Method for enhancing impermeability of composite solidified soil

Abstract

The invention provides a composite solidified soil anti-permeability performance enhancing method, and belongs to the technical field of soil solidification. Through particle screening or mixing treatment, a nano silicon dioxide modified curing agent solution is prepared, and the ion concentration and the pH value are controlled through a conductivity meter and a pH meter; ultrasonic oscillation equipment is adopted to assist mixing, a moisture content-compactness response matrix is established to dynamically adjust power density, molecular dynamics simulation software is utilized to construct a soil pore structure model, an evolution process is simulated through a Monte Carlo method, and a pore connectivity index is calculated to guide compaction treatment intensity adjustment. A multi-factor accelerated aging test system is established to regularly measure the change of the permeability coefficient and construct an attenuation matrix, and when the attenuation rate exceeds a threshold value, an early warning mechanism and a reinforcing treatment program are started, so that the technical problem that a systematic process parameter regulation and control method is lacked in the solidified soil preparation process is solved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of soil solidification, and in particular relates to a method for enhancing the anti-permeability of composite solidified soil. BACKGROUND

[0002] Solidified soil technology is an important foundation treatment method in geotechnical engineering. Traditional technology mainly uses inorganic cementitious materials such as cement and lime to mechanically mix with soil, and forms a cementing structure through hydration reaction of the solidifying agent to improve the engineering properties of the soil. In practical engineering applications, solidified soil is widely used in soft soil foundation treatment, slope stability, anti-seepage engineering, and contaminated soil remediation, and the preparation quality directly affects the long-term stability and safety of the project. However, the traditional solidified soil preparation technology mainly relies on experience to determine the mix proportion and process parameters, lacks quantitative analysis and systematic control of key factors such as soil particle size distribution, specific surface area, and pore structure, and cannot optimize the process according to the characteristics of different soil types. In the current engineering practice, due to the lack of systematic process parameter control method, the solidification effect is often uneven, and the anti-permeability is unstable, which seriously affects the engineering quality and use effect of the solidified soil. That is, there is a technical problem in the prior art that the solidified soil preparation process lacks a systematic process parameter control method. SUMMARY

[0003] Therefore, the present application provides a method for enhancing the anti-permeability of composite solidified soil, which can solve the technical problem in the prior art that the solidified soil preparation process lacks a systematic process parameter control method.

[0004] The application is implemented in the following manner: a composite solidified soil anti-permeability performance enhancement method is provided, particle size distribution of a soil sample is determined by a laser particle size analyzer, soil specific surface area data is obtained by a specific surface area tester, a particle size distribution-specific surface area correlation matrix is established, when the correlation matrix rank value is less than 3, particle screening processing is performed, when the matrix rank value is greater than 8, particle mixing processing is performed, a nano-SiO2 modified solidifying agent solution is prepared, the solidifying agent content is controlled at 5% to 15% of the soil mass, when the ion concentration exceeds 0.5 mol / L, the dilution ratio is adjusted, when the pH value deviates from the range of 7.0 to 9.0, a buffer is added for adjustment, the modified solidifying agent is mixed with the soil, ultrasonic oscillation equipment is used for auxiliary mixing for 60 to 120 minutes, the water content and compactness of the mixture are monitored in real time, a water content-compactness response matrix is established, when the matrix rank value change rate exceeds 20%, the ultrasonic power density is adjusted, a soil pore structure model is constructed by molecular dynamics simulation software, the pore structure evolution process is simulated by the Monte Carlo method, the pore connectivity index and pore size distribution parameters are calculated, the pore structure parameter matrix is decomposed into a pore morphology matrix, a connectivity matrix and a size distribution matrix, which are linear combinations of three low-rank matrices, when the pore connectivity index is less than 0.3, the tamping processing intensity is increased, the solidified soil sample is standardly cured, the curing temperature is controlled at 20±2℃, the relative humidity is kept above 95%, real-time monitoring is performed by a thermocouple temperature sensor and a humidity sensor, when the temperature deviation exceeds ±1℃ or the humidity is lower than 90%, an environmental regulation system is started, a multi-factor accelerated aging test system is established, three kinds of acceleration conditions of temperature cycle, wet-dry cycle and freeze-thaw cycle are set, the permeability coefficient is determined periodically by a permeability coefficient tester, a permeability coefficient-time decay matrix is constructed, when the decay rate exceeds 10% of the initial value, a performance early warning mechanism is triggered, at the same time, a reinforcement processing program is started and the curing parameters are adjusted.

[0005] The particle size distribution-specific surface area correlation matrix is established by matrix operation of the particle content in different particle size intervals and the corresponding specific surface area contribution value, and the matrix element represents the contribution degree of the particle size range to the total specific surface area.

[0006] The nano-SiO2 modified solidifying agent is prepared by dispersing nano-SiO2 particles in a cement-based solidifying agent, the nano-particles fill the soil pores and form chemical bonds with the soil particles, the ion concentration is obtained by the conductivity measurement principle of a conductivity meter, and the pH value is obtained by the potential measurement principle of a pH meter.

[0007] The ultrasonic oscillation device adopts an ultrasonic generator with a frequency of 20-40 kHz, and mechanical vibration is used to promote the uniform distribution of the curing agent in the soil.

[0008] The water content-density response matrix records the real-time correspondence between the water content and the density during the mixing process, and the matrix elements reflect the correlation strength between the two parameters.

[0009] The molecular dynamics simulation software adopts the LAMMPS package to establish an atomic-scale soil particle model, and the Monte Carlo method simulates the pore structure evolution process through random sampling statistics, and the pore size distribution parameters are obtained by the mercury injection method.

[0010] The pore connectivity index is defined as the ratio of connected pore volume to total pore volume, with a value range of 0-1, reflecting the connectivity of the pore network, and the pore volume data is obtained by CT scanning technology.

[0011] The multi-factor accelerated aging test system simulates various environmental effects that soil faces in actual engineering, shortens the evaluation period through accelerated testing, and the permeability coefficient is obtained by the constant head permeability test method.

[0012] The permeability coefficient-time decay matrix records the change law of the permeability coefficient at different ages, and the matrix row represents the time node and the list represents different test conditions.

[0013] It also includes a long-term performance prediction system based on an adaptive weight adjustment model, which dynamically adjusts model parameters using a hierarchical fusion weight adjustment function, and input parameters include cementation strength, compressive strength, elastic modulus, and Poisson's ratio data. The adaptive weight adjustment model is used to predict the evolution trend of impermeability performance within 50 years, and when the prediction result shows that the performance decay exceeds 30%, the corresponding maintenance strategy is developed.

[0014] The performance warning mechanism triggers a warning signal by setting a permeability coefficient change threshold, and automatically issues a warning when the permeability coefficient decay rate exceeds the set value. The reinforcement treatment program includes three treatment methods: increasing the curing agent content, increasing the curing temperature, and extending the curing time, and the corresponding treatment intensity is selected according to the warning level. The adjustment of the curing parameters includes increasing the curing temperature from the standard 20±2℃ to 25±2℃, increasing the relative humidity from 95% to more than 98%, extending the curing period to 1.5 times the original plan, and increasing the frequency of watering maintenance to every 12 hours.

[0015] The specific structure of the adaptive weight adjustment model is a multi-layer perceptron architecture, including an input layer, three hidden layers and an output layer, the input layer receives four parameters of cementing strength, compressive strength, elastic modulus and Poisson's ratio, the first hidden layer includes 64 neurons, the second hidden layer includes 32 neurons, the third hidden layer includes 16 neurons, and the output layer generates a long-term impermeability performance prediction value.

[0016] The hierarchical fusion weight adjustment function is calculated based on the rank value of the particle size distribution-specific surface area correlation matrix, the rank value of the water content-density response matrix and the pore connectivity index, when the fusion weight coefficient a is in [0, 0.3), the hierarchical fusion weight parameter of the adaptive weight adjustment model is adjusted in the form of reducing the weight of the first hidden layer, when the fusion weight coefficient a is in [0.3, 0.7], the weight of each hidden layer is balanced, and when the fusion weight coefficient a is in (0.7, 1.0], the weight of the third hidden layer is increased.

