Soft foundation in-situ solidification construction optimization method and system based on Beidou 3D positioning

Abstract

The invention relates to the technical field of soft foundation treatment engineering, in particular to a soft foundation in-situ solidification construction optimization method and system based on Beidou 3D positioning. The method comprises the following steps: obtaining high-precision 3D position coordinates through positioning optimization of multi-source data fusion; extracting soil solidification potential characteristics by using geologic characteristics of machine learning, and obtaining optimized construction parameter characteristics according to construction parameter recommendation of case reasoning; on the basis of the high-precision 3D position coordinates, the soil solidification potential features and the optimized construction parameter features, construction priority region features are obtained through construction region division based on multi-feature fusion; material effective coverage range features and mechanical operation efficiency features are combined, and material and mechanical collaborative features are obtained according to collaborative optimization of a control theory; and fusing the influence characteristics of the environment on the curing process, the construction priority region characteristics and the material and machine cooperation characteristics, and dynamically planning a construction plan. According to the invention, the precision, intelligence and high efficiency of the soft foundation construction process are realized.

Description

Technical Field

[0001] This invention relates to the field of soft soil treatment engineering technology, specifically to an optimized method and system for in-situ solidification construction of soft soil based on BeiDou 3D positioning. Background Technology

[0002] In-situ consolidation technology for soft soil foundations has become one of the mainstream technologies for soft soil foundation treatment in infrastructure construction such as roads, ports, and airport runways due to its core advantages such as ease of construction, cost-effectiveness, and low environmental disturbance. However, the existing in-situ consolidation technology system for soft soil foundations still has many core technical bottlenecks, which restrict the stability of construction quality and the improvement of project efficiency, as specifically manifested as follows: The positioning accuracy is significantly limited: Traditional construction positioning schemes mostly rely on a single Beidou positioning mode or a Beidou + base station differential positioning architecture. They fail to fully consider key influencing factors such as vibration interference during construction machinery operation and positioning deviations caused by terrain undulations in soft soil areas. This results in insufficient spatial positioning accuracy of the construction machinery operating end (such as curing agent spray head and mixing drill bit), which can easily induce quality defects such as curing agent spray deviation and inaccurate mixing depth control.

[0003] The geological feature characterization and solidification potential evaluation system is too simplistic: Existing technologies often only rely on a single geological parameter (such as soil moisture content) or empirical fitting formula to conduct soil solidification potential assessment, without fully considering the synergistic mechanism and spatial heterogeneity characteristics among multiple geological parameters. This makes it difficult to achieve accurate matching of construction parameters with geological conditions at different spatial locations, which can easily lead to problems such as blind construction plans and waste of solidification materials.

[0004] The ability to coordinate and control material diffusion and mechanical operation is weak: the determination of curing agent spraying parameters mostly relies on engineering experience and does not take into account the spatial heterogeneity of soil pore structure and the rheological properties of curing agent for targeted optimization design; mechanical operation parameters (such as operating speed and stirring speed) mostly adopt fixed setting mode, which fails to form a real-time match with the dynamic process of curing agent diffusion, which can easily lead to uneven spatial distribution of curing agent, insufficient effective coverage or excessive material consumption.

[0005] Insufficient dynamic response and quantitative assessment capabilities for environmental factors: Existing technologies for assessing environmental impact mostly adopt static analysis models (such as only considering temperature conditions on the day of construction) or qualitative description methods, failing to achieve quantitative characterization of the dynamic evolution of environmental factors and accurate calculation of influence weights. This makes it difficult to effectively cope with the adverse interference of sudden environmental changes such as temperature fluctuations and rainfall on the curing reaction process and curing effect.

[0006] Lack of dynamic adaptability in construction plans: The priority division of construction areas often adopts simple topographic division or uniform division strategies, failing to integrate core characteristics such as soil consolidation potential and the adaptability of construction parameters; the formulation of construction plans is mainly based on static planning, and does not fully incorporate uncertainties such as dynamic environmental changes and resource constraints, which can easily lead to project risks such as project delays and cost overruns.

[0007] In summary, existing in-situ hardening construction technologies for soft soil foundations still face significant technical bottlenecks in core aspects such as positioning accuracy, feature characterization, collaborative control, environmental response, and plan adaptation. There is an urgent need to develop a whole-process optimization method for in-situ hardening construction of soft soil foundations that can achieve high-precision positioning, accurate feature extraction, dynamic collaborative optimization, and adaptive plan adjustment, in order to overcome the limitations of existing technologies and improve the precision and intelligence of soft soil foundation hardening construction. Summary of the Invention

[0008] To address the shortcomings of existing methods and the needs of practical applications, and in order to solve the aforementioned problems, this invention provides an optimized method for in-situ solidification construction of soft soil foundations based on BeiDou 3D positioning, comprising the following steps: High-precision 3D location coordinates are obtained through multi-source data fusion and positioning optimization; soil consolidation potential features are extracted using geological features derived from machine learning, and optimized construction parameter features are obtained based on case-based reasoning; construction priority area features are obtained through multi-feature fusion for construction area division; material and machinery synergy features are obtained by combining material effective coverage range features and machinery operation efficiency features and through synergistic optimization based on control theory; and construction plans are dynamically planned by integrating environmental influence features on the consolidation process, construction priority area features, and material and machinery synergy features.

[0009] Optionally, the step of obtaining high-precision 3D position coordinates through multi-source data fusion positioning optimization includes the following steps: Kalman filtering was used to achieve primary hierarchical fusion of BeiDou and IMU data; DEM digital elevation data of the construction area was introduced to verify the elevation consistency of the positioning results after hierarchical fusion, and high-precision 3D position coordinates were obtained.

[0010] Optionally, the extraction of soil consolidation potential features using geological features derived from machine learning includes the following steps: A structured association dataset of 3D location and multi-dimensional geological parameters is constructed; based on the structured association dataset, feature selection and correlation analysis are performed to obtain core geological features; the core geological features are used to train an ensemble learning model and extract soil consolidation potential features.

[0011] Optionally, the step of recommending optimized construction parameter features based on case-based reasoning includes the following steps: Differentiated weights are assigned to search keywords, and historical matching cases are retrieved from the structured case library based on the weight assignment results; the case parameters in the historical matching cases are corrected, and optimized construction parameter features are extracted from the correction results.

[0012] Optionally, the step of obtaining construction priority area features by dividing the construction area based on high-precision 3D location coordinates, soil consolidation potential characteristics, and optimized construction parameter characteristics through multi-feature fusion includes the following steps: Using the 3D location coordinates of the construction area as a spatial reference, and integrating the soil consolidation potential characteristics and optimized construction parameter characteristics, a regional division evaluation index system is constructed. Based on the regional division evaluation index system, the construction area is divided into different priorities through cluster analysis, and the regional characteristics of construction priority areas are extracted.

[0013] Optionally, the step of combining the effective coverage characteristics of the material and the mechanical operation efficiency characteristics to obtain the material-mechanical synergy characteristics based on the synergistic optimization of control theory includes the following steps: Based on model predictive control theory, a collaborative optimization model for material diffusion and mechanical operation is constructed. Combining the characteristics of effective material coverage and mechanical operation efficiency with the collaborative optimization model, dynamic collaboration between mechanical operation and material diffusion is achieved through rolling optimization and feedback correction, thereby obtaining the collaborative characteristics of materials and machinery.

[0014] Optionally, the effective coverage characteristics of the material are obtained through material diffusion optimization via fluid dynamics, including the following steps: Based on the porous media seepage theory in fluid dynamics, a diffusion model of the curing agent in soft soil is constructed. Combining the soil pore structure, the rheological properties of the curing agent, and the operating parameters of the construction machinery, the diffusion model is used to simulate the diffusion process of the curing agent, optimize the operating parameters to ensure that the material effectively covers the target area, and obtain the characteristics of the effective coverage range of the material.

[0015] Optionally, the mechanical operation efficiency characteristics are obtained through mechanical behavior optimization using reinforcement learning, including the following steps: Based on the Q-learning algorithm, a mechanical operation optimization system is constructed. The operating parameters of the construction machinery are used as actions, and construction efficiency and construction quality are used as reward signals. The mechanical operation behavior is optimized through continuous iterative learning to obtain the mechanical operation efficiency characteristics.

