Method and system for underwater dredging optimization

Abstract

The invention relates to the technical field of deep foundation pit desilting, in particular to a method and system for underwater desilting optimization. The method comprises the following steps: integrating desilting influence parameters into a grid division mechanism to obtain a three-dimensional dynamic terrain grid model with an importance weight; through multi-source data fusion and spatial interpolation, obtaining a sludge density and moisture content spatial distribution diagram continuously distributed in the whole field; dynamic construction window parameters are obtained through multi-target search by combining the three-dimensional dynamic terrain grid model, the steady-state flow velocity field characteristics and the turbulence intensity characteristics and integrating an environment risk coefficient and an operation efficiency reduction coefficient; according to the sludge density, the moisture content spatial distribution diagram, the layering result along the section and the preliminary soil type identification, the optimal desilting equipment moving track is analyzed; and on the basis of the running state of the dredging equipment, the dynamic construction window parameters and the optimal moving track of the dredging equipment, an underwater dredging optimization scheme of deep foundation pit excavation is obtained through a fuzzy algorithm.

Description

Technical Field

[0001] This invention relates to the field of deep foundation pit dredging technology, and specifically to a method and system for optimizing underwater dredging. Background Technology

[0002] During deep foundation pit excavation, underwater silt deposition can easily lead to insufficient bearing capacity of the pit foundation, low equipment operating efficiency, and even cause environmental safety problems such as settlement of surrounding buildings and water pollution. Therefore, underwater dredging is a critical step in deep foundation pit construction. Existing deep foundation pit underwater dredging technologies mostly rely on traditional experience or single-dimensional technical methods, which have the following significant drawbacks: Fixed mesh generation: Traditional mesh generation only considers the single factor of terrain and adopts a post-processing approach of first generating the mesh and then incorporating other parameters. This approach cannot adapt to the real-time changes in terrain during deep foundation pit excavation. The mesh accuracy is insufficient in key areas (such as thick silt layers and core equipment operation areas), while the mesh is redundant in non-critical areas, affecting the accuracy of subsequent data analysis.

[0003] Inaccurate flow velocity data processing: Underwater flow velocities in deep foundation pits exhibit seasonal fluctuations and random noise (equipment disturbances, water flow eddies, etc.). Existing technologies employ single filtering methods (such as Kalman filtering and median filtering), which do not consider seasonal fluctuation characteristics. This can easily lead to the loss of turbulence signals or noise residue, making it impossible to provide reliable flow velocity field data for silt diffusion prediction and equipment operation stability analysis.

[0004] Insufficient characterization of silt distribution: Using a single data source (such as field sampling) or fixed parameter interpolation methods to construct the distribution of silt density and water content has the problems of limited coverage and insufficient accuracy. It does not consider the indirect influence of flow velocity on silt deposition and is difficult to accurately reflect the spatial heterogeneity of silt distribution.

[0005] The efficiency of sediment layer stratification and soil identification is low: it relies on manual drilling or single sonic detection, resulting in poor stratification accuracy and low efficiency. Soil identification depends on experience judgment and has not been integrated with silt characteristic data, which cannot meet the needs of automated operation.

[0006] Poor adaptability of load models: Traditional load models are fixed coefficient models that only consider a single factor (such as silt density), do not integrate multiple working condition parameters such as flow velocity and terrain, and do not realize dynamic updates of the model. They are difficult to adapt to the changing characteristics of deep foundation pit working conditions, which can easily lead to equipment overload damage or low operating efficiency.

[0007] Imbalance between risk and efficiency: In existing technologies, environmental risk assessment and operational efficiency analysis are carried out independently without being linked and integrated. This can easily lead to problems of prioritizing efficiency over risk or vice versa, making it impossible to balance operational economy and environmental safety.

[0008] Unreasonable trajectory planning: The standard path planning algorithm is used without considering working conditions such as silt distribution, layered operation, and soil type differences, which can easily lead to missed cleaning, repeated silt cleaning or equipment collision, resulting in high energy consumption.

[0009] Low accuracy in plan formulation: Deterministic decision-making methods based on fixed rules or single parameters do not fully consider the fuzziness and uncertainty of parameters in multiple stages, and cannot achieve deep integration of data from multiple stages, resulting in dredging plans that lack pertinence and feasibility.

[0010] Therefore, there is an urgent need for an underwater dredging optimization technology for deep foundation pit excavation that can integrate multi-source data, adapt to dynamic working conditions, balance multiple objective requirements, and accurately output optimization solutions to address the shortcomings of existing technologies. Summary of the Invention

[0011] To address the shortcomings of existing methods and the needs of practical applications, and in order to solve the aforementioned problems, this invention provides a method for optimizing underwater dredging, comprising the following steps: By integrating dredging impact parameters into a grid generation mechanism, a three-dimensional dynamic terrain grid model with importance weights is obtained. Through multi-source data fusion and spatial interpolation, a spatial distribution map of continuous silt density and moisture content across the entire field is obtained. Combining the three-dimensional dynamic terrain grid model, steady-state velocity field characteristics, turbulence intensity characteristics, and comprehensive environmental risk coefficients and operational efficiency reduction coefficients, dynamic construction window parameters are obtained through multi-objective search. Based on the spatial distribution map of silt density and moisture content, the stratification results along the profile, and preliminary soil type identification, the optimal dredging equipment movement trajectory is analyzed. Based on the dredging equipment operating status, the dynamic construction window parameters, and the optimal dredging equipment movement trajectory, an optimized underwater dredging scheme for deep foundation pit excavation is obtained through a fuzzy algorithm.

[0012] Optionally, the step of incorporating dredging impact parameters into the mesh generation mechanism to obtain a three-dimensional dynamic terrain mesh model with importance weights includes the following steps: A multi-dimensional parameter weighting system is established using the analytic hierarchy process (AHP). Based on this system, a coupled grid partitioning model is constructed, and a weight correction factor is introduced to adjust the unit size, resulting in a three-dimensional dynamic terrain grid model with importance weights.

[0013] Optionally, obtaining a spatial distribution map of silt density and water content that is continuously distributed across the entire field through multi-source data fusion and spatial interpolation includes the following steps: A Bayesian estimation method was used to fuse multi-source sample data; the variogram parameters of the Kriging interpolation were dynamically corrected using steady-state velocity field data; and a spatial distribution map of silt density and water content with continuous distribution across the entire field was obtained based on the fused sample data and the corrected variogram parameters.

[0014] Optionally, the process of obtaining dynamic construction window parameters through multi-objective search by combining a three-dimensional dynamic terrain mesh model, steady-state velocity field characteristics and turbulence intensity characteristics, as well as comprehensive environmental risk coefficients and operational efficiency reduction coefficients, includes the following steps: A multi-objective optimization model was constructed; NSGA-Ⅲ was used to solve the multi-objective optimization problem, and practical solutions that meet safety requirements were selected from the Pareto optimal solution set; based on the comprehensive optimal solution, the parameters of the three-dimensional dynamic construction window were decomposed.

[0015] Optionally, obtaining the steady-state velocity field characteristics and turbulence intensity characteristics includes the following steps: The STL method is used to separate the components of the velocity time series data, and the residual terms are filtered by adaptive threshold filtering. The steady-state velocity field is reconstructed based on the processed data, and then the turbulence intensity is calculated.

[0016] Optionally, calculating the comprehensive environmental risk coefficient and the operational efficiency reduction coefficient includes the following steps: An environmental constraint indicator system is constructed, and the risk of each individual indicator is calculated. A comprehensive environmental risk coefficient is obtained by combining the environmental constraint indicator system and the individual indicator risks. A quantitative reduction model is constructed based on the impact of environmental constraints on operational efficiency, and the operational efficiency reduction coefficient is extracted through the quantitative reduction model.

[0017] Optionally, the step of analyzing the optimal dredging equipment movement trajectory based on the spatial distribution map of silt density and moisture content, the stratification results along the profile, and the preliminary soil type identification includes the following steps: Based on the sediment layer stratification results and soil type identification data, the stratification operation sequence is determined; based on the stratification operation sequence and operation area, the optimal dredging equipment movement trajectory is analyzed using a trajectory planning algorithm.

[0018] Optionally, obtaining the stratification results and preliminary soil type identification along the profile includes the following steps: Interfacial features of sedimentary layers are extracted through spectral analysis; using these interfacial features, silt density, and water content data, an intelligent identification model is trained to achieve preliminary identification of stratification and soil types.

[0019] Optionally, the underwater dredging optimization scheme for deep foundation pit excavation, obtained through a fuzzy algorithm based on the operating status of the dredging equipment, the dynamic construction window parameters, and the optimal dredging equipment movement trajectory, includes the following steps: The input parameters are fuzzified; a fuzzy rule base is constructed, and fuzzy inference is performed by combining the fuzzification results and the fuzzy rule base; the fuzzy inference results are defuzzified to obtain an optimized underwater dredging scheme for deep foundation pit excavation.

[0020] This invention constructs a dynamic mesh using weighted integration, achieving accurate terrain representation and dynamic adaptation to working conditions. Silt distribution is obtained through Bayesian fusion and flow velocity correction interpolation, accurately reflecting the spatial heterogeneity of the silt. A dynamic construction window is searched using NSGA-III, combining multi-dimensional features to balance efficiency, risk, and cost. Optimal trajectories are planned based on layering and soil type identification to avoid omissions and collisions, improving dredging coverage and efficiency. A fuzzy algorithm integrates multi-parameter output schemes to handle parameter uncertainties, enhancing the scheme's relevance and feasibility. Overall, this invention achieves intelligent optimization of the entire dredging process, ensuring operational quality, efficiency, and safety, and adapting to complex deep foundation pit conditions.