[0017] The present application realizes accurate regulation and control based on soil characteristics by establishing a particle size distribution-specific surface area correlation matrix and guiding particle screening or mixing treatment according to the rank value variation law of the matrix, and solves the problem that traditional methods cannot perform differential treatment for different soil types. The present application solves the problems of uneven dispersion of the curing agent and difficult control of the mixing effect by preparing a nano-silicon dioxide modified curing agent and controlling the ion concentration and pH value, combining with ultrasonic oscillation assisted mixing technology, establishing a water content-density response matrix dynamic adjustment process parameter, significantly improving the sufficiency and uniformity of the curing reaction. The present application realizes whole-process system regulation and control from structure characterization to process control by constructing a soil pore structure model through molecular dynamics simulation, simulating the pore evolution process by using the Monte Carlo method, calculating the pore connectivity index and adjusting the tamping strength accordingly, and combining with multi-factor accelerated aging test to monitor the change of permeability, solving the technical problem of lack of systematic process parameter regulation method in the preparation process of the cured soil mentioned in the background technology. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 The flowchart of the method of the present application.

[0019] Figure 2 The 50-year permeability coefficient prediction curve graph of the cured soil in the embodiment.

[0020] Figure 3 The pore connectivity index time-varying evolution trend graph in the embodiment. DETAILED DESCRIPTION

[0021] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0022] As Figure 1 shown is a flow chart of a composite solidified soil anti-permeability performance enhancement method provided by the present application, and the method comprises the following steps:

[0023] S01, a laser particle size analyzer is used to determine the soil particle size distribution, a specific surface area tester is used to obtain the soil specific surface area data, and a particle size distribution-specific surface area correlation matrix is established, when the rank value of the correlation matrix is less than 3, particle screening processing is performed, and when the rank value of the matrix is greater than 8, particle mixing processing is performed;

[0024] S02, a nano-silica modified solidifying agent solution is prepared, the solidifying agent content is controlled to be 5-15% of the soil mass, an electric conductivity meter is used to determine the ion concentration of the solution, a pH meter is used to determine the pH value, when the ion concentration exceeds 0.5 mol / L, the dilution ratio is adjusted, and when the pH value deviates from the range of 7.0-9.0, a buffer is added for adjustment;

[0025] S03, the modified solidifying agent is fully mixed with the soil, ultrasonic oscillation equipment is used to assist mixing for 60-120 minutes, the water content and the compactness of the mixture are monitored in real time, a water content-compactness response matrix is established, and when the rank value change rate of the matrix exceeds 20%, the ultrasonic power density is adjusted;

[0026] S04, a soil micro-pore structure model is constructed by using a molecular dynamics simulation software, a Monte Carlo method is used to simulate the pore structure evolution process, a pore connectivity index and a pore size distribution parameter are calculated, a pore structure parameter matrix is decomposed into a linear combination of three low-rank matrices of a pore morphology matrix, a connectivity matrix and a size distribution matrix, and when the pore connectivity index is less than 0.3, the tamping treatment intensity is increased;

[0027] S05, standard curing is performed on the solidified soil sample, the curing temperature is controlled to be 20±2℃, and the relative humidity is maintained to be above 95%, real-time monitoring is performed by using a thermocouple temperature sensor and a humidity sensor, and when the temperature deviation exceeds ±1℃ or the humidity is lower than 90%, an environment adjusting system is started;

[0028] S06, a multi-factor accelerated aging test system is established, three kinds of acceleration conditions of temperature cycle, wet-dry cycle and freeze-thaw cycle are set, the permeability coefficient is determined periodically by using a permeability coefficient tester, a permeability coefficient-time decay matrix is constructed, when the decay rate exceeds 10% of the initial value, a performance early warning mechanism is triggered, a reinforcement processing program is started at the same time, and the curing parameters are adjusted;

[0029] S07, a long-term performance prediction system based on an adaptive weight adjustment model is established, a hierarchical fusion weight adjustment function is used to dynamically adjust the model parameters, the input parameters include cementing strength, compressive strength, elastic modulus, and Poisson's ratio data, the adaptive weight adjustment model is used to predict the evolution trend of the impermeability performance within 50 years, and a corresponding maintenance strategy is developed when the prediction result shows that the performance attenuation exceeds 30%.

[0030] Wherein, the particle size distribution-specific surface area correlation matrix is established by matrix operation of particle content in different particle size intervals and corresponding specific surface area contribution value, the matrix element represents the contribution degree of particle size range to total specific surface area, the particle size distribution is obtained by light scattering principle of laser particle size analyzer, and the specific surface area data is obtained by nitrogen adsorption method of specific surface area tester.

[0031] Wherein, the nano-silica modified curing agent is prepared by dispersing nano-silica particles in cement-based curing agent, nano-particles fill soil pores and form chemical bonding with soil particles, ion concentration is obtained by conductivity measurement principle of conductivity meter, and pH value is obtained by potential measurement principle of pH meter.

[0032] Wherein, the ultrasonic oscillation equipment uses an ultrasonic generator with a frequency of 20-40 kHz, and the uniform distribution of the curing agent in the soil is promoted by mechanical vibration, the water content is obtained by drying method, and the compactness is obtained by cutting ring method.

[0033] Wherein, the water content-compactness response matrix records the real-time correspondence between water content and compactness during mixing, and the matrix element reflects the correlation strength between the two parameters.

[0034] Wherein, the molecular dynamics simulation software uses LAMMPS package to establish an atomic scale soil particle model, and Monte Carlo method simulates the pore structure evolution process by random sampling statistics, and the pore size distribution parameters are obtained by mercury injection method.

[0035] Wherein, the pore connectivity index is defined as the ratio of connected pore volume to total pore volume, the value range is 0-1, and it reflects the connectivity of pore network, and the pore volume data is obtained by CT scanning technology.

[0036] Wherein, the multi-factor accelerated aging test system simulates various environmental effects that soil faces in actual engineering, shortens the evaluation period through accelerated test, the permeability coefficient is obtained by constant head permeability test method, the cementing strength is obtained by unconfined compressive strength test, the compressive strength is obtained by standard compression test, the elastic modulus is obtained by stress-strain curve calculation, and the Poisson's ratio is obtained by strain measurement calculation.

[0037] The permeability coefficient-time decay matrix records the change rule of the permeability coefficient at different ages, and the matrix row represents the time node and the list represents different test conditions.

[0038] The performance warning mechanism triggers a warning signal by setting a permeability coefficient change threshold value, and automatically issues a warning when the permeability coefficient decay rate exceeds the set value.

[0039] The reinforcement treatment program includes three treatment methods of increasing the dosage of curing agent, improving the curing temperature, and prolonging the curing time, and the corresponding treatment intensity is selected according to the warning level.

[0040] The adjustment of the curing parameters includes increasing the curing temperature from the standard 20±2℃ to 25±2℃, increasing the relative humidity from 95% to more than 98%, prolonging the curing period to 1.5 times the original plan, and increasing the frequency of watering maintenance to once every 12 hours, so as to promote the full progress of the curing reaction and improve the impermeability.

[0041] The maintenance strategy includes four steps of determining the maintenance time node, selecting the maintenance method, calculating the maintenance cost and evaluating the maintenance effect, and a personalized maintenance scheme is developed by comprehensive analysis of the prediction results and the actual engineering situation.

[0042] The specific structure of the adaptive weight adjustment model is a multi-layer perceptron architecture, including an input layer, three hidden layers and an output layer, wherein the input layer receives four parameters of cementing strength, compressive strength, elastic modulus and Poisson's ratio, the first hidden layer contains 64 neurons, the second hidden layer contains 32 neurons, and the third hidden layer contains 16 neurons, and the output layer generates a long-term impermeability performance prediction value. The hierarchical fusion weight parameters in the model are determined according to three parameters of the rank value of the particle size distribution-specific surface area correlation matrix, the rank value of the water content-density response matrix and the pore connectivity index.