[0016] Optionally, the dynamic planning of the construction schedule, based on the influence characteristics of the fusion environment on the curing process, the characteristics of construction priority areas, and the synergistic characteristics of materials and machinery, includes the following steps: Based on dynamic programming theory, the construction process is divided into multiple decision-making stages according to time phases; the state transition equation for optimizing the construction plan is constructed based on the regional characteristics of construction priority, material and machinery synergy characteristics, and environmental impact characteristics of each stage; and the dynamic programming construction plan is solved in reverse order based on the optimization objective.

[0017] This invention provides a spatial reference for the entire process using high-precision 3D coordinates, solves construction adaptability issues by utilizing soil consolidation potential and optimized construction parameter characteristics, avoids blind operations, clarifies the core of resource allocation by defining construction priority areas, improves overall efficiency, ensures construction quality and resource utilization based on the synergistic characteristics of materials and machinery, and finally uses dynamic planning construction plans to comprehensively address uncertainties such as environmental changes. By balancing schedule, cost, and quality objectives, it effectively solves core problems such as inaccurate positioning, parameter mismatch, and insufficient coordination in soft soil foundation construction, promoting efficient and high-quality construction.

[0018] Secondly, to efficiently execute the soft soil foundation in-situ solidification construction optimization method based on BeiDou 3D positioning provided by this invention, this invention also provides a soft soil foundation in-situ solidification construction optimization system based on BeiDou 3D positioning, including a processor, an input device, an output device, and a memory. The processor, input device, output device, and memory are interconnected. The memory stores a computer program containing program instructions. The processor is configured to call the program instructions to execute the soft soil foundation in-situ solidification construction optimization method based on BeiDou 3D positioning as described in the first aspect of this invention. The soft soil foundation in-situ solidification construction optimization system based on BeiDou 3D positioning of this invention has a compact structure and stable performance, and can stably execute the soft soil foundation in-situ solidification construction optimization method based on BeiDou 3D positioning provided by this invention, further improving the overall applicability and practical application capability of this invention. Attached Figure Description

[0019] Figure 1 A flowchart of an optimized construction method for in-situ solidification of soft soil based on BeiDou 3D positioning is provided in an embodiment of the present invention. Figure 2 This is a framework diagram of an optimized construction system for in-situ solidification of soft soil based on BeiDou 3D positioning, provided for an embodiment of the present invention. Detailed Implementation

[0020] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0021] Throughout this specification, references to an embodiment, example, or illustration mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, phrases appearing in various places throughout the specification, such as "in one embodiment," "in an embodiment," "an example," or "an illustration," do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in any suitable combination and / or sub-combination in one or more embodiments or examples. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0022] Please see Figure 1 To address the aforementioned problems, this invention provides an optimized construction method for in-situ solidification of soft soil foundations based on BeiDou 3D positioning, such as... Figure 1 As shown, in one embodiment, the method includes the following steps: S1. High-precision 3D position coordinates are obtained through positioning optimization by fusing multi-source data.

[0023] Based on the complementarity of multi-source heterogeneous data, by integrating BeiDou-3 satellite positioning data, inertial measurement unit (IMU) data, construction area reference station data, and mechanical attitude sensor data, an adaptive filtering algorithm is used to eliminate the noise effects of vibration, obstruction, and electromagnetic interference in the construction scene, thereby achieving high-precision 3D positioning of the construction machinery operating end (such as curing agent spraying head and mixing drill bit).

[0024] Specifically, the method of obtaining high-precision 3D position coordinates through multi-source data fusion positioning optimization includes the following steps: S11. Primary hierarchical fusion of BeiDou and IMU data is achieved through Kalman filtering.

[0025] Constructing the state and observation equations for a Kalman filter: State vector (Including 9 state variables: 3D position, velocity, and acceleration), the state equations are constructed based on the Newtonian kinematic model. A is the state transition matrix, and B is the control input matrix. For IMU acceleration input, The process noise was initially set to Gaussian white noise, and its variance was determined through offline calibration. The observation equation was constructed using the 3D position provided by BeiDou positioning as the observation value. H is the observation matrix. To mitigate observation noise, the variance is dynamically adjusted based on the signal-to-noise ratio of the BeiDou satellites. Through a Kalman filter prediction-update iterative process, the current position, velocity, and other state variables are first predicted using the state equations. Then, BeiDou observations are used to correct the prediction results, yielding a preliminary fused 3D position coordinate. Then, differential data from the base station are introduced for secondary correction, and the objective function is constructed using the weighted least squares method. ,in This is the initial fusion result. These are the theoretical values ​​corrected for the base station. For weights, weights Based on the dynamic allocation of the BeiDou satellite signal-to-noise ratio (SNR), the specific calculation formula is as follows: , Let be the signal-to-noise ratio of the i-th visible satellite. A higher signal-to-noise ratio indicates better satellite observation quality and a larger corresponding weight, thereby improving the stability and accuracy of the fusion results.

[0026] S12. Introduce DEM digital elevation data of the construction area, perform elevation consistency verification on the positioning results after layered fusion, and obtain high-precision 3D position coordinates.

[0027] Digital elevation data (DEM) of the construction area at the corresponding resolution is introduced to verify the elevation consistency of the layered fusion positioning results. First, the elevation value Z in the fused positioning results is extracted and compared with the corresponding DEM elevation value Z_DEM for the 3D location (X,Y). The elevation deviation ΔZ = |Z - Z_DEM| is calculated. A deviation threshold of 5cm is set (determined based on the positioning accuracy requirements for soft soil foundation construction). If ΔZ ≤ 5cm, the positioning result is considered valid. If ΔZ > 5cm, or if ΔZ at three consecutive sampling points exceeds 3cm, a feedback correction mechanism is triggered. The preprocessed original data is retrieved again, the process noise variance of the Kalman filter and the weighting coefficients of the weighted least squares method are adjusted, and the layered fusion process is executed again. After multiple iterations of verification until the elevation deviation meets the requirements, the final 3D position coordinates (X,Y,Z) of the construction machinery operating end are obtained.

[0028] S2. Use machine learning to extract soil consolidation potential features from geological features, and obtain optimized construction parameter features based on case-based reasoning for construction parameter recommendations.

[0029] The machine ensemble learning algorithm is used to fuse multi-source geological survey data to extract core features that are strongly correlated with soil solidification effect, such as solidification difficulty, strength improvement potential, and optimal solidifying agent dosage range, i.e., soil solidification potential features.

[0030] Specifically, the extraction of soil consolidation potential features using geological features derived from machine learning includes the following steps: S211. Construct a structured association dataset of 3D location and multi-dimensional geological parameters.

[0031] First, precise coordinate matching was carried out. Based on the BeiDou high-precision 3D coordinate system of the construction area, the borehole sampling data and in-situ test data were unified to the same coordinate reference plane through Gaussian projection transformation. The nearest neighbor interpolation method was used to correct the coordinate deviation of a small number of sampling points. Finally, a structured association dataset of 3D location (X,Y,Z) and multi-dimensional geological parameters was constructed to realize a one-to-one correspondence between geological parameters and spatial location.

[0032] To eliminate dimensional differences, Z-score standardization was used to process the geological parameters, standardizing all parameters to a range with a mean of 0 and a standard deviation of 1, ensuring that the weights of each parameter are balanced in subsequent model training.

[0033] S212. Based on the structured association dataset, perform initial feature selection and correlation analysis to obtain core geological features.

[0034] In this embodiment, the unconfined compressive strength after 28 days of curing is used as the dependent variable, and various geological parameters, including moisture content, void ratio, and liquid limit, are used as independent variables. The Pearson correlation coefficient method is used to calculate the pairwise correlation coefficient *r* for correlation analysis. The value of *r* ranges from -1 to 1, with a value closer to 1 indicating a stronger correlation and a value closer to 0 indicating a weaker correlation. During the calculation, the normality of the independent and dependent variables is first tested (using the Shapiro-Wilk test). If normality is not satisfied, the data undergoes logarithmic transformation. In this embodiment, based on engineering practice, parameters with an absolute value of *r* > 0.5 are selected as candidate features, ultimately determining moisture content, void ratio, organic matter content, and SPT blow count as the core geological features. Simultaneously, a scatter plot matrix is ​​used to visually verify the correlation trend between each core geological feature and the curing strength, excluding parameters with multicollinearity and ensuring the independence and effectiveness of the core geological features.

[0035] S213. Use the core geological features to train an integrated learning model and extract soil consolidation potential features.

