Dynamic compaction parameter optimization method and equipment for multi-modal data and medium

Abstract

The invention relates to the technical field of construction, in particular to a dynamic compaction parameter optimization method and equipment for multi-modal data and a medium. The method comprises the steps that vibration time history data and building attribute information in the dynamic compaction construction process are acquired; based on the vibration time history data, dynamically inverting stratigraphic fluctuation parameters of the construction area through a trained convolutional neural network model; based on the stratigraphic fluctuation parameters and the building attribute information, constructing a vibration propagation attenuation correction model fused with a building vibration sensitivity weight; for the current to-be-constructed tamping point, based on the first target and the second target, establishing a multi-target optimization model; and based on a prediction result provided by the vibration propagation attenuation correction model, a multi-objective optimization algorithm is adopted to solve and obtain the optimal tamping energy and the optimal tamping point azimuth angle of the current tamping point. According to the method, the prediction accuracy, the decision-making intelligence and the adaptability to complex dynamic working conditions are improved, so that the construction quality and the environment safety are collaboratively optimized.

Description

Technical Field

[0001] This invention relates to the field of construction technology, and in particular to a method, equipment, and medium for optimizing dynamic compaction parameters based on multimodal data. Background Technology

[0002] Dynamic compaction is a construction method that uses the enormous impact energy generated by a heavy hammer falling freely from a height to compact deep foundations. Due to its economy and efficiency, it is widely used in foundation treatment for large-scale projects such as airports, ports, and warehousing and logistics centers. However, the strong vibration waves generated during construction can propagate to the surrounding soil, potentially causing structural damage, functional failure, or discomfort to nearby buildings, structures, precision instruments, or underground pipelines, posing significant environmental safety risks and social problems. With the advancement of urban renewal and infrastructure construction, the demand for dynamic compaction in densely built-up areas or environmentally sensitive areas is increasing, and vibration control has become a core challenge restricting its safe and efficient application.

[0003] Currently, existing technical solutions have significant shortcomings in terms of prediction accuracy, decision-making intelligence, and adaptability to complex dynamic working conditions. The core challenge in dynamic compaction construction—"how to achieve the designed soil compaction density most efficiently and economically while ensuring the absolute safety of the surrounding environment"—still relies heavily on the personal experience of construction personnel for trial-and-error adjustments, lacking a scientific and intelligent decision support system.

[0004] Therefore, there is an urgent need for a robust parameter optimization method that can improve the accuracy of prediction and the intelligence of decision-making. Summary of the Invention

[0005] In view of the above problems, this disclosure provides a method, device and medium for optimizing dynamic compaction parameters of multimodal data to overcome or at least partially solve the above problems, with the aim of improving the accuracy of prediction, the intelligence of decision-making and the adaptability to complex dynamic working conditions, thereby synergistically optimizing construction quality and environmental safety.

[0006] The objective of this invention can be achieved through the following technical solutions: A first aspect of the present invention provides a method for optimizing dynamic compaction parameters of multimodal data, comprising: Obtain vibration time history data and building attribute information during dynamic compaction construction; Based on the vibration time history data, the ground fluctuation parameters of the construction area are dynamically inverted using a trained convolutional neural network model; based on the ground fluctuation parameters and the building attribute information, a vibration propagation attenuation correction model that integrates building vibration sensitivity weights is constructed. For the current tamping points to be constructed, a multi-objective optimization model is established based on the first and second objectives; Based on the prediction results provided by the vibration propagation attenuation correction model, the optimal impact energy and optimal azimuth of the current compaction point are obtained by using a multi-objective optimization algorithm.

[0007] further, The first objective is to maximize the predicted soil density increment within the influence area of ​​the current compaction point to be constructed; the second objective is to minimize the risk-weighted vibration value of all surrounding building locations.

[0008] further, The building attribute information includes basic attribute information and functional attribute information; the basic attribute information shown includes structural type, foundation type, construction year, and current status.