[0043] The training data set of the adaptive weight adjustment model includes the following steps: collecting solidification test data of different soil types, sorting the impermeability performance test results at each age, constructing the correspondence between input and output parameters, performing data cleaning and standardization, and dividing the training set, validation set and test set according to the ratio of 7:2:1. The training data covers three soil types of clay, sand and silt, the time span covers the age range of 1 day to 5 years, and the total data amount is 10000 groups of samples.

[0044] The adaptive weight adjustment model training step includes initializing network weight parameters, setting the learning rate to 0.001, using mean square error as the loss function, updating parameters using the Adam optimizer, setting the batch size to 32, training for 1000 rounds, validating every 50 rounds, stopping training when the validation set loss does not decrease for 10 consecutive rounds, and saving the optimal model parameters for prediction. During the training process, the early stopping mechanism is used to prevent overfitting.

[0045] The hierarchical fusion weight adjustment function is used to adjust the hierarchical fusion weight parameters of the adaptive weight adjustment model. The hierarchical fusion weight adjustment function is calculated based on the rank values of the particle size distribution-specific surface area correlation matrix, the rank values of the water content-density response matrix, and the pore connectivity index. A fusion weight coefficient is obtained. When the fusion weight coefficient belongs to the interval a∈[0, 0.3), the first hidden layer weight is reduced to adjust the hierarchical fusion weight parameters of the adaptive weight adjustment model. When the fusion weight coefficient belongs to the interval a∈[0.3, 0.7], the weights of each hidden layer are balanced to adjust the hierarchical fusion weight parameters of the adaptive weight adjustment model. When the fusion weight coefficient belongs to the interval a∈(0.7, 1.0], the third hidden layer weight is increased to adjust the hierarchical fusion weight parameters of the adaptive weight adjustment model.

[0046] The matrix rank feature extraction algorithm based on singular value decomposition plays an important role in the method of enhancing the impermeability of composite solidified soil. Its core principle is to decompose the complex multi-dimensional soil parameter matrix into several simplified low-rank matrix combinations, each of which represents an independent physical and chemical action mode. The singular value decomposition algorithm decomposes the original soil performance parameter matrix into the product of three low-rank matrices through mathematical transformation, corresponding to three basic modes of pore morphology characteristics, inter-particle cementation mode, and permeation path distribution.

[0047] In the calculation process, the singular value decomposition algorithm first constructs a high-dimensional matrix containing key parameters such as permeability coefficient, porosity, density, and cementation strength. Then, by calculating the singular value decomposition of the matrix, the main components reflecting the essential characteristics of the soil microstructure are extracted. When the matrix rank value is small, the soil structure is relatively simple, and the correlation between parameters is strong. At this time, dimensionality reduction processing can effectively simplify the analysis complexity; when the matrix rank value is large, the soil structure presents diversity, and needs to be fully described by dimensionality expansion processing.

[0048] The singular value decomposition algorithm shows outstanding applicability in the scheme, and the main reason is that the impermeability of the solidified soil is affected by the coupling of multiple factors, and the traditional single parameter analysis method cannot accurately reflect the real performance state. The singular value decomposition algorithm can effectively identify and separate the independent contribution of various factors, providing a scientific basis for accurate evaluation and optimization control of impermeability.

[0049] Through the application of the singular value decomposition algorithm, the quantitative description and accurate control of the micro-pore structure of the solidified soil can be realized, and the accuracy and reliability of the impermeability prediction can be significantly improved. At the same time, the singular value decomposition algorithm reduces the calculation complexity of the traditional multi-parameter analysis and improves the engineering practicability of the overall scheme.

[0050] The advantages of the singular value decomposition algorithm in data processing are reflected in the ability to handle high-dimensional parameter matrices while maintaining computational efficiency. Through singular value decomposition, the original complex matrix is simplified into three independent low-rank matrices, each representing the pore geometry, the physical and chemical connection state between particles, and the spatial distribution characteristics of the permeation channel. The economic benefits of the singular value decomposition algorithm mainly lie in reducing the number of repeated tests, shortening the parameter optimization period, and reducing engineering costs, providing a feasible technical path for large-scale engineering applications.

[0051] The specific implementation of the above steps is described in detail below.

[0052] The specific implementation of step S01 is to use laser particle size analysis technology and specific surface area determination technology to construct the correlation matrix of particle size distribution and specific surface area. The purpose of this step is to establish a quantitative description of the physical properties of soil particles. First, the particle size distribution of the soil sample is determined by the light scattering principle of the laser particle size analyzer. The light scattering principle is based on the physical phenomenon that different particle sizes scatter light at different angles. The particle size distribution is inversely calculated by measuring the scattering light intensity distribution. Then, the specific surface area data of the soil sample is measured by the nitrogen adsorption method of the specific surface area analyzer. The nitrogen adsorption method is based on the physical adsorption law of gas molecules on the surface of a solid. The specific surface area is calculated by measuring the adsorption amount at different pressures. Then, the particle content data in different particle size intervals and the corresponding specific surface area contribution values are matrix operated to construct the correlation matrix of particle size distribution and specific surface area. The matrix element values reflect the contribution of each particle size range to the total specific surface area. Finally, the rank value of the correlation matrix is calculated by the singular value decomposition algorithm. When the rank value is less than 3, it indicates that the particle distribution is too concentrated and needs to be screened to increase the particle size diversity. When the rank value is greater than 8, it indicates that the particle distribution is too dispersed and needs to be mixed to optimize the particle size distribution.

[0053] The specific implementation of step S02 is to prepare the nano-SiO2 modified curing agent solution and control its physical and chemical properties. The purpose of this step is to prepare the cured material with excellent cementation performance and chemical stability. First, the nano-SiO2 particles are dispersed in the cement-based curing agent to prepare the modified curing agent. The nano-particles are uniformly distributed in the solution by the dispersion effect of the surfactant. The curing agent content is controlled in the range of 5% to 15% of the soil mass to ensure both sufficient cementation strength and not affect the engineering properties of the soil. Then the ion concentration of the solution is measured by conductivity meter. The conductivity measurement principle is based on the migration characteristics of ions under the action of electric field. The ion concentration is converted by measuring the conductivity value of the solution. When the ion concentration exceeds 0.5 mol / L, the dilution ratio needs to be adjusted to prevent the ion concentration from being too high to affect the cementation effect of the curing agent. At the same time, the pH value of the solution is measured by pH meter. The pH meter determines the acidity and alkalinity of the solution based on the potential measurement principle through the potential difference between the glass electrode and the reference electrode. When the pH value deviates from the range of 7.0 to 9.0, the buffer needs to be added to adjust it to the appropriate range to ensure that the curing reaction occurs in the best chemical environment.

[0054] The specific implementation of step S03 is to use ultrasonic oscillation technology to promote the thorough mixing of the modified curing agent and the soil. The purpose of this step is to achieve uniform distribution of the curing agent in the soil and optimize the physical properties of the mixture. First, the modified curing agent is preliminarily mixed with the soil, and then an ultrasonic generator with a frequency of 20 to 40 kHz is used for ultrasonic oscillation treatment. The mechanical vibration effect of ultrasonic waves can destroy the aggregation structure between soil particles and promote the penetration of the curing agent into the soil micropores. The oscillation time is controlled in the range of 60 to 120 minutes to ensure sufficient mixing. The moisture content and density of the mixture are monitored in real time during the oscillation process. The moisture content is obtained by drying method, that is, the water content is calculated by measuring the mass difference before and after drying the sample. The density is obtained by the ring method, that is, the dry density of the sample is measured by the standard volume ring. A response matrix of moisture content and density is established to record the real-time correspondence between the two parameters. The matrix element value reflects the correlation strength between the parameters. The rank value of the response matrix is calculated by the singular value decomposition algorithm. When the matrix rank value change rate exceeds 20%, the ultrasonic power density needs to be adjusted to maintain the stability of the mixing effect.

