A method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform

Abstract

This invention discloses a method for optimizing sediment flux sampling paths based on a hydrological multimodal platform, belonging to the field of monitoring technology. The method includes: fitting vertical distribution functions of flow velocity and sediment concentration using historical regional data to establish a sediment flux field; uniformly distributing virtual measuring points across the water cross-section to form a grid; calculating the spatial gradient field of the sediment flux field; identifying regions with gradient change rates greater than a threshold as differential zones; and densifying the spacing between virtual measuring points; generating an initial sampling path based on the virtual measuring points; obtaining virtual values ​​of flow rate and suspended sediment flux; using the difference between the virtual values ​​and historical measured values ​​as the objective function, inputting iterative models for minimization, and outputting the monitoring path; obtaining the historical water level fluctuation range; classifying water levels according to the minimum grid interval; regenerating sampling paths for each level to form a path set; and storing the path set in a knowledge base for use by automated cableway programs to improve the problem of data distortion caused by environmental fluctuations in sampling paths during hydrological monitoring.

Description

Technical Field

[0001] This invention relates to the field of hydrological monitoring technology, specifically to a method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform. Background Technology

[0002] With the rapid development of modern automated hydrological monitoring technology and the widespread application of automated cableways, simultaneous monitoring of flow and suspended sediment flux at hydrological monitoring sections has become possible. Given the increasingly stringent requirements for timeliness, current automated sampling equipment is increasing monitoring frequency to obtain more comprehensive and timely data, providing crucial data support for hydrological research.

[0003] Existing automated cableway systems for hydrology typically guide hydrographs along pre-set paths for sampling. However, in practice, these systems struggle to capture the dynamic details of changes in flow and sediment.

[0004] The reasons for this were simulated and analyzed using a multimodal hydrological platform. The limiting factors are the opposite characteristics of flow velocity and the vertical distribution of suspended sediment: flow velocity is higher at the top and lower at the bottom, while suspended sediment is lower at the top and higher at the bottom. This makes the sensor susceptible to interference in high-sediment-laden flow environments. Secondly, when the hydrometer is in a high-sediment-concentration environment, suspended particles easily adhere to the sensor surface, potentially altering key parameters such as the refractive index and temperature of the liquid medium, leading to significant deviations in the measurement data.

[0005] Especially during time-sensitive data collection, systematic deviations can easily lead to data drift and distortion. Distortion of measurement data not only affects the accuracy of flow calculations, but in severe cases, it can cause the estimated suspended sediment flux to deviate from the true value, directly impacting the quality of hydrological monitoring reports and interfering with hydrological early warning systems. Therefore, optimizing the sampling path for quicksand flux and ensuring the effectiveness of automated sampling is a worthy area of ​​research. Summary of the Invention

[0006] The purpose of this invention is to provide a method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform, which aims to improve the problem that the sampling path is affected by environmental fluctuations and the acquired data is easily distorted during hydrological monitoring.

[0007] To address the aforementioned technical problems, this invention employs the following technical solution: a method for optimizing sediment flux sampling paths based on a hydrological multimodal platform, comprising the following steps: acquiring historical data of the target area; fitting vertical distribution functions of flow velocity and sediment concentration based on the historical data to establish a sediment flux field; uniformly distributing virtual measuring points on the water cross-section to form a grid; calculating the spatial gradient field of the sediment flux field; identifying regions where the gradient change rate exceeds a predetermined threshold as difference zones; and densifying the spacing between virtual measuring points within these difference zones; randomly generating initial sampling paths based on the virtual measuring points; acquiring virtual values ​​of flow rate and suspended sediment flux corresponding to the initial sampling paths; using the difference between the virtual values ​​and historical measured values ​​as the objective function; inputting the objective function into an iterative model to minimize the objective function and outputting the monitoring path; acquiring historical water level variation ranges; classifying water levels according to the minimum grid interval; and randomly generating sampling paths for each water level level to obtain monitoring paths for several water level levels and forming a path set; and solidifying the path set into a knowledge base, which is used for program calls in automated cableways.

[0008] As a preferred option, it also includes real-time monitoring of the target area to obtain hydrological data; setting analysis intervals, calculating the correlation coefficient between the current flux field and the historical flux field in each analysis interval, and reconstructing the monitoring path when the correlation coefficient is lower than the set threshold coefficient.

[0009] As a preferred method, when obtaining the vertical distribution of flow velocity, the underlying logarithmic law, power law function, and surface exponential law are used to obtain the segmented flow velocity and weights are introduced for coupling; when obtaining the vertical distribution of sediment concentration, the local Reynolds number is introduced through the Rouse formula, and the result of sediment concentration is corrected by the Reynolds number.