[0009] further, The dynamic inversion of ground fluctuation parameters in the construction area based on the vibration time history data using a trained convolutional neural network model includes: The vibration time history data from multiple measurement points are arranged according to the spatial location of the sensors to form a vibration feature map, which serves as the input to the convolutional neural network encoder-decoder structure. The output of the network is the stratum wave parameters, including the thickness, shear wave velocity, density, and damping ratio of each soil layer.

[0010] further, The vibration propagation attenuation correction model, which integrates building vibration sensitivity weights, is constructed based on the ground wave parameters and the building attribute information, including: The building vibration sensitivity weighting coefficient is calculated based on the building attribute information; the coefficient is used as a risk amplification factor and coupled into the two-dimensional non-uniform medium wave equation established based on the inverted stratum parameters; the output of the vibration propagation attenuation correction model is the risk-weighted vibration value.

[0011] further, The establishment of the multi-objective optimization model includes: The decision variables of the multi-objective optimization model are the impact energy E and the azimuth angle θ of the impact point. Based on safety constraints and equipment constraints, a multi-objective optimization model is constructed, including a first multi-objective optimization model based on a first objective and a second multi-objective optimization model based on a second objective. The first multi-objective optimization model is expressed as: Where X = [E, θ], Calculate the total number of grid points within the target area; This represents the soil compaction increment at the j-th grid point; The second multi-objective optimization model is expressed as: ;in For the number of buildings; This represents the risk-weighted vibration value.

[0012] Furthermore, the method of using a multi-objective optimization algorithm to solve the problem includes: using a constrained multi-objective particle swarm optimization algorithm and a fast surrogate model constructed through a deep neural network to predict vibration values; the multi-objective optimization algorithm outputs a Pareto optimal solution set, and the final implementation scheme is selected through a multi-attribute decision method.

[0013] A second aspect of the present invention provides a system for optimizing dynamic compaction parameters for multimodal data, comprising: The data acquisition module is used to acquire vibration time history data and building attribute information during the dynamic compaction construction process; The parameter optimization module is used to dynamically invert the ground fluctuation parameters of the construction area based on the vibration time history data through a trained convolutional neural network model; and to construct a vibration propagation attenuation correction model that integrates building vibration sensitivity weights based on the ground fluctuation parameters and the building attribute information. The multi-model optimization module is used to establish a multi-objective optimization model for the current compaction point to be constructed, based on the first objective and the second objective. The decision analysis module is used to obtain the optimal impact energy and optimal azimuth of the current tamping point based on the prediction results provided by the vibration propagation attenuation correction model and by using a multi-objective optimization algorithm.

[0014] A third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the dynamic compaction parameter optimization method for multimodal data as described in the first aspect.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored thereon, which, when executed by one or more processors, cause the processors to perform the dynamic compaction parameter optimization method for multimodal data as described in the first aspect.

[0016] The technical solution proposed in this application can bring the following beneficial effects: 1. This invention dynamically inverts the ground fluctuation parameters of the construction area through a trained convolutional neural network model, and constructs a vibration propagation attenuation correction model by combining building sensitivity weights. This achieves high-precision and forward-looking prediction of vibration impact from physical quantity to risk quantity, overcoming the limitations of traditional empirical formulas and static models.

[0017] 2. This invention models the construction parameter decision-making for each compaction point as a multi-objective optimization problem and uses an algorithm to solve it within seconds, achieving a real-time, quantitative, and optimal balance between quality and safety objectives, replacing the trial-and-error adjustment that relies on experience.

[0018] The above description is merely an overview of the technical solution disclosed herein. In order to better understand the technical means of this disclosure and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this disclosure more apparent and understandable, specific embodiments of this disclosure are described below. Attached Figure Description

[0019] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this disclosure. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a flowchart illustrating the steps of a method for optimizing dynamic compaction parameters based on multimodal data, as provided in the embodiments of this specification. Figure 2 This is a schematic diagram of the structure of a dynamic compaction parameter optimization system for multimodal data provided in the embodiments of this specification. Detailed Implementation

[0020] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. The technical solutions provided by various embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0021] This application provides a method, device, and medium for optimizing dynamic compaction parameters of multimodal data.

