Foam light soil strength detection device and prediction method

Abstract

The application discloses a kind of foamed light soil strength detection device and prediction method, the prediction method includes: preparation including the experimental group set of foamed light soil cylinder and cubic test block, after applying environmental erosion effect to experimental group set, cylinder test block is tested by radial compressive strength testing device, obtains radial compressive strength test data, cubic test block is tested by electro-hydraulic pressure testing machine, obtains overall compressive strength test data, with the setting parameter of environmental erosion effect and radial compressive strength test data as input, overall compressive strength test data as output, train random forest model, obtain the mapping relationship model between radial compressive strength and overall compressive strength, carry out drilling to foamed light soil roadbed site, and the radial compressive strength of different positions at different times is measured in hole site by radial compressive strength testing device, further predict the compressive strength of foamed light soil roadbed site by the mapping relationship model.

Description

Technical Field

[0001] This invention relates to the field of foamed lightweight soil testing, and more particularly to a foamed lightweight soil strength testing device and prediction method. Background Technology

[0002] Foamed lightweight soil, as a novel functional material, possesses core characteristics including lightweight, high fluid plasticity, microporous structural stability, and low carbon footprint, and has been widely applied in roadbed engineering. However, the long-term service performance of foamed lightweight soil roadbeds faces severe challenges from the coupled effects of complex natural environments, specifically: freeze-thaw cycle damage in seasonally frozen zones, wet-dry cycle damage in humid and hot environments, and chemical erosion, thus shortening the service life of foamed lightweight soil. Therefore, accurately understanding the service status of foamed lightweight soil roadbeds under complex environmental conditions is of great significance for ensuring traffic safety, preventing sudden disasters, extending service life, and reducing maintenance costs.

[0003] Traditional methods for monitoring the service condition of foamed lightweight soil subgrades primarily rely on structural loading tests and laboratory examinations, such as falling weight deflectometers and manual localized borehole sampling. However, these traditional methods suffer from low levels of automation, long testing cycles, and high costs, and are susceptible to human error, leading to inaccurate monitoring results. Furthermore, due to limited sampling data, these methods cannot comprehensively characterize the overall condition of the subgrade, cannot achieve continuous monitoring, and therefore cannot provide timely warnings of progressive disasters or detect sudden disasters, thus affecting the safety and long-term service performance of the subgrade.

[0004] To overcome the above problems, this invention provides a foamed lightweight soil strength testing device and prediction method. By combining the testing device and machine learning methods, the strength of foamed lightweight soil is predicted, thereby detecting the radial compressive strength of foamed lightweight soil and predicting the overall compressive strength. This provides reliable technical support for ensuring the safety and long-term service performance of foamed lightweight soil subgrades. Summary of the Invention

[0005] The purpose of this invention is to provide a device and method for testing and predicting the strength of foamed lightweight soil.

[0006] To achieve the above objectives, the present invention is implemented according to the following technical solution:

[0007] The first aspect of this invention provides a method for predicting the strength of foamed lightweight soil, comprising the following steps:

[0008] Step S1: Prepare an experimental set including cylindrical and cubic specimens of foamed lightweight soil;

[0009] Step S2: Apply environmental erosion effects to the experimental group, test the cylindrical specimens with a radial compressive strength testing device to obtain radial compressive strength test data, and test the cubic specimens with a pressure testing machine to obtain overall compressive strength test data;

[0010] Step S3: Train a random forest model using parameters of environmental erosion effects and radial compressive strength test data as inputs and overall compressive strength test data as outputs to obtain a mapping relationship model;

[0011] Step S4: Drill holes in the foamed lightweight soil subgrade at the site, and measure the radial compressive strength at the hole location using a radial compressive strength testing device. Then, predict the overall compressive strength of the foamed lightweight soil subgrade at the site using the mapping relationship model.

[0012] Preferably, the method for applying environmental erosion effects to the experimental group set in step S2 is to apply environmental erosion effects of the same intensity but different erosion types to different experimental groups.