[0010] Preferably, when identifying the aforementioned difference zone, the flow sand flux field is determined, and spatial interpolation is performed on all virtual measuring point data to form a continuous flux field distribution; the spatial gradient value at each virtual measuring point is obtained based on the flux field distribution; the gradient values ​​of all virtual measuring points in the flux field are statistically analyzed, and the dynamic threshold of the gradient change rate is determined by obtaining the average gradient and standard deviation; the spatial gradient value of each measuring point is compared with the dynamic threshold, and if the spatial gradient value at a certain measuring point exceeds the dynamic threshold, the area where the measuring point is located is marked as a difference zone.

[0011] Preferably, when densifying the spacing of virtual measuring points in the aforementioned difference zone, the original spacing of the virtual measuring points is set to be no less than 0.05 meters, and the densification spacing is set according to the magnitude of the local flux field gradient, so that the virtual measuring points are more densely packed in areas with higher gradients.

[0012] Preferably, the iterative model described above uses the Multi-Objective Genetic Algorithm (MOGA). The algorithm is implemented as follows: Step A: Randomly generate a certain number of individuals as first-generation individuals and establish an initial population. Step B: Determine the objective function in the quicksand flux sampling path optimization, calculate the objective function value for each first-generation individual, and select the best first-generation individuals as candidate individuals. Step C: Perform crossover and mutation operations on the candidate individuals to obtain second-generation individuals. Step D: Combine the second-generation individuals with the candidate individuals to form a new population, and repeat Step B until a preset termination condition is met.

[0013] By continuously updating the fitness of individuals in each iteration and gradually optimizing the population through selection, crossover, and mutation operations, the population converges towards the optimal solution. When the preset termination condition is reached, the process stops and outputs the optimal solution set, thereby determining the final optimization scheme for the quicksand flux sampling path.

[0014] A further technical solution is that the above-mentioned cross operation is to perform a tangential cross operation on the parent path in the difference section, and the above-mentioned mutation operation is to perform a directional mutation along the gradient direction on the path points in the difference section.

[0015] As a preferred option, when classifying water levels as described above, a variable interval strategy is adopted, with historical high-frequency water level intervals using 0.01m intervals and other intervals using 0.05m intervals. An inheritance mechanism for the path settings of adjacent water level levels is configured so that when the changing trends of adjacent water level levels are similar, the strategy parameters of adjacent water level levels are inherited.

[0016] A further technical solution involves reconstructing the monitoring path by accessing the knowledge base to call the new monitoring path and confirming the real-time water level of the virtual measuring point on the monitoring path. The knowledge base is then used to query the water level of the adjacent water level segments of the virtual measuring point to obtain the water level of the next level and the water level of the next level. The reconstructed monitoring path is verified by using the water level of the next level and the water level of the next level through a quadratic interpolation method.

[0017] Compared with the prior art, the beneficial effects of the present invention are:

[0018] This invention introduces a quicksand flux field based on flow velocity and sediment concentration. This field accurately reflects the quicksand flux characteristics at different water depths and locations. Furthermore, it identifies regions of difference based on the variations in quicksand flux across different areas. Virtual monitoring points in these regions are then densified to ensure monitoring density in key areas. Water level classification is performed using historical water level data, and an iterative model is employed for minimization to output the monitoring path. Simultaneously, during hydrological data monitoring, the correlation coefficient between the current quicksand flux and historical data is utilized to dynamically adjust the sampling path when the correlation decreases. This allows for path selection based on circumstances during monitoring, and optimization of the sampling path makes it more adaptable to changing environmental conditions. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0021] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this invention are only used to explain the relative positional relationship, movement, etc., in a specific working state. If the specific posture changes, the directional indication will also change accordingly. In this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be an electrical signal connection or a signal connection; it can also refer to the internal connection of two components or the interaction relationship between two components, unless otherwise explicitly limited. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0022] If the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Furthermore, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0023] refer to Figure 1 As shown, one embodiment of the present invention is a method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform, comprising the following steps: acquiring historical data of the target area, fitting vertical distribution functions of velocity and sediment concentration according to the historical data, and establishing a quicksand flux field. By combining the vertical distribution functions of velocity and sediment concentration, a dynamic quicksand flux field can be established within the target area.

[0024] Specifically, the flow velocity and sediment concentration in the target area fluctuate over time. By acquiring historical data from the target area, this historical data is used to provide long-term actual observation data, which serves as a reference for the rationality of establishing the shifting sand flux field.

[0025] The water flow velocity is higher at the surface and lower at the bottom. The sediment concentration distribution in the water body varies with depth, typically higher at the surface and lower at the bottom. Therefore, historical data needs to be fitted with both the vertical distribution function of flow velocity and the vertical distribution function of sediment concentration.

[0026] The aforementioned fitted vertical velocity distribution function is used to quantify the velocity variation patterns at different water depths, ensuring that the fitted vertical velocity distribution function covers the case where the velocity distribution is non-uniform at different depths. The fitted vertical sediment concentration distribution function is used to represent the sediment concentration status or trend at different depths in the water body. A sediment flux field is established by fitting the vertical velocity and sediment concentration distribution functions.