[0021] Secondly, to efficiently execute the underwater dredging optimization method provided by this invention, this invention also provides a system for underwater dredging optimization, 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 underwater dredging optimization method as described in the first aspect of this invention. The underwater dredging optimization system of this invention has a compact structure and stable performance, and can stably execute the underwater dredging optimization method provided by this invention, further improving the overall applicability and practical application capability of this invention. Attached Figure Description

[0022] Figure 1 A flowchart of a method for optimizing underwater dredging provided in an embodiment of the present invention; Figure 2 This is a system framework diagram for underwater dredging optimization provided in an embodiment of the present invention. Detailed Implementation

[0023] 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.

[0024] 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.

[0025] Please see Figure 1 To address the above problems, this invention provides a method for optimizing underwater dredging, such as... Figure 1 As shown, in one embodiment, the method includes the following steps: S1. Integrate the dredging impact parameters into the grid division mechanism to obtain a three-dimensional dynamic terrain grid model with importance weights.

[0026] First, the collected data on topography, machinery, soil, and fluids were categorized and organized. The min-max normalization method was used to unify the data dimensions, converting parameters with different dimensions such as elevation, operating radius, cohesion, and flow velocity to the [0,1] interval to avoid distortion in subsequent weight calculations due to dimensional differences. For outlier removal, the mean μ and standard deviation σ of each data type were calculated, and the range of outliers was defined based on the 3σ criterion. If data was missing, linear interpolation was used to supplement it, taking into account the characteristics of limited water flow in deep foundation pits and strong spatial correlation of data. For areas with a large number of missing data points, the mean of nearby measuring points was used to fill in the gaps, ensuring data integrity and reliability and providing high-quality data support for subsequent model construction.

[0027] Furthermore, the process of incorporating dredging impact parameters into a mesh generation mechanism to obtain a three-dimensional dynamic terrain mesh model with importance weights includes the following steps: S11. A multi-dimensional parameter weighting system is established using the analytic hierarchy process.

[0028] First, a three-layer structure model is established, consisting of the target layer (optimizing grid division accuracy), the criterion layer (construction adaptability, correlation of silt characteristics, degree of environmental impact), and the scheme layer (mechanical operation radius, soil cohesion, water flow velocity, and topographic elevation change rate). Secondly, considering the characteristics of deep foundation pit operations in confined spaces (limited equipment operating range, prone to blind spots, and concentrated silt accumulation), senior construction and algorithm engineers were organized to construct a judgment matrix, compare the impact of each parameter on the dredging operation, and finally determine the weights. In this embodiment, the weight allocation is as follows: mechanical operating radius 0.35, soil cohesion 0.25, water flow velocity 0.2, and terrain elevation change rate 0.2.

[0029] S12. Based on the multi-dimensional parameter weighting system, construct a coupled grid partitioning model, introduce a weight correction factor to adjust the unit size, and obtain a three-dimensional dynamic terrain grid model with importance weights.

[0030] A four-dimensional coupled mesh generation model of terrain elevation, mechanical operation parameters, soil mechanical properties, and fluid velocity is constructed. Based on the finite element mesh generation method, a weight correction factor λ is introduced to realize the dynamic adjustment of the element size.

[0031] Specifically, firstly, the reference unit size is set based on the total area of ​​the foundation pit and the required operational accuracy; then, the weighting correction factor λ for each parameter is calculated. , where ωi is the weight of the i-th parameter and xi is the normalized value of the parameter; finally, adjust the cell size: actual cell size = baseline cell size × (1-λ), to ensure that the region λ value corresponding to the parameter with higher weight is larger and the actual cell size is smaller, so as to achieve grid densification in key areas and grid simplification in non-key areas, and accurately represent the terrain features of the core dredging area.

[0032] Furthermore, after completing the initial 3D terrain mesh generation based on the four-dimensional coupled mesh generation model, a comprehensive mesh quality check is required to ensure that the mesh is suitable for subsequent data processing needs such as velocity field analysis and trajectory planning. The check indicators include: 1) element distortion rate; 2) element aspect ratio; 3) element interior angle range; and 4) mesh continuity. For substandard meshes, a mesh reconstruction method is used: for locally distorted elements, the element shape is optimized by adjusting node coordinates; for continuously substandard regions, the mesh density of that region is re-divided until all meshes meet the quality standards, ensuring that the mesh can accurately support the data analysis and calculation tasks in subsequent steps.

[0033] S2. By fusing multi-source data and spatial interpolation, a spatial distribution map of silt density and moisture content with continuous distribution across the entire field is obtained.

[0034] The distribution of underwater silt in deep foundation pits exhibits significant spatial heterogeneity, making it difficult to achieve full coverage from a single data source. This study first improves the reliability of sample data through multi-source data fusion, then uses Kriging interpolation to achieve continuous distribution characterization across the entire field. Simultaneously, flow velocity field data is introduced to correct interpolation parameters, improving distribution accuracy and adapting to the uneven deposition characteristics of silt in deep foundation pits.

[0035] Specifically, obtaining a spatial distribution map of silt density and water content that is continuously distributed across the entire field through multi-source data fusion and spatial interpolation includes the following steps: S21. Use Bayesian estimation to fuse multi-source sample data.

[0036] Based on a node coordinate system (including X, Y plane coordinates and Z elevation coordinates) of a three-dimensional dynamic terrain grid, a unique grid node number is assigned to each sampling point and sensor monitoring point. A three-dimensional data association table of silt density / moisture content-spatial coordinates-collection time is established by mapping the number, ensuring that all data are accurately anchored to the corresponding grid node, providing a sample basis with location information for subsequent spatial interpolation, and avoiding interpolation distortion caused by the disconnect between data and spatial location.

[0037] Based on this, a Bayesian estimation method is adopted to achieve the complementary advantages of dual data sources. The high precision of the sampled data is used to correct the systematic error of the sensor data, while the high coverage of the sensor data is used to make up for the spatial gap of the sampled data.

[0038] Specifically, the first step is preprocessing of prior and likelihood information: Field sampling data is used as prior information, and a prior distribution is set based on the accuracy of the sampling operation; sensor monitoring data is used as likelihood information, and a likelihood distribution is set by combining the sensor's factory accuracy and field calibration error. The second step is constructing the fusion model: Based on Bayes' theorem P(θ|D)=P(D|θ)P(θ) / P(D) (where θ is the true value of the fused data, D is the sensor monitoring data, P(θ) is the prior probability, P(D|θ) is the likelihood probability, and P(θ|D) is the posterior probability), a dual-source fusion model is constructed, and the posterior probability distribution of each grid node sample point is calculated through the model. The third step is solving for the fused data: The maximum likelihood estimation method is used to maximize the posterior probability to obtain the true value of the fused data for each sample point, satisfying: Fusion data = ( ×Sampling data+ × Sensor data) / ( + ), The overall error variance of the sensor. The sampling error variance is weighted according to the error variance of the two types of data. The smaller the error, the greater the weight. This fully combines the high precision advantage of the sampling data with the continuous coverage advantage of the sensor data, thereby significantly improving the reliability and spatial coverage of the sample data.

[0039] S22. Dynamically correct the variogram parameters of Kriging interpolation using steady-state velocity field data.

[0040] Based on steady-state velocity field data, the variogram parameters of Kriging interpolation are dynamically corrected to establish a linkage relationship between velocity and silt spatial correlation.

[0041] Specifically, the first step is the selection of the basic model for the variogram: considering the depositional characteristics of deep foundation pit silt (spatial correlation decreases exponentially with increasing distance), an exponential variogram is selected as the basic model, satisfying the following: (where γ(h) is the variogram value, h is the distance between sample points, C0 is the nugget value, C is the arch height, C0+C is the sill value, and a is the range). The sill value reflects the maximum spatial variability of silt characteristics and is the core parameter that determines the interpolation accuracy.

[0042] The second step is the flow velocity correlation correction logic: The higher the flow velocity within the deep foundation pit, the stronger the disturbance of the silt, the more uneven the spatial distribution of silt particles, and the weaker the spatial correlation between sample points. Therefore, the corresponding variogram sill value should be larger. The sill value correction formula is set as follows: , where v is the steady-state flow velocity value of the current grid node, v_max is the maximum steady-state flow velocity value monitored in the pit, and 10% is the flow velocity influence coefficient.

[0043] The third step is to verify the corrected parameters: select two regions with significant differences in flow velocity, and conduct interpolation experiments using the variogram parameters before and after correction. Compare the errors between the interpolation results and the measured data to verify the rationality of the correction logic.

[0044] S23. Based on the fused sample data and the corrected variogram parameters, a spatial distribution map of silt density and water content with continuous distribution across the entire field is obtained.

[0045] Based on the fused sample data and the corrected variogram parameters, the ordinary kriging interpolation method is used to achieve a continuous distribution characterization of silt density and water content across the entire site, adapting to the complex spatial morphology of deep foundation pits.

[0046] First, the data is grouped before interpolation: the fused sample data is grouped according to the deep pit partitions (such as the pit center area, edge area, and corner area), and a different interpolation radius is used in each group (the flow velocity is uniform in the center area, so the interpolation radius is set to 3m; the terrain is complex in the edge area, so the interpolation radius is set to 1.5m) to avoid interpolation distortion caused by regional characteristics differences.