[0013] Preferably, the method steps in step S3 are as follows:

[0014] S31 collects radial compressive strength test data, overall compressive strength test data, and environmental erosion effect setting parameters as a dataset. After cleaning the dataset, variational mode decomposition processing is performed.

[0015] S32 constructs features from the dataset after variational mode decomposition and inputs them into the random forest model for training.

[0016] S33 determines whether the random forest model has completed training. If not, it returns to S32 until training is complete.

[0017] Preferably, the method steps for performing variational mode decomposition in S31 are as follows:

[0018] S311 minimizes the bandwidth of each mode through an objective function, the expression of which is:

[0019]

[0020] In the formula, and The first The first mode and the first The frequency corresponding to each mode The total number of modes, For time, For the first The time-varying amplitude function of each mode, It is a natural constant. The imaginary unit, As a regularization factor;

[0021] in,

[0022]

[0023] In the formula, For the first Each mode in Time-varying amplitude, for The time-varying phase of time;

[0024] S312 introduces the objective function into the optimization process using the Lagrange multiplier method, expressed as:

[0025]

[0026] In the formula, The original time series data, These are Lagrange multiplier functions;

[0027] in,

[0028]

[0029]

[0030] In the formula, The total number of sampling points. For the first The sampled values ​​of each sampling point For the Dirac function, In the first The sampling time for each sampling point As the initial value, For the Lagrange multiplier function in The derivative at time, For integration variables;

[0031] S313 iteratively updates the modes and frequencies using the alternating direction method, expressed as:

[0032]

[0033]

[0034] In the formula, The total duration of the signal. Let be the integral variable of the signal duration, representing the time in the signal. Any point in time within;

[0035] S314 When the modal change is less than the preset threshold or the preset maximum number of iterations is reached, stop the iterative update and obtain each mode and modal frequency.

[0036] Preferably, the method steps of S32 are as follows:

[0037] S321 constructs the dataset after variational mode decomposition into a dataset. ,in Indicates the first One input, express The corresponding compressive strength values ​​of foamed lightweight soil;

[0038] S322 dataset Dataset generated using the K-nearest neighbor algorithm , the dataset and dataset Merge to obtain the final dataset. , the dataset The data is fed into a random forest algorithm for training.

[0039] The S323 random forest algorithm integrates the stress resistance prediction results of each decision tree to obtain the final stress resistance prediction value, expressed as:

[0040]

[0041] In the formula, This is the predicted final compressive strength value. For the total number of decision trees, For the first The compressive strength value predicted by the decision tree.

[0042] Preferably, the method for determining whether the random forest model has completed training in S33 is as follows:

[0043] The performance of out-of-bag samples on the random forest model is calculated to evaluate the training state, expressed as:

[0044]

[0045] In the formula, For OOB error rate, The total number of samples outside the bag. For indicator functions, For the first Out-of-bag prediction results for a single sample. For the first The true label of each sample;

[0046] If the rate of change of the OOB error rate in several consecutive iterations is less than a preset threshold, then the random forest model is considered to have completed training.

[0047] Preferably, the method for drilling holes in the foamed lightweight soil subgrade in step S4 involves setting a drilling cross section every 100 to 200 meters along the longitudinal direction of the subgrade, and drilling holes in two directions at the top, middle, and bottom of each drilling cross section. Specifically, the two directions are:

[0048] Start drilling from one side slope of the roadbed and drill to the centerline of the roadbed;

[0049] Start drilling from the other side of the roadbed slope and drill to the corresponding shoulder.

[0050] The second aspect of this invention provides a foamed lightweight soil strength testing device, specifically the radial compressive strength testing device described in the first aspect of this invention, comprising a workbench (1) and a base plate (8), wherein an oil tank (3) and a data acquisition, storage and processing platform (4) are fixed on the top of the workbench (1), and an oil pump (2) is provided on the top of the oil tank (3). The device is characterized in that...