[0027] The aforementioned quicksand flux is obtained by multiplying the flow velocity and sediment concentration, while the quicksand flux field is used to integrate quicksand flux-related information into a spatial distribution map. The distribution map represents the quicksand flux at different locations, in different water depths, and under different flow conditions.

[0028] Virtual measuring points are uniformly distributed on the water cross-section to form a grid. By calculating the spatial gradient field of the quicksand flux field, regions with gradient change rates greater than a predetermined threshold are identified as difference zones, and the spacing between virtual measuring points in the difference zones is increased.

[0029] The aforementioned virtual measuring points cover the entire spatial range of the water cross-section. A grid is formed by the uniformly distributed virtual measuring points, dividing the cross-section into several uniform boundary regions to ensure the uniformity of the sediment flux across the entire cross-section. It is important to note that the virtual measuring points do not actually collect samples; they are primarily used to simulate the sampling process to determine the variation of sediment flux at various locations. The data from the virtual measuring points relies on historical data, combined with the vertical distribution functions of velocity and sediment concentration, to simulate the velocity and sediment concentration at the water depth and location corresponding to the virtual measuring points.

[0030] Considering that the quicksand flux field is a spatial distribution composed of flow velocity and sediment concentration, the rate of change of quicksand flux can be inferred by obtaining the spatial gradient field of the quicksand flux field, thereby determining the variation of quicksand flux at different locations. That is, a faster change in quicksand flux in any region indicates a more complex quicksand characteristic in that region. By analyzing the spatial gradient field, regions with gradient change rates exceeding a set threshold are identified, facilitating the location of areas with significant quicksand flux variations and aiding in subsequent path setting for these areas.

[0031] Specifically, after identifying areas with large fluctuations in quicksand flux as differential zones, the spacing between virtual measuring points in these zones is increased to ensure that the distance between virtual measuring points within the differential zones is reduced, thereby increasing the sampling density and thus better capturing the changing characteristics of quicksand flux in these areas.

[0032] Among these, the large gradient change rate in the identified difference zone, and the possibility that such a large gradient change rate indicates drastic changes in hydrodynamics, bed characteristics, or sediment concentration, necessitates the following approach: An initial sampling path is randomly generated based on virtual measuring points. Virtual values ​​of flow rate and suspended sediment flux corresponding to this initial sampling path are obtained. The difference between the virtual values ​​and historical measured values ​​is used as the objective function. This objective function is then input into an iterative model to minimize the objective function and output the monitoring path.

[0033] Specifically, by randomly generating an initial sampling path, virtual values ​​of flow rate and suspended sediment flux corresponding to that path are calculated. These virtual values ​​are based on virtual measuring point locations along the initial path, and are therefore estimated parameters calculated using known parameters such as flow velocity and sediment concentration.

[0034] For reference, by comparing virtual values ​​with historical measured values ​​(i.e., real monitoring data) and calculating the difference, the difference reflects the degree of matching between the current sampling path's effectiveness and the actual data. Therefore, the difference can be used as the objective function. Specifically, the objective function formed by the difference can be the error between virtual and measured values, and the error can be expressed as mean squared error. Through an iterative process, the path is adjusted through operations such as crossover and mutation, continuously searching for the optimal solution that minimizes the objective function.

[0035] The historical water level fluctuation range is obtained, and the water level is classified according to the minimum grid interval. Sampling paths are randomly generated for each water level level to obtain monitoring paths under several water level levels and form a path set. The path set is then fixed in the knowledge base, which is used for program calls of the automated cableway.

[0036] For reference, considering that the distribution of shifting sediment flux may differ at lower or higher water levels, it is necessary to differentiate and optimize the sampling path through water level classification. Obtaining the historical water level variation range involves using historical water level data to determine the maximum variation in water level within a certain time frame. The historical water level variation range reflects the fluctuations in water level under different seasons, weather conditions, or environmental changes. Based on the historical water level variation range, the water level range is divided into multiple water level levels, which can be classified using grid intervals in this invention. Different sampling paths are generated at different water levels based on the actual situation.

[0037] The purpose of re-randomizing the sampling path is to optimize the current path. This process typically involves generating an initial sampling path based on virtual measuring points at the same water level, calculating the corresponding virtual values ​​of flow rate and suspended sediment flux, and minimizing the objective function by combining historical measured data to optimize the sampling path.

[0038] Based on the above embodiments, another embodiment of the present invention further includes, in order to ensure the effectiveness of the currently constructed path, real-time monitoring of the target area to obtain hydrological data; setting an analysis interval, calculating the correlation coefficient between the current flux field and the historical flux field in each analysis interval, and reconstructing the monitoring path when the correlation coefficient is lower than the set threshold coefficient.