[0047] Secondly, semi-variogram fitting: For each set of sample data, substitute the corrected sill value, range and other parameters, and use the weighted least squares method to fit the semi-variogram curve to ensure that the fitted curve can accurately reflect the spatial correlation of the sample points.

[0048] Then, the interpolation calculation is performed: using the nodes of the 3D dynamic terrain grid as interpolation points, for each interpolation point, all sample points within its interpolation radius are searched. The weighting coefficients of the sample points are calculated based on the fitted semi-variogram (the closer the spatial distance and the stronger the correlation, the greater the weight). The silt density and water content values ​​of the interpolation point are obtained by weighted summation, and the weighting coefficients satisfy the following conditions: ( (where is the weight of the i-th sample point), ensuring the unbiasedness of the interpolation result.

[0049] Finally, special treatment for edge areas: The edge areas of deep foundation pit corners and slopes have dense grid nodes and few sample points. A combination of local sample densification and data extrapolation from neighboring areas is adopted. First, 3-5 temporary sampling points are added in the edge area, and then extrapolation is performed by combining the interpolation results of neighboring areas to avoid interpolation gaps or distortions in the edge area.

[0050] This invention combines Bayesian fusion with Kriging interpolation. First, it improves the quality of multi-source data through fusion, and then introduces velocity field data to dynamically correct the variogram parameters of the interpolation. This deeply couples the velocity characteristics with the current distribution, enabling accurate full-field characterization of the density and water content of underwater silt in deep foundation pits. This provides core silt characteristic data support for subsequent sediment layer stratification and equipment trajectory planning, ensuring the pertinence and accuracy of subsequent steps.

[0051] S3. By combining a three-dimensional dynamic terrain mesh model, steady-state velocity field characteristics and turbulence intensity characteristics, as well as comprehensive environmental risk coefficient and work efficiency reduction coefficient, dynamic construction window parameters are obtained through multi-objective search.

[0052] Underwater flow velocity in deep foundation pits exhibits significant seasonal fluctuations (such as differences in water levels between the rainy and dry seasons) and random noise (such as equipment operation disturbances and water flow eddies). A combination of seasonal decomposition and adaptive threshold filtering based on standard deviation is adopted. First, the trend term, seasonal term, and residual term (including noise) of the flow velocity data are separated. Then, an adaptive threshold filter based on standard deviation is designed for the residual term to achieve accurate noise removal and extract the steady-state flow velocity field and turbulence intensity characteristics that reflect the actual working conditions.

[0053] Specifically, obtaining the steady-state velocity field characteristics and turbulence intensity characteristics includes the following steps: S311. The STL method is used to separate the components of the flow velocity time series data, and the residual terms are filtered by adaptive threshold filtering.

[0054] First, the core parameters are determined. Based on historical seasonal precipitation data for the area where the deep foundation pit is located, the seasonal cycle is determined through trend fitting analysis. If the area experiences concentrated rainfall during the rainy season and its impact on the foundation pit water level is significant (water level fluctuation ≥ 0.5m), the rainy season cycle is set to 30 days. Conversely, if rainfall is scarce during the dry season and water level fluctuations are stable (fluctuation ≤ 0.2m), the seasonal cycle is set to 60 days, ensuring that the cycle division closely matches the actual hydrological characteristics. The trend term smoothing parameter is set to 0.1. This parameter is based on the characteristic of frequent short-term fluctuations in flow velocity in deep foundation pits. Conventional hydraulic engineering smoothing parameters are mostly 0.3-0.5. Due to the confined space in deep foundation pits, water flow is easily affected by equipment operation and excavation disturbances, resulting in short-term fluctuations. Reducing the smoothing parameter better preserves the short-term variation patterns in the trend term.

[0055] The decomposition process is then executed: first, the trend term is fitted using Loess local weighted regression; then, the trend term is removed from the original data to obtain a residual sequence containing seasonal fluctuations and noise; finally, the seasonal term is obtained by periodic extraction from the residual sequence, and the remaining part is the residual term containing random noise and effective turbulence signal. After decomposition, the rationality of the components needs to be verified: the trend term should reflect the overall change in flow velocity within the pit (such as the gradual change in flow velocity with increasing excavation depth), and the seasonal term should be consistent with historical precipitation cycles to ensure the validity of the decomposition results.

[0056] Furthermore, the statistical characteristics of the residuals are calculated: the standard deviation σ of the residuals is calculated using a sliding window method (window size set to 20 sampling points). The sliding window can dynamically capture the local fluctuation characteristics of the residuals, avoiding threshold distortion caused by overall statistics. Then, an adaptive threshold is set: threshold... Where k is a dynamic correction coefficient, which is precisely adjusted according to the size of the deep foundation pit. Based on engineering practice experience, foundation pits with an area <1000㎡ are defined as narrow foundation pits (limited space, small turbulent signal amplitude, easily misjudged as noise), and k is set to 1.2; foundation pits with an area >5000㎡ are defined as open foundation pits (larger space, larger turbulent signal amplitude, the threshold can be appropriately increased to filter noise), and k is set to 1.5; for foundation pits with an area between 1000-5000㎡, the k value is determined by linear interpolation (e.g., for an area of ​​3000㎡, k=1.3). Finally, threshold verification is performed: flow velocity data during periods when no equipment is operating in the deep foundation pit are selected to test the signal retention rate under different k values, avoiding the loss of effective signals or noise residue due to improper threshold setting.

[0057] Furthermore, precise noise reduction is performed based on a set adaptive threshold to effectively separate noise from turbulent signals. First, point-by-point judgment is performed: for each data point in the residual term, its absolute value |x| is calculated. If |x| > T, it is determined to be random noise, and the average of the two nearest valid data points is used to replace the noise point; if |x| ≤ T, it is determined to be a valid turbulent signal, and the original data is retained. Second, signal smoothing is performed: the filtered data may contain local abrupt changes. A 5-point moving average method is used to smooth the filtered residual term, further eliminating residual minor noise while retaining the core characteristics of the turbulent signal (such as the peak value and period of turbulent fluctuations). Finally, the separation effect is verified: the correlation between the filtered residual term and the original residual term is calculated (correlation coefficient ≥ 0.8), and the filtering results are compared with those during periods without equipment operation to ensure that the core characteristics of the turbulent signal, such as amplitude and period, are not distorted.

[0058] S312. Reconstruct the steady-state velocity field based on the processed data, and then calculate the turbulence intensity.

[0059] By integrating trend terms and effective turbulence signals, and combining them with spatial coordinates, a precise three-dimensional velocity field is constructed.

[0060] First, component superposition is performed: extract the trend term velocity value and the turbulence signal value of the filtered residual term for each grid node, and calculate the steady-state velocity value of the node using the algebraic superposition method (steady-state velocity value = trend term velocity value + turbulence signal value). When superimposing, it is necessary to ensure that the time dimension of the two is consistent.

[0061] Next, spatial interpolation is performed: Due to the limited density of sensor deployment, some grid nodes do not have directly corresponding flow velocity data. Inverse distance weighted interpolation (IDW) is used to complete the data of the entire field. Based on the data of the 8 effective sensors closest to the target node, the weights are assigned inversely proportional to the square of the distance. The closer the distance, the greater the weight. The steady-state flow velocity value of the target node is then calculated.

[0062] Finally, a three-dimensional velocity field is constructed: the steady-state velocity values ​​of all grid nodes are bound to the corresponding spatial coordinates (X, Y, Z), and a dynamic velocity field model is constructed using three-dimensional visualization technology to clearly present the velocity differences in different areas within the foundation pit, providing accurate spatial velocity data for subsequent sludge diffusion prediction and equipment operation stability analysis.

[0063] Furthermore, based on the reconstructed steady-state velocity field and the filtered turbulence signal, the turbulence intensity at each grid node is quantitatively calculated. The turbulence intensity I satisfies: ,in Let U be the turbulent fluctuating velocity, and U be the average flow velocity. Calculation When calculating U, the residual data after continuous filtering of each grid node is selected, and its root mean square value is calculated as the turbulent pulsating velocity; when calculating U, the steady-state velocity value of the same grid node is selected, and the average velocity is calculated by the arithmetic mean method.

[0064] Then, the calculation results are corrected: considering that the space in the deep foundation pit is narrow and the water flow is prone to local eddies, the calculated turbulence intensity is corrected by boundary correction. The turbulence intensity of the nodes within 5 meters of the foundation pit boundary is adjusted by a correction factor of 0.9 (the water flow at the boundary is constrained, and the actual value of the turbulence intensity is slightly lower than the calculated value); the calculated value of the nodes in the central area of ​​the foundation pit remains unchanged.

[0065] Finally, the rationality of the results was verified: referring to the normal turbulence intensity range (0.1-0.3) for underwater operations in deep foundation pits, outliers exceeding this range were eliminated, and abnormal nodes were recalculated to ensure that the turbulence intensity values ​​of all nodes conform to the actual working conditions, providing accurate turbulence characteristic data for subsequent environmental risk assessment and equipment load modeling.

[0066] This invention combines seasonal decomposition with adaptive threshold filtering based on standard deviation, and then dynamically adjusts the threshold coefficient according to the size of the foundation pit space. It fully adapts to the flow velocity characteristics of the narrow space of deep foundation pits, effectively separates seasonal fluctuations, random noise and effective turbulence signals in the flow velocity data, and avoids the problems of turbulence signal loss or noise residue caused by traditional filtering methods. It provides accurate flow velocity field basic data for subsequent sludge diffusion prediction and equipment operation stability assessment, and is suitable for the special environment of water flow restriction in deep foundation pits.