[0051] The output end of the oil pump (2) is connected to an oil pipe (6), and the end of the oil pipe (6) is connected to an oil bladder (7). A stress sensor (15) and a strain sensor (16) are fixed on the surface of the oil bladder (7).

[0052] Electric push rods (9) and second end caps (11) are respectively installed on both sides of the top of the base plate (8). The output end of the electric push rod (9) is fixedly connected to the first end cap (10). A model groove (14) is provided between the first end cap (10) and the second end cap (11). The inner cavity of the model groove (14) is used to place a cylindrical test block (13).

[0053] The oil pipe (6) transmits the pressure output by the oil pump (2) to the oil bladder (7), which is placed inside the cylindrical test block (13) and applies pressure evenly to the inner wall through expansion;

[0054] The data acquisition, storage and processing platform (4) is used to acquire data from the stress sensor (15) and strain sensor (16).

[0055] Preferably, the electric push rod (9) is used to push the first end cap (10) toward the cylindrical test block (13), and the first end cap (10) and the second end cap (11) are used to block the through part of the cylindrical test block (13) to form a sealing structure.

[0056] Preferably, the first end cap (10) is provided with a groove (12) for fixing the end of the oil pipe (6).

[0057] The beneficial effects of this invention are:

[0058] This invention provides a strength testing device and prediction method for foamed lightweight soil. The strength testing device features a reasonable mechanical structure, simple and convenient operation, and a high degree of intelligent testing. The prediction model in the strength prediction method considers the mapping relationship between the radial compressive strength and compressive strength of foamed lightweight soil under complex environmental factors. Applying the foamed lightweight soil strength testing equipment and prediction method of this invention to the field can predict the service condition of foamed lightweight soil subgrades. This method is highly efficient, low-cost, and has small errors, enabling intelligent and continuous monitoring of foamed lightweight soil subgrades, providing a scientific basis for subgrade safety assessment and maintenance. Attached Figure Description

[0059] Figure 1 A flowchart of a method for predicting the strength of foamed lightweight soil provided by the present invention;

[0060] Figure 2 This is a flowchart of training a random forest model in an embodiment of the present invention;

[0061] Figure 3 This is a diagram showing the layout of boreholes for the foamed lightweight soil subgrade in an embodiment of the present invention.

[0062] Figure 4 A schematic diagram of the structure of a foamed lightweight soil strength testing device provided by the present invention;

[0063] Figure 5 for Figure 4 Enlarged schematic diagram of the oil bladder structure at point A;

[0064] Figure 6 for Figure 4 Enlarged schematic diagram of the first cap at point B.

[0065] Reference numerals: 1-Workbench, 2-Oil pump, 3-Oil tank, 4-Data acquisition, storage and processing platform, 5-Indicator, 6-Oil pipe, 7-Oil bladder, 8-Base plate, 9-Electric actuator, 10-First end cap, 11-Second end cap, 12-Groove, 13-Cylindrical foam lightweight soil test block, 14-Model groove, 15-Stress sensor, 16-Strain sensor. Detailed Implementation

[0066] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0067] In this document, the terms "upper," "lower," "inner," "outer," "front," "rear," "one end," and "the other end," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the present invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0068] In this document, unless otherwise explicitly specified and limited, the terms "installed," "equipped with," "connected," etc., should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, a direct connection, or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0069] In this article, "multiple" means two or more, that is, it includes two, three, four, five, etc.

[0070] Example 1:

[0071] Reference Figure 1 , Figure 2 As shown, the first aspect of the present invention provides a method for predicting the strength of foamed lightweight soil, comprising the following steps:

[0072] Step S1: Prepare an experimental set including cylindrical and cubic specimens of foamed lightweight soil;

[0073] It should be understood that, due to the significant differences in mechanical behavior between cylindrical and cubic foamed lightweight soil specimens, the compressive strength test in subsequent steps can reveal the correlation of strength characteristics of foamed lightweight soil under different stress modes. Furthermore, by preparing multiple sets of samples under different environmental erosion parameters, repeated tests can be conducted to reduce random errors and avoid the randomness of single-sample test results.

