A dynamic compaction construction intelligent monitoring and self-adaptive control method based on physical information neural network

Abstract

The present application relates to a kind of physical information neural network-based dynamic compaction construction intelligent monitoring and adaptive control method, comprising: vibration sensor array is laid in construction safety area, real-time collection ramming vibration signal and extraction time domain, frequency domain and space propagation characteristic;Physical information neural network model is constructed, soil elastic wave propagation equation is used as physical constraint, with vibration characteristic and monitoring point space coordinate as input, the three-dimensional distribution of soil physical property parameter is predicted;Based on the predicted soil parameter, the bearing capacity distribution and development trend of foundation are calculated using the theory of Terzaghi bearing capacity;Multi-objective optimization model is established with bearing capacity up to standard, minimum energy consumption and shortest time as target, and subsequent ramming parameters are optimized;Construction is carried out according to the optimized parameters, and data and model parameters are continuously updated.Compared with prior art, the present application solves the problem that soil state is not observable in traditional dynamic compaction construction, and optimization lacks theoretical support, realizes intelligent and accurate control of construction process.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction and foundation treatment technology, and in particular to an intelligent monitoring and adaptive control method for dynamic compaction construction based on physical information neural networks. Background Technology

[0002] Dynamic compaction is a widely used foundation treatment technology that improves the engineering properties of foundation soil through the powerful impact energy generated by a heavy hammer falling from a height. However, traditional dynamic compaction construction faces the following technical challenges:

[0003] (1) Problem of unobservable internal soil condition: Traditional monitoring methods are mainly based on surface settlement measurement, which cannot directly observe the density distribution, modulus change and bearing capacity development inside the soil. This leads to a lack of scientific basis in the construction process and seriously affects the construction quality control. The existing technology CN116399302A proposes a compaction monitoring method based on binocular vision and neural network, which realizes non-contact real-time measurement. However, its core technology is still limited to the observation of surface displacement, which cannot reflect the changes in the internal mechanical state of the soil. Moreover, it relies on visible light imaging, and the reliability of identification decreases under dust obstruction or complex lighting conditions.

[0004] (2) Difficulty in modeling physical mechanisms: The response mechanism of soil under dynamic compaction is extremely complex, involving multiple physical processes such as nonlinear large deformation, stress wave propagation, and pore water pressure changes. Traditional empirical formulas are difficult to accurately describe these complex physical phenomena. Although the existing technology CN112252292A has established a compaction degree calculation model using artificial neural networks, its input parameters are concentrated on the control parameters of the road roller and the data of surface sensors. It does not involve deep mechanisms such as the energy transfer of dynamic compaction impact and the internal wave response of the soil. Moreover, the model relies on historical data for training, which limits its generalization ability and makes it difficult to adapt to the significant differences in geological conditions during dynamic compaction construction.

[0005] (3) Lack of theoretical basis for construction parameter optimization: Existing technologies mainly rely on engineering experience to determine the number of tamping blows, energy level and interval time, and lack systematic optimization methods. This leads to low energy utilization efficiency and high construction costs.

[0006] (4) Insufficient real-time control capability: Traditional methods cannot adjust construction parameters according to real-time changes in soil condition, making it difficult to achieve adaptive control and affecting construction efficiency and quality stability.

[0007] In summary, existing technologies have failed to address core challenges in dynamic compaction construction, such as real-time sensing of soil internal conditions, coupled modeling of multiple physical processes, and dynamic optimization of construction parameters. Therefore, a technological approach that deeply integrates soil internal response monitoring, intelligent modeling, and real-time control is urgently needed to achieve precise and intelligent dynamic compaction construction. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the existing technology by providing an intelligent monitoring and adaptive control method for dynamic compaction construction based on physical information neural networks. This method solves the technical problems in traditional dynamic compaction construction, such as the unobservable soil state, difficulty in modeling physical mechanisms, lack of theoretical basis for parameter optimization, and insufficient real-time control capability.

[0009] The objective of this invention can be achieved through the following technical solutions:

[0010] This invention provides an intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network, comprising the following steps:

[0011] S1. Deploy a vibration sensor array within a safe range of 20-50 meters from the impact point to collect ground vibration signals generated by the impact in real time, and extract the time domain characteristics, frequency domain characteristics, and spatial propagation characteristics of the vibration waveform;

[0012] S2. Construct a physical information neural network model, using the soil elastic wave propagation equation as a physical constraint, and taking the vibration characteristics extracted in S1 and the spatial coordinates of the monitoring points as inputs to predict the spatial distribution of soil physical parameters.

[0013] S3. Based on the spatial distribution of soil physical property parameters predicted by the physical information neural network model, the three-dimensional distribution and development trend of foundation bearing capacity are calculated using Terzaghi bearing capacity theory.

[0014] S4. Based on the calculated three-dimensional distribution and development trend of foundation bearing capacity, establish a multi-objective optimization model with the goals of achieving bearing capacity standards, minimizing energy consumption, and minimizing construction time, and optimize subsequent construction parameters.

[0015] S5. Perform dynamic compaction construction based on the optimized construction parameters, and continuously update the vibration characteristic data and physical information neural network model parameters during the construction process to achieve adaptive control of the construction process.