[0055] The specific implementation of step S04 is to construct and analyze the micro-pore structure of the soil through molecular dynamics simulation and Monte Carlo method. The purpose of this step is to quantitatively evaluate the micro-structure characteristics of the soil after solidification treatment and its influence on the permeability. First, the LAMMPS program package is used to establish an atomic-scale soil particle model. Molecular dynamics simulation is based on Newton's mechanics to simulate the micro-structure and dynamic behavior of matter by calculating the interatomic interaction force. Then, the Monte Carlo method is used to simulate the evolution process of the pore structure. Monte Carlo method can effectively handle complex problems with multiple variable couplings by statistically simulating the evolution law of complex systems. Then, the pore connectivity index and pore size distribution parameters are calculated. The pore connectivity index is defined as the ratio of connected pore volume to total pore volume, with a value ranging from 0 to 1. The pore volume data is obtained through CT scanning technology. The pore size distribution parameters are obtained by mercury intrusion method based on the physical principle of mercury entering pores of different sizes under different pressures. Finally, the pore structure parameter matrix is decomposed into a linear combination of three low-rank matrices, namely the pore morphology matrix, the connectivity matrix, and the size distribution matrix, through the singular value decomposition algorithm. When the pore connectivity index is less than 0.3, the compaction treatment intensity needs to be increased to reduce the pore connectivity.

[0056] The specific implementation of step S05 is to standardize the curing of the solidified soil sample and monitor the environmental conditions in real time. The purpose of this step is to provide the best temperature and humidity environment for the solidification reaction to ensure the full development of the impermeability. First, set the curing environment parameters. The curing temperature is controlled within the range of 20±2℃ to ensure that the cement hydration reaction proceeds at an appropriate temperature, and the relative humidity is maintained above 95% to prevent surface water loss from affecting the solidification effect. Then, the curing temperature is monitored in real time by a thermocouple temperature sensor. Thermocouples measure temperature based on the Seebeck effect, which is the physical principle of generating an electromotive force when two different metals are in contact. At the same time, a humidity sensor is used to monitor the environmental humidity. The humidity sensor measures air humidity based on the principle of resistance or capacitance change after the material absorbs moisture. When the temperature deviation exceeds ±1℃ or the humidity is less than 90%, the environmental regulation system is automatically started. The environmental parameters are adjusted to the set range through devices such as heaters and humidifiers to ensure the stability of the environmental conditions during the entire curing process.

[0057] The specific implementation of step S06 is to establish a multi-factor accelerated aging test system to evaluate the long-term impermeability of solidified soil. The purpose of this step is to simulate the long-term environmental effects in actual engineering through accelerated tests and predict the performance degradation law. First, three accelerated aging conditions are set, including temperature cycling, wet-dry cycling, and freeze-thaw cycling. Temperature cycling simulates the influence of seasonal temperature difference by alternating high and low temperatures, wet-dry cycling simulates the rainfall and drought cycle by alternating drying and soaking, and freeze-thaw cycling simulates the freezing and thawing action in cold regions by alternating freezing and thawing. Then, the permeability coefficient of the sample is measured regularly by a permeability coefficient tester. The permeability coefficient measurement uses the constant head permeability test method based on Darcy's law, which is the permeation law of water flow through porous media. At the same time, mechanical parameters such as cementation strength, compressive strength, elastic modulus, and Poisson's ratio are measured. Cementation strength is measured by unconfined compressive strength test, compressive strength is obtained by standard compression test, elastic modulus is obtained by stress-strain curve calculation, and Poisson's ratio is obtained by strain measurement. Then, a decay matrix of permeability coefficient and time is constructed to record the change law of permeability coefficient at different ages. The matrix row represents the time node, and the list represents different test conditions. When the decay rate of permeability coefficient exceeds 10% of the initial value, the performance warning mechanism is triggered, and the reinforcement treatment program is started, including increasing the dosage of solidifying agent, improving the curing temperature, and prolonging the curing time. Adjust the curing parameters, including increasing the curing temperature from 20±2℃ to 25±2℃, increasing the relative humidity from 95% to more than 98%, prolonging the curing period to 1.5 times the original planned time, and increasing the frequency of watering maintenance to once every 12 hours.

[0058] Step S07 is an optional step, and its specific implementation is to establish a long-term performance prediction system based on adaptive weight adjustment model. The purpose of this step is to predict the evolution trend of the impermeability performance of the solidified soil within 50 years and develop a maintenance strategy through artificial neural network technology. The prediction system adopts a multi-layer perceptron architecture, including a deep neural network structure of an input layer, three hidden layers, and an output layer. The input layer receives four key parameters, namely cementation strength, compressive strength, elastic modulus, and Poisson's ratio, as model inputs, which can comprehensively reflect the mechanical performance state of the solidified soil. The first hidden layer contains 64 neurons responsible for extracting primary features of the input parameters, the second hidden layer contains 32 neurons for further abstraction and combination of features, and the third hidden layer contains 16 neurons to realize extraction and integration of advanced features. The output layer generates long-term impermeability performance prediction values, and outputs continuous numerical prediction results through regression analysis. The hierarchical fusion weight adjustment function dynamically adjusts the model parameters according to three parameters, namely the rank value of the particle size distribution and specific surface area correlation matrix, the rank value of the water content and density response matrix, and the pore connectivity index. When the fusion weight coefficient belongs to the interval 0 to 0.3, the model parameters are adjusted by reducing the weight of the first hidden layer. When the coefficient belongs to the interval 0.3 to 0.7, the model parameters are adjusted by balancing the weights of each hidden layer. When the coefficient belongs to the interval 0.7 to 1.0, the model parameters are adjusted by increasing the weight of the third hidden layer. When the prediction result shows that the performance decay exceeds 30%, the corresponding maintenance strategy is developed. The maintenance strategy development includes four steps: determining the maintenance time node, selecting the maintenance method, calculating the maintenance cost, and evaluating the maintenance effect.

[0059] The detailed steps for establishing the training data set of the long-term performance prediction system include five stages: data collection, data arrangement, relationship construction, data processing, and data division. In the data collection stage, data from solidification tests of different soil types are collected, covering three main soil types: clay, sand, and silt, to ensure the representativeness and comprehensiveness of the data. For each soil type, test data under different solidification agent contents and different curing conditions are collected to form diverse data samples. In the data arrangement stage, the permeability test results at each age are arranged, covering a time span from 1 day to 5 years, including performance data in the early solidification stage, the medium-term stability stage, and the long-term aging stage. The data at each time node include key parameters such as permeability coefficient, cementation strength, compressive strength, elastic modulus, and Poisson's ratio. In the relationship construction stage, the corresponding relationship between input and output parameters is constructed to establish the mapping relationship between input parameters and long-term permeability performance, ensuring that the model can learn the inherent rules between parameters. In the data processing stage, data cleaning and standardization processing are performed, including steps such as removing outliers, filling missing values, and correcting erroneous data. Standardization processing includes data normalization, parameter dimension unification, and numerical range adjustment. In the data division stage, the data set is divided into training set, validation set, and test set according to the ratio of 7:2:1. The training set is used for model parameter learning, the validation set is used for model performance evaluation, and the test set is used for final model effect verification. The entire training data set contains a total of 10,000 samples, ensuring the sufficiency and reliability of model training.