[0074] In this embodiment, a cylindrical specimen of foamed lightweight soil with a diameter of 100 mm and a height of 100 mm was prepared, along with a specimen with a size of 100 mm. 100mm Seven environmental erosion groups were set up for 100mm foamed lightweight soil cube test blocks, including control group, single dry-wet action group, single freeze-thaw action group, single chemical action group, dry-wet-freeze-thaw action group, dry-wet-chemical action group, and multiple action group. The number of cylindrical and cubic test blocks in each environmental erosion group was 25.

[0075] Step S2: Apply environmental erosion effects to the experimental group, test the cylindrical specimens with a radial compressive strength testing device to obtain radial compressive strength test data, and test the cubic specimens with a pressure testing machine to obtain overall compressive strength test data;

[0076] In this embodiment, the same degree of wet-dry cycle, freeze-thaw cycle, and chemical erosion were applied to the test blocks in each environmental erosion group. However, the environmental erosion effects applied to different environmental erosion groups were different, as shown in Tables 1 and 2, which are the specific implementation schemes for the environmental erosion schemes and environmental erosion conditions of each test group, respectively:

[0077] Table 1. Effects of Environmental Erosion

[0078]

[0079] Table 2. Environmental erosion conditions for each experimental group

[0080]

[0081] It should be explained that by artificially applying environmental erosion (wet-dry cycle, freeze-thaw cycle, and chemical erosion), it can be used to simulate the long-term performance degradation process of foamed lightweight soil subgrade in the natural environment, so that the test data is close to the actual engineering situation, avoids the limitation of only testing the strength of the original material, and ensures that the subsequent correlation model can reflect the real mechanical performance of the subgrade in complex environment.

[0082] In this embodiment, a 20mm diameter hole was drilled at the center of the cylindrical foamed lightweight soil specimen after environmental erosion, penetrating completely through it. During drilling, a three-dimensional adjustable clamp was used to vertically fix the specimen on the drilling platform, and a rubber anti-slip pad was added to the bottom. A laser positioning instrument was used to calibrate the central axis, while ensuring that the drill bit coincided with the specimen's axis. A diamond drill bit was used with a high-precision vertical drilling machine to drill at a speed of 200 rpm, penetrating 20mm to a depth from the upper surface of the cylindrical foamed lightweight soil to form a guide hole to prevent drill bit deviation. Further, the rotation speed was gradually increased to 400 rpm and maintained at 0.5 rpm. Feed at a constant speed of 1 mm / s, and withdraw the drill bit and clean up the debris to cool the drill bit every 30 mm until it reaches a depth of 140 mm; when it is close to penetrating, reduce the speed to 50 rpm to avoid chipping at the edge of the hole; after drilling is completed, use an internal wall grinder to finish the inner wall of the hole to remove burrs and micro-cracks.

[0083] The cube specimens were tested using a TYA-2000 electro-hydraulic pressure testing machine with the maximum test force set to 2000N. The cube specimens were subjected to standard loading according to the "Test Methods for Mechanical Properties of Ordinary Concrete" (GB50081-2002) to obtain the overall compressive strength of the cube specimens.

[0084] Finally, the radial compressive strength test data of cylindrical foamed lightweight soil specimens and the overall compressive strength test data of cubic specimens in the same group were compiled, totaling 350 groups.

[0085] Step S3: Train a random forest model using parameters of environmental erosion effects and radial compressive strength test data as inputs and overall compressive strength test data as outputs to obtain a mapping relationship model;

[0086] The specific steps of step S3 are as follows:

[0087] S31 collects radial compressive strength test data, overall compressive strength test data, and environmental erosion effect setting parameters as a dataset. After cleaning the dataset, variational mode decomposition processing is performed.

