the inverse of the covariance matrix; The normalization term determined by the determinant of the covariance matrix; It is a natural exponential function; This is the transpose of the time difference measurement error vector; S2.3, to Take the natural logarithm and compare it with the coordinates of the sound source. Find the partial derivative and combine it with the expected value. Construct the Fisher information matrix, invert it to obtain the CRLB, and combine it with the Jacobian matrix. CRLB can be simplified to: ； In the formula, For noise variance; It is the transpose of the Jacobian matrix; For fixed structure matrix ; It is an M×3 Jacobian matrix; S2.4 Integrate the microseismic events over the entire monitoring area to minimize the lower bound of the global positioning error, and obtain the optimization function.

5. The method for optimizing the layout of acoustic emission sensors for asphalt mixtures based on multi-parameter fusion according to claim 4, characterized in that: In step S2.2, the time difference data is modeled as an error-inclusive time difference model: ； in, It represents its theoretical truth value without error; This indicates the actual time difference collected. The noise is measured and follows a zero-mean Gaussian random distribution, independent of the time difference and the coordinates of the sound source.

6. The method for optimizing the layout of acoustic emission sensors for asphalt mixtures through multi-parameter fusion according to claim 4, characterized in that: The matrix expression is: ； In the formula, Theoretical to time difference The first-order partial derivatives with respect to the x, y, and z coordinates of the sound source; It is the symbol for partial differential.

7. The method for optimizing the layout of acoustic emission sensors for asphalt mixtures through multi-parameter fusion according to claim 1, characterized in that: In step S3, the improved genetic algorithm includes the following steps: S3.1 Randomly generate an initial population and use binary encoding to represent the acoustic emission sensor layout scheme; one chromosome represents a set of acoustic emission sensor layout schemes, and each gene in the chromosome corresponds to a candidate acoustic emission sensor position. A gene value of 1 indicates that an acoustic emission sensor is placed at that position, and a gene value of 0 indicates that no acoustic emission sensor is placed at that position. S3.2 Calculate the fitness value of the individual using the optimization function in step S2.4, wherein the gene values ​​at all positions on the chromosome satisfy the formula. , For chromosome number The gene value at each location, that is, the number of genes in the chromosome that indicate the presence of acoustic emission sensors, is equal to the actual number of acoustic emission sensors M to be deployed. S3.3 Termination Condition Judgment: Set an iteration termination condition. If the condition is met, output the current best individual and end the algorithm; otherwise, proceed to step S3.

4. The iteration termination condition adopts one of the following two criteria: Criterion 1: The difference between the fitness value of the best individual in the current population and the average fitness value of the population is less than a preset threshold; Criterion 2: After a certain number of iterations, the optimal fitness value of the population remains unchanged; S3.

4. Iteratively optimize the population using selection, crossover, and mutation operators. First, a truncation selection strategy is adopted, ranking individuals by fitness and retaining only the top 70% of superior individuals for the next generation, while the remaining 30% are eliminated. Then, crossover is performed on the selected individuals with a probability of 0.85 to generate new individuals. Finally, mutation is performed on the individual genes with a probability of 0.01 to generate a new population. Then, return to step S3.2 to evaluate the fitness of the next generation population. This process is repeated until the termination condition in step S3.3 is triggered.

Images