[0067] Deep foundation pit dredging operations need to balance environmental safety and operational efficiency. By integrating multiple environmental constraints and risk-efficiency linkage analysis, a comprehensive environmental risk coefficient is calculated by incorporating various environmental constraints such as water quality, settlement, and hydrology. Combined with equipment operating status and silt characteristic data, an efficiency reduction model is constructed to obtain the operational efficiency reduction coefficient, thereby realizing the linkage representation of risk and efficiency and adapting to the sensitive characteristics of deep foundation pit operations.

[0068] Specifically, calculating the comprehensive environmental risk coefficient and the operational efficiency reduction coefficient includes the following steps: S321. Construct an environmental constraint indicator system and calculate the risk of each indicator.

[0069] First, indicator screening was conducted: Considering the environmental impact mechanisms of underwater dredging in deep foundation pits (dredging disturbances can easily lead to sludge diffusion and water pollution; operational vibrations and water level changes may induce settlement of surrounding buildings; excessive water level fluctuations affect operational stability), a combination of correlation analysis and expert scoring was used to screen indicators. For correlation analysis, the correlation coefficients between candidate indicators (suspended solids concentration, COD, ammonia nitrogen, total phosphorus, building settlement, settlement rate, water level fluctuation amplitude, and water flow velocity) and the environmental risks of dredging operations were calculated, retaining indicators with a correlation coefficient ≥ 0.6. For expert scoring, environmental engineering experts and deep foundation pit construction experts were organized to score the importance, monitorability, and correlation of candidate indicators (out of 10 points), retaining indicators with an average score ≥ 7 points. In the embodiments, three main categories of core indicators were ultimately identified: water quality indicators (suspended solids concentration, COD) – suspended solids concentration directly reflects the degree of sludge diffusion, and COD characterizes the level of organic pollution in the water body. Both are water quality parameters that are most likely to exceed standards and have the widest impact during dredging operations; settlement indicators (maximum settlement of buildings) – deep foundation pits are mostly surrounded by existing buildings, and excessive settlement may lead to structural damage, making it a core indicator for safety risk management; and hydrological indicators (water level fluctuation range) – water level fluctuations directly affect the accuracy of the dredging equipment's operating depth, and large fluctuations may exacerbate the risk of slope instability.

[0070] Secondly, the entropy weight method is used to determine the indicator weights, objectively assigning weights based on the dispersion of the indicator data to avoid subjective experience bias. The specific implementation process is as follows: First, data standardization is performed, transforming all indicator data to the [0,1] interval through normalization; second, the information entropy e_j is calculated; the smaller the information entropy, the greater the dispersion of the indicator, and the higher the weight should be; third, the weights are determined. In this embodiment, based on typical engineering sample data of deep foundation pit dredging, the following weights were calculated: water quality index weight 0.4, settlement index weight 0.35, and hydrological index weight 0.25, ensuring that the weight allocation conforms to the actual influence of the indicators and the data dispersion characteristics.

[0071] Calculating single-indicator risk refers to quantifying the association between monitoring data and risk levels using membership functions. First, the constraint standards for each indicator are determined, strictly adhering to relevant national / industry standards. For example, the suspended solids concentration standard adopts the Class IV water standard in the "Surface Water Environmental Quality Standard" (GB3838-2002), and the COD standard also adopts the Class IV water standard. Second, a suitable membership function is selected. Considering that the risk of environmental indicators changes non-linearly with increasing monitoring values, a trapezoidal membership function is used to calculate the risk value of a single indicator. The function expression is divided into three segments: when the monitoring value X ≤ the standard value X0, the risk value μ(x) = 0 (no risk); when X0 < X ​​≤ X1 (exceeding warning value, X1 = 1.2X0), μ(x) = (X - X0) / (X1 - X0), and the risk increases linearly; when X > X1, μ(x) = 1 (high risk).

[0072] S322. By combining the environmental constraint index system and the single index risk, a comprehensive environmental risk coefficient is obtained.

[0073] First, a basic weighted summation calculation is performed, and the initial value of the comprehensive environmental risk coefficient R0 is calculated as follows: R0 = Σ(w_j × μ_j) (where w_j is the weight of the j-th indicator and μ_j is the risk value of the j-th indicator).

[0074] Secondly, turbulence intensity is introduced for risk correction. The greater the turbulence intensity, the stronger the disturbance of the silt by the water flow, the wider the range of suspended solids diffusion, and the higher the risk of water pollution. At the same time, turbulence may aggravate water level fluctuations, indirectly increasing the risk of settlement. A correction coefficient K1 = 1 + (I / I_max) is set, where I is the turbulence intensity of the current working area, and I_max is the maximum turbulence intensity monitored in the foundation pit. This coefficient is calibrated based on the silt diffusion test data of deep foundation pits. The corrected comprehensive environmental risk coefficient R = R0 × K1.

[0075] Finally, risk level thresholds were set: R≤0.3 indicates low risk (normal operation is possible), 0.3<R≤0.6 indicates medium risk (operation parameters need to be adjusted), and R>0.6 indicates high risk (operation needs to be suspended and control measures need to be taken), providing a clear risk assessment basis for subsequent optimization decisions.

[0076] S323. Construct a quantitative reduction model based on the influence of environmental constraints on work efficiency, and extract the work efficiency reduction coefficient through the quantitative reduction model.

[0077] Based on the impact of environmental constraints on operational efficiency, a quantitative reduction model is constructed to achieve a linkage between efficiency and risk, avoiding the limitation of traditional efficiency analysis that ignores environmental constraints.

[0078] First, an analysis of the impact patterns was conducted. In the example, through field tests and historical data statistics, the specific impacts of exceeding environmental standards on operational efficiency were clarified: 1. Exceeding the standard for suspended solids concentration necessitates reducing equipment operating speed (to avoid exacerbating disturbances). When the concentration exceeds the standard by 20%, the operating speed decreases by 15%, and efficiency decreases by 15% accordingly; when it exceeds the standard by 50%, the operating speed decreases by 40%, and efficiency decreases by 40%. 2. Exceeding the standard for building settlement necessitates suspending operations in high-risk areas (such as within 10m of buildings). Reducing the operating area will lead to a decrease in efficiency. When the settlement exceeds the standard by 20%, the operating area is reduced by 10%, and efficiency decreases by 10%; when it exceeds the standard by 50%, the operating area is reduced by 30%, and efficiency decreases by 30%. 3. Exceeding the standard for water level fluctuation requires frequent adjustments to the equipment operating depth, resulting in a reduction in effective operating time. When the fluctuation exceeds the standard by 50%, the effective operating time decreases by 20%, and efficiency decreases by 20%; when the fluctuation exceeds the standard by 100%, the effective operating time decreases by 50%, and efficiency decreases by 50%.

[0079] Secondly, an efficiency reduction coefficient model is constructed. The reduction coefficient η represents the ratio of the actual operating efficiency to the rated efficiency under environmental constraints, satisfying: η=1-Σ(w_j×r_j), where r_j is the efficiency reduction rate of the j-th indicator, and w_j is the indicator weight (consistent with the weight of the environmental risk coefficient to ensure that the weights of risk and efficiency are matched).

[0080] Furthermore, to improve the accuracy of the model, a soil type correction factor f is introduced: f=1.0 for silty clay region (high cohesion, efficiency reduction consistent with model calculation), f=0.9 for silty clay region (medium cohesion, slightly lower efficiency reduction), and f=1.1 for sandy soil region (low cohesion, easy diffusion, slightly higher efficiency reduction). The final corrected efficiency reduction coefficient η'=η×f ensures that the model is adapted to the dredging characteristics of different soil types.

[0081] The core objectives of deep foundation pit dredging are high efficiency, low risk, and low cost. A Pareto frontier search and risk-constrained decision-making strategy is adopted, with operational efficiency, environmental risk, and construction cost as optimization objectives. By combining data such as three-dimensional topography, flow velocity field, risk and efficiency coefficients, the Pareto optimal solution set is searched. Then, dynamic construction window parameters (operation time, operation area, equipment configuration) are obtained based on risk constraints to adapt to the multi-objective balance operation requirements of deep foundation pits.

[0082] Specifically, the process of obtaining dynamic construction window parameters through multi-objective search by combining a three-dimensional dynamic terrain mesh model, steady-state velocity field characteristics and turbulence intensity characteristics, as well as comprehensive environmental risk coefficients and operational efficiency reduction coefficients, includes the following steps: S331. Construct a multi-objective optimization model.

[0083] Specifically, the optimization objectives include: maximizing operational efficiency, where operational efficiency E satisfies: ,in This is a reduction factor for work efficiency. The rated operating efficiency of the equipment is given, and T represents the actual operating time. The calculation must be synchronized with the equipment's operating status parameters. If the equipment is in its optimal load range... It can be corrected by 1.05 times (characterizing the efficiency improvement when the load is adapted). Minimize environmental risk, with the comprehensive environmental risk coefficient R as the core indicator, and the objective is min(R); To minimize construction costs, a full-process cost accounting model is constructed, satisfying the following: total cost ,in C_F represents the comprehensive cost per unit time and the cost per unit load, which is related to the characteristics of the sludge. In this example, the sludge density is ≥1.8 g / cm³. 3 At that time, C_F = 15 yuan / t, density < 1.5 g / cm³3 At that time, C_F = 8 yuan / t, and F_avg is the average load of the equipment during the operation.