[0036] We construct a random forest-XGBoost ensemble model by employing parallel ensemble and result-weighted fusion to fully leverage the advantages of both machine learning models.

[0037] First, the dataset was partitioned. The 3D location-core geological feature-solidification intensity association dataset was randomly divided into training and testing sets in a 7:3 ratio. The model input consisted of standardized values ​​of four core geological features, and the output was the predicted unconfined compressive strength 28 days after solidification. For the random forest model, the initial number of decision trees was set to 100, the maximum depth to 10 layers, and the Gini coefficient was used as the feature splitting criterion. Hyperparameters such as the number of decision trees (50-200), maximum depth (5-15 layers), and minimum number of sample splits (2-10) were optimized using a grid search method. For the XGBoost model, the initial learning rate was set to 0.1, the tree depth to 6 layers, and the regularization parameter λ=1. The mean squared error (MSE) was used as the loss function, and key hyperparameters were also optimized using a grid search method. To avoid overfitting, a 5-fold cross-validation strategy was adopted: the training set was divided into 5 subsets, and the model was trained with 4 subsets and validated with 1 subset, repeated 5 times. The average validation error was calculated, and the hyperparameters were adjusted based on the error changes. Finally, a weighted fusion method was used to integrate the prediction results of the two models, with the weights determined based on the prediction accuracy on the test set.

[0038] Furthermore, based on the precision of soft soil foundation construction, the construction area is divided into grids, and the 3D coordinates of the center point of each grid are extracted. The standardized values ​​of candidate geological parameters for each grid are supplemented using spatial interpolation, forming a complete geological parameter input matrix for the construction area. This matrix is ​​then input into the trained ensemble model to obtain the predicted solidification intensity value for each grid. Based on this, soil solidification potential characteristics are extracted, including: The curing difficulty level is divided into 5 levels based on the predicted strength, combined with the engineering curing quality requirements: Level 1 (predicted strength ≥ 2.5 MPa, easiest to cure), Level 2 (2.0-2.5 MPa, easy to cure), Level 3 (1.5-2.0 MPa, medium difficulty), Level 4 (1.0-1.5 MPa, relatively difficult to cure), and Level 5 (< 1.0 MPa, extremely difficult to cure). Strength enhancement potential value is calculated as the difference between the predicted strength and the natural strength of the soil before solidification. The larger the difference, the better the strength enhancement effect after solidification. The optimal curing agent dosage range is determined by analyzing the feature importance of the integrated model, establishing the mapping relationship between each candidate geological parameter and the curing agent dosage, and by traversing the strength prediction results under different dosages, the range with the lowest dosage that meets the strength standard is determined. For example, when the moisture content is 30%-40% and the porosity is 1.2-1.5, the optimal dosage range is 12%-15%.

[0039] Finally, the three major features are associated with the 3D coordinates of the grid center point to construct a soil consolidation potential feature set covering the entire region, providing accurate spatial geological basis for subsequent construction parameter recommendations and regional priority division.

[0040] A case-based reasoning system is used to construct a case library for in-situ consolidation of soft soil. The core search criteria are the soil consolidation potential characteristics of the current construction area. Similar historical cases in the case library are matched, and historical construction parameters are corrected in combination with the current construction scenario. Finally, optimized construction parameter features are output. By leveraging successful historical experience, the system can quickly adapt to the current scenario and improve the rationality and reliability of construction parameters.

[0041] Specifically, the process of recommending optimized construction parameter features based on case-based reasoning includes the following steps: S221. Assign differentiated weights to the search keywords, and retrieve historical matching cases from the structured case library based on the weight assignment results.

[0042] First, a comprehensive collection of historical case studies was conducted, covering soft soil foundation in-situ stabilization projects across different regions, soil types, and construction scales to ensure the case study database's scenario coverage. The screening phase employed dual admission criteria: ① Data completeness: Case studies must include complete geological parameters, construction parameters, environmental conditions, and stabilization effect testing data; cases lacking core fields (such as 28-day strength after stabilization) were directly eliminated; ② Construction effectiveness: Only successful cases that achieved stabilization quality standards (strength ≥ 1.5 MPa) and were free of major construction accidents were retained to avoid interference from invalid experience.

[0043] The extraction of core features needed to balance relevance and practicality, ultimately identifying eight core features: soil physicochemical parameters (moisture content, void ratio), solidification potential parameters (solidification difficulty level), material parameters (solidifying agent type and key indicators such as viscosity), equipment parameters (machine model and power), operational parameters (spraying pressure, operating speed), and effect parameters (unconfined compressive strength after 28 days of solidification). Standardization employed a categorized approach: numerical features (moisture content, void ratio, spraying pressure, etc.) were uniformly mapped to the [0,1] interval using min-max normalization; categorical features (solidifying agent type, machine model) were converted to binary vectors using one-hot encoding (e.g., solidifying agent types are divided into cement-based, lime-based, and composite-based, corresponding to encodings of [1,0,0], [0,1,0], and [0,0,1]); and graded features (solidification difficulty levels 1-5) were converted to continuous values ​​of 1-5 using ordinal encoding. Finally, the standardized features are stored in a structured manner according to case ID, feature vector, construction parameters, and curing effect, thus constructing a structured case library that supports rapid retrieval.

[0044] Furthermore, based on engineering experience and feature importance analysis, differentiated weights were assigned to the search keywords: soil moisture content weight 0.35, porosity weight 0.35, and solidification difficulty level weight 0.3.

[0045] Subsequently, a feature vector A for the current scene is constructed. The moisture content, porosity, and solidification difficulty level of the soil solidification potential features in the current construction area are multiplied by their corresponding weights to form a three-dimensional weighted feature vector. At the same time, the corresponding features of each historical case are extracted from the case library, and a historical case feature vector B is constructed according to the same weight rules. Then, the cosine similarity algorithm is used to calculate the similarity between A and B. The higher the similarity, the more matched the case is.

[0046] To improve retrieval efficiency, the Kd-tree algorithm is first used to index and construct the feature vectors of the case library, quickly filtering out low-matching cases with similarity below 0.6. Then, the precise cosine similarity is recalculated for the remaining cases, and finally, the cases are sorted from high to low similarity to select the top 5 historical cases.

[0047] S222. Correct the case parameters in the historical matching cases, and extract the optimized construction parameter features from the correction results.

[0048] First, we identified the core differences between the current construction scenario and the top 5 historical cases, categorizing them into three main types: ① Material differences (hardener type, viscosity, density, etc.); ② Equipment differences (machine model, power, nozzle diameter, etc.); ③ Environmental differences (construction temperature, humidity, rainfall, etc.). Based on the degree of influence on construction parameters, we assigned adjusted weights to each difference dimension: material differences 0.4, equipment differences 0.3, and environmental differences 0.3. The weighted correction method is used to quantify and adjust historical construction parameters, satisfying the following condition: Corrected parameter = Historical case parameter × [1 + ∑(Difference coefficient × Corresponding dimension weight)]. The difference coefficient is determined based on the actual differences, such as: ① Material difference coefficient: If the current curing agent viscosity is 20% higher than the historical case, the viscosity difference coefficient is +0.2; if the current curing agent density is 5% lower than the historical case, the density difference coefficient is -0.05; ② Equipment difference coefficient: If the current mechanical power is 15% higher than the historical case, the power difference coefficient is +0.15; if the nozzle diameter is 10% larger than the historical case, the diameter difference coefficient is -0.1; ③ Environmental difference coefficient: If the current construction temperature is 5℃ lower than the historical case, the temperature difference coefficient is -0.1; if the current humidity is 10% higher than the historical case, the humidity difference coefficient is -0.05.

[0049] Furthermore, the weights of the Top 5 corrected cases are calculated and normalized based on the cosine similarity from the retrieval phase, satisfying the following: weight of a case = similarity of that case / ∑(similarity of Top 5 cases), with higher similarity resulting in greater weight. Subsequently, the weighted average method is used to calculate the final optimized parameters for the five corrected core construction parameters (spraying pressure, spraying flow rate, operating speed, mixing speed, and curing agent dosage), obtaining the optimized construction parameter characteristics.

[0050] Meanwhile, by using the 3D coordinates of the current construction area as a link, the optimized parameter feature set is bound to the corresponding spatial location, ensuring that the parameters can accurately match the construction needs of different spatial locations.

