Roadbed dynamic compaction construction quality control method and system based on rammer dynamic response

Abstract

The invention provides a roadbed dynamic compaction construction quality control method and system based on rammer dynamic response. The control method comprises the following steps that an acceleration signal of a rammer in the current ramming process is obtained; data processing is carried out on the acceleration signal of the rammer, a plurality of key parameters used for predicting the dynamic deformation modulus are obtained, and the data processing comprises filtering processing, time domain analysis and frequency domain analysis which are carried out in sequence; on the basis of the multiple key parameters and a predetermined dynamic deformation modulus prediction model, the dynamic deformation modulus of the tamping point completing the current time of tamping is obtained; and comparing the dynamic deformation modulus of the tamping point with a set dynamic deformation modulus design value, and sending a tamping stopping prompt or a tamping continuing prompt based on a comparison result. According to the control method provided by the invention, the dynamic compaction construction is subjected to quality control based on the acceleration signal obtained in real time and the dynamic deformation modulus prediction model, so that the uniform compaction of the roadbed is ensured, and the construction efficiency and the construction quality are improved.

Description

Technical Field

[0001] The invention belongs to the technical field of roadbed compaction construction quality control, and in particular relates to a roadbed compaction construction quality control method and system based on rammer power response. Background Art

[0002] Dynamic compaction machine is a kind of engineering machinery equipment widely used in foundation treatment. Dynamic compaction machine uses lifting equipment to lift the rammer to a certain height, then releases the rammer to make it fall freely, and uses the strong impact force generated by the rammer falling to compact the soil, thereby improving the strength of the foundation and reducing possible foundation settlement. The dynamic compaction machine acts on the foundation soil, with the purpose of improving the strength and uniformity of the foundation soil.

[0003] At present, the quality control of roadbed compaction construction mainly has the following problems:

[0004] (1) Traditional dynamic tamping construction quality control relies heavily on the operator’s experience and judgment of the construction site. For example, the specification requires the use of the average tamping amount of the last two hits as the tamping stop standard during the tamping process. This method is greatly affected by human factors, and the manual leveling method used cannot meet the high-precision construction requirements.

[0005] (2) The quality of compaction construction is usually evaluated through post-testing methods, such as static load test, standard penetration test, static penetration test and indoor geotechnical test. Although these methods can effectively evaluate the bearing capacity of the roadbed, the testing process is time-consuming and costly. In addition, this type of testing method is mostly carried out after the construction is completed, and it is impossible to monitor and adjust the compaction effect of the soil layer in real time during the construction process, which can easily lead to problems such as uneven compaction, over-compacting or under-compacting of the soil during the construction process.

[0006] (3) Traditional quality control methods fail to fully utilize the dynamic response data of the rammer and cannot achieve accurate and real-time compaction quality assessment. Summary of the invention

[0007] The purpose of the present invention is to provide a dynamic compaction construction quality control method that can monitor the dynamic response of the rammer in real time during the construction process, and analyze and adjust the construction parameters in real time based on the response signal, so as to ensure uniform compaction of the roadbed and improve construction efficiency and construction quality.

[0008] To achieve the above object, the present invention provides a roadbed dynamic tamping construction quality control method based on the dynamic response of the tamping hammer, and the control method comprises the following steps:

[0009] Step S10, obtaining an acceleration signal of the rammer during the current rammering process;

[0010] Step S20, performing data processing on the acceleration signal of the rammer to obtain a plurality of key parameters for predicting the dynamic deformation modulus, wherein the data processing comprises filtering processing, time domain analysis and frequency domain analysis performed in sequence, and the plurality of key parameters comprises acceleration peak value, rammer action time, acceleration kurtosis and acceleration main frequency;

[0011] Step S30, based on the multiple key parameters and a predetermined dynamic deformation modulus prediction model, obtaining the dynamic deformation modulus of the tamping point after the current tamping, wherein the dynamic deformation model prediction model is a multiple linear regression model;

[0012] Step S40, compare the dynamic deformation modulus of the tamping point with the set dynamic deformation modulus design value, and when the dynamic deformation modulus of the tamping point is greater than or equal to the dynamic deformation modulus design value, send a prompt to stop tamping; when the dynamic deformation modulus of the tamping point is less than the dynamic deformation modulus design value, send a prompt to continue tamping, and repeat steps S10 to S30.

[0013] In a specific implementation, the dynamic deformation model prediction model is:

[0014] E vd =β 0 +β 1 a m +β 2 t m +β 3 u+β 4 f m

[0015] Among them, E vd is the dynamic deformation modulus, a m is the peak acceleration, t m is the tamping hammer action time, u is the acceleration peak, f m is the acceleration frequency, β 0 ~β 4 is the fitting coefficient.