[0016] Furthermore, the time-domain features include peak amplitude. Valid value Duration ;

[0017] The frequency domain features include the dominant frequency. Frequency band energy Spectrum centroid ;

[0018] The spatial propagation characteristics include amplitude attenuation rate. Phase difference speed of transmission ,in: , These are the distances from the impact point. , The vibration amplitude at that location, , The unit is mm / s. , The unit is m. For dimensionless parameters, , These represent the sensors. i and j The radial distance between the impact point and the ramming point.

[0019] Furthermore, in S2, a physical information neural network model is constructed, using the soil elastic wave propagation equation as a physical constraint and the vibration characteristics extracted in S1 and the spatial coordinates of the monitoring points as input. The specific process for predicting the three-dimensional distribution of the soil density field, elastic modulus field, and Poisson's ratio field includes:

[0020] The vibration feature vector extracted from S1 and the spatial coordinate vector of the monitoring point are combined into an input vector;

[0021] A multi-layer feedforward neural network structure is constructed, which includes an input layer, multiple hidden layers, and an output layer. The hidden layers use the ReLU activation function for nonlinear transformation, and the output layer outputs three physical property parameters: soil density, elastic modulus, and Poisson's ratio.

[0022] During network training, the soil elastic wave propagation equation is used as a physical constraint. The residual between the network output and the physical equation is calculated using automatic differentiation technology. This residual is then incorporated into the overall loss function as the physical equation residual loss. The overall loss function consists of a weighted sum of data fitting loss, physical equation residual loss, boundary condition loss, and initial condition loss. The data fitting loss ensures that the network output is consistent with the measured data, the physical equation residual loss ensures that the prediction results conform to the physical laws of wave propagation, and the boundary condition loss and initial condition loss ensure the uniqueness and stability of the solution.

[0023] The overall loss function is minimized by a gradient-based optimization algorithm, and the network parameters are iteratively updated to finally obtain a trained physical information neural network model. This model can accurately predict the three-dimensional spatial distribution of soil physical parameters using vibration characteristics and spatial coordinates as input.

[0024] Furthermore, in S2, the input to the physical information neural network model is... ,in: For vibration characteristic vectors, It is a spatial coordinate vector;

[0025] The output of the physical information neural network model is ,in: Soil density field distribution Elastic modulus field distribution Poisson's ratio field distribution;

[0026] The loss function of the physical information neural network model is:

[0027]

[0028] in: For data fitting loss; This represents the residual loss in the physical equations. For boundary condition loss; Loss due to initial conditions; , , These are the weighting coefficients. This represents the total number of sample data points collected by the vibration sensor. It is the actual observed value of the vibration characteristic at the i-th position. It is the predicted value of the corresponding vibration characteristics by the physical information neural network model. This represents the residual obtained by substituting the field quantity u predicted by the neural network into the soil elastic wave control equation. It is the number of spatial points used when calculating physical residuals.

[0029] Furthermore, in S2, the soil elastic wave propagation equation is a three-dimensional elastic wave propagation equation, specifically as follows:

[0030]

[0031] in: It is a displacement vector. For soil density field distribution, For time, ∇ It is the Hamiltonian operator used to calculate spatial gradients and divergences. and Lamé constant, representing the elastic properties of soil, is related to the elastic modulus and Poisson's ratio as follows:

[0032]

[0033] in E This represents the elastic modulus of soil. n It represents the Poisson's ratio of soil.

[0034] Furthermore, in S3, the specific process of calculating the three-dimensional distribution and development trend of the foundation bearing capacity using Terzaghi's bearing capacity theory, based on the spatial distribution of soil physical property parameters predicted by the physical information neural network model, includes:

[0035] Based on the three-dimensional distribution data of soil density field, elastic modulus field, and Poisson's ratio field predicted by the physical information neural network model described in S2, the Lamé constant of the soil is calculated according to the elastic modulus and Poisson's ratio.

[0036] Based on the predicted soil density, the cohesion and internal friction angle of the soil are calculated using empirical relationships.

[0037] The calculated cohesion, internal friction angle, soil weight, foundation depth and width are used as input parameters, and the three-dimensional spatial distribution of foundation bearing capacity is calculated using Terzaghi's bearing capacity theory formula.

[0038] By analyzing the changes in the spatial distribution of bearing capacity under different construction stages, the development trend of foundation bearing capacity can be predicted.

[0039] Furthermore, in S3, the Terzaghi bearing capacity theoretical formula is:

[0040]

[0041] in: It refers to the bearing capacity of the foundation, measured in kPa. Cohesion, measured in kPa, is determined through empirical relationships. calculate, This is a reference value for cohesion, in kPa. It is the proportionality coefficient of cohesion as a function of density, with units of Pa / (kg / m³). 3 ), r This is the actual density of the soil, expressed in kg / m³. 3 , r 0 This is the reference density of the soil, in kg / m³. 3 ;

[0042] The weight of soil above the foundation is expressed in kN / m³. 3 ; The weight of the soil below the foundation, in kN / m³. 3 ; The foundation burial depth is expressed in meters (m). This is the base width, in meters (m). , , The bearing capacity coefficient is dimensionless and calculated based on the internal friction angle. The relationship between the internal friction angle and soil density is as follows: , This is the reference value for the friction angle, measured in rad. It is the proportionality coefficient of the friction angle as a function of density, and its unit is rad.