[0022] The first aspect of the present invention provides a method for optimizing dynamic compaction parameters of multimodal data, comprising the following steps: S101. Obtain vibration time history data and building attribute information during the dynamic compaction construction process; Among them, vibration time history data is a sequence of physical quantities that change over time, recorded by vibration sensors at the moment of impact and for a period of time thereafter; specifically including: A composite network layout scheme of "concentric rings + radial" is adopted. The inner ring (e.g., radii of 5m, 10m, 15m) is densely deployed around the compaction point to accurately capture the vibration source characteristics and near-field attenuation law. The outer ring is deployed at the foundations of all surrounding buildings that need protection or at the nearest point to directly monitor the target response. The middle ring (e.g., 30m, 50m, 80m) is radially deployed along the main vibration propagation path to calibrate the vibration attenuation characteristics in the soil layer. The vibration sensor uses accelerometers in the X, Y, and Z directions to record the spatiotemporal changes of the vibration vector and synchronizes the time through wireless Internet of Things technology to ensure the spatiotemporal consistency of the data.

[0023] The vibration sensor acquires the acceleration time history signal a(t), which is integrated once to obtain the velocity time history signal v(t), and integrated twice to obtain the displacement time history signal d(t). Therefore, the final "vibration time history data" is a multidimensional dataset that contains at least the time sequence of acceleration, velocity, and displacement at each measuring point in three directions.

[0024] The building attribute information includes basic attribute information and functional attribute information; the basic attribute information shown includes structural type, foundation type, construction year, and current status.

[0025] The structural types include brick-concrete structures, reinforced concrete frame structures, shear wall structures, and steel structures. Different types of structures have different natural frequencies and energy dissipation capacities.

[0026] Foundation types include isolated foundations, strip foundations, raft foundations, and pile foundations. The foundation type directly affects the coupling efficiency of vibrations transmitted from the ground to the building.

[0027] The construction date includes factors such as design standards and the degree of material aging.

[0028] The current situation includes the existing damage conditions obtained through visual inspection or inspection reports, such as crack width and distribution, which can be quantified using the "crack index".

[0029] Functional attribute information refers to information about the social importance of a building related to its function, such as: ordinary residential buildings, schools, hospitals, historical buildings (heritage sites), precision instrument factories, etc. The more important the function or the more sensitive the internal activities are to vibration, the higher the priority of its protection. This information can be obtained through on-site surveys in the early stages of construction, reviewing architectural design drawings and archival materials, and then stored in a structured manner in the system database.

[0030] S102. Based on the vibration time history data, the ground fluctuation parameters of the construction area are dynamically inverted using a trained convolutional neural network model, specifically including: The vibration time history data from multiple measurement points are arranged according to the spatial location of the sensors to form a vibration feature map, which serves as the input to the convolutional neural network encoder-decoder structure. The output of the network is the stratum wave parameters, including the thickness, shear wave velocity, density, and damping ratio of each soil layer.

[0031] Using raw acceleration time history data directly is too redundant and contains noise. Therefore, the triaxial acceleration signals acquired at each measurement point i are first preprocessed and feature extracted. Preprocessing includes denoising (such as wavelet thresholding) and baseline correction. Subsequently, key features characterizing vibration intensity, spectral properties, and attenuation patterns are extracted from the processed signals, such as: Peak velocity (PPVi) represents the maximum value of the combined velocity in three directions and is a core indicator for measuring vibration intensity. ; The dominant frequency (Fi) represents the frequency component where vibrational energy is most concentrated, and can be obtained through Fourier spectral analysis; Vibration duration (Ti) represents the duration for which the signal amplitude exceeds a certain threshold (such as 5% of the maximum value); Specific frequency band energy (Ei) represents the integrated energy within certain key frequency bands (such as 1-80Hz).