[0016] In a specific embodiment, a method for constructing a predetermined dynamic deformation modulus prediction model includes:

[0017] Perform multiple tamping tests at multiple test tamping points respectively, obtain collected data of each tamping test at each test tamping point, and obtain multiple groups of collected data, each group of the collected data includes a dynamic deformation modulus measurement value and an acceleration signal of the tamping hammer, and the dynamic deformation modulus measurement value and the acceleration signal of the tamping hammer in the same group of collected data have a corresponding relationship, and the number of the collected data is the sum of the number of tamping times of all the test tamping points;

[0018] Performing data processing on the acceleration sampling signal of the rammer in each of the collected data to obtain a plurality of factor groups, each of the factor groups including acceleration peak value, rammer action time, acceleration kurtosis and acceleration main frequency;

[0019] Combining the dynamic deformation modulus measurement value and the corresponding factor group in each of the collected data into one sample data to obtain a plurality of sample data;

[0020] Randomly extracting a first preset number of sample data from all sample data as a sample set, and using the sample set to calculate coefficients of an initial multivariate regression model to determine the multivariate regression model coefficients;

[0021] Based on the multivariate regression model coefficients, obtain the target multivariate regression model.

[0022] In a specific embodiment, the method for constructing the predetermined dynamic deformation modulus prediction model further includes a model verification process:

[0023] Randomly select a second preset number of sample data from all sample data as a validation set;

[0024] Substituting each factor group in the validation set into the target multivariate regression model to obtain a predicted value of the dynamic deformation modulus corresponding to each factor group;

[0025] Based on the dynamic deformation modulus measurement value and the dynamic deformation modulus prediction value corresponding to each factor group, the model prediction accuracy is calculated to obtain multiple model prediction accuracy values;

[0026] Get the average of the prediction accuracy values ​​of multiple models and use it as the prediction accuracy of the target multivariate regression model;

[0027] The prediction accuracy of the target multivariate regression model is compared with a preset threshold. When the prediction accuracy of the target multivariate regression model is less than or equal to the preset threshold, the verified target multivariate regression model meets the requirements, and the target multivariate regression model is used as the dynamic deformation modulus prediction model.

[0028] In a specific embodiment, when the prediction accuracy of the target multivariate regression model is greater than a preset threshold, the construction method further includes:

[0029] Randomly extracting a third preset number of sample data from all sample data as a sample set, and using the sample set to calculate coefficients of the initial multivariate regression model to determine new multivariate regression model coefficients, wherein the third preset number is greater than the first preset number, and the difference between the third preset number and the first preset number is between 10 and 20;

[0030] Based on the new multivariate regression model coefficients, a new target multivariate regression model is obtained;

[0031] The new target multivariate regression model was validated.

[0032] In a specific implementation, the data processing includes:

[0033] Step (1), filtering and noise reduction processing of the acceleration signal of the rammer;

[0034] Step (2): based on the acceleration signal after filtering and noise reduction, obtain the calculation data interval of the current impact acceleration, wherein the calculation data interval is t 0 Time to t 1 The data interval corresponding to the time, t 0 Time and t 1 The acceleration value corresponding to the moment is 0 and t 0 Time to t 1 At this moment, the acceleration value increases from 0, exceeds the peak value, and then decreases to 0;

[0035] Step (3), performing time domain analysis and frequency domain analysis on the data set corresponding to the calculation data interval of the current impact acceleration, and obtaining the acceleration peak value a according to multiple calculation formulas m , tamping hammer action time t m , acceleration kurtosis u, acceleration main frequency f m ,in:

[0036] Get the peak acceleration a m The calculation formula is:

[0037] in, is the acceleration time domain data, j represents the jth acceleration value in the calculation interval;

[0038] Get the rammer action time t m The calculation formula is: m =t 1 -t 0

[0039] Among them, t 0 and t 1 They are the starting time point and ending time point of the calculation data interval respectively;

[0040] The calculation formula for obtaining the kurtosis u is:

[0041] Among them, n represents the amount of acceleration data in the calculated data interval, is the acceleration time domain data, represents the mean of the calculated data interval, and s represents the variance of the calculated data interval;

[0042] Get the main frequency f m The calculation formula is: m = argmax P(A(f k ));

[0043] Among them, A(f k ) is the acceleration time domain data The frequency domain signal obtained by Fourier transform, P(A(f k )) is used to calculate the power spectral density of the frequency domain signal, argmaxP(A(f k )) is the frequency corresponding to the maximum value of the power spectral density.

[0044] In a specific implementation, the calculation formulas for calculating the mean of a data interval and the variance of a data interval are as follows:

[0045]

[0046] in, represents the mean of the calculated data interval, n represents the amount of acceleration data in the calculated interval, is the acceleration time domain data;

[0047]

[0048] Among them, s represents the variance of the calculated data interval, n represents the amount of acceleration data in the calculated data interval, is the acceleration time domain data, Represents the mean of the calculated data interval.

[0049] In a specific implementation, the step of calculating coefficients of the initial multivariate regression model using the sample set to determine the coefficients of the multivariate regression model includes:

[0050] Step (a), substituting the l sample data in the school-based set into the initial multivariate regression model, the resulting model form is:

[0051] y i =β 0 +β 1 x i1 +β 2 x i2 +β 3 x i3 +β 4 x i4 +ε i (i=1, 2, ..., l)

[0052] Where: i represents the error term, β 0~β 4 is the corresponding fitting coefficient, l represents the sample size;

[0053] Step (b), written in expanded form according to the model form:

[0054]

[0055] Where: Y is the dependent variable matrix, X is the independent variable matrix, and β is the coefficient matrix;

[0056] Step (c), calculating the residual sum of squares according to a preset formula, the preset formula is:

[0057]

[0058] Among them, Y T is the transposed matrix of the dependent variable matrix Y, X T is the transposed matrix of the independent variable matrix X, β T is the transposed matrix of the coefficient matrix β;

[0059] Step (d), taking the derivative of the residual sum of squares to obtain an equation after taking the derivative of the residual sum of squares, wherein the equation is:

[0060]

[0061] Step (e), based on the residual sum of squares being 0 and the equation after the residual sum of squares is derived, the coefficient matrix is ​​calculated to obtain the coefficients of the multivariate regression model.