[0043] Furthermore, in S4, based on the calculated three-dimensional distribution and development trend of the foundation bearing capacity, a multi-objective optimization model is established with the goals of achieving bearing capacity standards, minimizing energy consumption, and minimizing construction time. The specific process of optimizing subsequent construction parameters includes:

[0044] Based on the predicted trend of foundation bearing capacity and the current spatial distribution of bearing capacity in S3, an optimization problem with multiple conflicting objective functions is constructed.

[0045] The multi-objective optimization model includes: a function designed to penalize the failure of the predicted bearing capacity to reach the design target value; a function to calculate the total energy consumption of all subsequent compaction blows; and a function to calculate the total time of all subsequent compaction intervals.

[0046] The number of subsequent tamping blows, the energy value of each tamping blow, and the interval time after each tamping blow were determined as decision variables, and constraints that conform to the scope of engineering practice were set for the decision variables.

[0047] A multi-objective evolutionary algorithm is used to solve this optimization problem, and a set of Pareto optimal solutions that achieve the best balance among multiple objectives is generated through iterative calculation.

[0048] Finally, an optimal solution is selected from the Pareto optimal solution set, and its corresponding subsequent construction parameter sequence is output, including the specific number of tamping blows, the energy configuration of each tamping blow, and the interval time arrangement.

[0049] Furthermore, in S4, the multi-objective optimization model is:

[0050]

[0051] in: As a penalty for insufficient load-bearing capacity, This is the preset target value for foundation bearing capacity. It is based on the current foundation bearing capacity value predicted by the model;

[0052]

[0053] Total energy consumption It is the energy value applied by the i-th impact;

[0054] Total construction time. It is the interval time required after the i-th tamping is completed;

[0055] As decision variables, The total number of tamping blows;

[0056] The constraints include: range of tamping times. single impact energy kJ; Intermittent time range h.

[0057] Furthermore, the specific process of implementing dynamic compaction based on optimized construction parameters, and continuously updating vibration characteristic data and physical information neural network model parameters during construction to achieve adaptive control of the construction process includes:

[0058] The subsequent compaction operation is carried out according to the optimized construction parameter sequence output by S4, and new ground vibration signals are collected in real time through the vibration sensor array deployed by S1 during each compaction process.

[0059] The newly acquired vibration signals are processed to extract new vibration feature parameters, and the vibration feature dataset in S1 is updated and expanded.

[0060] An incremental learning approach is adopted, using newly acquired vibration feature data and spatial coordinates as training samples to update the parameters of the physical information neural network model constructed by S2 and optimize its prediction accuracy.

[0061] Based on the updated physical information neural network model, the bearing capacity calculation and trend prediction of S3 are re-executed, and the subsequent construction parameters are dynamically adjusted through the multi-objective optimization model of S4.

[0062] By collecting data in real time, continuously updating the model, and cyclically optimizing parameters, a closed-loop feedback control process is formed, enabling intelligent monitoring and adaptive control of the dynamic compaction construction process.

[0063] Compared with the prior art, the present invention has the following beneficial effects:

[0064] First, by deploying a vibration sensor array and extracting multi-dimensional features of the vibration waveform, a comprehensive perception of energy propagation and soil response signals during the compaction process was achieved, solving the technical problem that the internal state of the soil cannot be directly observed in traditional methods, and providing a reliable data foundation for subsequent analysis.

[0065] Second, by constructing a physical information neural network model, the physical laws of soil elastic wave propagation are embedded as constraints in the learning process. This effectively combines vibration monitoring data with the inherent physical mechanism, solving the difficulty of accurately modeling the constitutive relationship of complex soils and significantly improving the accuracy of soil parameter inversion and state prediction.

[0066] Third, based on the soil parameters predicted by the physical information neural network, Terzaghi's bearing capacity theory is used to scientifically assess the development state of the foundation bearing capacity and establish a multi-objective optimization model. This transforms the decision-making of construction parameters from relying on experience to an optimization process based on model prediction and theoretical calculation, thus solving the problem of lack of theoretical basis for parameter optimization.

[0067] Fourth, by continuously collecting new data and updating the model while executing optimized parameters, a closed-loop control loop of perception, prediction, optimization, execution, and feedback is formed, realizing dynamic adjustment of construction parameters and autonomous optimization of the process. This solves the problem of insufficient real-time control capability in traditional dynamic compaction construction and significantly improves construction quality and efficiency. Attached Figure Description

[0068] Figure 1 This is a flowchart illustrating the technical solution of the present invention. Detailed Implementation

[0069] Physics-Informed Neural Networks (PINNs), as an emerging machine learning method, can embed physical laws into neural networks to achieve high-precision predictions under small sample conditions. This invention introduces PINNs technology into the field of dynamic compaction construction, providing a new approach to solving the technical challenges in existing technologies.

[0070] This invention presents an intelligent monitoring and adaptive control method for dynamic compaction construction based on physical information neural networks. This method involves deploying a vibration sensor array within the safe construction area to collect ground vibration signals caused by compaction in real time and extracting characteristic parameters of the vibration waveform. A PINNs model integrating the physical equations of soil wave propagation is constructed, using vibration characteristics and spatial coordinates as inputs to predict the three-dimensional distribution of the soil's density, modulus, and Poisson's ratio fields. The development of foundation bearing capacity is calculated based on Terzaghi's bearing capacity theory. Finally, a multi-objective optimization algorithm is used to determine the optimal construction parameters. This method achieves accurate prediction of the soil's internal state and intelligent control of the construction process.

[0071] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Component models, material names, connection structures, circuit structures, control methods, algorithms, and other features not explicitly described in this technical solution are considered common technical features disclosed in the prior art.