[0039] Hydrological data is acquired through real-time monitoring of the target area, and reasonable analysis intervals are set. Within each analysis interval, the correlation coefficient between the current flux field and the historical flux field is calculated. For reference, the generation and optimization of sampling paths are based on a shifting sediment flux field constructed from historical data, which reflects the statistical distribution and dynamic patterns of factors such as water flow and sediment concentration over a longer timescale. The correlation coefficient primarily quantitatively assesses the similarity between the spatial distribution characteristics of shifting sediment flux under the current hydrological environment and its historical distribution characteristics.

[0040] A low correlation coefficient indicates a significant difference between the flux field formed by the current actual hydrological data and the historical field distribution patterns. Since this significant difference can be interpreted as a change in current hydrological conditions, the sampling path originally optimized based on historical characteristics may no longer represent the complex current shifting sediment characteristics. Therefore, in this invention, this situation is considered a failure of the current path, requiring re-optimization of the monitoring path.

[0041] It should be noted that the occurrence of significant differences does not necessarily indicate that the path has failed. However, in order to reduce the potential risks of sampling and improve the reliability of the data, this application still considers this condition to require re-optimization of the path.

[0042] Furthermore, when reconstructing the monitoring path, the knowledge base is retrieved to call the new monitoring path, and the real-time water level of the virtual measuring point of the monitoring path is confirmed. The water level of the adjacent water level segment of the virtual measuring point is retrieved from the knowledge base to obtain the water level of the previous level and the water level of the next level. The reconstructed monitoring path is verified by using the previous level and the water level of the next level through quadratic interpolation to verify whether it is transmitted.

[0043] The upper-level water level is selected based on the nearest real-time water level of the virtual measuring point, which is greater than or equal to the current water level. The lower-level water level is selected based on the nearest real-time water level of the virtual measuring point, which is less than or equal to the current water level. After determining the upper and lower-level water levels, a third water level point that is close to the current measuring point and has a clear water level change trend is obtained from the knowledge base. The water level data of the upper-level, lower-level, and third water level point are obtained. The interpolated water level is calculated using existing Lagrange interpolation. Finally, the interpolated water level is compared with the real-time water level to determine whether the absolute difference is less than 0.1 meters. If the absolute difference is less than 0.1 meters, the transmission is successful; otherwise, it indicates that the transmission has not been successful.

[0044] To facilitate understanding, an example is provided here. When reconstructing the monitoring path, the real-time water level of the virtual monitoring point along the path is obtained. The knowledge base is then used to query neighboring water level segments. Based on the location and water depth information of the virtual monitoring point, the corresponding water level classification data for its area is retrieved. When querying neighboring water level segments, i.e., the water level of the previous and next levels;

[0045] For example, a monitoring section was constructed in a river segment using a hydrological multimodal platform, and several virtual measuring points were deployed on this section. Based on the selected virtual measuring point's location on the section and the real-time monitored water level, the knowledge base was queried to obtain nearby water level data, thus acquiring the next-level and next-level water levels for this virtual measuring point. Specifically, the selected virtual measuring point is located at position 10 on the section with a water level of 3.15 meters; the obtained next-level water level is located at position 8 on the section with a water level of 3.0 meters; and the next-level water level is located at position 12 on the section with a water level of 3.4 meters. Another data point relatively close to the current measuring point was selected as the third water level point, and its location on the section was determined to be position 14 with a water level of 3.5 meters.

[0046] Therefore, the following information can be obtained from the three sets of known data points formed by the previous water level, the next water level, and the third water level.

[0047] The data point for the next higher water level is (x0=8, H0=3.0).

[0048] The data point for the next water level is (x1=12, H1=3.4).

[0049] The data points for the third water level are (x2=14, H2=3.5).

[0050] The three sets of known data points formed by the upper level water level, the lower level water level and the third water level are imported into the Lagrange quadratic interpolation formula to construct the interpolation function: F(x)=H0·L0(x)+H1·L1(x)+H2·L2(x);

[0051] The basis functions that can be determined are as follows:

[0052] L0(x)=((x−x1)·(x−x2)) / ((x0−x1)·(x0−x2));

[0053] L1(x)=((x−x0)·(x−x2)) / ((x1−x0)·(x1−x2));

[0054] L2(x)=((x−x0)·(x−x1)) / ((x2−x0)·(x2−x1));

[0055] Using the position of the current virtual measuring point on the cross section as the independent variable, i.e., X=10, and substituting it into the above basis functions, we get L0(10)=0.3333, L1(10)=1.0, L2(10)=-0.3333. The results are then imported into the Lagrange quadratic interpolation formula to construct the interpolation function.

[0056] The interpolation function F(x) = 3.0 × 0.3333 + 3.4 × 1.0 + 3.5 × (–0.3333) ≈ 3.2333 indicates that the verification interpolated water level is 3.2333 meters. The current real-time water level at the virtual measuring point is 3.15 meters. If the absolute difference between the verification interpolated water level of 3.2333 and the real-time water level of the virtual measuring point of 3.15 meters is less than 0.1 meters, then the transmission is considered successful.