[0060] The training process of the adaptive weight adjustment model adopts the standard training process of deep learning, including four links: parameter initialization, hyperparameter setting, training execution, and model saving. In the parameter initialization stage, the random initialization method is used to set the network weight parameters to ensure the randomness and convergence of model training. In the hyperparameter setting stage, the learning rate is set to 0.001 to ensure stable convergence of the model, the mean square error is used as the loss function to quantify the difference between the predicted value and the true value, the Adam optimizer is used for parameter update to realize adaptive learning rate adjustment, the batch size is set to 32 to balance the training efficiency and memory usage, and the number of training rounds is set to 1000 to ensure sufficient learning of the model. In the training execution stage, the model performance on the validation set is evaluated every 50 rounds, and the early stopping mechanism is triggered when the validation set loss does not decrease for 10 consecutive rounds to prevent overfitting. In the model saving stage, the model parameters with the best performance on the validation set are saved for actual prediction, ensuring the generalization ability and prediction accuracy of the model.

[0061] It should be noted that the key technical ideas of the present application mainly reflect four aspects. The first key technical idea is a matrix rank feature extraction algorithm based on singular value decomposition. This algorithm can decompose a complex multi-dimensional soil parameter matrix into several simplified low-rank matrix combinations, and each low-rank matrix represents an independent physical and chemical action mode. Compared with the traditional single parameter analysis method, the singular value decomposition algorithm can effectively identify and separate the independent contributions of various influencing factors, realize the quantitative description and accurate control of the soil microstructure, and significantly improve the accuracy and reliability of the anti-permeability performance prediction. The second key technical idea is a multi-layer perceptron architecture of an adaptive weight adjustment model. This architecture realizes accurate modeling of complex parameter relationships through the nonlinear mapping capability of a deep neural network. Compared with traditional linear regression or simple neural networks, the multi-layer perceptron architecture can capture the complex nonlinear relationships between input parameters, and realize more accurate long-term performance prediction through hierarchical feature extraction and fusion. The third key technical idea is a multi-factor accelerated aging test system. This system simulates the actual engineering environment through temperature cycling, wet-dry cycling and freeze-thaw cycling. Compared with the traditional single environmental aging test, the multi-factor test system can more comprehensively evaluate the long-term performance degradation law of the solidified soil under complex environment, and improve the engineering applicability of performance prediction. The fourth key technical idea is a dynamic adjustment mechanism of hierarchical fusion weight adjustment function. This mechanism dynamically optimizes the model weight distribution according to the real-time changes of soil microstructure parameters. Compared with the traditional model with fixed weights, the dynamic adjustment mechanism can adapt to different soil conditions and environmental changes, improve the adaptive ability and prediction stability of the model. The synergistic effect of these four key technical ideas forms a complete soil anti-permeability performance enhancement and prediction system. Through feature extraction by singular value decomposition algorithm, intelligent modeling by multi-layer perceptron, comprehensive evaluation by multi-factor test and adaptive optimization by dynamic adjustment, the whole chain technology integration from micro-mechanism analysis to macro-performance prediction is realized. Compared with existing technologies, the present application has the significant advantages of high prediction accuracy, wide application range and strong engineering practicability.

[0062] It should be noted that in the conventional solidified soil preparation process, the ion concentration and pH value of the solidifying agent solution often lack precise control, resulting in unstable solidification reaction rate, affecting the formation quality of the cementing structure and the final impermeability. The present application determines the ion concentration of the solidifying agent solution in real time by conductivity meter, adjusts the dilution ratio in time when it exceeds 0.5 mol / L, monitors the pH value by pH meter and adjusts it by adding buffer when it deviates from the range of 7.0-9.0, ensuring that the solidification reaction proceeds in the most suitable chemical environment. At the same time, the addition of nano-silicon dioxide particles not only fills the soil pores but also forms chemical bonds with soil particles, significantly improving the cementing strength and compactness, providing a reliable chemical reaction basis for excellent impermeability. The curing of the solidified soil in the prior art often lacks precise environmental control, and the fluctuations of temperature and humidity can significantly affect the progress of the solidification reaction and the development of the final performance, resulting in large differences in the performance of different batches of solidified soil. The present application establishes a standard curing environment control system, accurately controls the curing temperature at 20±2℃, and maintains the relative humidity above 95%. Real-time monitoring is carried out through thermocouple temperature sensor and humidity sensor, and when the temperature deviation exceeds ±1℃ or the humidity is lower than 90%, the environmental regulation system is automatically started, ensuring that the solidification reaction proceeds fully under stable environmental conditions. At the same time, the established multi-factor accelerated aging test system can simulate various environmental effects in actual engineering, and the performance change is monitored in real time through the permeability coefficient-time decay matrix, and when the decay rate exceeds 10% of the initial value, the early warning mechanism is triggered and the reinforcement treatment program is started, realizing the active maintenance and long-term stable control of the performance of the solidified soil.

[0063] Specifically, the principle of the present application is that the key to solving the problem of lack of systematic process parameter control method in the preparation process of solidified soil lies in the establishment of a multi-level structure feature recognition and process parameter optimization system. First, the particle size distribution of soil is determined by a laser particle size analyzer, and the specific surface area data of soil are obtained by a specific surface area analyzer, and a particle size distribution-specific surface area correlation matrix is established, which can quantitatively reflect the contribution of particles of different particle sizes to the total specific surface area. When the rank value of the matrix is less than 3, it indicates that the soil structure is relatively simple and needs to be screened, and when the rank value is greater than 8, it indicates that the structure is complex and needs to be mixed, realizing the differentiated pretreatment based on the characteristics of the soil. Second, the application of nano-silicon dioxide modified solidifying agent combined with ultrasonic oscillation technology fills the soil pores through the high specific surface area and dispersibility of nano-particles, forming a more compact cementation network. The establishment of the water content-density response matrix can reflect the correlation changes of the two key parameters in the mixing process in real time, guide the dynamic adjustment of the ultrasonic power density, and ensure the uniform distribution of the solidifying agent in the soil. Third, the atomic scale soil particle model constructed by the molecular dynamics simulation software can accurately describe the geometric shape and connectivity characteristics of the pore structure. The definition and calculation of the pore connectivity index provide a quantitative basis for the adjustment of the compaction treatment strength, realizing the accurate control of the pore structure. Finally, the multi-factor accelerated aging test system simulates the actual engineering environment through temperature cycling, wet-dry cycling, freeze-thaw cycling, etc. The establishment of the permeability coefficient-time attenuation matrix can monitor the time-dependent changes of the solidification effect, provide feedback information for the continuous optimization of the process parameters, and form a complete closed-loop control system.

[0064] A specific embodiment 1 of the present application is provided below, and the specific implementation of each step in the embodiment 1 is described in detail as follows.

[0065] The specific implementation of step S01 is to construct a particle size distribution-specific surface area correlation matrix using laser particle size analysis technology and specific surface area measurement technology. This step involves matrix operations and singular value decomposition calculations. The particle size distribution-specific surface area correlation matrix M ps is specifically represented as follows:

[0066]

[0067] In the formula, p i is the particle content percentage of the i-th particle size interval, s j is the specific surface area contribution value corresponding to the j-th particle size interval, m is the total number of particle size intervals, and n is the number of specific surface area measurement points. The rank value R ps of the matrix is calculated by a singular value decomposition algorithm:

[0068] M ps = U∑VT ;

[0069] In the formula, U is a left singular vector matrix, ∑ is a singular value diagonal matrix, V T is the transpose of a right singular vector matrix, and the matrix rank value R ps = rank(∑). The parameter acquisition method is as follows: p i The laser particle size analyzer is used to measure and obtain by light scattering principle, including the following steps: 1. Disperse the soil sample in deionized water to prepare a suspension; 2. Measure the scattering light intensity distribution at different angles by the laser particle size analyzer; 3. Calculate the particle size distribution according to the Mie scattering theory.s j The specific surface area analyzer is used to measure and obtain by nitrogen adsorption method, including the following steps: 1. Dry the soil sample at 105°C to constant weight; 2. Perform nitrogen adsorption test at liquid nitrogen temperature; 3. Calculate the specific surface area value according to the BET theory. When R ps <3, perform particle screening processing; when R ps >8, perform particle mixing processing.