[0032] For each extracted feature (such as PPVi), its measurement value is assigned to the corresponding grid cell based on the true geographic coordinates of each sensor node. Since the number of sensors is limited, most grid cells do not have direct measurement values. In this case, spatial interpolation algorithms (such as Kriging interpolation or inverse distance-weighted interpolation) are used to estimate the feature values ​​of these blank grid cells. For example, the PPVi value of a grid cell without measurement points can be calculated by weighting the PPVi values ​​of several surrounding measurement points based on distance. PPV_grid=Σ(PPVi / di) / Σ(1 / di), where di is the distance from the center of the grid to the i-th measurement point.

[0033] By summing the interpolation results of all measurement points for a certain feature (such as PPVi), a two-dimensional image reflecting the continuous spatial distribution of that feature is obtained, namely, a "vibration feature map". Multiple features can be extracted to generate multi-channel feature maps.

[0034] The convolutional neural network encoder-decoder structure is used to establish a complex nonlinear mapping from "vibration feature map" (characterizing the spatial pattern of surface vibration response) to "stratum wave parameter vector" (characterizing the properties of underground medium); in, The encoder section consists of alternating convolutional and pooling layers. For example: The first layer uses 64 3×3 convolution kernels to convolve the input H×W×C feature map to extract low-level features (such as edges and gradients), followed by ReLU activation function and 2×2 max pooling, which halves the size.

[0035] Subsequent layers: The number of convolutional kernels is increased layer by layer (128, 256), the receptive field is enlarged, and more abstract high-level features are extracted (such as the macroscopic pattern of vibration wave propagation and the shape of anomaly regions). Pooling layers continuously reduce the spatial size of the feature map and compress information.

[0036] Ultimately, the encoder "encodes" the input two-dimensional feature map into a highly abstract one-dimensional feature vector.

[0037] The decoder section consists of fully connected layers. The one-dimensional feature vector passes through one or more fully connected layers, ultimately mapping to a fixed-length output vector. Each dimension of this vector corresponds to a formation fluctuation parameter.

[0038] The output parameters typically include: the thickness of each soil layer (h1, h2, ...); the shear wave velocity of each soil layer (Vs1, Vs2, ...): characterizing the soil layer stiffness, which is a key parameter for the propagation speed of vibration waves; the density of each soil layer (ρ1, ρ2, ...); and the damping ratio of each soil layer (ξ1, ξ2, ...): characterizing the soil layer's ability to dissipate vibration energy. These parameters together define the non-uniform geological physical model used for wave propagation calculations.

[0039] During construction, real-time collected and processed vibration feature maps are input into a trained convolutional neural network (CNN). After forward propagation (which takes only milliseconds), the network can directly output the estimated ground parameter vector. After each construction zone is completed or a certain amount of new data is accumulated, the new real data pairs (feature maps) are added to the training set, and the CNN undergoes short-term incremental training. This allows the model to adapt to changes in ground parameters after soil compaction, achieving adaptive model updates.

[0040] further, Based on the aforementioned ground wave parameters and building attribute information, a vibration propagation attenuation correction model integrating building vibration sensitivity weights is constructed, specifically including: The building vibration sensitivity weighting coefficient is calculated based on the building attribute information; the coefficient is used as a risk amplification factor and coupled into the two-dimensional non-uniform medium wave equation established based on the inverted stratum parameters; the output of the vibration propagation attenuation correction model is the risk-weighted vibration value.

[0041] First, calculate the building vibration sensitivity weighting coefficient: Based on the building attribute information, calculate a comprehensive building vibration sensitivity weighting coefficient βi for the i-th building. The purpose is to transform the building's objective attributes into a quantitative indicator of subjective risk tolerance. ; The weights for each item are assigned based on the survey results, for example: The structural type weight w_type includes brick-concrete structure = 1.2, reinforced concrete frame = 1.0, and steel structure = 0.8.