[0057] Based on the above embodiments, another embodiment of the present invention is that, when obtaining the vertical distribution of flow velocity, the underlying logarithmic law, power law function, and surface exponential law are called to obtain the segmented flow velocity and weights are introduced for coupling; when obtaining the vertical distribution of sediment concentration, the local Reynolds number is introduced through the Rouse formula, and the result of sediment concentration is corrected by the Reynolds number.

[0058] For reference, the traditional method is to use the least squares method to fit the changes in water velocity and sediment concentration with water depth. It is relatively simple overall, and the main independent variable is limited to water depth. However, its parameter effect is not ideal when presenting hydrodynamic behavior and sand distribution.

[0059] For reference, this implementation primarily employs various flow velocity and sediment concentration models to describe conditions under different circumstances. The use of multiple model fitting methods provides a foundation for subsequent iterative optimization using genetic algorithms. Specifically, the surface exponential law describes the surface velocity distribution. The middle layer uses a power-law function to describe the velocity distribution, which, by representing the velocity distribution at a larger scale, is used to represent a large area at a certain height from the bottom. The bottom logarithmic law is mainly used for the turbulent boundary layer, where the velocity distribution of the fluid follows a logarithmic law; therefore, the bottom logarithmic law is used to represent the flow region near the bottom.

[0060] To facilitate reading and understanding of the specific methods of this invention, examples are provided here, and the underlying logarithmic law, power law function, and surface exponential law related models are used to describe the invention, providing implementation reference for those skilled in the art.

[0061] One possible approach is to use the following general model for the underlying logarithmic law:

[0062] In the formula, K is a constant used to simulate turbulence, representing the frictional velocity between the fluid and the rough bottom surface, and its value is between 0.3 and 0.5. This represents the vertical distance from the current position of the bottom layer. The rough bottom affects the height. Bottom roughness height to altitude Logarithmic change of flow velocity at a point.

[0063] One possible approach is to adopt the following general model for power-law functions:

[0064] In the formula, It is the reference speed at position h. For environmental correlation coefficient, This represents the vertical distance from the current position of the middle layer. To indicate reference height The velocity at a certain point follows a power law with height. It changes with the increase.

[0065] One possible approach is to adopt the following general model for the surface exponential law:

[0066] In the formula, The maximum surface velocity, It is the exponential decay rate. It is a reference characteristic height. By adjusting the attenuation rate, the output conclusions can reflect the effects of hydrodynamic characteristics and factors such as friction near the water surface and fluid viscosity on velocity attenuation. The surface features show that the surface water flow velocity gradually decreases with increasing water depth.

[0067] Based on this, the formula for fitting the vertical velocity distribution function is to weight the segmented velocities according to their weights, and the weighting formula can be... The weighting coefficients ω1, ω2 and ω3 were determined through historical data evaluation.

[0068] The Rouse formula is a model that estimates particle concentration variation with height based on particle settling velocity and vertical eddy viscosity of fluid. The Rouse number in the formula primarily represents parameters related to particle and fluid characteristics. Since the Rouse formula is quite common in this field, it will not be elaborated upon further.

[0069] This invention primarily considers that the Rouse number is limited by the ratio of particle settling velocity to friction velocity; while the local Reynolds number is typically the ratio of inertial force to viscous force when fluid flows around an object. By observing changes in the local Reynolds number, the flow state can be observed to some extent. Generally, the settling velocity under transitional flow conditions needs to be 10% to 20% lower than that under pure laminar flow, while the friction velocity at the bed surface may not be sufficient to reflect local hydrodynamic characteristics under transitional flow conditions, requiring an increase of 10% to 20%. In other words, by using local Reynolds adjustment to adjust the settling velocity and friction velocity, the Rouse number in the Rouse formula is corrected to some extent, thus achieving a correction of the sediment concentration result.

[0070] Based on the above embodiments, another embodiment of the present invention is that, when identifying the above-mentioned difference zone, the flux field of quicksand is determined, and spatial interpolation is performed on all virtual measuring point data to form a continuous flux field distribution; the spatial gradient value at each virtual measuring point is obtained based on the flux field distribution.

[0071] The gradient values ​​of all virtual measurement points in the flux field are statistically analyzed, and the dynamic threshold of the gradient change rate is determined by obtaining the average gradient and standard deviation. The spatial gradient value of each measurement point is compared with the dynamic threshold. If the spatial gradient value at a certain measurement point exceeds the dynamic threshold, the area where the measurement point is located is marked as a difference zone.

[0072] Among them, by calculating the average and standard deviation of the gradient values ​​of all virtual measuring points, the dynamic threshold can reflect the current flux field distribution. Compared with the set fixed threshold, the dynamic threshold can reduce the risk of failure caused by changes in environmental conditions or large data fluctuations.