[0051] S3. Based on high-precision 3D location coordinates, soil consolidation potential characteristics, and optimized construction parameter characteristics, construction priority area characteristics are obtained through multi-feature fusion for construction area division.

[0052] Specifically, the method of obtaining construction priority area features by dividing the construction area based on high-precision 3D location coordinates, soil consolidation potential characteristics, and optimized construction parameter characteristics through multi-feature fusion includes the following steps: S31. Using the 3D location coordinates of the construction area as a spatial reference, and integrating the soil consolidation potential characteristics and optimized construction parameter characteristics, a regional division evaluation index system is constructed.

[0053] Based on the core needs of soft soil foundation construction resource allocation and engineering practice experience, a regional division evaluation index system is constructed to ensure that the indicators are both relevant and quantifiable. The selection criteria for the indicators are as follows: ① Solidification difficulty level (weight 0.3): As a core indicator, it directly determines the construction difficulty and quality control complexity. The higher the level, the higher the construction cost, and construction in low-difficulty areas should be prioritized; ② Strength improvement potential value (weight 0.25): Reflects the improvement effect of soil bearing capacity after construction. The higher the potential, the greater the construction value, and it is the core value-oriented indicator for priority division; ③ Solidifying agent dosage (weight 0.2): Related to the cost of construction materials. The higher the dosage, the higher the cost per unit area, and cost and construction sequence need to be balanced; ④ Construction complexity score (weight 0.15): Comprehensively considers the impact of mechanical operation space and terrain undulation on operation (score range 1-10 points, the higher the score, the higher the complexity), which directly affects construction efficiency; ⑤ Terrain slope (weight 0.1): The greater the slope, the worse the stability of mechanical operation, which is prone to uneven solidification, and it needs to be used as an auxiliary constraint indicator.

[0054] Furthermore, the weights are rigorously determined using the analytic hierarchy process (AHP): First, a hierarchical model is constructed (target layer: regional priority division; criteria layer: 5 evaluation indicators; scheme layer: each construction area); second, a pairwise comparison matrix is ​​constructed using expert scoring (using a 1-9 scale, e.g., if the difficulty level of solidification is more important than the terrain slope, the scale is 5); then, the maximum eigenvalue and eigenvector of the matrix are calculated to obtain the initial weights; finally, a consistency check is performed (requiring a consistency ratio CR < 0.1). If the requirement is not met, the expert scores are revised until the check passes, ensuring that the weight allocation is scientific and reasonable and avoiding subjective experience bias.

[0055] Considering the differences in the dimensions and value ranges of various indicators, the min-max standardization method is used to uniformly map them to the [0,1] interval, eliminating dimensional interference. Differentiated design of standardization formulas for different types of indicators: ① Grade-based indicators (fixing difficulty level, potential for intensity improvement): For example, if the curing difficulty level is 1-5, and the level of a certain area is 2, then the standardized value = (2-1) / (5-1) = 0.25; ② Numerical indicators (curing agent dosage, terrain slope): For positive indicators (the higher the curing agent dosage, the higher the complexity), positive standardization is used; for negative indicators (the larger the absolute value of the terrain slope, the more unfavorable it is, so the absolute value is taken first and then standardized); ③ Scoring indicators (construction complexity score): directly according to... Mapping (based on a scoring range of 1-10 points). After standardization, the 3D coordinates (X, Y, Z) of a 1m×1m grid in the construction area are used as unique identifiers to establish a one-to-one correspondence between coordinates and the standardized values ​​of each indicator, thus avoiding spatial misalignment.

[0056] Then, a weighted fusion calculation is performed to satisfy the following: comprehensive evaluation score. (in To solidify the standardized values ​​for difficulty levels, Standardized value for strength enhancement potential. This is the standardized value for the curing agent dosage. This is a standardized value for the construction complexity score. (This is a standardized value for terrain slope). A higher score indicates lower construction difficulty, higher value, and higher construction priority for the area. The contribution percentage of each indicator score is recorded simultaneously during the fusion process to provide a basis for subsequent priority feature analysis.

[0057] S32. Based on the aforementioned regional division evaluation index system, the construction area is divided into different priorities through cluster analysis, and the characteristics of the construction priority areas are extracted.

[0058] Using gridded 3D coordinates (X, Y, Z) and a comprehensive evaluation score S as the clustering sample dataset, the clustering results are designed to possess both spatial correlation and priority attributes. First, the elbow rule is used to determine the optimal number of clusters K, followed by K-means clustering. The implementation process is as follows: ① Initialize cluster centers: Select K initial centers from the sample set using a random sampling method; ② Calculate the distance: Use Euclidean distance to calculate the distance between each sample and each cluster center, satisfying: ,in As cluster center; ③ Sample allocation: Each sample is assigned to the nearest cluster center, forming K initial clusters; ④ Update centers: Calculate the mean of all samples within each cluster and use it as the new cluster center; ⑤ Iterative convergence: Repeat steps ②-④ until the cluster centers converge. After clustering, set priority score intervals based on the actual project requirements: high priority region (S∈[0.7,1.0]), medium priority region (S∈[0.3,0.7]), and low priority region (S∈[0,0.3]). If K=4, the medium priority region can be further subdivided into medium-high and medium-low sub-priorities to accommodate more refined resource allocation needs.

[0059] Furthermore, based on the clustering results, multi-dimensional feature extraction is performed to form a feature set for construction priority areas that combines spatial attributes, quantitative indicators, and engineering implications. Specific extraction content includes: 3D spatial extent characteristics: The spatial boundary coordinates of each priority region are extracted using the convex hull algorithm, and the area and volume of the region are calculated by the grid integration method to obtain the coordinates of the center point of the region and the spatial distribution heat map; Quantitative scoring characteristics: Statistical analysis of the average comprehensive score, score standard deviation, and average standardized value of each individual indicator for each priority region; Core project characteristics: Summarize the key attributes of each priority area, such as high priority areas: low curing difficulty, high potential for strength improvement, low construction complexity, and gentle terrain slope, suitable for prioritizing the allocation of high-quality machinery and personnel; medium priority areas: balanced indicators, can be promoted according to conventional resource allocation; low priority areas: high curing difficulty, low strength potential, and high construction complexity, it is recommended to allocate special equipment for construction later. Resource demand characteristics: Combining the characteristics of mechanical operation efficiency and optimized construction parameters, the construction time per unit area, the number of machines required, and the consumption of curing agent for each priority area are estimated, providing a quantitative basis for subsequent resource allocation.

[0060] Finally, all features are integrated according to the structure of priority level - 3D spatial range - indicators - engineering features - resource requirements to form a standardized set of construction priority area features.

[0061] S4. Combining the effective coverage characteristics of materials and the efficiency characteristics of mechanical operation, the synergistic characteristics of materials and machinery are obtained through synergistic optimization based on control theory.

[0062] First, the effective coverage characteristics of the material are obtained through material diffusion optimization using fluid dynamics, including the following steps: S411. Based on the porous media seepage theory in fluid dynamics, a diffusion model of the solidifying agent in soft soil is constructed.

[0063] Based on the seepage theory of porous media in fluid dynamics, a three-dimensional diffusion model of the solidifying agent is constructed with Darcy's law and the continuity equation as its core, fully adapting to the heterogeneous and complex pore structure characteristics of soft soil. Darcy's law describes the relationship between the solidifying agent seepage velocity and the hydraulic gradient, with the core formula being: (v is the seepage velocity, k is the soil permeability coefficient, Δh is the head difference between the two ends of the seepage path, and L is the seepage path length). Considering the spatial variability of the void ratio in soft soil, an empirical formula is used. , Let be the baseline permeability coefficient, e be the void ratio, and α be the fitting coefficient. A dynamic mapping relationship between the permeability coefficient and the void ratio is established to achieve differentiated assignment of the diffusion coefficient at different spatial locations.

[0064] At the same time, a continuity equation is introduced. ρ is the density of the curing agent, φ is the soil porosity, and t is the time. Based on the non-Newtonian fluid characteristics of the curing agent (using a power-law model to describe the change in viscosity with shear rate), a complete three-dimensional diffusion control equation is constructed.

[0065] S412. Combining the soil pore structure, the rheological properties of the curing agent, and the operating parameters of the construction machinery, the diffusion model is used to simulate the diffusion process of the curing agent, and the operating parameters are optimized to ensure that the material effectively covers the target area, thereby obtaining the characteristics of the effective coverage range of the material.