[0062] The present invention also provides a roadbed strong compaction construction quality control system based on the dynamic response of the rammer, the control system comprising:

[0063] Rammer, used for strong tamping of roadbed;

[0064] A plurality of acceleration sensors, used for collecting acceleration signals of the rammer;

[0065] A controller is communicatively connected with the plurality of acceleration sensors respectively, and the controller is used in the control method described in steps S10 to S40 described above.

[0066] In a specific embodiment, a plurality of acceleration sensors are mounted on the rammer and are evenly distributed around the central axis of the rammer, and the distance from each acceleration sensor to the central axis of the rammer is the same as the distance from each acceleration sensor to the contour line of the rammer.

[0067] The beneficial effects of the present invention include at least:

[0068] 1. The present invention provides a method for controlling the quality of roadbed compaction construction based on the dynamic response of a rammer, the control method comprising the following steps: step S10, obtaining the acceleration signal of the rammer during the current tamping process; step S20, performing data processing on the acceleration signal of the rammer to obtain a plurality of key parameters for predicting the dynamic deformation modulus, the data processing comprising filtering processing, time domain analysis and frequency domain analysis performed in sequence, the plurality of key parameters comprising acceleration peak value, rammer action time, acceleration kurtosis and acceleration main frequency; step S30, obtaining the dynamic deformation modulus of the tamping point that has completed the current tamping based on the plurality of key parameters and a predetermined dynamic deformation modulus prediction model, the dynamic deformation model prediction model being a multivariate linear regression model; step S40, The dynamic deformation modulus of the tamping point is compared with the set design value of the dynamic deformation modulus. When the dynamic deformation modulus of the tamping point is greater than or equal to the design value of the dynamic deformation modulus, a prompt to stop tamping is sent; when the dynamic deformation modulus of the tamping point is less than the design value of the dynamic deformation modulus, a prompt to continue tamping is sent, and steps S10 to S30 are repeated. In the present invention, the dynamic deformation modulus of the tamping point after the tamping is completed is obtained by real-time acquisition of the rammer acceleration signal and a dynamic deformation modulus prediction model between multiple key parameters and the dynamic deformation modulus of the soil based on the test tamping data. After comparing it with the design value of the dynamic deformation modulus, it is possible to determine whether to continue tamping the tamping point based on the comparison result, thereby realizing real-time monitoring of the soil tamping quality and adjustment of construction parameters during construction.

[0069] 2. The present invention sets multiple acceleration sensors to accurately and comprehensively obtain the acceleration signal when the rammer hits, and analyzes the acceleration signal to extract the corresponding statistical characteristics of acceleration peak value, rammer action time, acceleration kurtosis and acceleration main frequency, and directly establishes a mathematical relationship between the corresponding statistical characteristics and the dynamic deformation modulus of the soil to construct a dynamic deformation modulus prediction model. In the process of constructing the dynamic deformation modulus prediction model, the prediction accuracy of the dynamic deformation modulus prediction model can be improved by increasing the number of sample sets, thereby ensuring that the prediction accuracy of the dynamic deformation modulus prediction model is high, and the reinforcement condition of the soil after each impact can be obtained more accurately.

[0070] In addition to the above-described purposes, features and advantages, the present invention has other purposes, features and advantages. The present invention will be further described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 A schematic diagram of the steps of a method for controlling the quality of roadbed compaction construction based on the dynamic response of a rammer provided in one embodiment of the present invention;

[0072] Figure 2 A schematic diagram of a rammer equipped with an acceleration sensor provided in one embodiment of the present invention;

[0073] Figure 3 This is a schematic diagram of an acceleration signal collected by the acceleration sensor provided by the present invention. DETAILED DESCRIPTION

[0074] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0075] See also Figure 1 and Figure 2 The present invention provides a method for controlling the quality of roadbed compaction construction based on the dynamic response of a rammer, and the control method comprises the following steps:

[0076] Step S10, obtaining the acceleration signal of the rammer during the current rammering process.

[0077] In the present invention, the acceleration signal is collected by the acceleration sensor 102 installed on the rammer 101 .

[0078] In the present invention, the number of acceleration sensors installed on the rammer may be 1, 2, 3 or any other number.

[0079] Preferably, there are multiple acceleration sensors 20 , and the multiple acceleration sensors 20 are evenly distributed around the central axis of the rammer 10 .

[0080] More preferably, the distance between each acceleration sensor 20 and the central axis of the rammer 10 is the same as the distance between each acceleration sensor 20 and the outer contour of the rammer 10 .

[0081] Please refer to Figure 2 , Figure 2 A top view of a rammer provided by the present invention with an acceleration sensor installed. In the present invention, the number of the acceleration sensors is four, which are evenly distributed around the central axis of the rammer, and the distances between the four acceleration sensors and the central axis of the rammer are the same as the distances between the four acceleration sensors and the outer contour of the rammer.