[0072] Example 1

[0073] The technical objectives to be achieved by the technical solution of this embodiment include: (1) constructing a physical mapping relationship between vibration characteristics and the internal state of soil, so as to realize accurate prediction of soil density distribution and bearing capacity development that cannot be directly observed; (2) developing a PINNs modeling method based on physical constraints to ensure that the prediction results conform to the basic laws of soil mechanics and improve the prediction accuracy under small sample conditions; (3) establishing a multi-objective optimization algorithm for construction parameters to scientifically determine the optimal number of tamping blows, energy configuration and time arrangement, so as to maximize energy utilization efficiency.

[0074] This invention is achieved through the following technical solution:

[0075] This invention provides an intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network, referencing... Figure 1 Specifically, it includes the following steps:

[0076] S1: Deploy a three-dimensional vibration sensor array within a safe range of 20-50 meters from the impact point to collect ground vibration signals generated by the impact and extract the time domain, frequency domain, and spatial propagation characteristics of the vibration waveform.

[0077] In specific implementation, in S1, a three-ring three-dimensional vibration sensor array is deployed within a safe range of 20-50 meters from the impact point. GeoSpace's GS-20DX triaxial velocity sensor is used, with a frequency response range of 4.5Hz-400Hz and a sensitivity of 28.8 V / (m / s). The sensor array is distributed in three rings, positioned at 25 meters, 40 meters, and 55 meters respectively, with four measuring points in each ring, forming a spatial layout of 12 monitoring points.

[0078] In specific implementation, in S1, the three-ring sensor deployment scheme is as follows: near-field ring ( ) Capture high-intensity vibration signals to obtain shallow dynamic response characteristics of soil; mid-field ring ( ) Collects medium-intensity vibration signals to reflect changes in the physical parameters of the middle layer of soil; far-field ring ( m) Record the attenuated vibration signal to obtain the deep propagation characteristics of the soil. Each sensor is buried at a depth of 0.6m, with a 10cm thick fine sand cushion layer at the bottom, and is protected by a high-strength steel cover with an IP68 sealing rating, capable of withstanding 100g impact acceleration.

[0079] In specific implementation, in S1, the system uses a sampling frequency of 1000Hz to acquire vibration signals, employs a 24-bit high-precision A / D converter, and has a dynamic range greater than 120dB. The acquired raw vibration signals are first preprocessed, including steps such as removing DC components, bandpass filtering to remove high-frequency noise and low-frequency interference, and outlier detection and rejection.

[0080] In specific implementation, in S1, vibration feature extraction includes three types of key vibration feature parameters: time-domain features include peak amplitude. Valid value Duration Frequency domain characteristics include the dominant frequency. Frequency band energy Spectrum centroid Spatial propagation characteristics include amplitude attenuation rate. Phase difference speed of transmission .in , These are the distances from the impact point. , The vibration amplitude at that location, , The unit is mm / s. , The unit is m. For dimensionless parameters, , These represent the sensors. i and j The radial distance between the impact point and the vibration sensor. In practice, this is achieved by selecting two vibration sensors at different locations (e.g., the distance from the impact point). =25 meters and =40 meters), the actual measured distance difference With the corresponding amplitude value ( , Substitute directly into the formula To calculate the amplitude attenuation rate.

[0081] S2: Construct a Physical Information Neural Network (PINNs) model, using the soil elastic wave propagation equation as a physical constraint and the vibration feature vector and three-dimensional spatial coordinates as inputs to predict the spatial distribution of soil physical parameters.

[0082] In S2, a physical information neural network model is constructed. The network input includes an 8-dimensional vibration feature vector and 3-dimensional spatial coordinates, and the output consists of three physical parameters: soil density, elastic modulus, and Poisson's ratio at the corresponding location. The network adopts an encoder-decoder structure. The encoder part uses a three-layer fully connected neural network. The first layer maps the 8-dimensional vibration features to a 64-dimensional hidden representation. The second layer expands to 128 dimensions and enhances the feature representation capability. The third layer compresses to 64 dimensions, and the fourth layer further compresses to a 32-dimensional encoding vector. The ReLU activation function is used between each layer. The decoder part connects the 32-dimensional encoding vector with the 3-dimensional spatial coordinates to form a 35-dimensional input vector. This 35-dimensional input vector is then passed through a 64-dimensional fully connected network to finally output a 3-dimensional soil parameter vector.

[0083] In practical implementation, in S2, the differential equation for the propagation of elastic waves in the soil is used as a physical constraint condition: ,in It is a displacement vector. and Let Lamé constant be . For soil density field distribution, For time, ∇ It is the Hamiltonian operator used to calculate spatial gradients and divergences. Automatic differentiation techniques are used to calculate the gradient of the network output parameters with respect to spatial coordinates, thereby calculating wave velocity and vibration attenuation patterns. The physical constraint loss function is defined as the mean square error between the predicted vibration attenuation and the actual observed attenuation.

[0084] and The relationship between elastic modulus and Poisson's ratio is as follows:

[0085]

[0086] in E This represents the elastic modulus of soil. n It represents the Poisson's ratio of soil.