[0084] Secondly, a multi-dimensional constraint system is defined. In addition to the original constraints, key working condition constraints are added: equipment operating depth constraints, such as the operating depth being less than or equal to 90% of the current silt layer thickness and greater than or equal to 0.5m above the base, based on the sediment layer stratification results; dredging quality constraints, such as the residual silt thickness being less than 5cm; and equipment operating radius constraints. The constraints are integrated in the form of a set of inequalities to ensure the feasibility of the model solution.

[0085] S332. Use NSGA-Ⅲ to perform multi-objective optimization and select practical solutions that meet safety requirements from the Pareto optimal solution set.

[0086] First, parameter calibration was completed: considering the number of working condition combinations for deep foundation pit dredging, the population size was set to 100, the number of iterations to 50, the crossover probability to 0.8, and the mutation probability to 0.1. Second, population initialization and constraint integration were optimized: real-number encoding was used, with each individual in the population corresponding to a set of working time-working area-equipment load parameter combinations. Then, the core search process was executed: the initial population was first non-dominated sorted, and the crowding distance of each individual was calculated (representing the distribution density of the individual in the solution set; the larger the crowding distance, the better the diversity of solutions). Based on the sorting results and crowding distance, parent individuals were selected using roulette wheel selection, and offspring individuals were generated using simulated binary crossover (SBX) and polynomial mutation. Constraint verification was performed on the offspring individuals, eliminating those that violated the working condition constraints, merging the qualified individuals with their parents, and repeating the sorting and selection process to finally obtain the Pareto optimal solution set.

[0087] Furthermore, practical solutions that meet safety requirements are selected from the Pareto optimal solution set and sorted using a multi-attribute decision method to ensure that the final solution balances efficiency and cost.

[0088] First, risk constraint screening is performed: extract the comprehensive environmental risk coefficient R corresponding to each optimal solution group, and strictly screen solutions with R≤0.3 (low risk threshold); if the number of solutions after screening is <5 groups (insufficient diversity of solutions), the threshold can be temporarily relaxed to R≤0.4 (lower limit of medium risk), and the risk level of the relaxed solutions is marked simultaneously. Risk control measures need to be added later; if the number of solutions after screening is >15 groups (excessive computation), cluster analysis (K-means algorithm, K=5) is used to merge similar solutions and retain representative solutions.

[0089] Secondly, the Top-Approximation-Ideal-Solution (TOPSIS) method is used for multi-attribute ranking: a decision matrix is ​​constructed, with the selected solutions as rows and operational efficiency E, environmental risk R, and construction cost C as columns, inputting the quantitative index values ​​of each solution; the decision matrix is ​​standardized; positive and negative ideal solutions are determined: the positive ideal solution is the virtual solution with the largest E, smallest R, and smallest C, and the negative ideal solution is the virtual solution with the smallest E, largest R, and largest C; the Euclidean distance between each solution and the positive and negative ideal solutions is calculated, and the relative proximity is solved, satisfying: , The distance between the solution and the ideal solution. To determine the distance from the negative ideal solution, sort the solutions in descending order of relative proximity and select the solution with the highest proximity as the overall optimal solution. If there are two solutions with a proximity difference ≤ 0.02, then the solution with smaller R is preferred.

[0090] S333. Based on the comprehensive optimal solution, decompose the parameters of the three-dimensional dynamic construction window.

[0091] First, the operation time window is generated: combining the seasonal variation of the flow field and environmental risk data, the core operation period and avoidance period are divided. For example, the period with the highest operation efficiency corresponding to the optimal solution is the non-flood season (September to May of the following year), which is further refined to 6:00-18:00 every day (during this period, the water level is stable, the turbulence intensity I≤0.15, and the environmental risk is the lowest); the avoidance period is the flood season (June to August) and 0:00-4:00 every day (the water level fluctuation range ≥0.3m, and the operation accuracy decreases); at the same time, the dynamic adjustment conditions of the time window are marked: if the water level fluctuation range >0.3m is detected in real time, the time window is immediately triggered to shrink, and the operation is suspended for 2 hours until the water level stabilizes.

[0092] Next, the operation area window is generated: based on the 3D terrain grid and silt density distribution, the grid coordinate range of the core operation area is defined (e.g., X∈[10m,50m], Y∈[20m,60m], Z∈[0m,3m]), and sub-regions are divided according to silt density: high-priority sub-regions (density ≥1.8g / cm³). 3 The grid cell size is 0.3m × 0.3m, requiring priority operation; the medium priority sub-region (1.5 ≤ density < 1.8g / cm³) 3 (Unit size 0.5m × 0.5m), low-priority sub-regions (density < 1.5g / cm³) 3 (Unit size 1m×1m); synchronously marked area avoidance range: the corner slope area of ​​the foundation pit (slope > 15°), the area within 10m of the adjacent building (high risk of settlement), a low-speed operation strategy should be adopted.

[0093] Finally, an equipment configuration window is generated: combining soil type identification results and equipment operating status data, the equipment type and operating parameters for each sub-region are clarified; for example, the high-priority sub-region (silty clay, density 1.8-2.0 g / cm³) 3 Equipped with a grab bucket dredging machine, with an optimal load of 3.5t, a rotation speed of 1232r / min, and an operating depth of 1.2m; medium priority sub-area (silty clay, density 1.5-1.8g / cm³). 3 Equipped with a suction-type sludge pump, with an optimal power of 25kW, a speed of 1300r / min, and an operating depth of 1.0m; low-priority sub-area (silty soil, density 1.2-1.5g / cm³). 3 Configure a small vacuum cleaner with an optimal power of 10kW, a speed of 1100r / min, and an operating depth of 0.8m; at the same time, specify the equipment quantity configuration: configure 2 main machines + 1 backup machine during the core operation period, and configure 1 main machine during non-core operation period to ensure a balance between efficiency and cost.

[0094] The NSGA-Ⅲ algorithm is used to search the Pareto front. Data such as 3D terrain and velocity field are incorporated into the algorithm as constraints. The algorithm is combined with risk constraints and the ranking method of approximating the ideal solution to select the optimal solution. This realizes the dynamic and multi-objective optimization of the construction window, which is conducive to the balance optimization of multiple objectives in deep foundation pit dredging. The dynamic construction window parameters are adapted to the real-time changes of foundation pit conditions, providing a scientific decision-making basis for subsequent equipment trajectory planning and operation implementation, and ensuring the scientific and efficient nature of dredging operations.

[0095] S4. Based on the spatial distribution map of silt density and moisture content, the stratification results along the profile, and the preliminary soil type identification, analyze the optimal dredging equipment movement trajectory.

[0096] Due to factors such as excavation disturbance and water deposition, the sedimentary layers in deep foundation pits have blurred stratification interfaces, making soil type identification difficult. By extracting the interface features of the sedimentary layers through spectral analysis (the acoustic reflection spectra of different soil layers differ), and combining this with silt density and water content data, an intelligent identification model is trained to achieve preliminary identification of stratification and soil type. This overcomes the limitations of traditional manual drilling for stratification and is adapted to the complex sedimentary layer characteristics of deep foundation pits.

[0097] Specifically, obtaining the stratification results and preliminary soil type identification along the profile includes the following steps: S411. Extract the interface features of the sediment layer through spectral analysis.

[0098] First, a spectrum analysis is performed. The acoustic signal is segmented according to the detection depth, and the spectrum is obtained by performing an FFT transform on the signal of each depth unit. The cosine similarity algorithm is used to calculate the spectrum similarity between adjacent depth units. The closer the similarity is to 1, the more similar the soil characteristics of the two depth units are. Conversely, there may be an interface.

[0099] Next, candidate points for the interface are screened, and a similarity mutation threshold is set. Based on the pilot data of deep foundation pit geological exploration, it is usually 0.5. That is, the similarity between adjacent depth units drops sharply from greater than 0.8 to less than 0.3, and the mutation amplitude is greater than 0.5. Depth points with similarity below this threshold are candidate points for the interface. At the same time, the depth coordinates of the candidate points are recorded to form a preliminary set of candidate points.

[0100] The results were then corrected: density data at the corresponding depth were extracted from the spatial distribution map of silt density, and a density-depth curve was plotted. If the density abrupt change point matched the depth of the interface candidate point with a depth greater than 80%, it was determined to be a real interface; if there was no obvious density abrupt change at the candidate point, it was determined to be a false candidate point and was removed.

[0101] Finally, interface smoothing is performed: the actual interface of the same profile should be a continuous curve. If there are individual interface points with large depth deviations, cubic spline interpolation is used to correct the depth of the point to ensure that the interface curve is continuous and smooth, and to avoid local abnormal points affecting the overall layering effect.

[0102] Furthermore, based on the multi-dimensional characteristics of known soil types, a standardized feature library is constructed to provide reliable training and comparison for intelligent identification, ensuring the integrity and representativeness of the features. First, data collection is conducted: based on a detailed geological survey report of the area where the foundation pit is located, typical soil samples are selected (common soil types for deep foundation pits include silty clay, silty clay, silt, sand, and residual soil); each sample group contains two parts of data: one is the time-domain and frequency-domain characteristics obtained from acoustic detection, and the other is the silt density and moisture content data measured in the field. Second, feature standardization is performed: the feature dimensions of different soil types vary significantly (e.g., the spectral energy range is 10). 2 -10 4 Its density ranges from 1.2 to 2.0 g / cm³. 3 All features were processed using Z-score normalization, converting them into standard normal distribution data with a mean of 0 and a variance of 1 to avoid the impact of dimensional differences on model training accuracy. Then, a feature library structure was constructed: the feature library was organized in tabular form, with fields including sample number, soil type name (label), time-domain features (peak value, amplitude variance, etc., 4 items), frequency-domain features (dominant frequency, spectral energy, etc., 3 items), density, moisture content, sampling depth, sampling time, and sample reliability level (Class A: laboratory-calibrated samples; Class B: field-precise sampling samples; Class C: routine monitoring samples). Finally, a feature library update mechanism was established: subsequent additions of soil samples or supplementary detection data are preprocessed according to the same standards and then included in the feature library to ensure that the feature library adapts to changes in soil characteristics in the area where the foundation pit is located.