[0088] S32 constructs features from the dataset after variational mode decomposition and inputs them into the random forest model for training.

[0089] S33 determines whether the random forest model has completed training. If not, it returns to S32 until training is complete.

[0090] In some embodiments, the method steps for performing variational mode decomposition processing described in S31 are specifically as follows:

[0091] S311 minimizes the bandwidth of each mode through an objective function, the expression of which is:

[0092]

[0093] In the formula, and The first The first mode and the first The frequency corresponding to each mode The total number of modes, For time, For the first The time-varying amplitude function of each mode, It is a natural constant. The imaginary unit, As a regularization factor;

[0094] in,

[0095]

[0096] In the formula, For the first Each mode in Time-varying amplitude, for The time-varying phase of time;

[0097] S312 introduces the objective function into the optimization process using the Lagrange multiplier method, expressed as:

[0098]

[0099] In the formula, The original time series data, These are Lagrange multiplier functions;

[0100] in,

[0101]

[0102]

[0103] In the formula, The total number of sampling points. For the first The sampled values ​​of each sampling point For the Dirac function, In the first The sampling time for each sampling point As the initial value, For the Lagrange multiplier function in The derivative at time, For integration variables;

[0104] S313 iteratively updates the modes and frequencies using the alternating direction method, expressed as:

[0105]

[0106]

[0107] In the formula, The total duration of the signal. Let be the integral variable of the signal duration, representing the time in the signal. Any point in time within;

[0108] S314 When the modal change is less than the preset threshold or the preset maximum number of iterations is reached, stop the iterative update and obtain each mode and modal frequency.

[0109] In some embodiments, the method steps of S32 are specifically as follows:

[0110] S321 constructs the dataset after variational mode decomposition into a dataset. ,in Indicates the first One input, express The corresponding compressive strength values ​​of foamed lightweight soil;

[0111] S322 dataset Dataset generated using the K-nearest neighbor algorithm , the dataset and dataset Merge to obtain the final dataset. , the dataset The data is fed into a random forest algorithm for training.

[0112] The S323 random forest algorithm integrates the stress resistance prediction results of each decision tree to obtain the final stress resistance prediction value, expressed as:

[0113]

[0114] In the formula, This is the predicted final compressive strength value. For the total number of decision trees, For the first The compressive strength value predicted by the decision tree.

[0115] In some embodiments, the method for determining whether the random forest model has completed training in S33 is as follows:

[0116] The performance of out-of-bag samples on the random forest model is calculated to evaluate the training state, expressed as:

[0117]

[0118] In the formula, For OOB error rate, The total number of samples outside the bag. For indicator functions, For the first Out-of-bag prediction results for a single sample. For the first The true label of each sample;

[0119] If the rate of change of the OOB error rate in several consecutive iterations is less than a preset threshold, then the random forest model is considered to have completed training.

[0120] Step S4: Drill holes in the foamed lightweight soil subgrade at the site, and measure the radial compressive strength at the hole location using a radial compressive strength testing device. Then, predict the overall compressive strength of the foamed lightweight soil subgrade at the site using the mapping relationship model.

[0121] like Figure 3 As shown, this is a schematic diagram of the layout of the strength detection device in this embodiment. A total of 6 monitoring points are set up on the cross section. Two boreholes are drilled in the upper, middle and lower parts of the roadbed cross section. The drilling directions are as follows: (1) Drilling from one side slope of the roadbed to the center line of the roadbed; (2) Drilling from the other side slope of the roadbed to the corresponding shoulder.

[0122] Specifically, holes drilled in different directions must have the same starting height and inclination angle.

[0123] Furthermore, when the radial compressive strength testing device is applied to the roadbed site, oil bladders are placed at corresponding positions on the roadbed and at the center of the roadbed, and the oil bladders are connected to other parts of the device through oil pipes. The other parts of the device are installed on both sides of the left and right slope toes of the roadbed.