[0082] In the present invention, four acceleration sensors are symmetrically installed on the rammer, and the installation positions are located at the symmetrical points of the rammer, so as to comprehensively monitor the acceleration changes of the rammer.

[0083] It should be noted that in order to avoid signal distortion or delay, the connection between the acceleration sensor and the rammer should be firm during installation, and the acceleration sensor layout point can effectively avoid external interference. In this example, each acceleration sensor is installed by bolt fixing, and the collected acceleration signal is transmitted to the controller in the cab in real time through the wireless data transmission module.

[0084] It can be understood that the number of acceleration sensors is four, and the number of acquired acceleration signals is also four. In the following, after the acceleration signals are processed, the data used for analysis is the average value of the four data processing results.

[0085] Preferably, the acceleration sensor has high sensitivity and a wide frequency response range, such as a sensitivity of 100 mV / g and a frequency response range of 0.1 Hz to 20000 Hz.

[0086] Step S20, performing data processing on the acceleration signal of the rammer to obtain multiple key parameters for predicting the dynamic deformation modulus, the data processing includes filtering processing, time domain analysis and frequency domain analysis performed in sequence, and the multiple key parameters include acceleration peak value, rammer action time, acceleration kurtosis and acceleration main frequency.

[0087] In the present invention, the data processing includes the following steps:

[0088] Step (1), filtering and noise reduction processing of the acceleration signal of the rammer.

[0089] In the present invention, the filtering and noise reduction processing is specifically: using a low-pass filtering method to filter and reduce noise on the acceleration signal, and the filter cutoff frequency is set to 500 Hz to remove high-frequency noise and external interference signals.

[0090] Step (2): based on the acceleration signal after filtering and noise reduction, obtain the calculation data interval of the current impact acceleration, wherein the calculation data interval is t 0 Time to t 1 The data interval corresponding to the time, t 0 Time and t 1 The acceleration value corresponding to the moment is 0 and t 0 Time to t 1 At this moment, the acceleration value increases from 0, exceeds the peak value, and then decreases to 0.

[0091] See also Figure 3 , Figure 3 The figure shows the acceleration signal of the tamping hammer corresponding to multiple tamping points (after filtering and noise reduction). Figure 3 The upper gray filled area corresponds to the calculated data interval.

[0092] When a plurality of acceleration signals are acquired, the method further includes acquiring an average acceleration signal of the plurality of acceleration signals and using the average acceleration signal as basic data for acquiring a calculation data interval in step (b).

[0093] Take the present invention as an example with four acceleration sensors to calculate the average value of the four signals. As the final rammer acceleration value, step (2) obtains the calculation data interval based on the final rammer acceleration value, that is, the acceleration signal after filtering and noise reduction is the average acceleration signal of the four acceleration signals after filtering and noise reduction.

[0094] Preferably, there are multiple acceleration sensors, and the acceleration signal after filtering and noise reduction processing is an average acceleration signal.

[0095] Step (3), performing time domain analysis and frequency domain analysis on the data set corresponding to the calculation data interval of the current impact acceleration, and obtaining the acceleration peak value a according to multiple calculation formulas m , tamping hammer action time t m , acceleration kurtosis u, acceleration main frequency f m ,in,

[0096] Get the peak acceleration a m The calculation formula is:

[0097] in, is the acceleration time domain data, and j represents the jth acceleration value in the calculated data interval.

[0098] In the present invention, Real-time data representing the average signal of multiple accelerometers.

[0099] Get the rammer action time t m The calculation formula is: m =t 1 -t 0

[0100] Among them, t 0 and t 1 are the starting time point and ending time point of the calculation data interval respectively.

[0101] The calculation formula for obtaining the kurtosis u is:

[0102] Where n represents the amount of acceleration data in the calculation interval, is the acceleration time domain data, represents the mean of the calculation interval, and s represents the variance of the calculation interval.

[0103] In the present invention, the calculation formulas for calculating the mean of the calculation interval and the calculation interval variance are as follows:

[0104]

[0105] Get the main frequency f m The calculation formula is: m ×argmax P(A(fk )).

[0106] Among them, A(f k ) is the acceleration time domain data The frequency domain signal obtained by Fourier transform, P(A(f k )) is used to calculate the power spectral density of the frequency domain signal, argmaxP(A(f k )) is the frequency corresponding to the maximum value of the power spectral density.

[0107] Step S30, based on the multiple key parameters and a predetermined dynamic deformation modulus prediction model, the dynamic deformation modulus of the tamping point that completes the current tamping is obtained, and the dynamic deformation model prediction model is a multiple linear regression model.

[0108] Preferably, the dynamic deformation modulus prediction model is:

[0109] E vd =β 0 +β 1 a m +β 2 t m +β 3 u+β 4 f m

[0110] Among them, E vd is the dynamic deformation modulus, a m is the peak acceleration, t m is the tamping hammer action time, u is the kurtosis, f m is the main frequency, β 0 ~β 4 is the fitting coefficient.

[0111] It should be noted that in the dynamic deformation modulus prediction model, β 0 ~β 4 is a specific value.

[0112] The present invention can obtain the dynamic deformation modulus of the tamping point after each tamping in real time through a dynamic deformation modulus prediction model, thereby reducing the cost of later quality inspection, improving construction quality and efficiency, and ensuring the overall uniformity of roadbed tamping reinforcement.