[0073] Specifically, the dynamic threshold needs to be adjusted according to the changes in the statistical characteristics of the entire quicksand flux field. When the quicksand characteristics of the entire environment fluctuate, the fixed threshold will usually show a large number of abnormal areas, making it impossible to obtain a valid basis for the entire path.

[0074] Dynamic anomaly identification can select areas with more pronounced fluctuations as difference zones within batch fluctuations, effectively reducing computational redundancy caused by fluctuations and better selecting high-risk areas for labeling in various complex scenarios. Therefore, dynamic threshold setting and comparison methods can improve situations where failures or blurred key areas occur due to changes in environmental conditions or large data fluctuations.

[0075] To facilitate understanding, an example is provided here. This embodiment takes the flow sand flux data collected by virtual measuring points on a certain monitoring section as an example, and obtains a continuous flux field through spatial interpolation. Eight measuring points are selected as an example for the description.

[0076] First, obtain the gradient values ​​at 8 measurement points, assuming the gradient values ​​at the 8 points are 1.5, 1.6, 1.4, 0.8, 1.7, 1.3, 4.5, and 1.6. Then, determine the mean gradient of the 8 points, which is approximately 1.9. Next, obtain the standard deviation of the 8 points, which is approximately 0.9. The sum of the mean gradient and the standard deviation is set as the dynamic threshold, which is 2.8.

[0077] Then, each gradient value is compared with the dynamic threshold, and the measuring points with gradient values ​​greater than 2.8 are selected. The measuring points with gradient values ​​between 1.3 and 1.7 are all below 2.8 and do not belong to the abnormal change area, so they are not marked as the difference area. The measuring point with a gradient value of 4.5 is greater than the dynamic threshold. Therefore, the change in quicksand flux at the measuring point corresponding to the gradient value of 4.5 is very drastic, and this area is marked as the difference area.

[0078] Based on the above embodiments, another embodiment of the present invention is that when the spacing of the virtual measuring points in the above-mentioned difference region is densified, the original spacing of the virtual measuring points is not less than 0.05 meters, and the densification spacing is set according to the magnitude of the local flux field gradient, so that the virtual measuring points are more dense in the region with higher gradient.

[0079] The original spacing refers to the initial spacing between virtual measuring points on a water cross-section. A spacing of 0.05 meters or more is mainly used in areas with gentle flux fluctuations and slow local flux changes, where an original sampling density of 0.05 meters or more is sufficient to describe the data characteristics. Similarly, the densified spacing is set according to the magnitude of the local flux field gradient, mainly to further reduce the measuring point spacing in areas with high flux gradients through dynamic adjustment.

[0080] There is no limit to the encryption spacing here because the more sampling points within the same distance range, the higher the sampling accuracy. If adding one sampling point at a 0.05-meter interval fails to capture the details of a sudden increase in local quicksand flux, it indicates that the encryption sampling density is insufficient, and more sampling points need to be added within the same distance range.

[0081] Based on the above embodiments, another embodiment of the present invention is that the iterative model is a multi-objective genetic algorithm (MOGA). On the one hand, using MOGA for optimization allows the sampling path to be adaptively adjusted through the evolutionary process, which is beneficial for selecting paths that better meet actual needs from numerous virtual test points. If necessary, path reconstruction and inheritance can also be performed through a knowledge base, directly using overlapping paths to reduce the computational load on the server.

[0082] Specifically, the objective function described above is processed in the iterative model as follows:

[0083] Step A involves randomly generating a certain number of individuals as the first generation and establishing an initial population. Initially, N sampling path individuals are generated, each consisting of a set of sequentially arranged virtual measurement points. Each sampling path individual is randomly generated and does not necessarily represent an optimized path; in principle, the individuals only need to provide an initial foundation for later evolution.

[0084] Step B involves determining the objective function in the quicksand flux sampling path optimization, calculating the objective function value for each generation of individuals, and selecting the best-performing individuals as candidate individuals. This is achieved by comparing the simulated error of quicksand flux at all virtual measuring points along the path with the actual measurement data to obtain an error index. Then, based on the distances between the virtual measuring points along the path, the total travel length of the path is obtained. The objective function value is the product of the travel length and the error index, weighted separately, to obtain the objective function value for each individual, and selecting the best-performing candidate individuals.

[0085] Step C involves performing crossover and mutation operations on the candidate individuals to obtain second-generation individuals. Treating the candidate individuals as parents, a crossover operation is performed. For example, if the path points of parent individual 1 are {P1, P3, P5, P7, P9} and the path points of parent individual 2 are {P2, P4, P6, P8, P10}, then a crossover position is selected, and some subsequences are exchanged to generate a new individual. For instance, after crossing parent individual 1 and parent individual 2, a new individual A{P1, P3, P6, P8, P10} can be obtained.