[0124] In some embodiments, environmental variables at the roadbed site are directly obtained, specifically: historical freeze-thaw cycle counts are statistically analyzed using meteorological data; borehole sampling is used to analyze water content and groundwater salt ion concentration (Cl-, ...). Collect corrosive media from the roadbed at the site and test the pH value and concentration of corrosive ions;

[0125] The on-site environmental variables are then converted to dimensions consistent with the environmental erosion effect in step S2.

[0126] In some embodiments, the degree of erosion can also be inferred by obtaining material damage at the roadbed site, specifically by matching the loss index with a dimension consistent with the environmental erosion effect.

[0127] Furthermore, the overall compressive strength of the roadbed at the site is predicted based on the measured radial compressive strength and environmental erosion parameters.

[0128] Example 2:

[0129] Reference Figure 4 , Figure 5 and Figure 6As shown, the second aspect of the present invention also provides a foamed lightweight soil strength testing device, specifically the radial compressive strength testing device described in claim 1, comprising a workbench 1 and a base plate 8. An oil tank 3 and a data acquisition, storage, and processing platform 4 are fixed to the top of the workbench 1. An oil pump 2 is provided on the top of the oil tank 3. The oil pump 2 has an output end connected to an oil pipe 6, and an oil bladder 7 is connected to the end of the oil pipe 6. A stress sensor 15 and a strain sensor 16 are fixed to the surface of the oil bladder 7. Electric actuators 9 and second end caps 11 are respectively installed on both sides of the top of the base plate 8. A first end cap 10 is fixedly connected to the output end of the electric actuator 9. A model groove 14 is provided between the first end cap 10 and the second end cap 11. The inner side of the model groove 14 is used to place a cylindrical test block 13. The oil pipe 6 transmits the pressure output by the oil pump 2 to the oil bladder 7. The oil bladder 7 is placed inside the cylindrical test block 13, and the pressure is evenly applied to the inner wall through expansion. The data acquisition, storage, and processing platform 4 can acquire data from the stress sensor 15 and the strain sensor 16.

[0130] like Figure 4 and Figure 6 As shown, the electric actuator 9 is used to push the first end cap 10 toward the cylindrical test block 13. The first end cap 10 and the second end cap 11 are used to block the upper and lower parts of the through portion of the cylindrical test block 13, respectively, to form a sealing structure. The first end cap 10 is provided with a groove 12 for fixing the end of the oil pipe 6.

[0131] It should be explained that the workbench 1 provides a stable mounting platform for all components. The oil pump 2, as the power core, can control the pressure output and provide a continuous and stable pressure source to ensure that the cylindrical test block 13 is tested for radial compressive strength under controllable pressure, thus ensuring the reliability of the test data. The oil tank 3 can provide sufficient hydraulic oil for the oil pump 2, ensuring its continuous and stable operation. The data acquisition and storage platform 4 can acquire the stress and strain data of the cylindrical test block 13 in real time and can perform efficient processing and storage. The prompter 5 can issue an alarm sound when the acquired stress-strain data reaches the specified range for early warning.

[0132] The working principle of this device is as follows: Vaseline is applied to the inner wall of the model groove 14 to reduce friction. The cylindrical test block 13 to be tested is installed into the model groove 14, where the model groove 14 provides rigid constraint for the cylindrical test block 13. The oil bladder 7 is then placed inside the cylindrical test block 13. The electric actuator 9 is activated, pushing the first end cap 10 towards the cylindrical test block 13. Further, the end of the oil pipe 6 connected to the oil bladder 7 is engaged with the groove 12 on the first end cap 10, ensuring a tight fit between the first end cap 10 and the upper part of the hollow portion of the cylindrical test block 13. Simultaneously, the second end cap 11 is installed at the other end of the cylindrical test block 13, forming a sealing structure through the combined action of the first end cap 10 and the second end cap 11. The oil pump 2 on the top of the workbench 1 is turned on. The oil pump 2 draws hydraulic oil from the oil tank 3 and delivers it to the oil bladder 7 through the oil pipe 6. As the hydraulic oil is injected, the oil bladder 7 gradually expands. When the oil bladder 7 expands, it applies radial pressure evenly to the inner wall of the cylindrical specimen 13. During the pressure application process, stress sensor 15 and strain sensor 16 monitor stress and strain data in real time. The electrical signals from stress sensor 15 and strain sensor 16 are transmitted to data acquisition, storage, and processing platform 4 for display. After the detection is completed, oil pump 3 is turned off, and the hydraulic oil in oil bladder 7 flows back to oil tank 3 through oil pipe 6. The operator removes the first end cap 10 and the second end cap 11 from both ends of the cylindrical specimen 13.