[0042] The age-related weights w_age include the construction year: <10 years = 1.0, 10-30 years = 1.1, >30 years = 1.3.

[0043] The function-sensitive weights w_func include historical sites / hospitals = 1.5, schools = 1.3, residences = 1.0, and warehouses = 0.7.

[0044] The current condition weights w_cond include no visible cracks = 1.0, minor cracks = 1.2, and significant damage = 1.5.

[0045] After calculation, βi of all buildings is normalized using the max-min normalization method so that its range is [0, 1].

[0046] The wave equation for a two-dimensional inhomogeneous medium is expressed as: ; Where u is the displacement field vector, which is the quantity to be determined; It is the density of spatial variation; and It is the spatially varying Lamé constant, which is related to the inverted shear wave velocity Vs and Poisson's ratio; It is the viscous damping coefficient, which is related to the damping ratio ξ obtained by inversion; It is the gradient operator, and I is the identity matrix.

[0047] After discretizing the two-dimensional non-uniform medium wave equation (e.g., using the finite difference method), the physical vibration field (displacement, velocity, acceleration) of the entire site can be numerically solved under given source (impact) conditions.

[0048] At this point, the two-dimensional non-uniform medium wave equation calculates the predicted physical vibration value V_physical,i (usually peak velocity) without considering building risk. Then, it is corrected using βi to obtain the risk-weighted vibration value V_risk,i, which is the final output of the model: V_risk,i = V_physical,i * (1 + γ * βi); where γ is the risk amplification factor, a constant set according to engineering experience and regulatory requirements (e.g., γ=0.5). For a very sensitive building (βi is high), even if the physical vibration is not large, its "risk vibration value" will be amplified, thus subjecting it to stricter constraints in subsequent optimization.

[0049] Based on the above, a digital twin simulation system can be obtained that can receive construction parameters (impact energy E, azimuth angle θ) and the current stratum model, and output the predicted density increment and risk-weighted vibration values ​​at each building location.

[0050] S103. For the current tamping points to be constructed, establish a multi-objective optimization model based on the first and second objectives; The first objective is to maximize the predicted soil density increment within the area affected by the current tamping point; the second objective is to optimize quality, and the optimization algorithm needs to select the construction parameters that can most effectively tampe the soil around the point.

[0051] The second objective is to minimize the risk-weighted vibration values ​​of all surrounding building locations; the first objective is construction safety, in which case the optimization algorithm needs to select the construction parameters that have the least risk of vibration impact on the surrounding building complex.

[0052] The establishment of the multi-objective optimization model includes: The decision variables of the multi-objective optimization model are the impact energy E and the azimuth angle θ of the impact point; Impact energy (E) is typically achieved by adjusting the mass and drop height of the hammer, and is measured in kilojoules (kJ). It is the most important factor affecting compaction effect and vibration intensity.

[0053] The azimuth angle (θ) of the tamping point represents the rotational offset angle (unit: degrees) of the tamping point relative to its original design position on the horizontal plane. By rotating the tamping machine or adjusting its position, the main direction of vibration energy propagation can be changed, thereby avoiding sensitive buildings.

[0054] Based on safety constraints and equipment constraints, a multi-objective optimization model is constructed, including a first multi-objective optimization model based on a first objective and a second multi-objective optimization model based on a second objective. The first multi-objective optimization model is expressed as: Where X = [E, θ]; The total number of grid points is calculated within the target area, which is usually determined based on experience or experimentation, such as a circular area centered on the tamping point with a radius of 1.5 times the diameter of the tamping hammer; This represents the soil compaction increment at the j-th grid point; It is obtained through a density increment prediction model, which is expressed as follows: ; r represents the horizontal distance from grid point j to the tamping point; Indicates the equivalent radius of the ram; Represents the azimuth angle in the decision variables; This represents the azimuth angle of grid point j relative to the tamping point; , Empirical coefficients related to soil types are calibrated through indoor tests or historical data. The density of the soil is represented by S102 inversion; The second multi-objective optimization model is expressed as: ;in For the number of buildings; Risk-weighted vibration value; It is directly derived from the Vibration Propagation Attenuation Correction Model (VPACM) constructed by S102; for a given X (i.e., a specific E and θ), VPACM can calculate the predicted physical vibration value V_physical,i(X) at building i.