[0086] It should be noted that when applying mutation operations to some new individuals, the main process is to select any individual at a random location to replace a measurement point with an adjacent but unselected measurement point, thereby forming a new sampling route. After the second-generation individual set is combined with crossover and mutation operations, multiple new paths will be generated, resulting in several second-generation individuals. The second-generation individuals may vary in error and path length compared to their parent generation.

[0087] Step D involves forming a new population from the second-generation individuals and candidate individuals, and repeating Step B until the preset termination condition is met. The new second-generation individuals are then combined with the aforementioned candidate individuals to form a new population. The objective function value for each individual is calculated again according to Step B, and outstanding individuals are selected for further crossover and mutation. After the third generation of iterations, several individuals emerge whose overall objective function performance tends to have lower errors and more reasonable paths. The fitness is continuously updated until the termination condition is met, at which point a set of optimized optimal sampling paths is output.

[0088] The termination condition is usually reaching the maximum number of iterations. If necessary, a condition can be set where the fitness improvement falls below a certain threshold for several consecutive generations. It should be noted that validated and effective paths, especially those that remain stable and effective after multiple generations, can generally be directly used to avoid redundant calculations. Similarly, newly generated paths generally need to be compared with existing paths in the knowledge base. If the similarity is high, some parameters can be directly "inherited," thus reducing the server load during iteration.

[0089] By continuously updating the fitness of individuals in each iteration and gradually optimizing the population through selection, crossover, and mutation operations, the algorithm converges towards the optimal solution. When a preset termination condition is reached, the algorithm stops and outputs the optimal solution set. The termination condition can be the maximum number of iterations or reaching a fitness threshold. By using the termination condition, the computational load is kept within a certain controllable range, effectively determining the final optimized scheme for the quicksand flux sampling path.

[0090] Furthermore, the above-mentioned crossover operation is a tangential crossover operation performed on the parent path in the difference section, and the above-mentioned mutation operation is a directional mutation performed on the path points in the difference section along the gradient direction.

[0091] The use of tangential intersection primarily involves recombining the two parent paths in sections of difference, while preserving their respective local geometric characteristics. This aims to allow the offspring paths to leverage their strengths and compensate for their weaknesses. Mutation is used because drastic flux changes can prevent local paths from fully capturing subtle shifts. By applying minute perturbations along the current path's flux gradient, directional mutation is achieved, enabling the path to better track the local upward or downward trends in the flux.

[0092] In practice, the sampling points of the current parent path in the difference zone are first determined based on the spatial gradient value. For the sampling points in the difference zone, several adjacent measurement points are obtained and a local curve is fitted to determine the tangent direction of the local curve at that point. Then, in the difference zone of the two parent paths, the corresponding segments are selected, and the path points in the selected segments are matched according to the tangent direction. The two matched paths are then combined along the tangent direction of the difference zone to cross and combine the parent paths. That is, a local segment is taken from one parent and another local segment is taken from the other parent to combine and generate a new path segment, thus completing the cross operation.

[0093] For a path point within the difference zone, the local quicksand flux gradient vector around it is calculated. An appropriate offset along the gradient direction is set, and the coordinates of the current path point are adjusted along the gradient direction. For example, for a parent path point P4 within the difference zone, its quicksand flux gradient points in the R direction, with a gradient magnitude of G. The variable offset length / unit of a single gradient is set to 0.01 meters, then the variable offset of P4 is 0.01*G. If g is 3, the variable offset distance is 0.03 meters. In this case, P4 is moved 0.03 meters along the R direction, forming a new point P4′. Theoretically, P4′ better reflects the local quicksand flux variation trend from low to high. Using P4′ as a path point optimizes the adaptability of the sampling path to real hydrological conditions.

[0094] Based on the above embodiments, another embodiment of the present invention is that, during the water level classification, a variable interval strategy is adopted, using a 0.01m interval for historical high-frequency water level intervals and a 0.05m interval for other intervals. An inheritance mechanism for the path settings of adjacent water level levels is configured so that when the changing trends of adjacent water level levels are similar, the strategy parameters of adjacent water level levels are inherited. The finer water level intervals in high-frequency areas are mainly to facilitate better capture of water level changes by the path, which helps optimize the monitoring of shifting sediment. Larger intervals are used in non-high-frequency areas, mainly to reduce the number of water level levels and the overall parameter configuration. The method of inheriting parameters from adjacent water level segments facilitates a relatively natural transition of the path in changing areas and avoids path discontinuities caused by abrupt strategy changes.

[0095] In this specification, terms such as "one embodiment," "another embodiment," "embodiment," and "preferred embodiment" refer to specific features, structures, or characteristics described in connection with that embodiment, which are included in at least one embodiment described in the general description of this application. The appearance of the same term in multiple places in the specification does not necessarily refer to the same embodiment. Furthermore, when a specific feature, structure, or characteristic is described in connection with any embodiment, the intention is to suggest that implementing such a feature, structure, or characteristic in conjunction with other embodiments also falls within the scope of this invention.