[0087] In specific implementation, in S2, the network is configured with learnable reference density parameters (1800 kg / m³) and reference modulus parameters (30 MPa) to ensure that the output parameters are within a reasonable physical range. The Adam optimizer is used for network training, with an initial learning rate of 0.001 and a training cycle of 3000 epochs. The total loss function is a weighted sum of the data fitting loss and the physical constraint loss. ,in The fitting loss for the observed data, For the residual loss of the physical equation, For boundary condition loss, For initial conditional loss, the weighting coefficients , , . This represents the total number of sample data points collected by the vibration sensor. It is the actual observed value of the vibration characteristic at the i-th position. It is the predicted value of the corresponding vibration characteristics by the physical information neural network model. This represents the residual obtained by substituting the field quantity u predicted by the neural network into the soil elastic wave control equation. It is the number of spatial points used when calculating physical residuals.

[0088] S3: Based on the soil physical parameters predicted by PINNs, the three-dimensional distribution and development trend of foundation bearing capacity are calculated using Terzaghi's bearing capacity theory;

[0089] In practical implementation, in S3, the trained PINNs model can predict the density, modulus, and Poisson's ratio distribution at any location within the soil. Based on these predicted parameters, Terzaghi's bearing capacity theory is used to calculate the foundation bearing capacity: , It is the bearing capacity of the foundation, measured in kPa. Cohesion, measured in kPa, is determined through empirical relationships. calculate, This is a reference value for cohesion, in kPa. It is the proportionality coefficient of cohesion as a function of density, with units of Pa / (kg / m³). 3 ), r This is the actual density of the soil, expressed in kg / m³. 3 , r 0 This is the reference density of the soil, in kg / m³. 3 ; The weight of soil above the foundation is expressed in kN / m³. 3 , The weight of the soil below the foundation, in kN / m³. 3 ; The foundation burial depth is expressed in meters (m). This is the base width, in meters (m). , , The bearing capacity coefficient is dimensionless and calculated based on the internal friction angle. The relationship between the internal friction angle and soil density is: . This is the reference value for the friction angle, measured in rad. It is the proportionality coefficient of the friction angle as a function of density, and its unit is rad.

[0090] S4: Establish a multi-objective optimization model with the objectives of achieving bearing capacity standards, minimizing energy consumption, and minimizing time to optimize subsequent construction parameters;

[0091] In practical implementation, in S4, a multi-objective optimization model is established with the goals of achieving the bearing capacity standard, minimizing energy consumption, and minimizing construction time: ,in: As a penalty for insufficient load-bearing capacity, Total energy consumption The total construction time is the decision variable. The optimization constraints include: number of impacts (3-15 times), single impact energy (50-500kJ), and interval time (0.5-24 hours).

[0092] In specific implementation, in S4, the NSGA-II algorithm is used to solve the multi-objective optimization problem. The algorithm parameters are set as follows: population size 100, maximum number of iterations 500, crossover probability 0.9, mutation probability 0.1, and tournament selection strategy. A cumulative simulation model of the impact effect is established to predict the cumulative effect of multiple impacts. The formula for calculating the bearing capacity increment is: ,in To strike energy, This refers to energy conversion efficiency. The formula for calculating the consolidation effect coefficient is: ,in This refers to the interval time.

[0093] S5: Execute construction based on optimized parameters and continuously update the PINNs model to achieve adaptive control.

[0094] In practical implementation, S5 establishes a model update mechanism based on incremental learning. After each compaction, the vibration feature dataset is automatically updated, and the PINNs model parameters are updated using incremental learning. Hyperparameters are adjusted based on Bayesian optimization, and prediction uncertainties are assessed in real time to adjust the control strategy. The system has a learning update function: after all construction at each compaction point is completed, the measured bearing capacity data is fed back to the PINNs model to continuously optimize prediction accuracy.

[0095] The specific process includes:

[0096] After each tamping operation, new ground vibration signals are collected through the vibration sensor array deployed in S1, vibration characteristic parameters are extracted, and these new data are added to the existing vibration characteristic dataset to achieve dynamic expansion of the dataset.

[0097] An incremental learning algorithm is used to update the parameters of the physical information neural network model constructed by S2 with newly collected vibration feature data and corresponding spatial coordinates as new training samples, thereby improving the model's adaptability to the current construction conditions without losing existing knowledge.

[0098] Based on the Bayesian optimization framework, key hyperparameters such as the network learning rate are dynamically adjusted according to the model’s performance on new data, thereby optimizing the training process and improving the model’s convergence performance.

[0099] By calculating the posterior distribution of the prediction results, the uncertainty of the physical information neural network model output is evaluated in real time, and the conservatism of the control strategy is adjusted according to the degree of uncertainty.

[0100] After all the construction at each compaction point is completed, the measured bearing capacity data is obtained through in-situ testing. This measured data is then fed back to the physical information neural network model as a truth label to further optimize the model parameters and continuously improve the accuracy of soil parameter prediction and bearing capacity calculation.

[0101] In practice, the final step also includes a prediction result verification step: by comparing the soil state predicted by PINNs with the in-situ test results of standard penetration tests, static cone penetration tests, and load tests, the accuracy of the model is verified. A prediction accuracy evaluation system is established to achieve quantitative evaluation and quality control of the prediction results.

[0102] The specific process includes:

[0103] Based on the predicted soil physical properties and foundation bearing capacity obtained from S2 and S3, the measured soil parameters and bearing capacity data at the corresponding locations are obtained through various in-situ testing methods such as standard penetration test, static cone penetration test and load test.