[0055] Subsequently, the risk-weighted value was calculated based on the building's sensitivity weighting coefficient βi: ; Where γ is the risk amplification factor (e.g., 0.5); It can control the vibration risk of the most sensitive and affected building, ensuring the system's safety baseline.

[0056] Furthermore, the constraints of the optimization algorithm for the multi-objective optimization model include: Safety constraint: V_i(X) ≤ V_allow,i, for all i = 1, 2, ..., N_b. Where V_allow,i is the allowable risk vibration value for the i-th building, which can be set according to specifications, building safety assessment reports, or owner requirements; Equipment constraints: E_min ≤ E ≤ E_max, |θ| ≤ θ_max. E_min and E_max are determined by the physical capabilities of the dynamic compaction machine; θ_max is the maximum permissible azimuth offset, typically limited by the construction site layout and the machine's operating range.

[0057] S104. Based on the prediction results provided by the vibration propagation attenuation correction model, the optimal impact energy and optimal azimuth of the current tamping point are obtained by using a multi-objective optimization algorithm.

[0058] The solution is obtained by using a multi-objective optimization algorithm, including: using a constrained multi-objective particle swarm optimization algorithm and a fast surrogate model constructed by a deep neural network to predict vibration values; the multi-objective optimization algorithm outputs a Pareto optimal solution set, and the final implementation scheme is selected by a multi-attribute decision method.

[0059] Constrained multi-objective particle swarm optimization algorithms specifically include: Population initialization: Randomly generate a set of particles, where the position vector of each particle represents a candidate solution X. k = [E k θ k The velocity vector is randomly initialized.

[0060] Fitness assessment: For each particle X k It is necessary to calculate its two objective function values ​​f1(X). k f2(X) and f2(X) k The calculations also include the degree of constraint violation. This is the most time-consuming part of the calculation.

[0061] Constraint Domination Ranking: To compare the quality of solutions, the following constraints are set: Solution A dominates solution B if and only if: 1) A is no worse than B in all objectives; 2) A is better than B in at least one objective; 3) The degree of constraint violation of A is no greater than that of B; feasible solutions (with zero constraint violations) and solutions with smaller constraint violations are preferred.

[0062] External archive maintenance: Use an external archive set to store all currently found non-dominated solutions (Pareto optimal solutions); employ techniques such as adaptive grid methods to maintain the distribution and diversity of the archives and prevent convergence to local frontiers.

[0063] Particle state update: Each particle updates its velocity and position based on its individual historical best position and the population's global best position (selected from external archives).

[0064] Fast proxy models built with deep neural networks replace complex physical models, achieving millisecond-level speeds. and Predictions, specifically including: Input layer: decision variables [E, θ], current stratum key parameters, building location, and βi.

[0065] Output layer: All grid points Predicted value (used for calculation) ) and each building Predicted value (used for calculation) ).

[0066] Training data: Before construction, a large number of sampling calculations were performed in the parameter space using offline high-fidelity VPACM and density models to generate tens of thousands to hundreds of thousands of (input, output) data pairs.

[0067] Training: Use the dataset to train a deep fully connected neural network until its prediction accuracy meets the requirements (e.g., average relative error with the physical model <5%).

[0068] Iteration and Termination: The evaluation, sorting, and updating process is repeated until the preset number of iterations or convergence criteria are reached. Ultimately, the solution set stored in the external archive is the approximate Pareto optimal solution set, where each solution represents a different trade-off between quality and safety.

[0069] The final implementation scheme is selected using a multi-attribute decision-making method because the optimization algorithm outputs a Pareto optimal set, not a single solution. A final implementation scheme needs to be chosen from this set. The steps using a multi-attribute decision-making method, such as TOPSIS, are as follows: First, for each solution in the Pareto solution set, calculate its two objective values. and .