[0066] Specifically, finite element analysis software was used to numerically simulate the diffusion process. By batch calculating the diffusion results under different combinations of injection pressure and flow rate, the diffusion path, coverage area and concentration distribution cloud map of the curing agent corresponding to each combination were obtained.

[0067] Furthermore, the optimization objective is clearly defined as ensuring the coverage area completely encompasses the target construction area with a uniformity error of <10%, where the uniformity error is calculated using the coefficient of variation: CV = σ / μ (σ is the standard deviation of the curing agent concentration within the calculation domain, and μ is the mean concentration). Particle swarm optimization (PSO) is employed to optimize parameters. The injection pressure and flow rate are set as two-dimensional particle codes, and the fitness function is set as CV + λ·(1-CR) (CR is the coverage achievement rate, and λ is the weighting coefficient, taken as 0.6). The particle swarm size is set to 30, the learning factors c1 = c2 = 2, and the inertia weight ω decreases linearly from 0.9 to 0.4, with 50 iterations. The optimal combination of injection pressure and flow rate parameters that minimizes the fitness function is obtained through iterative calculation, while invalid solutions exceeding the equipment's operating limits are eliminated. Then, the optimal injection parameters optimized by PSO are substituted into the diffusion model for final simulation, outputting complete three-dimensional diffusion results of the curing agent.

[0068] The extraction of effective coverage features for materials needs to be based on practical engineering requirements: First, the effective coverage area is defined, using the concentration threshold corresponding to a curing agent concentration greater than or equal to the designed dosage as the standard. Spatial areas meeting this condition are extracted, and the coverage area and diffusion depth are statistically analyzed using domain grid coordinates, along with the 3D boundary coordinates of the covered area. Second, the uniformity of concentration distribution is quantified by calculating the coefficient of variation (CV) of the curing agent concentration within the effective coverage area. A CV < 5% is considered excellent, and 5%-10% is considered good. Characteristic parameters of the concentration distribution cloud map (such as maximum concentration, minimum concentration, and average concentration) are also calculated. Third, diffusion efficiency is evaluated by calculating the effective utilization rate of the curing agent. Finally, the above features (coverage area, diffusion depth, 3D boundary coordinates, concentration CV, and effective utilization rate) are integrated into a feature set of the material's effective coverage area.

[0069] Secondly, the mechanical operation efficiency characteristics are obtained through mechanical behavior optimization using reinforcement learning, including the following steps: S421. Construct a mechanical operation optimization system based on the Q-learning algorithm.

[0070] Based on the dynamic scenario requirements of soft soil foundation construction, a high-dimensional and quantifiable state space S and action space A are constructed to ensure comprehensive state perception and reasonable action adjustment.

[0071] The state space S contains six core quantitative parameters: ① Mechanical 3D position: Beidou high-precision positioning results serve as the spatial reference; ② Operating speed: ranging from 0.2 to 0.8 m / s, collected in real time by the mechanical speed sensor; ③ Stirring speed: ranging from 50 to 200 r / min, associated with mechanical control system data; ④ Spraying angle: ranging from 0 to 90 degrees, quantified by the attitude sensor; ⑤ Material coverage uniformity: characterized by the concentration coefficient of variation (CV) (0-100%, the smaller the CV, the better the uniformity); ⑥ Terrain slope: calculated based on DEM data (-15 degrees to 15 degrees, a common slope range in soft soil areas), reflecting differences in operating resistance.

[0072] Action Space A focuses on the fine-tuning of the mechanical components, with each adjustment setting taking into account the equipment's performance limits and construction accuracy requirements: operating speed adjustment ±0.1m / s, stirring speed adjustment ±5r / min, and spray angle adjustment ±3 degrees; at the same time, clear action constraints are defined: the adjusted parameters must fall within the equipment's safe operating range.

[0073] Furthermore, a dynamic reward function that takes into account multiple objectives is constructed. The system dynamically allocates weights to adapt to different construction priorities, and the reward / penalty values ​​are quantified based on project losses. Specifically: ① Construction efficiency reward. Based on the target amount of work done per unit time, when the target is met... =10, when the standard is not met =5, if the amount of work done per unit time exceeds the target value by 120%. =12; ②Quality reward The standard for compliance is a material coverage uniformity CV < 10%. =15, when the standard is not met =0, if CV < 5% (uniformity is good). =18; ③ Energy consumption penalty Energy consumption thresholds are determined based on machinery model and construction specifications; exceeding the threshold will result in... =8, not exceeding the limit =0. The weights α, β, and γ are dynamically adjusted: α=0.5, β=0.3, γ=0.2 in the early stage of construction; β=0.6, α=0.2, γ=0.2 in the middle stage of construction; γ=0.3, α=0.4, β=0.3 in the later stage of construction; and β is increased by 0.1 for high-priority areas to strengthen quality control.

[0074] S422. Using the operating parameters of the construction machinery as actions and construction efficiency and construction quality as reward signals, the mechanical operation behavior is optimized through continuous iterative learning to obtain the mechanical operation efficiency characteristics.

[0075] A progressive training process of initialization-exploration-update-convergence is adopted, and an experience replay mechanism is introduced to improve training stability and efficiency. First, the Q-table is initialized: the Q-table dimension is the state space dimension × action space dimension. Considering the data scale after quantization of state parameters, small random numbers of 0.01-0.1 are used for initialization, and the upper limit of the Q-value is set to 30. During training, the machine acts as an agent and selects actions according to the ε-greedy strategy: initially ε=0.3, ε decays by 0.01 every 100 iterations until ε=0.05. Data acquisition and update: the machine collects the current state s in real time, and after executing action a, obtains the reward r and the new state. ,Will Stored in the experience pool, 32 samples are randomly selected each time an update is performed, and the formula is updated using Q: Iterative optimization is performed, where the learning rate η is initially set to 0.1, and decreases by 0.01 every 200 iterations until it stabilizes at 0.01. The discount factor λ = 0.9.

[0076] Based on the converged Q-table, and combined with spatial positioning information and engineering calculations, a multi-dimensional, spatialized mechanical operation efficiency feature set is constructed. First, the optimal operation parameters are screened: for each state s corresponding to the 3D position of the machine, the action a with the largest Q value in the Q-table is retrieved. If there are multiple actions with a Q value difference of less than 5%, the action with the lowest energy consumption is selected as the optimal action, and the corresponding operation parameters (operating speed v, stirring speed n, and spray angle θ) are extracted.

[0077] Then, the core efficiency indicators were calculated: ① Work volume per unit time: Combining the mechanical working width, curing depth, and working speed v, to meet the following requirements. =v×width×depth calculation; ②Energy consumption: Real-time power P is collected by mechanical power sensor, combined with operation time t, and calculated according to the formula E=P×t, and the energy consumption per unit of project is statistically calculated simultaneously.

[0078] Finally, spatial correlation of features is achieved: using the timestamp of Beidou 3D positioning as the link, the optimal operating parameters (v,n,θ), the amount of work per unit time, the energy consumption per unit of work, and the mechanical 3D operation trajectory (X,Y,Z) are precisely bound together to form a mechanical operation efficiency feature set of spatial position-operating parameters-efficiency index.

[0079] Furthermore, the method of combining the effective coverage characteristics of the material and the efficiency characteristics of mechanical operation, and obtaining the material-mechanical synergy characteristics based on the synergistic optimization of control theory, includes the following steps: S431. Based on model predictive control theory, construct a collaborative optimization model for material diffusion and mechanical operation.

[0080] Based on the coupling characteristics of soft soil foundation construction, the control variables and objective functions are precisely defined, and a linear quadratic control (LQC) collaborative model that balances accuracy and practicality is constructed.

[0081] The core basis for selecting control variables is the construction impact weight: the mechanical operation speed v and the mixing speed n are the core input variables. The two have a strong coupling relationship with the material diffusion effect and need to be optimized simultaneously.

[0082] The control objectives focus on the core engineering requirements: material coverage uniformity U, calculated based on the coefficient of variation of concentration distribution, U = 1 - CV, where CV is the coefficient of variation. The closer U is to 1, the better the uniformity. The material utilization rate per unit time η (the ratio of the total amount of curing agent to the total amount sprayed within the effective coverage area, reflecting the degree of material waste) and the weighting of these two parameters are also considered. Based on the dynamic allocation of construction priorities, high-priority areas must be given priority in ensuring curing quality. In low-priority regions, efficiency can be prioritized, and weight reversal can be set to... ; take the medium priority area Achieve balanced optimization.