[0104] The prediction results of the physical information neural network model are systematically compared with the measured data obtained from various in-situ tests. The error between the predicted value and the measured value is calculated to verify the accuracy of the model prediction.

[0105] Establish a prediction accuracy evaluation system that includes multiple quantitative evaluation indicators, and evaluate the accuracy of the prediction results for soil density, elastic modulus, Poisson's ratio and bearing capacity respectively;

[0106] The reliability of the physical information neural network model is graded based on the accuracy assessment results, and the comparison results are fed back to the model training process of S2 to provide a basis for optimizing model parameters.

[0107] When the prediction accuracy does not meet the engineering requirements, the model retraining mechanism is triggered. The prediction performance is improved by increasing training data or adjusting the network structure, thereby achieving closed-loop control of the dynamic compaction construction quality.

[0108] Application Example 1

[0109] Foundation treatment project in an industrial park

[0110] 1. Project Overview

[0111] This application example uses a 15,000-square-meter foundation treatment project in an industrial park. The construction area is a rectangular site of 150m × 100m, and the design requirement is a foundation bearing capacity of 200 kPa. The original soil is soft plastic silty clay with a natural moisture content of [missing information]. Liquid limit index internal friction angle Initial bearing capacity The pressure is only 85 kPa. The impact points are arranged in a square grid with a spacing of 6 m × 6 m, totaling 420 impact points. A 3000 kJ dynamic compaction machine is used for construction, with a hammer mass of 30 t and a maximum drop height of 10 m.

[0112] 2. Sensor system deployment

[0113] Detailed monitoring was conducted at a representative impact point P-125 located in the center of the site, with coordinates (75 m, 50 m). The sensor deployment scheme was as follows: First ring ( ): 4 measuring points, azimuth angle , , , Second ring ( Four measuring points, azimuth angle The third ring ( ): 4 measuring points, azimuth angle A total of 12 three-dimensional vibration sensors, model GS-20DX, with a frequency response range of 4.5Hz-400Hz and a sensitivity of 28.8 V / (m / s). Installation requirements include a burial depth of 0.6 m for each sensor, a 10 cm thick fine sand pad at the bottom, and a high-strength steel protective cover with an IP68 sealing rating, capable of withstanding 100g impact acceleration.

[0114] 3. Vibration data acquisition and feature extraction

[0115] First impact data (energy 3000 kJ):

[0116] Measuring point T1 ( Recorded data: Peak amplitude mm / s; RMS value mm / s; duration s; main frequency Hz; band energy J·s; Spectral centroid Hz;

[0117] Measuring point T5 ( Record data: mm / s; mm / s; s; Hz; J.s; Hz;

[0118] Measuring point T9 ( Record data: mm / s; mm / s; s; Hz; J.s; Hz;

[0119] The amplitude attenuation rate is:

[0120]

[0121] Calculated based on the time difference of arrival of the wave crest P wave velocity m / s, S wave velocity m / s.

[0122] 4. PINNs Model Construction and Training

[0123] The input feature vector is constructed as The network structure is 11→64→128→64→32→3; the activation function is ReLU; the optimizer is Adam, the learning rate is set to 0.001; the batch size is set to 32; and the number of training epochs is 3000.

[0124] The loss function weights are set as follows: data fitting loss weight is 1.0; physical constraint loss weight is 1.0; boundary condition loss weight is 0.1; continuity loss weight is 0.1. The training process is monitored as follows: Round 500: Round 1000: Round 1500: Round 2000: Round 3000: .

[0125] 5. Soil parameter prediction results

[0126] Based on the trained PINNs model, predict the distribution of soil parameters around point P-125:

[0127] The predicted soil density at a depth of 1.0 m is: kg / m³; Elastic modulus: MPa; Poisson's ratio: ;

[0128] The prediction result at a depth of 2.0 m is as follows: kg / m³; MPa; ;

[0129] Prediction results at a depth of 3.0 m: kg / m³; MPa; .

[0130] 6. Bearing capacity calculation

[0131] Based on Terzaghi's bearing capacity theory, the bearing capacity at various depths was calculated:

[0132] Calculation of internal friction angle: ;

[0133] Bearing capacity coefficient: , , ;

[0134] Bearing capacity calculation (foundation width) burial depth ): ;

[0135] Considering a safety factor of 2.5, the design bearing capacity is 157.6 kPa, which is less than the target value of 200 kPa, so further compaction is required.

[0136] 7. Multi-objective optimization calculation

[0137] Optimize goal setting: Penalty for insufficient load-bearing capacity Minimize energy consumption; Minimize construction time. The NSGA-II algorithm parameters are set as follows: population size: 100; maximum number of iterations: 500; crossover probability: 0.9; mutation probability: 0.1. The optimization converges after 472 iterations, obtaining the Pareto optimal solution set.

[0138] Option A (Balanced Option): Number of subsequent compaction blows: 6; Energy configuration: [2400, 2100, 1800, 1500, 1200, 900] kJ; Interval time: [2, 4, 8, 12, 16, 24] hours; Predicted final bearing capacity: 202.5 kPa; Total energy consumption: 10800 kJ; Total construction time: 66 hours;

[0139] Option B (Rapid Solution): Number of subsequent compaction blows: 4; Energy configuration: [2700, 2400, 2100, 1800] kJ; Interval time: [1, 2, 4, 8] hours; Predicted final bearing capacity: 201.2 kPa; Total energy consumption: 9000 kJ; Total construction time: 15 hours;

[0140] Option C (Energy-Saving Option): Number of subsequent compaction blows: 8; Energy configuration: [2100, 1900, 1700, 1500, 1300, 1100, 900, 700] kJ; Interval time: [3, 6, 9, 12, 15, 18, 21, 24] hours; Predicted final bearing capacity: 203.8 kPa; Total energy consumption: 11200 kJ; Total construction time: 108 hours.