[0070] The specific implementation of step S02 is the same as the foregoing, and will not be described in detail here.

[0071] The specific implementation of step S03 is to use ultrasonic oscillation technology to promote the full mixing of the modified curing agent and the soil. This step involves the construction and calculation of the water content and density response matrix M wd The specific representation of the matrix M

[0072]

[0073] In the formula, w i is the water content at the i-th time point, d j is the density at the j-th time point, l is the number of water content measurements, and k is the number of density measurements. The matrix rank value change rate ΔR wd is calculated according to the following formula:

[0074]

[0075] In the formula, R wd (t) is the matrix rank value at the current time, and R wd (t-1) is the matrix rank value at the previous time. The parameter acquisition method is as follows: w i The drying method is used to measure and obtain, including the following steps: 1. Dry the sample in a 105°C oven to constant weight; 2. Calculate the mass difference before and after drying; 3. Calculate the water content according to the formula In the formula, m wet is the mass of the wet sample, and m dry is the mass of the dry sample.dj The ring knife method is used to obtain, including steps 1: using a standard ring knife to cut the soil sample; step 2: measuring the mass of the soil sample in the ring knife; step 3: calculating the compactness according to the formula The compactness is calculated, where m s is the dry mass of the soil sample, V s is the volume of the ring knife. When ΔR wd > 20%, adjust the ultrasonic power density.

[0076] The specific implementation of step S04 is to construct and analyze the micro-pore structure of the soil through molecular dynamics simulation and the Monte Carlo method, which involves pore connectivity index calculation and matrix decomposition. The formula for calculating the pore connectivity index φ c is as follows:

[0077]

[0078] In the formula, V connected is the connected pore volume, V total is the total pore volume. The pore structure parameter matrix M pore is decomposed into a linear combination of three low-rank matrices by the singular value decomposition algorithm:

[0079] M pore = M shape + M connect + M size ;

[0080] In the formula, M shape is the pore morphology matrix, M connect is the connectivity matrix, and M size is the size distribution matrix. The pore size distribution parameter D p is determined by the mercury intrusion method, and its distribution function is:

[0081]

[0082] In the formula, f(D p ) is the pore size distribution density function, V pore is the cumulative pore volume, D p is the pore size, dV pore is the pore volume element, and d(logD p ) is the logarithmic pore size element. The parameters are obtained as follows: V connected and V total are obtained by CT scanning technology, including steps 1: high-resolution CT scanning of the soil sample; step 2: identifying the pore structure through image processing technology; and step 3: calculating the connected pore and total pore volumes. D pThe mercury intrusion method is used to obtain, including steps 1: placing the sample in the mercury intrusion instrument; step 2: gradually increasing the pressure to measure the amount of mercury entering; step 3: calculating the pore size according to the Young-Laplace equation , wherein γ is the surface tension of mercury, θ is the contact angle, and P is the applied pressure. When φ c <0.3, increase the intensity of the tapping treatment.

[0083] The specific implementation of step S05 is the same as the foregoing, and will not be described in detail here.

[0084] The specific implementation of step S06 is to establish a multi-factor accelerated aging test system to evaluate the long-term impermeability of the solidified soil. This step involves the construction of a permeability coefficient and time decay matrix. The permeability coefficient and time decay matrix M kt is specifically represented as follows:

[0085]

[0086] , wherein k ij is the permeability coefficient at the jth time point under the ith test condition, c is the total number of test conditions, and t is the total number of time measurement points. The permeability coefficient decay rate λ k is calculated as follows:

[0087]

[0088] , wherein k0 is the initial permeability coefficient, and k t is the permeability coefficient at time t. The parameter acquisition method is as follows: k ij The constant head permeation test method is used to obtain, including steps 1: preparing a standard sample and installing it in the permeation instrument; step 2: measuring the seepage flow under constant water head conditions; and step 3: calculating the permeability coefficient according to Darcy's law , wherein Q is the seepage flow, L is the sample height, A is the sample cross-sectional area, h is the water head height, and τ is the permeation time. When λ k >10%, the performance early warning mechanism is triggered.

[0089] The specific implementation of step S07 is to establish a long-term performance prediction system based on an adaptive weight adjustment model. This step involves the calculation of a hierarchical fusion weight adjustment function. The calculation formula of the hierarchical fusion weight coefficient α fusion is as follows:

[0090] α fusion = β1R ps + β2R wd + β3φ c + ξ;

[0091] Wherein, β1, β2, β3 are weight coefficients, and ξ is an adjustment factor. The hierarchical fusion weight adjustment function f adjust According to the value of α fusion Segmented adjustment is performed according to the value of α

[0092] When α fusion ∈[0, 0.3), f adjust = γ1W1;

[0093] When α fusion ∈[0.3, 0.7], f adjust = γ2(W1+W2+W3) / 3;

[0094] When α fusion ∈(0.7, 1.0], f adjust = γ3W3;

[0095] Wherein, W1, W2, W3 are respectively the first, second and third hidden layer weights, and γ1, γ2, γ3 are adjustment coefficients. Among them, the parameter acquisition method is: β1, β2, β3 are obtained by least square fitting, including steps 1: collecting historical test data to construct training samples; step 2: determining the weight coefficient value through regression analysis; step 3: verifying the effectiveness and stability of the coefficient. The range of ξ is 0.05 to 0.15. γ1, γ2, γ3 are adjustment coefficients, the value ranges are respectively 0.8 to 1.2, 0.9 to 1.1, and 0.7 to 1.3, and are determined by empirical fitting. When the prediction performance decay exceeds 30%, the corresponding maintenance strategy is formulated.

[0096] It needs to be explained that the construction of the particle size distribution and specific surface area correlation matrix is based on the multiple linear regression theory, and the matrix element p i s j establishes the quantitative relationship between the particle distribution in different particle size ranges and the corresponding specific surface area contribution;

[0097]

[0098] The matrix can comprehensively reflect the influence mechanism of soil particle physical properties on the overall performance. Compared with the traditional single parameter description method, the matrix method can capture the complex nonlinear relationship between particle size distribution and specific surface area, and provide more accurate quantitative basis for soil improvement. The singular value decomposition algorithm is based on linear algebra theory, which can decompose high-dimensional complex matrix into the product form of three orthogonal matrices;

[0099] M ps = U∑V T ;

[0100] Wherein, the non-zero elements of the diagonal matrix ∑ are singular values σ i , and the matrix rank value is determined by the number of singular values;

[0101] R ps = rank(∑) = count(σ i >∈);

[0102] where ∈ is the numerical tolerance, usually taken as 10 -12 Compared with the traditional matrix analysis method, the algorithm has the advantages of good numerical stability and high calculation accuracy, and can effectively identify the potential correlation pattern between soil parameters, providing scientific guidance for particle size distribution optimization. The moisture content-density response matrix is based on the principles of material physics, and the matrix element w i d j reflects the dynamic coupling relationship between water migration and soil compaction during the curing process;

[0103]

[0104] Compared with the traditional static parameter control method, dynamic matrix monitoring can significantly improve the uniformity and stability of the mixing effect. The pore connectivity index is based on the seepage theory of porous media, and quantitatively describes the connectivity of the pore network through the ratio of connected pore volume to total pore volume;

[0105]

[0106] Compared with the traditional porosity parameter, the index can more accurately reflect the effectiveness of the pore structure, providing a key evaluation index for permeability optimization. The pore structure matrix decomposition is based on the structure analysis theory, which decomposes the complex three-dimensional pore network into a linear superposition of three independent characteristic matrices;

[0107] M pore = M shape + M connect + M size ;

[0108] This decomposition method can analyze the independent contribution of each structural feature to the permeability, and has stronger pertinence and controllability compared with the overall analysis method, providing accurate guidance for microstructure control. The permeability coefficient decay matrix is based on the material aging theory, and the matrix element k ij builds the decay law of permeability performance over time;

[0109]

[0110] This matrix can predict the long-term performance change of the material in a short time, significantly shortening the evaluation period compared with the traditional long-term test method, and improving the efficiency of engineering decision-making. The hierarchical fusion weight adjustment function is based on the multi-parameter fusion theory, and constructs a comprehensive evaluation index through weighted combination;

[0111] alpha fusion = beta1R ps + beta2R wd + beta3phi c + xi

[0112] The function can dynamically adjust the model parameters according to the real-time changes of the soil microstructure, has stronger self-adaptive ability and prediction stability compared with the fixed weight model, and significantly improves the performance prediction accuracy under complex working conditions.