[0133] In this embodiment, the cylindrical test block 13 has a length of 100mm, an outer diameter of 100mm, and an inner diameter of 20mm. The model groove 14 has an inner diameter of 101mm and an outer diameter of 120mm. The lengths of the first end cap 10 and the second end cap 11 are both 40mm. The pressure output rate of the oil pump 2 to the oil bladder 7 is as follows: (1) when the pressure of the oil bladder 7 is less than 1MPa, the rate is 0.03±0.005MPa / s; (2) when the pressure of the oil bladder 7 is not less than 1MPa, the rate is 0.05±0.005MPa / s. The data acquisition and storage processing platform 4 has a data acquisition and transmission frequency of not less than 5 times / s and a data storage capacity of not less than 200,000 sets. Among them, when the strain data acquired in real time by the signal acquisition and storage processing platform 4 is not less than 5 When the time comes, indicator 5 will emit a warning sound, and oil pump 2 will automatically stop outputting pressure.

[0134] The radial compressive strength of cylindrical foamed lightweight soil specimen 13 is:

[0135]

[0136] In the formula, The radial compressive strength of cylindrical foamed lightweight soil specimen 13 is given. The peak pressure generated by oil bladder 7 on the inner wall of cylindrical foam lightweight soil specimen 13. The contact area between the inner wall of the cylindrical test block 13 and the oil bladder 7 is 1256.6. .

[0137] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

Claims

1. A method of predicting the strength of a foamed lightweight soil, characterized by, Includes the following steps: Step S1: Prepare an experimental set including cylindrical and cubic specimens of foamed lightweight soil; Step S2: Apply environmental erosion effects to the experimental group, test the cylindrical specimens with a radial compressive strength testing device to obtain radial compressive strength test data, and test the cubic specimens with a pressure testing machine to obtain overall compressive strength test data; Step S3: Using parameters of environmental erosion effects and radial compressive strength test data as input, and overall compressive strength test data as output, train a random forest model to obtain a mapping relationship model, specifically as follows: S31 collects radial compressive strength test data, overall compressive strength test data, and environmental erosion effect setting parameters as a dataset. After cleaning the dataset, variational mode decomposition processing is performed. The method for variational mode decomposition is as follows: S311 minimizes the bandwidth of each mode through an objective function, the expression of which is: In the formula, and The first The first mode and the first The frequency corresponding to each mode The total number of modes, For time, For the first The time-varying amplitude function of each mode, It is a natural constant. The imaginary unit, As a regularization factor; in, In the formula, For the first Each mode in Time-varying amplitude, for The time-varying phase of time; S312 introduces the objective function into the optimization process using the Lagrange multiplier method, expressed as: In the formula, The original time series data, These are Lagrange multiplier functions; in, In the formula, The total number of sampling points. For the first The sampled values ​​of each sampling point For the Dirac function, In the first The sampling time for each sampling point As the initial value, For the Lagrange multiplier function in The derivative at time, For integration variables; S313 iteratively updates the modes and frequencies using the alternating direction method, expressed as: In the formula, The total duration of the signal. Let be the integral variable of the signal duration, representing the time in the signal. Any point in time within; S314 When the modal change is less than the preset threshold or the preset maximum number of iterations is reached, stop the iterative update and obtain each mode and modal frequency; S32 constructs features from the dataset after variational mode decomposition and inputs them into the random forest model for training; specifically: S321 constructs the dataset after variational mode decomposition into a dataset. ,in Indicates the first One input, express The corresponding compressive strength values ​​of foamed lightweight soil; S322 dataset Dataset generated using the K-nearest neighbor algorithm , the dataset and dataset Merge to obtain the final dataset. , the dataset The data is fed into a random forest algorithm for training. The S323 random forest algorithm integrates the stress resistance prediction results of each decision tree to obtain the final stress resistance prediction value, expressed as: In the formula, This is the predicted final compressive strength value. For the total number of decision trees, For the first The compressive strength value predicted by each decision tree; S33 determines whether the random forest model has completed training. If not, it returns to S32 until training is completed. Step S4: Drill holes in the foamed lightweight soil subgrade at the site, and measure the radial compressive strength at the hole location using a radial compressive strength testing device. Then, predict the overall compressive strength of the foamed lightweight soil subgrade at the site using the mapping relationship model.