[0096] Although the invention has been described herein with reference to several illustrative embodiments, it should be understood that many other modifications and implementations can be devised by those skilled in the art, which will fall within the scope and spirit of the principles disclosed herein. More specifically, various variations and modifications can be made to the components and / or layout of the subject matter combination within the scope of the disclosure and claims. Besides variations and modifications to the components and / or layout, other uses will be apparent to those skilled in the art.

Claims

1. A method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform, characterized in that, Includes the following steps: Historical data of the target area is obtained, and vertical distribution functions of flow velocity and sediment concentration are fitted according to the historical data to establish a sediment flux field. Virtual measuring points are uniformly distributed on the water cross section to form a grid. By calculating the spatial gradient field of the sediment flux field, areas with gradient change rate greater than a predetermined threshold are identified as difference zones, and the spacing of virtual measuring points in the difference zones is increased. The initial sampling path is randomly generated based on virtual measuring points. The virtual values ​​of flow and suspended sediment flux corresponding to the initial sampling path are obtained. The difference between the virtual value and the historical measured value is used as the objective function. The objective function is input into the iterative model to minimize the objective function and output the monitoring path. The historical water level fluctuation range is obtained, and the water level is classified according to the minimum grid interval. Sampling paths are randomly generated for each water level level to obtain monitoring paths under several water level levels and form a path set. The path set is then fixed in a knowledge base, which is used for program calls of automated cableways.

2. The method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform according to claim 1, characterized in that, It also includes real-time monitoring of target areas to obtain hydrological data; setting analysis intervals, calculating the correlation coefficient between the current flux field and the historical flux field in each analysis interval, and reconstructing the monitoring path when the correlation coefficient is lower than the set threshold coefficient.

3. The method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform according to claim 1, characterized in that, When obtaining the vertical distribution of flow velocity, the underlying logarithmic law, power law function, and surface exponential law are called to obtain the segmented flow velocity and weights are introduced for coupling. When obtaining the vertical distribution of sediment concentration, the local Reynolds number is introduced through the Rouse formula, and the sediment concentration result is corrected by the Reynolds number.

4. The method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform according to claim 1, characterized in that: When identifying the difference zone, the flux field of the quicksand is determined, and spatial interpolation is performed on all virtual measuring point data to form a continuous flux field distribution; the spatial gradient value at each virtual measuring point is obtained based on the flux field distribution. The gradient values ​​of all virtual measurement points in the flux field are statistically analyzed, and the dynamic threshold of the gradient change rate is determined by obtaining the average gradient and standard deviation. The spatial gradient value of each measurement point is compared with the dynamic threshold. If the spatial gradient value at a certain measurement point exceeds the dynamic threshold, the area where the measurement point is located is marked as a difference zone.

5. The method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform according to claim 1, characterized in that: When densifying the spacing of virtual measuring points within the difference zone, the original spacing of the virtual measuring points is set to be no less than 0.05 meters, and the densification spacing is set according to the magnitude of the local flux field gradient, so that the virtual measuring points are more densely packed in areas with higher gradients.

6. The method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform according to claim 1, characterized in that, The iterative model is based on the multi-objective genetic algorithm MOGA; the objective function is processed in the iterative model as follows: Step A: Randomly generate a certain number of individuals as the first generation of individuals and establish the initial population; Step B: Determine the objective function in the optimization of the flux sampling path in quicksand, calculate the objective function value for each generation of individuals, and select the best generation of individuals as candidate individuals; Step C: Perform crossover and mutation operations on the candidate individuals to obtain second-generation individuals; Step D: Combine the second-generation individuals with the candidate individuals to form a new population, and repeat step B until the preset termination condition is met.

7. The method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform according to claim 6, characterized in that, The crossover operation is a tangential crossover operation performed on the parent path in the difference section, and the mutation operation is a directional mutation performed on the path points in the difference section along the gradient direction.

8. The method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform according to claim 1, characterized in that, When classifying water levels, a variable interval strategy is adopted, with historical high-frequency water level intervals using 0.01m intervals and other intervals using 0.05m intervals. Configure an inheritance mechanism for path settings of adjacent water level levels so that when the changing trends of adjacent water level levels are similar, the strategy parameters of adjacent water level levels are inherited.

9. The method for optimizing the sampling path of quicksand flux based on a hydrological multimodal platform according to claim 2, characterized in that, When reconstructing the monitoring path, the knowledge base is retrieved to call the new monitoring path, and the real-time water level of the virtual measuring points of the monitoring path is confirmed. The water level of the adjacent water level segment of the virtual measuring point is retrieved from the knowledge base to obtain the water level of the next level. The reconstructed monitoring path is verified by using the upper and lower water levels and the quadratic interpolation method.

Images