[0113] It should be noted that the variables involved in the application are explained in detail as shown in Table 1.

[0114] Table 1: Variable explanation table

[0115]

[0116]

[0117] In order to better understand and implement the application, the following provides an embodiment 2 of a specific application scenario of the application: the soil of a certain clay layer with a thickness of 2.5m is mainly high liquid limit clay, the natural moisture content is 32.8%, the liquid limit is 58.4%, the plastic limit is 26.1%, and the initial permeability coefficient is 4.2x10 -7 m / s, which does not meet the engineering requirement of 1.0x10 -8 m / s standard. The technical team adopts a composite solidified soil anti-permeability performance enhancement method to treat the soil.

[0118] First, the implementation of step S01 is performed. The technical team uses a Mastersizer 3000 laser particle size analyzer to determine the particle size distribution of the soil sample, and the light scattering measurement result shows that the clay content is 67.3%, the silt content is 28.9%, and the sand content is 3.8%. The ASAP 2460 specific surface area analyzer is used to determine the specific surface area of the soil by nitrogen adsorption method, which is 43.7m 2 / g. The particle size distribution and specific surface area correlation matrix is constructed, and the particle content of three particle size intervals below 0.002mm, 0.002-0.05mm and 0.05-2.0mm is respectively matrix operated with the corresponding specific surface area contribution value 29.4m 2 / g, 12.6m 2 / g, 1.7m 2 / g. The rank value of the correlation matrix is calculated by the singular value decomposition algorithm to be 2.3, which is less than the threshold value of 3, indicating that the particle distribution is too concentrated, and particle screening treatment needs to be performed. The technical team adds 15% of quartz sand with a particle size of 0.1-0.5mm to make the particle size distribution more reasonable.

[0119] Then, the step S02 of preparing the nano-SiO2 modified curing agent solution is implemented. The technical team disperses the nano-SiO2 particles with a particle size of 20 nm in the ordinary Portland cement of grade 42.5 at a proportion of 3%, to prepare the modified curing agent. The curing agent content is controlled at 12% of the dry weight of the soil, and the solution ion concentration is 0.42 mol / L measured by the DDS-307A conductivity meter, meeting the requirement of less than 0.5 mol / L. The solution pH value is 8.2 measured by the PHS-3C pH meter, which is within the suitable range of 7.0-9.0, and no buffer needs to be added for adjustment.

[0120] Then, the step S03 of ultrasonic mixing treatment is implemented. The technical team mixes the modified curing agent with the treated soil according to the designed proportion, and uses an ultrasonic generator with a frequency of 25 kHz and a power density of 150 W / m 2 for oscillation treatment for 90 minutes. The changes in the water content and the compactness are monitored in real time during the oscillation process. The initial water content is 28.3%, and the compactness is 1.67 g / cm 3 . After oscillation for 60 minutes, the water content decreases to 26.8%, and the compactness increases to 1.73 g / cm 3 . The response matrix of the water content and the compactness is established, and the rank value of the matrix changes from 3.2 to 3.8, with a change rate of 18.8%, which does not exceed the threshold of 20%, and the original ultrasonic power density is maintained.

[0121] The step S04 of analyzing the micro-pore structure by molecular dynamics simulation is implemented. The technical team uses the LAMMPS program package to establish a soil particle model containing 51200 atoms, and the simulation box size is 45 nm x 45 nm x 30 nm. The pore structure evolution process is simulated for 1 million steps by the Monte Carlo method, and the pore connectivity index is calculated to be 0.38. The

[0122] The pore size distribution parameters are measured by the Micromeritics AutoPore V mercury porosimeter, and the main pore size is concentrated in the range of 50-200 nm, with an average pore size of 127 nm. The connected pore volume accounts for 38% of the total pore volume obtained by CT scanning technology. The pore structure parameter matrix is decomposed into three low-rank matrix combinations of pore morphology matrix, connectivity matrix, and size distribution matrix, and the pore connectivity index is greater than the threshold value of 0.3, and no additional tamping treatment intensity is needed.

[0123] Step S05 implements standard curing control. The technical team places the prepared solidified soil sample in a temperature and humidity controllable curing room, and monitors the environmental parameters in real time through a K-type thermocouple temperature sensor and an SHT30 humidity sensor. The curing temperature is controlled at 20.5℃, and the relative humidity is maintained at 96.2%. On the 3rd day of curing, the temperature sensor shows that the temperature rises to 21.8℃, which exceeds the ±1℃ deviation range, and the environmental regulation system automatically starts to cool down, adjusting the temperature to 20.2℃. On the 7th day of curing, the humidity sensor shows that the relative humidity drops to 88.7%, which is lower than the 90% threshold, and the humidifier automatically starts to increase the humidity to 95.8%.

[0124] Step S06 establishes a multi-factor accelerated aging test system. The technical team sets the temperature cycle conditions to be -10℃~40℃, completing one cycle every 24 hours; the wet-dry cycle conditions are 12 hours of immersion in water and 12 hours of drying at 60℃; and the freeze-thaw cycle conditions are 12 hours of freezing at -18℃ and 12 hours of thawing at 20℃. The TST-55 permeability coefficient tester is used to measure the change in permeability coefficient every 7 days, with an initial permeability coefficient of 8.6×10 -9 m / s. As shown in Table 2, the change in permeability coefficient under different aging conditions is shown.

[0125] Table 2 Change in permeability coefficient under different aging conditions

[0126]

[0127] On the 21st day of aging, the freeze-thaw cycle condition has a permeability coefficient decay rate of 12.3%, which exceeds the 10% early warning threshold, triggering the performance early warning mechanism of the system. The technical team initiates the reinforcement treatment program, increasing the solidifying agent content from 12% to 15%, increasing the curing temperature from 20±2℃ to 25±2℃, increasing the relative humidity from 95% to 98%, extending the curing period to 1.5 times the original plan, and increasing the frequency of watering maintenance to every 12 hours.

[0128] The related mechanical parameters are also measured simultaneously. The unconfined compressive strength test shows that the cementation strength is 1.87MPa, the standard compressive strength is 2.34MPa, the elastic modulus is 186MPa, and the Poisson's ratio is 0.32. As shown in Table 3, the development of each mechanical parameter at different ages is shown.

[0129] Table 3 Age development of mechanical parameters

[0130] Age (days) Cementing strength (MPa) Compressive strength (MPa) Elastic modulus (MPa) Poisson's ratio 3 0.98 1.24 95 0.38 7 1.45 1.82 142 0.35 14 1.73 2.18 169 0.33 28 1.87 2.34 186 0.32 56 2.03 2.51 201 0.31 90 2.12 2.63 212 0.30

[0131] Step S07 establishes a long-term performance prediction system. The technical team builds an adaptive weight adjustment model of a multi-layer perceptron architecture, and the input layer receives four parameters: cementing strength 2.12 MPa, compressive strength 2.63 MPa, elastic modulus 212 MPa, and Poisson's ratio 0.30. The first hidden layer has 64 neurons to extract primary features, the second hidden layer has 32 neurons for feature combination, and the third hidden layer has 16 neurons to realize advanced feature integration. According to the rank value 2.8 of the particle size distribution and specific surface area correlation matrix, the rank value 3.8 of the water content and density response matrix, and the pore connectivity index 0.38, the fusion weight coefficient is calculated as 0.52. The coefficient belongs to the interval of 0.3-0.7, and the model parameters are adjusted by balancing the weights of each hidden layer.