[0113] The multiple key parameters obtained in step S20 are substituted into the dynamic deformation modulus prediction model to calculate the dynamic deformation modulus of the tamping point that has completed the current tamping. The dynamic deformation model can be used to judge the compaction quality of the tamping point after the current tamping. If it does not meet the requirements, the tamping point needs to be tamped. If it meets the requirements, the construction of the tamping point is completed.

[0114] The method for constructing the predetermined dynamic deformation modulus prediction model comprises:

[0115] Step 1) perform multiple tamping tests at multiple test tamping points respectively, obtain collected data of each tamping test at each test tamping point, and obtain multiple groups of collected data, each group of the collected data includes a dynamic deformation modulus measurement value and an acceleration signal of the tamping hammer, and the dynamic deformation modulus measurement value and the acceleration signal of the tamping hammer in the same group of collected data have a corresponding relationship, and the number of collected data is the sum of the number of tamping times of all test tamping points.

[0116] The number of test points is determined based on the number of samples. In principle, the more test points, the better. This way, more groups of data are collected, and the subsequent model accuracy will be higher.

[0117] In the present invention, the number of test tamping points is 10 to 20.

[0118] In the present invention, each test tamping point is subjected to 10 to 15 tamping tests.

[0119] In this way, the number of groups of collected data is 100 to 300 groups.

[0120] It can be understood that when there are multiple acceleration sensors on the rammer, there are also multiple acceleration signals of the rammer in each set of collected data.

[0121] In the present invention, 10 tamping points are selected for tamping test, each tamping point is tamped 15 times, the weight of the tamping hammer is 20 tons, and the drop distance of each tamping is 1.5m; the number of acceleration sensors is 4, evenly distributed, and the frequency is set to 20000Hz.

[0122] In the present invention, after each tamping point is completed, use E vd The dynamic modulus tester measures the dynamic deformation modulus of the soil. The acceleration signal of each tamping point corresponds to the dynamic deformation modulus value measured after the tamping of the tamping point, providing a basis for the subsequent prediction model construction.

[0123] Step 2) performing data processing on the acceleration signal of each rammer in the collected data to obtain a plurality of factor groups, each of which includes acceleration peak value, rammer action time, acceleration kurtosis and acceleration main frequency.

[0124] The data processing in this step is exactly the same as that in step S20 and will not be described again here.

[0125] Step 3) combining the dynamic deformation modulus measurement value and the corresponding factor group in each of the collected data into one sample data to obtain a plurality of sample data.

[0126] In the present invention, each sample data includes a dynamic deformation modulus measurement value, and an acceleration peak value, a ram action time, an acceleration kurtosis and an acceleration main frequency corresponding to the dynamic deformation modulus measurement value.

[0127] For the sake of ease of understanding, let's take an example. Assume that the dynamic deformation modulus measured after the first tamping point is completed for the first time is M. After data processing of the acceleration signal obtained by the first tamping point for the first time, the obtained acceleration peak value, tamping hammer action time, acceleration kurtosis and acceleration main frequency are N1, N2, N3 and N4 respectively. Then, M and N1, N2, N3 and N4 together constitute a sample data.

[0128] Step 4) randomly extracting a first preset number of sample data from all the sample data as a sample set, and using the sample set to calculate the coefficients of the initial multivariate regression model to determine the multivariate regression model coefficients.

[0129] Preferably, the first preset number is 1 / 2 to 4 / 5 of the total amount of sample data.

[0130] In this embodiment, the total amount of sample data is 150, and the first preset number is 75-120.

[0131] The step of calculating the coefficients of the initial multivariate regression model using the sample set to determine the coefficients of the multivariate regression model comprises:

[0132] Step (a), substituting the l sample data in the school-based set into the initial multivariate regression model, the resulting model form is:

[0133] y i =β 0 +β 1 x i +β 2 x i2 +β 3 x i3 +β 4 x i4 +ε i (i=1, 2, ..., l)

[0134] Where: i represents the error term, β 0 ~β 4 is the corresponding fitting coefficient, and l represents the sample size.

[0135] It should be noted that in the model form, β 0 ~β 4 is the independent variable that needs to be solved.

[0136] In the present invention, the characteristic value acceleration peak a is selectedm , tamping hammer action time t m , acceleration kurtosis u and acceleration main frequency f m As the independent variable x i1 ~x i4 , dynamic deformation modulus E vd As the dependent variable y i , construct the initial multiple linear regression model.

[0137] Step (b), written in expanded form according to the model form:

[0138]

[0139] Among them: Y is the dependent variable matrix, X is the independent variable matrix, and β is the coefficient matrix.

[0140] It should be noted that, in this expansion form, Y and X are known numbers, and β is an unknown number to be solved.

[0141] Step (c), calculating the residual sum of squares according to a preset formula, the preset formula is:

[0142]

[0143] Among them, Y T is the transposed matrix of the dependent variable matrix Y, X T is the transposed matrix of the independent variable matrix X, β T is the transposed matrix of the coefficient matrix β.

[0144] Step (d), taking the derivative of the residual sum of squares to obtain an equation after taking the derivative of the residual sum of squares, wherein the equation is:

[0145]

[0146] Step (e), based on the residual sum of squares being 0 and the equation after the residual sum of squares is derived, the coefficient matrix is ​​calculated to obtain the coefficients of the multivariate regression model.