[0070] Second, construct a decision matrix and normalize it; divide each element by the square root of the sum of the squares of all elements in its column. After normalization, all values ​​for each objective are scaled to the same scale.

[0071] Third, determine the positive ideal solution ( maximum, Minimum) and negative ideal solution ( Minimum, maximum).

[0072] Fourth, calculate the Euclidean distance Da between each solution and the positive ideal solution, and the Euclidean distance Db between each solution and the negative ideal solution; the formula for calculating the Euclidean distance is existing technology and will not be elaborated here.

[0073] Fifth, calculate the relative closeness of each solution to the ideal solution. Ci = Db / (Da + Db), the closer Ci is to 1, the more ideal the solution is.

[0074] The solution with the highest Ci value can be selected as the final solution; alternatively, construction management personnel can be allowed to intervene and adjust the weights of the two objectives in TOPSIS to express a preference for "quality" or "safety", thereby achieving flexible decision-making through human-machine collaboration.

[0075] At this point, the system has completed the calculation process from perceived data to optimal decision parameters [E,θ], thus preparing for intelligent execution.

[0076] A second aspect of the present invention provides a dynamic compaction parameter optimization system 200 for multimodal data, comprising: The data acquisition module 201 is used to acquire vibration time history data and building attribute information during the dynamic compaction construction process; The parameter optimization module 202 is used to dynamically invert the ground fluctuation parameters of the construction area based on the vibration time history data through a trained convolutional neural network model; and to construct a vibration propagation attenuation correction model that integrates building vibration sensitivity weights based on the ground fluctuation parameters and the building attribute information. The multi-model optimization module 203 is used to establish a multi-objective optimization model based on the first objective and the second objective for the current compaction point to be constructed. The decision analysis module 204 is used to obtain the optimal impact energy and optimal azimuth of the current tamping point by using a multi-objective optimization algorithm based on the prediction results provided by the vibration propagation attenuation correction model.

[0077] A third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the dynamic compaction parameter optimization method for multimodal data as described in the first aspect.

[0078] A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored thereon, which, when executed by one or more processors, cause the processors to perform the dynamic compaction parameter optimization method for multimodal data as described in the first aspect.

[0079] This embodiment can divide the method into functional modules based on the above method example. For example, each function can be assigned to a separate module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0080] When each functional module is divided according to its corresponding function, the vehicle may include: a data acquisition module, a parameter optimization module, a multi-model optimization module, a decision analysis module, etc. It should be noted that all relevant content of each step involved in the above method embodiments can be referenced from the functional descriptions of the corresponding functional modules, and will not be repeated here.

[0081] This embodiment also provides a computer-readable storage medium (including but not limited to disk storage, CD-ROM, optical storage, etc.) storing computer program code. When the computer program code is run on a computer, the computer executes the above-mentioned related method steps to implement the dynamic compaction parameter optimization method for multimodal data provided in the above embodiment.

[0082] This embodiment also provides a computer program product. When the computer program product is run on a computer, it causes the computer to perform the aforementioned steps to implement the method for optimizing dynamic compaction parameters of multimodal data provided in the above embodiment. The beneficial effects of the above embodiments can be found in the corresponding methods described above, and will not be repeated here.

[0083] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0084] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of modules or units is merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. In the description of this disclosure, it should be understood that if terms such as "upper," "lower," "front," "rear," "left," and "right" are used to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, they are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the indicated position or element must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this disclosure.

[0085] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0086] The above are merely embodiments of this disclosure and are not intended to limit the scope of this disclosure. Various modifications and variations can be made to this disclosure by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of the claims of this disclosure.

