Intelligent identification modeling method and system for non-standard runoff plot of power transmission and transformation project

Abstract

The invention provides a power transmission and transformation project non-standard runoff plot intelligent identification modeling method and system, and relates to the technical field of data processing.The method comprises the steps that twinborn modeling is conducted according to a real-time monitoring data set, non-standard runoff plot identification division is conducted, water and soil loss coupling feature optimization is conducted according to a first runoff plot distribution model, and a second runoff plot distribution model is obtained; performing multi-node direct and indirect water and soil loss accident deduction on the second runoff plot distribution model according to a power transmission and transformation project construction scheme and a future meteorological data set, and performing multi-node global disturbance optimization on the second runoff plot distribution model according to direct and indirect loss deduction maps to obtain a third runoff plot distribution model; the technical problem that in the prior art, non-standard runoff plot division of a power transmission and transformation project is unreasonable, and consequently water and soil loss prevention and control and project safety are insufficient is solved. The technical effects of accurately identifying and dividing the non-standard runoff plots of the power transmission and transformation project and improving water and soil loss prevention and control and project safety are achieved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a method and system for intelligent identification and modeling of non-standard runoff zones in power transmission and transformation projects. Background Technology

[0002] Power transmission and transformation projects are characterized by long lines, wide areas, complex terrain, and strong construction disturbances. They typically cross complex terrain areas such as mountains, hills, and valleys. Construction involves a large number of operations such as foundation excavation, tower foundation construction, road opening, and conductor erection. These operations frequently disturb the surface, change the landform structure and runoff path, and can easily lead to changes in the surface runoff collection characteristics, thus forming non-standard runoff phenomena. That is, the runoff formation and confluence process does not follow the traditional hydrological zoning rules, resulting in local confluence anomalies, concentrated soil erosion, and increased risk of slope instability.

[0003] In existing technologies, runoff plot delineation methods mostly rely on fixed topographic zoning or expert experience, focusing on static indicators such as slope, flow direction, and catchment area. They lack comprehensive consideration of dynamic construction disturbances and time-varying meteorological factors, resulting in the inability to accurately identify the distribution characteristics and variation patterns of non-standard runoff plots. This leads to problems such as unreasonable water and soil conservation design and prevention measures, which in turn affect engineering safety and ecological stability.

[0004] Existing technologies have technical problems such as unreasonable division of non-standard runoff zones in power transmission and transformation projects, which leads to insufficient soil and water conservation and project safety. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for intelligent identification and modeling of non-standard runoff zones in power transmission and transformation projects, in order to solve the technical problem that the unreasonable division of non-standard runoff zones in power transmission and transformation projects leads to insufficient soil and water conservation and engineering safety.

[0006] In view of the above problems, this application provides a method and system for intelligent identification and modeling of non-standard runoff zones in power transmission and transformation projects.

[0007] The first aspect of this application provides an intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects. The method includes: constructing an engineering area model by performing twin modeling based on real-time monitoring datasets of the power transmission and transformation project area; identifying and dividing non-standard runoff zones based on the engineering area model to obtain a first runoff zone distribution model; optimizing the water and soil erosion coupling features based on the first runoff zone distribution model to obtain a second runoff zone distribution model; performing multi-node direct water and soil erosion accident simulations on the second runoff zone distribution model based on the power transmission and transformation project construction plan and future meteorological datasets of the power transmission and transformation project area to obtain a first water and soil erosion simulation map; performing multi-node indirect water and soil erosion accident simulations on the second runoff zone distribution model based on the power transmission and transformation project construction plan and future meteorological datasets to obtain a second water and soil erosion simulation map; and performing multi-node global perturbation optimization on the second runoff zone distribution model based on the first and second water and soil erosion simulation maps to obtain a third runoff zone distribution model.