[0083] The objective function is precisely defined as follows: A smaller J value indicates a better synergistic effect.

[0084] The constraint system covers three dimensions: ① Mechanical performance constraints: operating speed v∈[0.2,0.8]m / s, mixing speed n∈[50,200]r / min; ② Material delivery constraints: jet flow rate q∈[5,20]L / min; ③ Construction specification constraints, including mixing speed and operating speed matching constraints, and soil compatibility constraints.

[0085] S432. Combining the characteristics of the effective coverage area of ​​the material and the characteristics of the mechanical operation efficiency, the collaborative optimization model achieves dynamic coordination between mechanical operation and material diffusion through rolling optimization and feedback correction, thereby obtaining the material-mechanical coordination characteristics.

[0086] A rolling optimization framework based on Model Predictive Control (MPC) is adopted to achieve accurate prediction and parameter optimization in dynamic scenarios. Firstly, the prediction time domain is set to 5 seconds, and the core prediction algorithm uses a Kalman filter to construct the state vector. The process noise is calibrated based on construction uncertainties (such as velocity fluctuations caused by mechanical vibration and diffusion deviations caused by soil heterogeneity, which are set as a diagonal matrix, with the diagonal elements being the variances of each state quantity, and the initial values ​​are determined through offline experiments), and the observation noise is set based on the sensor accuracy.

[0087] A three-dimensional diffusion model of the curing agent (inputs: injection flow rate q, soil porosity, output: coverage uniformity U) and a mechanical operation efficiency model (inputs: v, n, output: workload per unit time, indirectly related to material utilization rate η) are embedded into the state transition process of a Kalman filter. Coupled prediction is achieved through state vector propagation. Finally, the optimal combination of control variables that minimizes J in the prediction time domain is obtained. .

[0088] Furthermore, the feedback correction core employs an adaptive PID control algorithm. Initial parameters are determined using the Ziegler-Nichols tuning method, and a priority adaptation mechanism is introduced: in high-priority regions, Kp and Ki are increased to enhance correction sensitivity; in low-priority regions, Kd is increased to reduce adjustment frequency. The iterative optimization process is as follows: after each 5-second prediction optimization, the predicted value is compared with the actual monitored value, and the comprehensive error E is calculated. If E ≥ 5%, the control variables (v, n, q) are corrected using the PID algorithm, and the data is re-input into the collaborative model for the next round of prediction optimization; this continues until E < 5% for three consecutive iterations, and the error of a single index meets the requirements, at which point the current stage of correction stops, and the system enters a stable operating state.

[0089] Based on the closed-loop optimization results, multi-dimensional feature extraction is carried out to form a standardized collaborative feature set, providing core support for subsequent adjustments to the construction plan. Feature extraction covers four core dimensions: Optimal combination of cooperative parameters: includes core operating parameters And adaptability conditions, clarify the precise matching values ​​under the 3D coordinates (X,Y,Z), soil solidification difficulty level, and environmental risk level of different construction areas; Cooperative error characteristics: detailed analysis of operating parameter error, material diffusion error, comprehensive cooperative error E, and error fluctuation range and maximum deviation value; Collaborative stability rating: A 10-point scale is used for quantification. The scoring rules are as follows: comprehensive error E < 3% (4 points), 3%-5% (3 points); parameter adjustment frequency < 2 times / min (3 points), 2-5 times / min (2 points), > 5 times / min (1 point); continuous stable running time > 30min (3 points), 10-30min (2 points), < 10min (1 point). A total score of 8-10 is excellent, 5-7 is good, and < 5 is in need of optimization. Spatial correlation features: By constructing a correlation index with BeiDou 3D positioning timestamps and spatial coordinates (X,Y,Z), the above features are precisely bound with construction priority area features (high / medium / low priority) and environmental impact features (risk level), clarifying the collaborative optimization focus of different spatial locations (e.g., high priority areas focus on stability and uniformity, while low priority areas focus on efficiency and cost).

[0090] The final integrated material and mechanical collaborative feature set is labeled with the data source, accuracy level and applicable scenarios of each feature field, ensuring that it can directly support the dynamic adjustment of the construction plan.

[0091] S5. Integrate the characteristics of the influence of the environment on the curing process, the characteristics of the construction priority area, and the characteristics of material and machinery synergy to dynamically plan the construction schedule.

[0092] First, based on the ARIMA model in time series analysis, dynamic environmental parameters during the construction process are collected. The changing patterns of environmental factors (temperature, humidity, wind speed, and rainfall) over time and their correlation with the curing effect are analyzed. The degree of influence of environmental factors on the curing process is assessed, and environmental impact characteristics (key influencing factors, influence weights, and risk levels) are extracted. The impact of dynamic environmental changes on the curing process is quantified, providing a basis for subsequent adjustments to the construction plan.

[0093] Specifically, the characteristics of the impact of the environment on the curing process are obtained, including the following steps: Using the official start time of construction as the time reference point, UTC timestamps were used to integrate multi-source data to construct a four-dimensional time series dataset encompassing time, environment, construction, and effect. Data integration required adherence to strict matching rules: ① Environmental parameters needed to be correlated with the monitoring averages for the corresponding time period; ② Construction content needed to be recorded in detail, including the 3D coordinates of the work area, curing agent dosage, and mechanical operation parameters for the corresponding time period, forming structured work log entries; ③ Curing strength data needed to be correlated with the corresponding construction time period and 3D area according to four key ages: 1d, 3d, 7d, and 28d, clearly defining the strength growth data under different environmental conditions.

[0094] Next, the ADF test is used to determine the stationarity of the time series of each environmental parameter. The significance level is set at 0.05, and the null hypothesis is that the series has a unit root (non-stationarity). If the p-value is < 0.05, the null hypothesis is rejected, and the series is considered stationary; if the p-value is ≥ 0.05, differencing is required. For non-stationary series (such as monthly fluctuations in rainfall and temperature), first-order differencing is used first, followed by a re-ADF test until the series reaches a stationary state. If the series is still non-stationary after first-order differencing, second-order differencing is performed, and the differencing order d is recorded. In this embodiment, Z-score standardization is also performed on all environmental parameter series to unify the data to an interval with a mean of 0 and a standard deviation of 1, avoiding interference from dimensional differences in model training.

[0095] Furthermore, an ARIMA(p,d,q) model is constructed (p is the autoregressive order, d is the differencing order, and q is the moving average order). The input is the processed time series of four environmental parameters (temperature, humidity, wind speed, and rainfall), and the output is the growth rate of curing intensity for the corresponding time period. Based on this, parameter optimization adopts the AIC criterion (Akaike Information Criterion), and the optimal parameter combination is selected by a traversal method: the value range of p is set to 0-3, the value range of q is set to 0-3, and the value of d is the differencing order (0, 1, or 2) determined in the preprocessing stage. The AIC value corresponding to each (p,d,q) combination is calculated, and the combination with the smallest AIC value is selected as the optimal model parameters. Subsequently, the influence weights of each environmental factor on the growth rate of curing intensity were quantified by analysis of variance (ANOVA): a multiple linear regression model was constructed with four sets of environmental parameters as independent variables and curing intensity growth rate as dependent variable. The standardized regression coefficients of each independent variable were calculated, and the percentage of the absolute value of the standardized regression coefficients was used as the influence weights. At the same time, the significance of the influence of each factor was judged by the F test, and non-significant factors with P value ≥ 0.05 were eliminated.

[0096] Furthermore, combining the results of the analysis of variance with engineering practice, key environmental factors (such as temperature and rainfall) with a p-value < 0.05 and an impact weight greater than 0.2 were selected, while secondary factors (such as wind speed) with an impact weight less than 0.1 were eliminated to simplify the subsequent decision-making process.

[0097] In this embodiment, the threshold setting needs to take into account construction specifications, historical data statistics and indoor test results: ① Temperature threshold: Through indoor tests, the growth curve of the curing strength of soft soil at different temperatures was obtained. It was determined that when the temperature is below 5℃, the curing reaction rate decreases by more than 50% and the curing strength compliance rate is less than 60%. Therefore, the low temperature threshold is set at 5℃; ② Rainfall threshold: Statistical analysis of historical construction data shows that when the daily rainfall is >10mm, rainwater easily washes away the uncured curing agent, resulting in a loose curing layer and a strength loss of more than 30%. Therefore, the rainfall threshold is set at 10mm.