Claims

1. A method for optimizing dynamic compaction parameters based on multimodal data, characterized in that, include: Obtain vibration time history data and building attribute information during dynamic compaction construction; Based on the vibration time history data, the ground fluctuation parameters of the construction area are dynamically inverted using a trained convolutional neural network model; based on the ground fluctuation parameters and the building attribute information, a vibration propagation attenuation correction model that integrates building vibration sensitivity weights is constructed. For the current tamping points to be constructed, a multi-objective optimization model is established based on the first and second objectives; Based on the prediction results provided by the vibration propagation attenuation correction model, the optimal impact energy and optimal azimuth of the current compaction point are obtained by using a multi-objective optimization algorithm.

2. The method for optimizing dynamic compaction parameters for multimodal data according to claim 1, characterized in that, The first objective is to maximize the predicted soil density increment within the influence area of ​​the current compaction point to be constructed; the second objective is to minimize the risk-weighted vibration value of all surrounding building locations.

3. The method for optimizing dynamic compaction parameters for multimodal data according to claim 1, characterized in that, The building attribute information includes basic attribute information and functional attribute information; the basic attribute information shown includes structural type, foundation type, construction year, and current status.

4. The method for optimizing dynamic compaction parameters for multimodal data according to claim 1, characterized in that, The dynamic inversion of ground fluctuation parameters in the construction area based on the vibration time history data using a trained convolutional neural network model includes: The vibration time history data from multiple measurement points are arranged according to the spatial location of the sensors to form a vibration feature map, which serves as the input to the convolutional neural network encoder-decoder structure. The output of the network is the stratum wave parameters, including the thickness, shear wave velocity, density, and damping ratio of each soil layer.

5. The method for optimizing dynamic compaction parameters for multimodal data according to claim 1, characterized in that, The vibration propagation attenuation correction model, which integrates building vibration sensitivity weights, is constructed based on the ground wave parameters and the building attribute information, including: The building vibration sensitivity weighting coefficient is calculated based on the building attribute information; the coefficient is used as a risk amplification factor and coupled into the two-dimensional non-uniform medium wave equation established based on the inverted stratum parameters; the output of the vibration propagation attenuation correction model is the risk-weighted vibration value.

6. The method for optimizing dynamic compaction parameters for multimodal data according to claim 1, characterized in that, The establishment of the multi-objective optimization model includes: The decision variables of the multi-objective optimization model are the impact energy E and the azimuth angle θ of the impact point. Based on safety constraints and equipment constraints, a multi-objective optimization model is constructed, including a first multi-objective optimization model based on a first objective and a second multi-objective optimization model based on a second objective. The first multi-objective optimization model is expressed as: Where X = [E, θ], Calculate the total number of grid points within the target area; This represents the soil compaction increment at the j-th grid point; The second multi-objective optimization model is expressed as: ;in For the number of buildings; This represents the risk-weighted vibration value.

7. The method for optimizing dynamic compaction parameters for multimodal data according to claim 6, characterized in that, The method of using a multi-objective optimization algorithm includes: using a constrained multi-objective particle swarm optimization algorithm and a fast surrogate model constructed through a deep neural network to predict vibration values; the multi-objective optimization algorithm outputs a Pareto optimal solution set, and the final implementation scheme is selected through a multi-attribute decision method.

8. A dynamic compaction parameter optimization system for multimodal data, characterized in that, include: The data acquisition module is used to acquire vibration time history data and building attribute information during the dynamic compaction construction process; The parameter optimization module is used to dynamically invert the ground fluctuation parameters of the construction area based on the vibration time history data through a trained convolutional neural network model; and to construct a vibration propagation attenuation correction model that integrates building vibration sensitivity weights based on the ground fluctuation parameters and the building attribute information. The multi-model optimization module is used to establish a multi-objective optimization model for the current compaction point to be constructed, based on the first objective and the second objective. The decision analysis module is used to obtain the optimal impact energy and optimal azimuth of the current tamping point based on the prediction results provided by the vibration propagation attenuation correction model and by using a multi-objective optimization algorithm.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores instructions that, when executed by one or more processors, cause the processors to perform the dynamic compaction parameter optimization method for multimodal data as described in any one of claims 1-7.

Images