[0147]

[0148] Step 5) Based on the multivariate regression model coefficients, obtain the target multivariate regression model.

[0149] Substitute the multivariate regression model coefficients determined in the previous step into the multivariate linear regression equation to obtain the target multivariate regression model, specifically:

[0150] E vd =β 0 +β 1 a m +β 2 t m +β3 u+β 4 f m

[0151] Among them, E vd is the dynamic deformation modulus, a m is the peak acceleration, t m is the tamping hammer action time, u is the kurtosis, f m is the main frequency, β 0 ~β 4 is the fitting coefficient.

[0152] It should be noted that the β 0 ~β 4 are four specific values, which may be related to β in the dynamic deformation modulus prediction model. 0 ~β 4 The same may also be different, depending on the verification results of the model prediction accuracy. When the prediction accuracy of the model meets the requirements, the β in the target multivariate regression model 0 ~β 4 and β in the dynamic deformation modulus prediction model 0 ~β 4 Similarly, when the prediction accuracy of the model does not meet the requirements, the β 0 ~β 4 and β in the dynamic deformation modulus prediction model 0 ~β 4 same.

[0153] Of course, in the present invention, the fitting coefficients can also be calculated based on experience using a sample set with sufficient data to ensure that the prediction accuracy of the target multivariate regression model meets the requirements. Therefore, the verification step of the model prediction accuracy is an optional step.

[0154] Step 6) Verify the prediction accuracy of the target multivariate regression model.

[0155] This step specifically includes:

[0156] Step 6-1: randomly select a second preset number of sample data from all sample data as a validation set.

[0157] In the present invention, the second preset number is 1 to 10, and preferably, the second preset number is 5.

[0158] Step 6-2: Substitute each factor group in the validation set into the target multivariate regression model to obtain a plurality of dynamic deformation modulus prediction values ​​corresponding to each factor group.

[0159] Step 6-3: Calculate the prediction accuracy based on the dynamic deformation modulus measurement value and the dynamic deformation modulus prediction value corresponding to each factor group to obtain multiple prediction accuracy values.

[0160] In the present invention, the calculation formula of the prediction accuracy is: absolute value of the difference between the dynamic deformation modulus measured value and the dynamic deformation modulus predicted value / dynamic deformation modulus measured value*100%.

[0161] Step 6-4: Obtain the average of multiple prediction accuracy values ​​and use it as the prediction accuracy of the target multivariate regression model.

[0162] Assuming that the second preset number is 5, a total of five prediction accuracy values ​​are obtained in the previous step, namely A1, A2, A3, A4 and A5. The sum of A1, A2, A3, A4 and A5 is divided by 5 to obtain the average value of the multiple prediction accuracy values.

[0163] Step 6-5: compare the prediction accuracy of the target multivariate regression model with a preset threshold. When the prediction accuracy of the target multivariate regression model is less than or equal to the preset threshold, the verified target multivariate regression model meets the requirements and is used as the dynamic deformation modulus prediction model.

[0164] When the result of the model verification process is that the prediction accuracy of the target multivariate regression model is greater than a preset threshold, the construction method further includes:

[0165] A third preset number of sample data is randomly selected from all the sample data as a sample set, and the sample set is used to calculate the coefficients of the initial multivariate regression model to determine new multivariate regression model coefficients, wherein the third preset number is greater than the first preset number, and the difference between the third preset number and the first preset number is between 10 and 20.

[0166] Based on the new multivariate regression model coefficients, a new target multivariate regression model is obtained.

[0167] The new target multivariate regression model was validated.

[0168] For ease of understanding, an example is given, assuming that the first preset number is 75 and the second preset number is 5. First, 75 sample data are randomly selected from all the sample data as the sample set, and the dynamic deformation modulus measurement value in the sample data is used as the Y value, and the acceleration peak value, the rammer action time, the acceleration kurtosis and the acceleration main frequency are substituted into the initial multivariate regression model as the X value to obtain 75 equations, and the four coefficients in the initial multivariate regression model are determined by calculating the residual sum of squares and the derivative of the residual sum of squares. The four coefficients are substituted into the initial multivariate regression model to obtain the target multivariate regression model; then 5 sample data are randomly selected from all the sample data as the validation set, and the acceleration peak value, the rammer action time, the acceleration kurtosis and the acceleration main frequency in the sample data of the validation set are substituted into the target multivariate regression model as the X value to obtain the dynamic deformation modulus prediction value, and then according to The dynamic deformation modulus measurement value and the dynamic deformation modulus prediction value in the sample data of the validation set are used to calculate the model prediction accuracy value, and the average of the 5 model prediction accuracy values ​​is obtained and used as the prediction accuracy of the target multivariate regression model. The prediction accuracy of the target multivariate regression model is compared with the preset threshold. If it is less than or equal to the preset threshold, the requirement is met and the target multivariate regression model is used as the dynamic deformation modulus prediction model. If it is greater than the preset threshold, 85 sample data are randomly selected from all the sample data as a new sample set, and the new sample set is used to determine the four coefficients in the initial multivariate regression model, and a new target multivariate regression model is obtained. The new target multivariate regression model is then verified to determine whether it meets the requirements. If it still does not meet the requirements, the number of new sample sets is increased to 95, ... until the verified target multivariate regression model meets the requirements.