[0103] S412. Using the interface features, silt density and moisture content data, train an intelligent recognition model to achieve preliminary identification of stratification and soil type.

[0104] A multi-feature fusion intelligent recognition model is built based on support vector machine (SVM). The model's recognition accuracy is improved through parameter optimization, which can meet the recognition needs of complex soil types in deep foundation pits.

[0105] First, the training and test sets are divided: 70% of the samples are randomly selected from the feature library as the training set and 30% as the test set. When dividing, ensure that the proportion of samples of each soil type in the training and test sets is consistent to avoid model bias caused by uneven sample distribution.

[0106] Next, the basic parameters of the model were set: Considering the multi-classification requirements for soil type identification, a one-to-one multi-class SVM model was adopted, with the radial basis function chosen as the kernel function. The initial penalty coefficient C was set to 10, and the initial gamma value was set to 0.1. Then, the model parameters were optimized: a grid search method combined with 10-fold cross-validation was used to optimize the parameters. The parameter search range was C∈[1,100], gamma∈[0.01,10]. Each search selected 20 parameter combinations, and the average accuracy of each parameter combination was calculated using 10-fold cross-validation. The parameter combination with the highest accuracy was selected as the optimal parameters (e.g., after optimization, C=15, gamma=0.05).

[0107] Finally, the interface recognition results and intelligent model output were integrated to complete the sedimentary layer stratification and soil type identification. The results were then verified and corrected using geological survey data to ensure reliability. First, profile stratification was performed: for each exploration profile (arranged along the grid lines, with a profile spacing ≤2m to ensure full field coverage), sedimentary layers were divided according to the interface recognition results, and the top and bottom depths of each layer were determined (e.g., the first layer has a top depth of 0m and a bottom depth of 1.2m, the second layer has a top depth of 1.2m and a bottom depth of 3.5m). The mean feature value was extracted for each layer (taking the average feature value of all depth units within the layer to reduce the influence of local anomalies). Simultaneously, the mean values ​​of silt density and water content within the corresponding depth range were extracted to form a comprehensive feature vector for each layer. Next, preliminary soil type identification is performed: the comprehensive feature vector of each layer is input into the trained intelligent recognition model to obtain the preliminary soil type identification for each layer (e.g., the first layer is identified as silty clay, and the second layer as silty clay); the recognition confidence level of the model output is recorded (0-1 range, the higher the confidence level, the more reliable the identification result); soil layers with a confidence level <0.7 are marked as pending verification, and then the results are verified and corrected: borehole data with known soil type distribution in the geological survey report are extracted, the layering at the borehole is compared with the soil type identification results, and the matching accuracy is calculated; for mismatched areas (e.g., the model identifies it as sandy soil, but the survey report identifies it as silty soil), the reasons are analyzed (e.g., feature extraction deviation, insufficient model training samples), the acoustic features and silt characteristics data of the area are re-extracted, and the soil type identification is manually corrected; for the soil layers to be verified, the soil type is determined by on-site supplementary sampling (borehole core sampling or grab sampling), and the identification results are corrected. Finally, the output results are as follows: stratification and soil type identification maps of each detection profile are drawn, and the top and bottom depths, soil type names, and identification confidence levels of each layer are marked; based on the profile results, spatial interpolation is used to complete the distribution of sedimentary layers across the entire field, generating a three-dimensional sedimentary layer and soil type distribution map, clearly presenting the spatial distribution pattern of sedimentary layers in the deep foundation pit, providing a core basis for subsequent equipment selection and adjustment of operating parameters.

[0108] Furthermore, the step of analyzing the optimal dredging equipment movement trajectory based on the spatial distribution map of silt density and moisture content, the stratification results along the profile, and the preliminary soil type identification includes the following steps: S421. Determine the stratification sequence based on the sedimentary layer stratification results and soil type identification data.

[0109] First, based on the spatial distribution data of silt density, the K-means clustering algorithm was used to divide the regions by priority, with the cluster centers set at 1.8 g / cm³. 3 (High priority), 1.6g / cm 3 (Medium priority), 1.3g / cm 3 (Low priority), the final classification standard is: high priority area (sludge density ≥ 1.8 g / cm³) 3The dredging is difficult and energy-intensive, requiring priority to avoid sediment hardening; and the medium-priority areas (1.5 ≤ silt density < 1.8 g / cm³) require dredging. 3 The difficulty level is similar to conventional dredging, and the work should be carried out in the conventional order. Low-priority areas (sludge density < 1.5 g / cm³) are also included. 3 (The dredging is easy and can be done later). After dividing the area, the number and distribution range of grid cells in each priority area should be marked. The principle of working order should be established from high to medium to low, and from the center to the edge for the same priority, so as to avoid efficiency loss caused by frequent cross-regional movement.

[0110] Furthermore, the thickness standards for layered dredging are clarified: based on the operating depth range of the dredging equipment, the sediment layer is divided into several working layers at 0.5m / layer. The core sequence of dredging is to proceed from top to bottom and layer by layer. After each layer is dredged, ultrasonic thickness monitoring is required to confirm that the residual silt thickness is ≤5cm (meeting the dredging quality standards) before proceeding to the next layer. For surface or bottom sediment layers with a thickness <0.5m, the operating parameters are adjusted according to the actual thickness to avoid over-excavation and damage to the base bearing layer.

[0111] Secondly, soil type-appropriate operation strategies were formulated: Based on the soil type identification results, operation speed and dredging density were set differently: For silty clay areas (high cohesion, easy to clump, low risk of diffusion), a low-speed dense dredging strategy was adopted, with the operation speed controlled at 0.8 m / s (based on equipment movement speed limit calibration), and the dredging trajectory overlap rate was set at 20% (to ensure no omissions), avoiding sludge splashing caused by high-speed operation; For silty clay areas (medium cohesion, with both fluidity and clumping properties), a medium-speed uniform dredging strategy was adopted, with an operation speed of 1.2 m / s and a trajectory overlap rate of 15%, balancing efficiency and dredging quality; For sandy soil areas (low cohesion, high fluidity, easy to diffuse pollution), a high-speed dredging strategy was adopted, with an operation speed of 1.5 m / s and a trajectory overlap rate of 10%, while using anti-diffusion baffle equipment to avoid sand diffusion caused by dredging disturbance; For mixed soil areas containing residual soil, the soil type distribution boundaries need to be additionally marked, and a segmented adaptation strategy should be adopted to ensure that the optimal operation method is used for different soil types.

[0112] S422. Based on the hierarchical operation sequence and operation area, the optimal dredging equipment movement trajectory is analyzed using a trajectory planning algorithm.

[0113] Based on the standard A* algorithm, and considering the uneven distribution of silt and limited space in deep foundation pits, this paper improves the algorithm's adaptability to different working conditions by refining the heuristic function and constraints. First, the heuristic function design is optimized: a silt resistance correction factor is introduced, constructing an improved heuristic function h(n) = ω1 × d(n,g) + ω2 × r(n), where d(n,g) is the Euclidean distance from node n to the target node g, with a weight ω1 = 0.6 to ensure the shortest path characteristic of the trajectory; r(n) is the silt resistance coefficient at the location of node n, with a weight ω2 = 0.4 to reflect the influence of silt characteristics on the trajectory; the silt resistance coefficient r(n) is calculated based on silt density data and satisfies: k is the calibrated value of the drag coefficient, taken as... Based on field test data, it was determined that the larger the silt density ρ(n), the greater the resistance coefficient, ensuring that the algorithm prioritizes the path with less silt resistance when planning the trajectory, thereby reducing equipment energy consumption.

[0114] Secondly, spatial constraints are added: taking into account the limited space of deep foundation pits, three constraints are embedded: first, the equipment operating radius constraint; second, the minimum turning radius constraint; and third, the boundary safety distance constraint. Trajectory nodes that do not conform to actual operations are eliminated through these constraints.

[0115] Furthermore, based on the work area, layering order, and improved A* algorithm, the initial movement trajectory is accurately generated to ensure the integrity of trajectory coverage and operational safety. First, the trajectory start and end points are set: strictly adhering to the principle of high priority, the geometric center grid node of the high-priority area is selected as the trajectory start point (this node has the highest silt density, and prioritizing silt removal reduces interference from subsequent operations); the edge grid nodes of the low-priority area are selected as the trajectory end point (facilitating rapid equipment removal after operation); if multiple high-priority sub-regions exist, the shortest path is used to connect the centers of each sub-region to avoid invalid cross-region movement. Next, the trajectory generation process is executed: the grid nodes of the work area are used as the search space of the algorithm, the heuristic function and constraints of the improved A* algorithm are input, and the search step size is set to 0.3m (to adapt to the grid cell size and ensure trajectory accuracy); during the generation process, the validity of the trajectory nodes needs to be verified in real time: if a node exceeds the target work area, violates the boundary constraints, or is in an invalid area, the search direction is automatically adjusted and the path is replanned; after the trajectory is generated, the coverage integrity needs to be verified by the grid traversal method to ensure that all grid nodes in the target area are covered by the trajectory (number of covered nodes / total number of nodes ≥ 99%). If there are uncovered nodes (such as small corners at the edge of the area), the trajectory segments are supplemented by local path completion to avoid omissions.