[0098] Furthermore, based on thresholds, environmental impact risk levels are divided into three levels: Level 1 (low risk): temperature 5℃-35℃, rainfall ≤5mm, humidity 40%-70%, curing strength compliance rate ≥95%; Level 2 (medium risk): temperature 3℃-5℃ or 10mm ≥ rainfall >5mm or humidity >70% / <40%, curing strength compliance rate 70%-95%; Level 3 (high risk): temperature <3℃ or rainfall >10mm or wind speed >8m / s, curing strength compliance rate <70%.

[0099] The final integrated set of environmental impact characteristics includes: Key influencing factors (name, weight of influence, significance P-value); Threshold standards for each factor (judgment basis, supported by experimental / statistical data); Risk level classification standard (corresponding environmental conditions and range of curing strength compliance rate). Risk response recommendations (such as suspending outdoor operations for Level 3 risks and increasing the curing agent dosage by 5%-10% for Level 2 risks) are provided. At the same time, by linking timestamps with the 3D coordinates of the construction area, accurate environmental risk data can be provided for dynamic adjustments to the subsequent construction plan.

[0100] Furthermore, the dynamic planning of the construction schedule, based on the influence characteristics of the fusion environment on the curing process, the characteristics of construction priority areas, and the synergistic characteristics of materials and machinery, includes the following steps: S51. Based on dynamic programming theory, the construction process is divided into multiple decision-making stages according to time phases.

[0101] Based on the accuracy requirements of construction progress control and the timeliness of environmental forecast data, a dynamic time segmentation method is adopted to determine the time stages: small soft soil foundation construction projects are divided into stages by hours, medium-sized projects by days, and large projects by weeks, to ensure that the stage division adapts to the control needs of different project scales.

[0102] State variables A full-dimensional quantitative representation needs to be achieved, specifically defined as follows: Current construction area: The 3D grid coordinates (X,Y,Z) and priority level (high / medium / low) of the associated construction priority area feature set are used to define the spatial range of the current operation; Completed work volume: The volumetric method is used according to the construction measurement specifications, that is, the volume of solidified soil = Σ (area of ​​a certain area × solidification depth), and the completion percentage of each priority area is marked simultaneously; Remaining Machinery / Materials Resources: Machinery resources are broken down into the number of machines of each model and remaining operating hours, while materials resources are broken down into the reserves of each type of curing agent, the amount in transit, and the safety stock threshold. All data are based on real-time statistical ledger data. Environmental risk level: The real-time risk level (levels 1-3) of the environmental impact characteristics set is directly adopted, and the key environmental parameters (temperature, rainfall) of the corresponding time period are associated.

[0103] Decision variables It is necessary to balance feasibility and refinement, specifically including: Current phase of construction area selection: Clearly define the priority areas and specific 3D sub-areas to be selected; Machinery allocation quantity: The number of different types of machinery is allocated according to the complexity of regional construction and the amount of work required per unit time; Work schedule: Detail the work plan by time period based on the environmental risk level (e.g., core work is arranged during low-risk periods (9:00-17:00), and auxiliary work is adjusted during high-risk periods), and simultaneously mark the emergency trigger conditions for work interruption (e.g., work must be stopped immediately if rainfall is >10mm).

[0104] S52. Based on the regional characteristics of construction priority, material and machinery synergy, and environmental impact characteristics of each stage, construct the state transition equation for construction plan optimization.

[0105] Based on the regionally differentiated unit-time engineering quantity in the synergistic characteristics of materials and machinery, a multi-dimensional state transition equation is constructed. This enables dynamic iteration of various state parameters. In this embodiment, the specific transition logic is as follows: Completed transfer of work volume: If the decision variables at the current stage Select high-priority area A, allocate 2 machines of model M1 (work volume Q = 50 m³ / h), work time t = 8 hours, and assume no downtime due to environmental interference. Then the formula for transferring the completed work volume is: Meanwhile, the completion percentage of area A is 800 / (1000×1)×100%=80%; Transfer of remaining mechanical resources: The energy consumption per unit time of model M1 machinery is E=20kW·h / h, and the initial remaining working time of a single machine is T0=100 hours. Therefore, the formula for transferring the remaining working time of two machines after 8 hours of operation is: Simultaneously update mechanical energy consumption statistics; Transfer of surplus material resources: If the unit consumption of curing agent in area A is C = 150 kg / m³, then the material consumption for 800 m³ of work is 800 × 150 = 120,000 kg = 120 t. If the initial surplus material is 500 t, then the transfer formula is: If the remaining quantity is lower than the safety stock threshold (e.g., 100t), a material replenishment warning will be triggered. Environmental risk level transition: If the current environmental risk level is Level 1 (low risk), and based on environmental time series forecast data, the predicted risk level for the next stage is also Level 1, then... If rainfall is predicted, the risk level will be upgraded to level 3, and the relocation logic will be adjusted accordingly (e.g., the operation time will be shortened to 0).

[0106] Constraint checks need to be embedded during the state transition process: if the number of machines allocated exceeds the number of currently available machines (e.g., if there is only 1 available machine, the decision is to allocate 2 machines), the allocation number is automatically corrected to 1 machine, the transition result is recalculated, and the feasibility of the transition equation is ensured.

[0107] S53. Based on the optimization objective, the dynamic programming construction plan is solved in reverse order.

[0108] Specifically, a multi-objective weighted objective function that takes into account schedule, cost, and quality is constructed. The weights are determined using the analytic hierarchy process. (make sure The core objective of weight allocation adaptation engineering is: Scenarios with priority given to project timelines (e.g., construction must be completed before the flood season): ; Cost-priority scenarios (e.g., limited budget): ; Quality-priority scenarios (such as core engineering areas): .

[0109] The quantitative details of each target are as follows: Project timeline target: The total planned construction period (e.g., 30 days). The actual cumulative construction period is calculated by adding up the duration of each phase of work plus adjusting for downtime. ), The closer to 0, the better the schedule control; Cost target: The total budgeted cost (including machinery costs, material costs, labor costs, and management fees). This represents the actual cumulative cost. , The closer the value is to 1, the more severe the cost overrun. Quality target: Q is the quality compliance rate, which is statistically analyzed by regional sampling inspection (Q = area of ​​quality compliance area / total construction area × 100%). The sampling density is set according to the construction specifications. The higher the Q, the better the quality.

[0110] Furthermore, the constraints employ a combination of hard and soft constraints: Environmental risk constraints: In Level 3 risk (temperature < 3℃, rainfall > 10mm, or wind speed > 8m / s), open-air curing operations are strictly prohibited; only indoor preparation and equipment maintenance are permitted. In Level 2 risk (temperature 3℃-5℃, rainfall ≥ 10mm > 5mm, or humidity > 70% / < 40%), operation time must be shortened (≤ 6 hours / day), and the frequency of quality inspections must be increased. In Level 1 risk, normal operation is permitted (≤ 10 hours / day). Hard constraints on resources: Machinery usage ≤ number of currently available machines (including machines on the way, calculated 2 days in advance), material consumption ≤ (current reserves + on-the-way replenishment - safety stock threshold), safety stock threshold is set based on the maximum consumption in 3 days (e.g., if the average daily consumption is 50t, then safety stock = 150t). Hard quality constraints: Overall quality compliance rate Q≥95%, compliance rate of individual priority areas≥90%. If a sampling inspection of a certain area fails to meet the standards, a rework process must be triggered and the cost of quality must be included. Soft constraints on project duration: actual project duration (Allow a 10% flexibility in the project schedule to cope with unforeseen disruptions).

[0111] Furthermore, a combined strategy of reverse-order solving and rolling optimization is adopted to ensure both solution efficiency and plan adaptability. The specific process of reverse-order solving is as follows: Initialize the terminal state: Set the ideal state of the last stage T (completed work volume = total work volume, remaining mechanical / material resources ≥ safety threshold, environmental risk level 1), and the objective function value F = 0 (optimal state). Reverse iterative computation: Working backward from stage T-1 to stage 1, for each possible state in each stage. Enumerate all feasible decision variables (Region selection, machine allocation, and operation time combination that satisfy the constraints), calculate the corresponding state transition equations. , combined Find the optimal objective function value and calculate the current state. Choice Decision Given the objective function value, select the decision with the smallest objective function value as the optimal decision for that state, and store the optimal decision path synchronously. State pruning optimization: Prune the state space at each stage to remove non-optimal states where the objective function value exceeds a threshold (e.g., F > 0.5), thereby reducing computational load and ensuring solution efficiency.