[0132] As shown in Figure 2 , the prediction results show that the impermeability of the solidified soil remains relatively stable within the first 10 years, with a permeability coefficient of 1.2×10 -8 m / s or lower. After the 15th year, a slow upward trend begins, and by the 30th year, the predicted permeability coefficient is 1.8×10 -8 m / s. As shown in Figure 3 , the evolution of the pore connectivity index over time shows that the connectivity index remains below 0.4 within the first 20 years and begins to increase significantly after the 25th year, reaching 0.65 by the 50th year. The prediction results show that the performance degradation will exceed the 30% threshold at the 32nd year, so the technical team develops a corresponding maintenance strategy and determines to perform the first maintenance at the 30th year using supplementary grouting, with a maintenance cost budget of 15% of the initial treatment cost.

[0133] During the training data set establishment process, the technical team collected 10,000 sample data sets of clay, sand, and silt soil types, covering a time span of 1 day to 5 years. According to a 7:2:1 ratio, 7,000 groups were divided into a training set, 2,000 groups into a validation set, and 1,000 groups into a test set. Model training used parameter settings of a learning rate of 0.001, a batch size of 32, and 1,000 training rounds, using the Adam optimizer and mean squared error loss function. At the 780th round, the validation set loss did not decrease for 10 consecutive rounds, triggering the early stop mechanism to complete training.

[0134] The final permeability coefficient of the solidified soil treated by the method is stabilized at 8.6×10 -9 m / s, which is nearly 50 times lower than the initial value of 4.2×10 -7 m / s, the cementing strength reaches 2.12 MPa, and the compressive strength reaches 2.63 MPa, all meeting the requirements of engineering design. The technical team selected 15 monitoring points in the construction area for a period of 6 months of field monitoring, and the average measured permeability coefficient was 9.3×10 -9The results of the method are basically consistent with the laboratory results, verifying the reliability and practicability of the method.

[0135] Compared with the traditional single cement solidification method, the method realizes multi-parameter coupling analysis through singular value decomposition algorithm, can quantitatively describe the microstructure characteristics of soil and accurately control the solidification process. The application of nano-SiO2 modified solidification agent significantly improves the chemical bonding strength of the cementation interface, the ultrasonic oscillation technology ensures the uniform distribution of the solidification agent, and the multi-factor accelerated aging test system can accurately evaluate the long-term performance attenuation law. The adaptive weight adjustment model realizes the 50-year long-term performance prediction through deep learning technology, which provides a scientific basis for engineering maintenance decision-making. These technical innovations comprehensively improve the impermeability and engineering reliability of solidified soil from the aspects of micro-mechanism, macro-performance and long-term prediction.

[0136] The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which shall be covered within the protection scope of the present application.

Claims

1. A method for enhancing the impermeability of composite solidified soil, characterized in that, The particle size distribution of soil samples was determined using a laser particle size analyzer, and the soil specific surface area data was obtained using a specific surface area analyzer. A particle size distribution-specific surface area correlation matrix was established. When the rank value of the correlation matrix was less than 3, particle sieving was performed; when the rank value was greater than 8, particle mixing was performed. A nano-SiO2 modified solidifying agent solution was prepared, with the solidifying agent dosage controlled at 5% to 15% of the soil mass. When the ion concentration exceeded 0.5 mol / L, the dilution ratio was adjusted; when the pH value deviated from the range of 7.0 to 9.0, a buffer was added for adjustment. The modified solidifying agent was thoroughly mixed with the soil, and ultrasonic oscillation was used to assist mixing for 60 to 120 minutes. The changes in moisture content and density of the mixture were monitored in real time, and a moisture content-density response matrix was established. When the rate of change of the matrix rank value exceeded 20%, the ultrasonic power density was adjusted. A soil pore structure model was constructed using molecular dynamics simulation software, and Monte Carlo simulation was performed. The Loehn method simulates the evolution of pore structure, calculates the pore connectivity index and pore size distribution parameters, and decomposes the pore structure parameter matrix into a linear combination of three low-rank matrices: pore morphology matrix, connectivity matrix, and size distribution matrix. When the pore connectivity index is less than 0.3, the compaction intensity is increased. The solidified soil samples are then subjected to standard curing, with the curing temperature controlled at 20±2℃ and the relative humidity maintained above 95%. The temperature and humidity are monitored in real time using thermocouple temperature and humidity sensors. When the temperature deviation exceeds ±1℃ or the humidity is below 90%, the environmental control system is activated. A multi-factor accelerated aging test system is established, with three accelerated conditions: temperature cycling, wet-drying cycling, and freeze-thaw cycling. The permeability coefficient is measured periodically using a permeability coefficient meter, and a permeability coefficient-time decay matrix is ​​constructed. When the decay rate exceeds 10% of the initial value, a performance warning mechanism is triggered, and a reinforcement treatment program is initiated while the curing parameters are adjusted.

2. The method for enhancing the impermeability of composite solidified soil according to claim 1, characterized in that, The particle size distribution-specific surface area correlation matrix is ​​established by performing matrix operations on the particle content of different particle size ranges and the corresponding specific surface area contribution values. The matrix elements represent the degree of contribution of the particle size range to the total specific surface area. The particle size distribution is obtained by measuring the light scattering principle of a laser particle size analyzer, and the specific surface area data is obtained by measuring the nitrogen adsorption method of a specific surface area analyzer.

3. The method for enhancing the impermeability of composite solidified soil according to claim 2, characterized in that, The nano-SiO2 modified curing agent is prepared by dispersing nano-SiO2 particles in a cement-based curing agent. The nanoparticles fill the soil pores and form chemical bonds with the soil particles. The ion concentration is obtained by measuring the conductivity of a conductivity meter, and the pH value is obtained by measuring the potential of a pH meter.

4. The method for enhancing the impermeability of composite solidified soil according to claim 3, characterized in that, The ultrasonic oscillation device uses an ultrasonic generator with a frequency of 20 to 40 kHz to promote the uniform distribution of the curing agent in the soil through mechanical vibration. The moisture content is obtained by drying method and the density is obtained by ring cutter method.

5. The method for enhancing the impermeability of composite solidified soil according to claim 4, characterized in that, The moisture content-density response matrix records the real-time correspondence between moisture content and density during the mixing process, and the matrix elements reflect the correlation strength between the two parameters.

6. The method for enhancing the impermeability of composite solidified soil according to claim 5, characterized in that, The molecular dynamics simulation software uses the LAMMPS package to establish an atomic-scale soil particle model. The Monte Carlo method simulates the evolution of pore structure through random sampling and statistics, and the pore size distribution parameters are obtained by mercury intrusion porosimetry.

7. The method for enhancing the impermeability of composite solidified soil according to claim 6, characterized in that, The pore connectivity index is defined as the ratio of the volume of connected pores to the total pore volume, with a value ranging from 0 to 1. It reflects the degree of connectivity of the pore network and the pore volume data is obtained through CT scanning technology.

8. The method for enhancing the impermeability of composite solidified soil according to claim 7, characterized in that, The multi-factor accelerated aging test system simulates various environmental effects faced by soil in actual engineering projects. The evaluation cycle is shortened through accelerated testing, and the permeability coefficient is obtained by constant head permeability test method.

9. The method for enhancing the impermeability of composite solidified soil according to claim 8, characterized in that, The permeability coefficient-time decay matrix records the variation of permeability coefficient at different ages, with rows representing time points and columns representing different experimental conditions.

10. The method for enhancing the impermeability of composite solidified soil according to claim 9, characterized in that, It also includes establishing a long-term performance prediction system based on an adaptive weight adjustment model, using a hierarchical fusion weight adjustment function to dynamically adjust model parameters, with input parameters including cementation strength, compressive strength, elastic modulus, and Poisson's ratio data.

Images