[0169] Step S40, compare the dynamic deformation modulus of the tamping point with the set dynamic deformation modulus design value, and when the dynamic deformation modulus of the tamping point is greater than or equal to the dynamic deformation modulus design value, send a prompt to stop tamping; when the dynamic deformation modulus of the tamping point is less than the dynamic deformation modulus design value, send a prompt to continue tamping, and repeat steps S10 to S30.

[0170] In this example, during the construction process, the dynamic deformation modulus design value is set is 80MPa, if the predicted value of the dynamic deformation modulus of the tamping point is higher than or equal to If the predicted value of the dynamic deformation modulus of the tamping point is lower than When the driver receives a prompt to continue tamping, he needs to control the tamping hammer to continue tamping until the design requirements are met.

[0171] See also Figure 2The present invention also provides a roadbed tamping construction quality control system based on the dynamic response of the tamping hammer, the control system comprising:

[0172] Rammer 101, used for performing strong tamping operation on the roadbed;

[0173] A plurality of acceleration sensors 102, used for collecting acceleration signals of the rammer;

[0174] A controller is communicatively connected to the plurality of acceleration sensors 102 , and is used to execute the control method described in steps S10 to S40 above.

[0175] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the controller described above can refer to the contents in the aforementioned method embodiment and will not be repeated here.

[0176] Preferably, a plurality of acceleration sensors 102 are mounted on the rammer 101 and are evenly distributed around the central axis of the rammer 101 , and the distance from each acceleration sensor 102 to the central axis of the rammer 101 and the distance to the contour line of the rammer 101 are the same.

[0177] In the present invention, the arrangement point of the acceleration sensor 102 can effectively avoid external interference, thereby avoiding signal distortion or delay.

[0178] In the present invention, the number of acceleration sensors 102 is four.

[0179] In the present invention, the acceleration sensor 102 and the rammer 101 are fixed by bolts.

[0180] Preferably, the controller is installed in the cab.

[0181] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions and substitutions can be made without departing from the concept of the present invention, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A method for controlling the quality of roadbed compaction construction based on the dynamic response of a rammer, characterized in that: The control method comprises the following steps: Step S10, obtaining an acceleration signal of the rammer during the current rammering process; Step S20, performing data processing on the acceleration signal of the rammer to obtain a plurality of key parameters for predicting the dynamic deformation modulus, wherein the data processing comprises filtering processing, time domain analysis and frequency domain analysis performed in sequence, and the plurality of key parameters comprises acceleration peak value, rammer action time, acceleration kurtosis and acceleration main frequency; Step S30, based on the multiple key parameters and a predetermined dynamic deformation modulus prediction model, obtaining the dynamic deformation modulus of the tamping point after the current tamping, wherein the dynamic deformation model prediction model is a multiple linear regression model; Step S40, compare the dynamic deformation modulus of the tamping point with the set dynamic deformation modulus design value, and when the dynamic deformation modulus of the tamping point is greater than or equal to the dynamic deformation modulus design value, send a prompt to stop tamping; when the dynamic deformation modulus of the tamping point is less than the dynamic deformation modulus design value, send a prompt to continue tamping, and repeat steps S10 to S30.

2. The roadbed tamping construction quality control method based on the tamping hammer dynamic response according to claim 1 is characterized in that: The dynamic deformation model prediction model is: E vd =β0+β1a m +β2t m +β3u+β4f m Among them, E vd is the dynamic deformation modulus, a m is the peak acceleration, t m is the tamping hammer action time, u is the acceleration peak, f m is the main frequency of acceleration, β0~β4 are fitting coefficients.

3. The roadbed tamping construction quality control method based on the tamping hammer dynamic response according to claim 2 is characterized in that: A method for constructing a predetermined dynamic deformation modulus prediction model, comprising: Perform multiple tamping tests at multiple test tamping points respectively, obtain collected data of each tamping test at each test tamping point, and obtain multiple groups of collected data, each group of the collected data includes a dynamic deformation modulus measurement value and an acceleration signal of the tamping hammer, and the dynamic deformation modulus measurement value and the acceleration signal of the tamping hammer in the same group of collected data have a corresponding relationship, and the number of the collected data is the sum of the number of tamping times of all the test tamping points; Performing data processing on the acceleration sampling signal of the rammer in each of the collected data to obtain a plurality of factor groups, each of the factor groups including acceleration peak value, rammer action time, acceleration kurtosis and acceleration main frequency; Combining the dynamic deformation modulus measurement value and the corresponding factor group in each of the collected data into one sample data to obtain a plurality of sample data; Randomly extracting a first preset number of sample data from all sample data as a sample set, and using the sample set to calculate coefficients of an initial multivariate regression model to determine the multivariate regression model coefficients; Based on the multivariate regression model coefficients, obtain the target multivariate regression model.