[0116] Based on this, a genetic algorithm is used to perform secondary optimization on the initial trajectory, achieving a dual-objective optimization of minimum energy consumption and shortest time, thereby improving the economy and efficiency of the trajectory. First, the optimization objectives and parameters are defined: the core optimization objectives are minimizing the total trajectory length and minimizing the operation time (total trajectory length / equipment moving speed + dredging time per layer); the node coordinates of the initial trajectory are used as the encoding objects of the genetic algorithm, and real-number encoding is used to ensure encoding accuracy. Next, the core optimization process is executed: Step 1, population initialization: based on the initial trajectory, candidate trajectories are generated by randomly perturbing the trajectory node coordinates, forming the initial population together with the initial trajectory; Step 2, fitness function calculation: a comprehensive fitness function F=α×(L_max-L) / L_max+β×(T_max-T) / T_max is constructed, where L is the trajectory length, T is the operation time, L_max and T_max are the length and time of the initial trajectory, and α=0.6 and β=0.4 are weighting coefficients, balancing energy consumption and efficiency objectives; the higher the fitness value, the better the trajectory; Step 3, selection and crossover. The mutation process involves: using roulette wheel selection to choose individuals with high fitness values ​​as parents; using single-point crossover to recombine the trajectory node coordinates of parent individuals to generate offspring individuals; using random mutation to slightly adjust the node coordinates of some offspring individuals to introduce new trajectory schemes; the fourth step is iterative convergence: repeating the above steps until the iteration is complete, selecting the individual with the highest fitness value as the optimal trajectory after optimization; after optimization, further processing of trajectory details is required: removing duplicate trajectory segments, adjusting the position of turning nodes (correcting acute angle turns to smooth arcs with an arc radius ≥ the minimum turning radius of the device) to ensure that the trajectory is smooth and easy to execute.

[0117] S5. Based on the operating status of the dredging equipment, the dynamic construction window parameters, and the optimal dredging equipment movement trajectory, an optimized underwater dredging scheme for deep foundation pit excavation is obtained through a fuzzy algorithm.

[0118] First, taking the rated load of the dredging equipment as the core benchmark, and combining the characteristics of sludge in deep foundation pit dredging, environmental parameters as auxiliary factors, and terrain constraints, a multi-parameter coupled initial load model is constructed. The core of the model incorporates four key operating condition parameters, and the logic and correlation of each parameter's influence on the load are as follows: First, sludge density (ρ): the higher the density, the stronger the adhesion between sludge particles, the greater the equipment's excavation / suction resistance, and the load shows a positive correlation, with the initial correlation coefficient set at 0.35; second, sludge moisture content (w): the higher the moisture content, the stronger the sludge's fluidity, the lower the resistance, and the load shows a negative correlation, with the initial correlation coefficient set at -0.25; third, steady-state flow velocity (v): the higher the flow velocity, the stronger the disturbance of the water flow to the equipment operation, requiring additional power to resist the water flow impact, and the load shows a positive correlation, with the initial correlation coefficient set at 0.2; fourth, terrain slope (α): the greater the slope, the greater the gravitational force that the equipment needs to overcome when moving or operating, and the load shows a positive correlation, with the initial correlation coefficient set at 0.2.

[0119] Furthermore, the initial load model satisfies: ,in For the rated load of the equipment, Based on the silt density, As the baseline moisture content, The maximum monitored flow velocity in the foundation pit, The initial correlation coefficients for each parameter are k1=0.35, k2=-0.25, k3=0.2, and k4=0.2.

[0120] Considering the dynamic characteristics of deep foundation pit conditions, such as the decrease in silt thickness during dredging, changes in flow velocity with excavation depth, and real-time updates to the terrain, the Recursive Least Squares (RLS) method is used to achieve real-time dynamic updates of the model coefficients, ensuring that the model always adapts to the current working conditions. First, data preprocessing rules are set to remove outliers from the real-time collected equipment load data, using the 3σ criterion to filter valid data. Second, core parameters are determined, and a forgetting factor λ = 0.9 is set based on the rate of change of deep foundation pit conditions. Finally, the update frequency is set, and the update formula satisfies: , , ,in The updated coefficient vector, For real-time load observations, Let P be the operating condition parameter vector. k Let be the covariance matrix.

[0121] Furthermore, based on the requirements for safe equipment operation and the operational specifications for deep foundation pit dredging, a multi-dimensional constraint system is constructed to prevent equipment overload damage, inefficient operation, or damage to the foundation pit structure. Specific constraints are as follows: Load constraints, setting the load range based on the equipment's rated parameters. ,in For minimum effective load, The system employs several constraints: a rated maximum load; a speed constraint with an upper limit set at the equipment's rated maximum speed and a lower limit set at 60% of the rated speed; an operating depth constraint, based on 3D terrain mesh data and sediment layer stratification results, setting an upper limit of 0.5m above the pit foundation (to avoid damaging the bearing layer during excavation) and a lower limit of 90% of the current silt layer thickness (to ensure dredging effectiveness and avoid excessive silt residue); and an energy consumption constraint, setting an upper limit for energy consumption per unit dredging volume to indirectly control inefficient equipment operation. These constraints are incorporated into the model using a system of inequalities to ensure that all calculated operating parameters meet the constraints.

[0122] Finally, based on the updated adaptive load model and constraints, the optimal operating parameters of the equipment are accurately calculated to achieve the goals of working condition adaptation, optimal load, and maximum efficiency. Specifically, the first step involves inputting working condition parameters by extracting core data from the current operating area in real time: sludge density ρ and water content w, steady-state flow velocity v, and terrain slope α. The second step is the optimal load calculation, which substitutes the above working condition parameters into the updated load model. The initial load value is calculated. If the initial value exceeds... The first step is to determine the optimal load based on the constraints. The second step is to match the operating parameters. Based on the operating efficiency curve (load-speed-operating depth-dredging volume correlation curve) provided by the equipment manufacturer, the optimal speed and operating depth are matched in reverse from the optimal load. At the same time, the parameters are adjusted in combination with the soil type identification results. For example, if the operating area is silty clay (high cohesion), the speed is appropriately reduced by 10% and the operating depth is increased by 0.2m to improve the dredging effect. The third step is to integrate and output the parameters. The optimal load, speed, and operating depth are integrated into a set of equipment operating status parameters, with an explanation of the applicable operating area range and working conditions, providing basic data for subsequent optimization decisions.

[0123] Deep foundation pit dredging operations involve multiple stages and parameters, and each parameter has a certain degree of fuzziness and uncertainty (such as spatial variation of silt characteristics and random fluctuations in equipment operation). By introducing fuzzy algorithms and through processes such as fuzzification, fuzzy rule reasoning, and defuzzification, core data from multiple stages can be integrated to output targeted optimization solutions for underwater dredging of deep foundation pit excavation, thereby achieving accurate fusion of data from multiple stages and effective handling of uncertainties.

[0124] Specifically, the underwater dredging optimization scheme for deep foundation pit excavation, obtained through a fuzzy algorithm based on the operating status of the dredging equipment, the dynamic construction window parameters, and the optimal dredging equipment movement trajectory, includes the following steps: S51. Fuzzyize the input parameters.

[0125] In this embodiment, eight key input parameters are precisely selected based on the core influencing factors of the entire deep foundation pit dredging process, forming a multi-dimensional input system. These include: equipment operating status (load, speed), construction window parameters (reasonableness of work time, regional risk level), trajectory parameters (trajectory coverage, energy consumption), and real-time monitoring data (water quality compliance rate, settlement deformation). To ensure the accuracy of the fuzzy processing, quantitative classification standards are established for each parameter based on parameter thresholds and industry standards: for example, load is divided into low load (≤30% of rated load), medium load (30%-70% of rated load), and high load (≥70% of rated load) based on 30% and 70% of the equipment's rated load; water quality compliance rate is classified according to the "Surface Water Environmental Quality Standard" (GB3838-2002) into excellent (≥95% compliance), good (85%-95% compliance), qualified (75%-85% compliance), and unqualified (<75% compliance). Furthermore, the precise values ​​of each parameter are converted into fuzzy values ​​using a triangular membership function, satisfying: μ(x)=max(0,1-|xa| / b) (where a is the center value of the fuzzy linguistic variable and b is the interval half-width).

[0126] S52. Construct a fuzzy rule base, and perform fuzzy inference by combining the fuzzification processing result with the fuzzy rule base.

[0127] First, based on operational experience and industry standards in deep foundation pit dredging, rules are constructed according to four dimensions: equipment operation optimization, construction window adjustment, trajectory correction, and risk emergency response. Second, the preconditions and output results of the rules are clearly defined. The preconditions cover combination scenarios with multiple input parameters, and the output results clearly define specific adjustment actions and quantitative parameters.