[0141] 8. Construction Execution and Verification

[0142] The actual implementation result after choosing option A for construction was as follows:

[0143] Second compaction (energy 2400kJ, 2-hour interval): Measured settlement: 11.2 cm; PINNs predicted bearing capacity: 165.3 kPa;

[0144] Third impact (energy 2100kJ, 4-hour interval): measured settlement: 8.7 cm; PINNs predicted bearing capacity: 178.9 kPa; new samples added, retrained.

[0145] 7th impact (energy 900kJ, 24-hour interval): measured settlement: 2.1 cm; PINNs predicted bearing capacity: 201.8 kPa.

[0146] Final verification test:

[0147] Load test: Characteristic value of bearing capacity: kPa; PINNs predicted value: 201.8 kPa; prediction error: 1.6%.

[0148] The above description of the embodiments is provided to enable those skilled in the art to understand and use the invention. It will be apparent to those skilled in the art that various modifications can be made to these embodiments, and the general principles described herein can be applied to other embodiments without inventive effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the invention should be within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring and adaptive control of dynamic compaction construction based on physical information neural networks, characterized in that, Includes the following steps: S1. Deploy a vibration sensor array within a safe range of 20-50 meters from the impact point to collect ground vibration signals generated by the impact in real time, and extract the time domain characteristics, frequency domain characteristics, and spatial propagation characteristics of the vibration waveform; S2. Construct a physical information neural network model, using the soil elastic wave propagation equation as a physical constraint, and taking the vibration characteristics extracted in S1 and the spatial coordinates of the monitoring points as inputs to predict the spatial distribution of soil physical parameters. S3. Based on the spatial distribution of soil physical property parameters predicted by the physical information neural network model, the three-dimensional distribution and development trend of foundation bearing capacity are calculated using Terzaghi bearing capacity theory. S4. Based on the calculated three-dimensional distribution and development trend of foundation bearing capacity, establish a multi-objective optimization model with the goals of achieving bearing capacity standards, minimizing energy consumption, and minimizing construction time, and optimize subsequent construction parameters. S5. Perform dynamic compaction construction based on the optimized construction parameters, and continuously update the vibration characteristic data and physical information neural network model parameters during the construction process to achieve adaptive control of the construction process; In S2, a physical information neural network model is constructed, using the soil elastic wave propagation equation as a physical constraint and the vibration characteristics extracted in S1 and the spatial coordinates of the monitoring points as input. The specific process of predicting the three-dimensional distribution of the soil density field, elastic modulus field, and Poisson's ratio field includes: The vibration feature vector extracted from S1 and the spatial coordinate vector of the monitoring point are combined into an input vector; A multi-layer feedforward neural network structure is constructed, which includes an input layer, multiple hidden layers, and an output layer. The hidden layers use the ReLU activation function for nonlinear transformation, and the output layer outputs three physical property parameters: soil density, elastic modulus, and Poisson's ratio. During network training, the soil elastic wave propagation equation is used as a physical constraint. The residual between the network output and the physical equation is calculated using automatic differentiation technology. This residual is then incorporated into the overall loss function as the physical equation residual loss. The overall loss function consists of a weighted sum of data fitting loss, physical equation residual loss, boundary condition loss, and initial condition loss. The data fitting loss ensures that the network output is consistent with the measured data, the physical equation residual loss ensures that the prediction results conform to the physical laws of wave propagation, and the boundary condition loss and initial condition loss ensure the uniqueness and stability of the solution. The overall loss function is minimized by a gradient-based optimization algorithm, and the network parameters are iteratively updated to finally obtain a trained physical information neural network model. This model can accurately predict the three-dimensional spatial distribution of soil physical parameters using vibration characteristics and spatial coordinates as input.

2. The intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network according to claim 1, characterized in that, The time-domain features include peak amplitude. Valid value Duration ; The frequency domain features include the dominant frequency. Frequency band energy Spectrum centroid ; The spatial propagation characteristics include amplitude attenuation rate. Phase difference speed of transmission ,in: , These are the distances from the impact point. , The vibration amplitude at that location, , The unit is mm / s. , The unit is m. For dimensionless parameters, , These represent the sensors. i and j The radial distance between the impact point and the ramming point.

3. The intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network according to claim 1, characterized in that, In S2, the input to the physical information neural network model is ,in: For vibration characteristic vectors, It is a spatial coordinate vector; The output of the physical information neural network model is ,in: Soil density field distribution Elastic modulus field distribution Poisson's ratio field distribution; The loss function of the physical information neural network model is: in: For data fitting loss; This represents the residual loss in the physical equations. For boundary condition loss; Loss due to initial conditions; , , These are the weighting coefficients. This represents the total number of sample data points collected by the vibration sensor. It is the actual observed value of the vibration characteristic at the i-th position. It is the predicted value of the corresponding vibration characteristics by the physical information neural network model. This represents the residual obtained by substituting the field quantity u predicted by the neural network into the soil elastic wave control equation. It is the number of spatial points used when calculating physical residuals.