[0112] Environmental time series forecasting data adopts a fusion approach of short-term high-precision forecasting and long-term trend forecasting: short-term forecasts use ARIMA model results to adjust work plans for the near term; long-term forecasts use trend extrapolation to control the overall project schedule.

[0113] The adjustment of the plan needs to achieve a closed loop of prediction-adjustment-verification. For example, if rainfall is predicted on the 5th day (environmental risk level is upgraded to level 3), then the decision variables for this stage are adjusted (the construction area is changed to auxiliary cleaning work in a low-priority area, the number of machines allocated is reduced by 50%, and the working time is adjusted to 4 hours). The objective function value after adjustment is calculated. If F≤0.2 (meets the optimization requirements), the adjustment plan is determined; if F>0.2, further optimization is carried out (such as allocating rain protection equipment in advance, or adjusting the amount of work before and after the stage).

[0114] The adjusted optimal construction plan should include three core elements: Time dimension: Work area, number of machines, work duration, material consumption, and personnel allocation for each stage (day / hour); Spatial dimension: the specific construction sequence of each priority area, the operation sequence of 3D sub-areas, and the location of quality inspection points; Emergency response: Emergency response plans for sudden environmental changes (such as heavy rain, low temperatures), equipment failures, and material shortages (such as backup machinery allocation routes, emergency material supply channels, and rework processes).

[0115] After the plan is output, a final verification is required: verify whether the plan meets all constraints and whether the objective function value reaches the optimum (F≤0.3) by simulating the construction process. If not, iterate again until the requirements are met, and finally complete the optimization of the entire construction process.

[0116] Please see Figure 2In this embodiment, to efficiently execute the soft soil foundation in-situ solidification construction optimization method based on BeiDou 3D positioning provided by the present invention, the present invention also provides a soft soil foundation in-situ solidification construction optimization system based on BeiDou 3D positioning, comprising: an input device, an output device, a processor, and a memory, wherein the input device, output device, processor, and memory are interconnected, and the memory contains program instructions for the steps of the soft soil foundation in-situ solidification construction optimization method based on BeiDou 3D positioning. The soft soil foundation in-situ solidification construction optimization system based on BeiDou 3D positioning of the present invention has a compact structure and stable performance, and can stably execute the soft soil foundation in-situ solidification construction optimization method based on BeiDou 3D positioning of the present invention, further improving the overall applicability and practical application capability of the present invention.

[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the present invention.

Claims

1. An optimized construction method for in-situ consolidation of soft soil foundations based on BeiDou 3D positioning, characterized in that, Includes the following steps: High-precision 3D position coordinates are obtained through positioning optimization by fusing multi-source data; By utilizing geological features derived from machine learning, soil consolidation potential features are extracted, and optimized construction parameters are recommended based on case-based reasoning. Based on high-precision 3D location coordinates, soil consolidation potential characteristics, and optimized construction parameter characteristics, construction priority area characteristics are obtained by dividing the construction area through multi-feature fusion. By combining the effective coverage characteristics of materials and the efficiency characteristics of mechanical operation, the synergistic characteristics of materials and machinery are obtained through synergistic optimization based on control theory. By integrating the characteristics of the influence of the environment on the curing process, the regional characteristics of construction priority, and the synergistic characteristics of materials and machinery, a dynamic construction plan is developed.

2. The optimized construction method for in-situ solidification of soft soil foundation based on BeiDou 3D positioning according to claim 1, characterized in that, The method of obtaining high-precision 3D position coordinates through multi-source data fusion positioning optimization includes the following steps: Primary hierarchical fusion of BeiDou and IMU data is achieved through Kalman filtering; By introducing DEM digital elevation data of the construction area, the elevation consistency of the positioning results after layer fusion is verified to obtain high-precision 3D position coordinates.

3. The optimized construction method for in-situ solidification of soft soil foundation based on BeiDou 3D positioning according to claim 1, characterized in that, The method of extracting soil consolidation potential features using geological features through machine learning includes the following steps: Construct a structured dataset linking 3D location with multi-dimensional geological parameters; Based on the structured associated dataset, initial feature selection and correlation analysis are performed to obtain core geological features; The core geological features are used to train an ensemble learning model and extract soil consolidation potential features.

4. The optimized construction method for in-situ solidification of soft soil foundation based on BeiDou 3D positioning according to claim 1, characterized in that, The process of recommending optimized construction parameter features based on case-based reasoning includes the following steps: Differentiated weights are assigned to search keywords, and historical matching cases are retrieved from the structured case library based on the weight assignment results; The case parameters in the historical matching cases are corrected, and the optimized construction parameter features are extracted from the corrected results.

5. The optimized construction method for in-situ consolidation of soft soil foundation based on BeiDou 3D positioning according to claim 1, characterized in that, The method for obtaining construction priority area features based on high-precision 3D location coordinates, soil consolidation potential characteristics, and optimized construction parameter characteristics through multi-feature fusion for construction area division includes the following steps: Using the 3D location coordinates of the construction area as a spatial reference, and integrating the soil consolidation potential characteristics and optimized construction parameter characteristics, a regional division evaluation index system is constructed. Based on the aforementioned regional division evaluation index system, the construction area is divided into different priorities through cluster analysis, and the characteristics of the construction priority areas are extracted.

6. The optimized construction method for in-situ solidification of soft soil foundation based on BeiDou 3D positioning according to claim 1, characterized in that, The method of combining the effective coverage characteristics of the material and the mechanical operation efficiency characteristics, and obtaining the material-mechanical synergy characteristics based on the synergistic optimization of control theory, includes the following steps: Based on model predictive control theory, a collaborative optimization model for material diffusion and mechanical operation is constructed. By combining the characteristics of effective material coverage and mechanical operation efficiency with the aforementioned collaborative optimization model, dynamic synergy between mechanical operation and material diffusion is achieved through rolling optimization and feedback correction, thereby obtaining the material-mechanical synergy characteristics.

7. The optimized construction method for in-situ consolidation of soft soil foundation based on BeiDou 3D positioning according to claim 6, characterized in that, The effective coverage characteristics of the material are obtained through material diffusion optimization using fluid dynamics, including the following steps: Based on the porous media seepage theory in fluid dynamics, a diffusion model of solidifying agent in soft soil is constructed. By combining soil pore structure, curing agent rheological properties, and construction machinery operating parameters, the diffusion model is used to simulate the diffusion process of the curing agent, optimize operating parameters to ensure that the material effectively covers the target area, and obtain the characteristics of the effective coverage range of the material.

8. The optimized construction method for in-situ solidification of soft soil foundation based on BeiDou 3D positioning according to claim 6, characterized in that, The mechanical operation efficiency characteristics are obtained through mechanical behavior optimization using reinforcement learning, including the following steps: Based on the Q-learning algorithm, a mechanical operation optimization system is constructed. Using the operating parameters of construction machinery as actions and construction efficiency and construction quality as reward signals, the mechanical operation behavior is optimized through continuous iterative learning to obtain the mechanical operation efficiency characteristics.

9. The optimized construction method for in-situ solidification of soft soil foundation based on BeiDou 3D positioning according to claim 1, characterized in that, The influence characteristics of the fusion environment on the curing process, the characteristics of construction priority areas, and the synergistic characteristics of materials and machinery are used to dynamically plan the construction schedule, including the following steps: Based on dynamic programming theory, the construction process is divided into multiple decision-making stages according to time phases; Based on the regional characteristics of construction priority, material and machinery synergy, and environmental impact characteristics of each stage, a state transition equation for construction plan optimization is constructed. Based on the optimization objective, the construction plan is solved by dynamic programming in reverse order.

10. An optimized construction system for in-situ solidification of soft soil foundations based on BeiDou 3D positioning, characterized in that, The soft soil foundation in-situ solidification construction optimization system based on BeiDou 3D positioning includes: an input device, an output device, a processor, and a memory. The input device, output device, processor, and memory are interconnected. The memory includes program instructions, which are used to execute the soft soil foundation in-situ solidification construction optimization method based on BeiDou 3D positioning as described in any one of claims 1-9.

Images