2. The method for predicting the strength of foamed lightweight soil according to claim 1, characterized in that, The method for applying environmental erosion effects to the experimental group set in step S2 is to apply environmental erosion effects of the same intensity but different erosion types to different experimental groups.

3. The method for predicting the strength of foamed lightweight soil according to claim 1, characterized in that, The method for determining whether a random forest model has completed training, as described in S33, is as follows: The performance of out-of-bag samples on the random forest model is calculated to evaluate the training state, expressed as: In the formula, For OOB error rate, The total number of samples outside the bag. For indicator functions, For the first Out-of-bag prediction results for a single sample. For the first The true label of each sample; If the rate of change of the OOB error rate in several consecutive iterations is less than a preset threshold, then the random forest model is considered to have completed training.

4. The method for predicting the strength of foamed lightweight soil according to claim 1, characterized in that, The method for drilling holes in the foamed lightweight soil subgrade in step S4 involves setting up a borehole cross section every 100 to 200 meters along the longitudinal direction of the subgrade, and drilling holes in two directions at the top, middle, and bottom of each borehole cross section. Specifically, the two directions are: Start drilling from one side slope of the roadbed and drill to the centerline of the roadbed; Start drilling from the other side of the roadbed slope and drill to the corresponding shoulder.

5. A foamed lightweight soil strength testing device using the method described in claims 1 to 4, wherein the radial compressive strength testing device includes a workbench (1) and a base plate (8), an oil tank (3) and a data acquisition, storage and processing platform (4) are fixed on the top of the workbench (1), and an oil pump (2) is provided on the top of the oil tank (3), characterized in that, The output end of the oil pump (2) is connected to an oil pipe (6), and the end of the oil pipe (6) is connected to an oil bladder (7). A stress sensor (15) and a strain sensor (16) are fixed on the surface of the oil bladder (7). Electric push rods (9) and second end caps (11) are respectively installed on both sides of the top of the base plate (8). The output end of the electric push rod (9) is fixedly connected to the first end cap (10). A model groove (14) is provided between the first end cap (10) and the second end cap (11). The inner cavity of the model groove (14) is used to place a cylindrical test block (13). The oil pipe (6) transmits the pressure output by the oil pump (2) to the oil bladder (7), which is placed inside the cylindrical test block (13) and applies pressure evenly to the inner wall through expansion; The data acquisition, storage and processing platform (4) is used to acquire data from the stress sensor (15) and strain sensor (16).

6. The foamed lightweight soil strength testing device according to claim 5, characterized in that, The electric push rod (9) is used to push the first end cap (10) toward the cylindrical test block (13). The first end cap (10) and the second end cap (11) are used to block the through part of the cylindrical test block (13) to form a sealed structure.

7. The foamed lightweight soil strength testing device according to claim 5, characterized in that, The first end cap (10) is provided with a groove (12) for fixing the end of the oil pipe (6).

Images