[0128] For example, if the load is low, the speed is low, the reasonableness of the operation time is poor, the trajectory coverage rate is medium, and the water quality compliance rate is good, the optimization plan is to increase the equipment speed by 15%, adjust the operation time to the core period, and increase the trajectory overlap rate to 18%. If the area risk level is high, the settlement deformation is medium, the energy consumption is high, and the trajectory coverage rate is low, the optimization plan is to suspend the operation in that area, switch to a low-risk sub-area, re-plan the trajectory starting point (the center of the high-priority area), and reduce the equipment load by 20%. At the same time, rule priorities are set: risk emergency response rules (such as excessive settlement and unqualified water quality) have the highest priority (priority 1), followed by equipment operation optimization rules (priority 2), and construction window and trajectory adjustment rules have the lowest priority (priority 3) to avoid conflicts when multiple rules are triggered and to ensure the consistency of decision-making logic.

[0129] Furthermore, the Mamdani reasoning method is used to conduct a complete reasoning process to ensure that the reasoning results are consistent with actual working conditions. The first step is rule matching, which involves matching the fuzzified input parameters (such as load = medium load (μ=0.67), rotation speed = medium rotation speed (μ=0.72), risk level of work area = low risk (μ=0.85), trajectory coverage = high (μ=0.91), water quality compliance rate = excellent (μ=0.93)) one by one with the preconditions in the rule base to filter out all triggering rules that meet the conditions. The second step is trigger intensity calculation, which determines the intensity of each triggering rule by taking the minimum value of each precondition fuzzy value. For example, if the fuzzy values ​​of a rule's preconditions are 0.67, 0.72, and 0.85, the trigger intensity is 0.67. The third step is result aggregation, which aggregates the output fuzzy values ​​of all triggering rules by taking the maximum value of each rule by taking the maximum value of the rule by taking the maximum value of the rule, forming a fuzzy output set. The set contains the fuzzy features of each adjustment dimension (such as the fuzzy value of increasing rotation speed 0.67, the fuzzy value of extending work time 0.75, and the fuzzy value of maintaining trajectory 0.91), clarifying the priority of each optimization direction.

[0130] S53. Defuzzify the fuzzy inference results to obtain an optimized underwater dredging scheme for deep foundation pit excavation.

[0131] The centroid method is used to process the fuzzy output set, converting fuzzy features into precise optimized parameter values. First, the calculation formula for the centroid method is defined: (Simplified to discrete scenarios) Where xi is the quantized value of the output parameter, and μ(xi) is the corresponding fuzzy value); secondly, specific conversion examples are determined for different output types. For example, the fuzzy output set for reducing the rotational speed is slightly reduced (μ=0.3, corresponding to a 5% reduction), moderately reduced (μ=0.6, corresponding to a 10% reduction), and significantly reduced (μ=0.1, corresponding to a 20% reduction). Substituting these values ​​into the formula, we get x*=(5%×0.3+10%×0.6+20%×0.1) / (0.3+0.6+0.1). 1) = 9.5%, ultimately determining a 10% reduction in rotational speed (rounded to commonly used engineering precision); adjusting the fuzzy output of the working area to shift northeast (μ = 0.7, corresponding to coordinates X + 5m, Y + 3m) and shrinking the area range (μ = 0.2, corresponding to a 2m reduction in the X range), combined with the working area coordinates (X ∈ [10m, 50m], Y ∈ [20m, 60m]), the precisely adjusted area is calculated to be X ∈ [15m, 48m], Y ∈ [23m, 60m]. Through this process, the optimization direction of the fuzzy output is transformed into a set of precisely executable parameters, ensuring the practicality of the solution.

[0132] Please see Figure 2 In an embodiment, to efficiently execute the underwater dredging optimization method provided by the present invention, the present invention also provides a system for underwater dredging optimization, 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 underwater dredging optimization method. The underwater dredging optimization system of the present invention has a compact structure and stable performance, and can stably execute the underwater dredging optimization method of the present invention, further improving the overall applicability and practical application capability of the present invention.

[0133] In this embodiment, the processor may be a central processing unit, but it can also be other general-purpose processors, digital signal processors, application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (OPGs), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor. Input devices can be used to acquire data. Output devices can be used to output the results obtained by storing program instructions contained in a computer program in the memory provided by this invention. The memory may include read-only memory and random access memory (RAM), and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory (RAM).

[0134] In one possible implementation, the memory may include a stored program area and a stored data area. The stored program area may store the operating system and applications required for at least one function; the stored data area may store data created during use. Furthermore, the memory may include read-only memory and random access memory, and provides instructions and data to the processor. The memory stores the operating system and operating instructions, executable modules, or data structures, or subsets thereof, or extended sets thereof. The operating instructions may include various operation instructions for implementing various operations. The operating system may include various system programs for implementing various basic tasks and handling hardware-based tasks.

[0135] The embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for optimizing underwater dredging.

[0136] The storage medium can include various media that can store program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.

[0137] 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. A method for underwater dredging optimization, characterized in that, The method comprises the following steps: The dredging influence parameters are integrated into the grid division mechanism to obtain a three-dimensional dynamic terrain grid model with importance weight; Through multi-source data fusion and spatial interpolation, a continuous distribution of silt density and water content spatial distribution map is obtained; The three-dimensional dynamic terrain grid model, the steady flow velocity field characteristics and the turbulence intensity characteristics, the comprehensive environmental risk coefficient and the operation efficiency reduction coefficient are combined, and a dynamic construction window parameter is obtained through multi-objective search; According to the silt density and the water content spatial distribution map, the stratification results along the profile and the preliminary soil type identification, the optimal dredging equipment moving track is analyzed; Based on the dredging equipment operating state, the dynamic construction window parameter and the optimal dredging equipment moving track, a underwater dredging optimization scheme for deep foundation pit excavation is obtained through a fuzzy algorithm.

2. The method for underwater dredging optimization of claim 1, wherein, The method for integrating the dredging influence parameters into the grid division mechanism to obtain a three-dimensional dynamic terrain grid model with importance weight comprises the following steps: An analytic hierarchy process is used to establish a multi-dimensional parameter weight system; Based on the multi-dimensional parameter weight system, a coupled grid division model is constructed, a weight correction factor is introduced to adjust the unit size, and a three-dimensional dynamic terrain grid model with importance weight is obtained.

3. The method for underwater dredging optimization of claim 1, wherein, The method for obtaining a continuous distribution of silt density and water content spatial distribution map through multi-source data fusion and spatial interpolation comprises the following steps: Bayesian estimation method is used to fuse multi-source sample data; The variogram parameters of the Kriging interpolation are dynamically corrected by using the steady flow velocity field data; Based on the fused sample data and the corrected variogram parameters, a continuous distribution of silt density and water content spatial distribution map is obtained.

4. The method for underwater dredging optimization of claim 1, wherein, The method for obtaining a dynamic construction window parameter by combining the three-dimensional dynamic terrain grid model, the steady flow velocity field characteristics and the turbulence intensity characteristics, the comprehensive environmental risk coefficient and the operation efficiency reduction coefficient, and through multi-objective search comprises the following steps: A multi-objective optimization model is constructed; NSGA-III is used to carry out multi-objective optimization solution, and a practical solution meeting the safety requirements is selected from the Pareto optimal solution set; Based on the comprehensive optimal solution, the three-dimensional dynamic construction window parameters are disassembled.

5. The method for underwater dredging optimization of claim 4, wherein, The method for obtaining the steady flow velocity field characteristics and the turbulence intensity characteristics comprises the following steps: STL method is used to separate the flow velocity time series data, and the residual term is filtered by adaptive threshold filtering; Based on the processed data, the steady flow velocity field is reconstructed, and the turbulence intensity is calculated.

6. The method for underwater dredging optimization of claim 4, wherein, The method for calculating the comprehensive environmental risk coefficient and the operation efficiency reduction coefficient comprises the following steps: An environmental constraint index system is constructed, and a single index risk is calculated; Based on the environmental constraint index system and the single index risk, a comprehensive environmental risk coefficient is obtained; A quantitative reduction model is constructed based on the influence law of environmental constraints on operation efficiency, and the operation efficiency reduction coefficient is extracted through the quantitative reduction model.

7. The method for underwater dredging optimization of claim 1, wherein, The method for analyzing the optimal dredging equipment moving track according to the silt density and the water content spatial distribution map, the stratification results along the profile and the preliminary soil type identification comprises the following steps: According to the stratification results and the soil type identification data, a stratification operation sequence is determined; Based on the hierarchical operation sequence and operation area, the optimal dredging equipment moving track is analyzed by a trajectory planning algorithm.

8. The method for underwater dredging optimization of claim 7, wherein, The cross-section profile-based hierarchical result and preliminary soil type identification are obtained, including the following steps: The interface features of the sedimentary layer are extracted by spectrum analysis; The intelligent recognition model is trained by using the interface features, silt density and moisture content data to achieve hierarchical and preliminary soil type identification.

9. The method for underwater dredging optimization of claim 1, wherein, Based on the dredging equipment operation state, the dynamic construction window parameters and the optimal dredging equipment moving track, the underwater dredging optimization scheme for deep foundation pit excavation is obtained by a fuzzy algorithm, including the following steps: The input parameters are fuzzified; A fuzzy rule base is constructed, and fuzzy reasoning is performed in combination with the fuzzification result and the fuzzy rule base; The fuzzy reasoning result is defuzzified to obtain the underwater dredging optimization scheme for deep foundation pit excavation.

10. A system for underwater dredging optimization, characterized by, The system for underwater dredging optimization includes an input device, an output device, a processor and a memory, which are connected to each other, and the memory includes program instructions for executing the method for underwater dredging optimization according to any one of claims 1-9.

Images