4. The intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network according to claim 1, characterized in that, In S2, the soil elastic wave propagation equation is a three-dimensional elastic wave propagation equation, and the soil elastic wave propagation equation is specifically as follows: in: It is a displacement vector. For soil density field distribution, For time, ∇ It is the Hamiltonian operator used to calculate spatial gradients and divergences. and Lamé constant, representing the elastic properties of soil, is related to the elastic modulus and Poisson's ratio as follows: in E This represents the elastic modulus of soil. ν It represents the Poisson's ratio of soil.

5. The intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network according to claim 1, characterized in that, In S3, the specific process of calculating the three-dimensional distribution and development trend of foundation bearing capacity using Terzaghi's bearing capacity theory, based on the spatial distribution of soil physical property parameters predicted by the physical information neural network model, includes: Based on the three-dimensional distribution data of soil density field, elastic modulus field, and Poisson's ratio field predicted by the physical information neural network model described in S2, the Lamé constant of the soil is calculated according to the elastic modulus and Poisson's ratio. Based on the predicted soil density, the cohesion and internal friction angle of the soil are calculated using empirical relationships. The calculated cohesion, internal friction angle, soil weight, foundation depth and width are used as input parameters, and the three-dimensional spatial distribution of foundation bearing capacity is calculated using Terzaghi's bearing capacity theory formula. By analyzing the changes in the spatial distribution of bearing capacity under different construction stages, the development trend of foundation bearing capacity can be predicted.

6. The intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network according to claim 5, characterized in that, In S3, the Terzaghi bearing capacity theoretical formula is: in: It refers to the bearing capacity of the foundation, measured in kPa. Cohesion, measured in kPa, is determined through empirical relationships. calculate, This is a reference value for cohesion, in kPa. It is the proportionality coefficient of cohesion as a function of density, with units of Pa / (kg / m³). 3 ), ρ This is the actual density of the soil, expressed in kg / m³. 3 , ρ 0 This is the reference density of the soil, in kg / m³. 3 ; The weight of soil above the foundation is expressed in kN / m³. 3 ; The weight of the soil below the foundation, in kN / m³. 3 ; The foundation burial depth is expressed in meters (m). This is the base width, in meters (m). , , The bearing capacity coefficient is dimensionless and calculated based on the internal friction angle. The relationship between the internal friction angle and soil density is as follows: , This is the reference value for the friction angle, measured in rad. It is the proportionality coefficient of the friction angle as a function of density, and its unit is rad.

7. The intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network according to claim 1, characterized in that, In S4, based on the calculated three-dimensional distribution and development trend of the foundation bearing capacity, a multi-objective optimization model is established with the goals of achieving bearing capacity standards, minimizing energy consumption, and minimizing construction time. The specific process of optimizing subsequent construction parameters includes: Based on the predicted trend of foundation bearing capacity and the current spatial distribution of bearing capacity in S3, an optimization problem with multiple conflicting objective functions is constructed. The multi-objective optimization model includes: a function designed to penalize the failure of the predicted bearing capacity to reach the design target value; a function to calculate the total energy consumption of all subsequent compaction blows; and a function to calculate the total time of all subsequent compaction intervals. The number of subsequent tamping blows, the energy value of each tamping blow, and the interval time after each tamping blow were determined as decision variables, and constraints that conform to the scope of engineering practice were set for the decision variables. A multi-objective evolutionary algorithm is used to solve this optimization problem, and a set of Pareto optimal solutions that achieve the best balance among multiple objectives is generated through iterative calculation. Finally, an optimal solution is selected from the Pareto optimal solution set, and its corresponding subsequent construction parameter sequence is output, including the specific number of tamping blows, the energy configuration of each tamping blow, and the interval time arrangement.

8. The intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network according to claim 1, characterized in that, In S4, the multi-objective optimization model is: in: As a penalty for insufficient load-bearing capacity, This is the preset target value for foundation bearing capacity. It is based on the current foundation bearing capacity value predicted by the model; Total energy consumption It is the energy value applied by the i-th impact; Total construction time. It is the interval time required after the i-th tamping is completed; As decision variables, The total number of tamping blows; The constraints include: range of tamping times. single impact energy kJ; Intermittent time range h.

9. The intelligent monitoring and adaptive control method for dynamic compaction construction based on a physical information neural network according to claim 1, characterized in that, The specific process of implementing dynamic compaction based on optimized construction parameters, and continuously updating vibration characteristic data and physical information neural network model parameters during construction to achieve adaptive control of the construction process includes: The subsequent compaction operation is carried out according to the optimized construction parameter sequence output by S4, and new ground vibration signals are collected in real time through the vibration sensor array deployed by S1 during each compaction process. The newly acquired vibration signals are processed to extract new vibration feature parameters, and the vibration feature dataset in S1 is updated and expanded. An incremental learning approach is adopted, using newly acquired vibration feature data and spatial coordinates as training samples to update the parameters of the physical information neural network model constructed by S2 and optimize its prediction accuracy. Based on the updated physical information neural network model, the bearing capacity calculation and trend prediction of S3 are re-executed, and the subsequent construction parameters are dynamically adjusted through the multi-objective optimization model of S4. By collecting data in real time, continuously updating the model, and cyclically optimizing parameters, a closed-loop feedback control process is formed, enabling intelligent monitoring and adaptive control of the dynamic compaction construction process.

Images