4. The roadbed tamping construction quality control method based on the tamping hammer dynamic response according to claim 3 is characterized in that: The method for constructing the predetermined dynamic deformation modulus prediction model also includes a model verification process: Randomly select a second preset number of sample data from all sample data as a validation set; Substituting each factor group in the validation set into the target multivariate regression model to obtain a plurality of dynamic deformation modulus prediction values ​​corresponding to each factor group; Based on the dynamic deformation modulus measurement value and the dynamic deformation modulus prediction value corresponding to each factor group, the model prediction accuracy is calculated to obtain multiple model prediction accuracy values; Get the average of the prediction accuracy values ​​of multiple models and use it as the prediction accuracy of the target multivariate regression model; The prediction accuracy of the target multivariate regression model is compared with a preset threshold. When the prediction accuracy of the target multivariate regression model is less than or equal to the preset threshold, the verified target multivariate regression model meets the requirements, and the target multivariate regression model is used as the dynamic deformation modulus prediction model.

5. The method for controlling the quality of roadbed compaction construction based on the dynamic response of the rammer according to claim 4 is characterized in that: When the prediction accuracy of the target multivariate regression model is greater than a preset threshold, the construction method further includes: Randomly extracting a third preset number of sample data from all sample data as a sample set, and using the sample set to calculate coefficients of the initial multivariate regression model to determine new multivariate regression model coefficients, wherein the third preset number is greater than the first preset number, and the difference between the third preset number and the first preset number is between 10 and 20; Based on the new multivariate regression model coefficients, a new target multivariate regression model is obtained; The new target multivariate regression model was validated.

6. The method for controlling the quality of roadbed compaction construction based on the dynamic response of the rammer according to any one of claims 3 or 5, characterized in that: The data processing includes: Step (1), filtering and noise reduction processing of the acceleration signal of the rammer; Step (2), based on the acceleration signal after filtering and noise reduction, obtaining the calculation data interval of the current impact acceleration, the calculation data interval is the data interval corresponding to the time t0 to the time t1, the acceleration values ​​corresponding to the time t0 and the time t1 are 0, and from the time t0 to the time t1, the acceleration value increases from 0 to exceed the peak value and then decreases to 0; Step (3), performing time domain analysis and frequency domain analysis on the data set corresponding to the calculation data interval of the current impact acceleration, and obtaining the acceleration peak value a according to multiple calculation formulas m , tamping hammer action time t m , acceleration kurtosis u, acceleration main frequency f m ,in: Get the peak acceleration a m The calculation formula is: in, is the acceleration time domain data, j represents the jth acceleration value in the calculation interval; Get the rammer action time t m The calculation formula is: m =t1-t0 Among them, t0 and t1 are the starting time point and the ending time point of the calculation data interval respectively; The calculation formula for obtaining the kurtosis u is: Among them, n represents the amount of acceleration data in the calculated data interval, is the acceleration time domain data, represents the mean of the calculated data interval, and s represents the variance of the calculated data interval; Get the main frequency f m The calculation formula is: m = argmax P(A(f k )); Among them, A(f k ) is the acceleration time domain data The frequency domain signal obtained by Fourier transform, P(A(f k )) is used to calculate the power spectral density of the frequency domain signal, argmaxP(A(f k )) is the frequency corresponding to the maximum value of the power spectral density.

7. The roadbed tamping construction quality control method based on the tamping hammer dynamic response according to claim 6 is characterized in that: The calculation formulas for calculating the mean of a data interval and the variance of a data interval are as follows: in, represents the mean of the calculated data interval, n represents the amount of acceleration data in the calculated interval, is the acceleration time domain data; Among them, s represents the variance of the calculated data interval, n represents the amount of acceleration data in the calculated data interval, is the acceleration time domain data, Represents the mean of the calculated data interval.

8. The roadbed tamping construction quality control method based on the tamping hammer dynamic response according to claim 6 is characterized in that: The step of calculating the coefficients of the initial multivariate regression model using the sample set to determine the coefficients of the multivariate regression model comprises: Step (a), substituting the l sample data in the school-based set into the initial multivariate regression model, the resulting model form is: y i =β0+β1x i1 +β2x i2 +β3x i3 +β4x i4 +e i (i=1,2,...,l) Where: i represents the error term, β0~β4 are the corresponding fitting coefficients, and l represents the number of samples; Step (b), written in expanded form according to the model form: Where: Y is the dependent variable matrix, X is the independent variable matrix, and β is the coefficient matrix; Step (c), calculating the residual sum of squares according to a preset formula, the preset formula is: Among them, Y T is the transposed matrix of the dependent variable matrix Y, X T is the transposed matrix of the independent variable matrix X, β T is the transposed matrix of the coefficient matrix β; Step (d), taking the derivative of the residual sum of squares to obtain an equation after taking the derivative of the residual sum of squares, wherein the equation is: Step (e), based on the residual sum of squares being 0 and the equation after the residual sum of squares is derived, the coefficient matrix is ​​calculated to obtain the coefficients of the multivariate regression model.

9. A roadbed tamping construction quality control system based on the dynamic response of the tamping hammer, characterized in that: The control system comprises: Rammer, used for strong tamping of roadbed; A plurality of acceleration sensors, used for collecting acceleration signals of the rammer; A controller is communicatively connected to the plurality of acceleration sensors respectively, and the controller is used to execute the control method described in steps S10 to S40 of claim 1.

10. The roadbed compaction construction quality control system based on the rammer dynamic response according to claim 9, characterized in that: A plurality of acceleration sensors are mounted on the rammer and are evenly distributed around the central axis of the rammer. The distance between each acceleration sensor and the central axis of the rammer is the same as the distance between each acceleration sensor and the contour line of the rammer.

Images