[0008] A second aspect of this application provides an intelligent identification and modeling system for non-standard runoff zones in power transmission and transformation projects. The system includes: an engineering area model construction module, used for performing twin modeling based on real-time monitoring datasets of the power transmission and transformation project area to construct an engineering area model; a first runoff zone distribution model acquisition module, used for identifying and dividing non-standard runoff zones based on the engineering area model to obtain a first runoff zone distribution model; a second runoff zone distribution model acquisition module, used for optimizing water and soil erosion coupling features based on the first runoff zone distribution model to obtain a second runoff zone distribution model; and a first water and soil erosion projection map acquisition module, used for obtaining water and soil erosion projection maps based on the data from the power transmission and transformation project area. The substation construction plan and future meteorological dataset are used to perform multi-node direct soil erosion accident simulation on the second runoff area distribution model to obtain a first soil erosion simulation map; the second soil erosion simulation map acquisition module is used to perform multi-node indirect soil erosion accident simulation on the second runoff area distribution model based on the substation construction plan and the future meteorological dataset to obtain a second soil erosion simulation map; the third runoff area distribution model acquisition module is used to perform multi-node global perturbation optimization on the second runoff area distribution model based on the first soil erosion simulation map and the second soil erosion simulation map to obtain a third runoff area distribution model.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: The method provided in this application embodiment constructs an engineering area model by performing twin modeling based on real-time monitoring datasets of the power transmission and transformation project area; identifies and divides non-standard runoff zones based on the engineering area model to obtain a first runoff zone distribution model; optimizes the water and soil erosion coupling characteristics based on the first runoff zone distribution model to obtain a second runoff zone distribution model; performs multi-node direct water and soil erosion accident simulation on the second runoff zone distribution model based on the power transmission and transformation project construction plan and future meteorological datasets to obtain a first water and soil erosion simulation map; performs multi-node indirect water and soil erosion accident simulation on the second runoff zone distribution model based on the power transmission and transformation project construction plan and future meteorological datasets to obtain a second water and soil erosion simulation map; and performs multi-node global perturbation optimization on the second runoff zone distribution model based on the first and second water and soil erosion simulation maps to obtain a third runoff zone distribution model. This achieves the technical effect of accurately identifying and dividing non-standard runoff zones in power transmission and transformation projects, improving water and soil erosion control and engineering safety.

[0010] The above description is merely an overview of the technical solution of this application. To enable a clearer understanding of the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0012] Figure 1 A flowchart illustrating the intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects provided in this application.

[0013] Figure 2 A schematic diagram of the structure of the intelligent identification and modeling system for non-standard runoff areas in power transmission and transformation projects provided in this application.

[0014] Explanation of reference numerals in the attached figures: Module 11 for building the engineering area model, Module 12 for obtaining the first runoff plot distribution model, Module 13 for obtaining the second runoff plot distribution model, Module 14 for obtaining the first soil and water loss projection map, Module 15 for obtaining the second soil and water loss projection map, and Module 16 for obtaining the third runoff plot distribution model. Detailed Implementation

[0015] This application provides a method and system for intelligent identification and modeling of non-standard runoff zones in power transmission and transformation projects. It addresses the technical problem of inadequate soil and water conservation and engineering safety caused by unreasonable division of non-standard runoff zones in existing technologies. The method achieves the technical effect of accurately identifying and dividing non-standard runoff zones in power transmission and transformation projects, thereby improving soil and water conservation and engineering safety.

[0016] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0017] Example 1, as Figure 1 As shown, this application provides a method for intelligent identification and modeling of non-standard runoff communities in power transmission and transformation projects. The method includes: Based on the real-time monitoring dataset of the power transmission and transformation project area, twin modeling is performed to construct a model of the project area.

[0018] Specifically, multiple monitoring devices deployed throughout the power transmission and transformation project area provide comprehensive real-time monitoring of the construction and operation areas. These devices monitor various dimensions of data, including topography, surface conditions, and hydrology. For example, lidar and total stations monitor topographic data such as elevation, slope, and aspect; soil moisture sensors monitor surface soil moisture, and a 3D laser scanner assesses surface roughness. High-resolution cameras mounted on drones capture images and analyze vegetation cover information to obtain surface data including soil moisture, surface roughness, and vegetation cover. Flow meters measure the flow velocity and volume of rivers or drainage channels in real time, and rain gauges measure rainfall to obtain hydrological parameters such as runoff and rainfall. The collected multidimensional data is cleaned using methods such as geographic coordinate unification, noise filtering, and interpolation correction to obtain a real-time monitoring dataset, which includes topographic, surface, and hydrological data.

[0019] Digital twin modeling, based on real-time monitoring datasets, refers to a method of digitally modeling the characteristics and data of power transmission and transformation engineering areas by mapping real-time monitoring datasets in a virtual space using digital twin technology. The core of this method lies in constructing a virtual model that is synchronized with the physical entity in real time. By integrating real-time data and physical mechanism models, a precise mapping is formed, supporting simulation analysis and decision optimization. Specifically, using topographic and surface data from real-time monitoring datasets, and based on data elevation and surface models, 3D Geographic Information System (GIS) and BIM technologies are used to generate 3D topographic and surface models of the engineering area. Using real-time hydrological monitoring data, a hydrological-geomorphological coupling model is employed to dynamically simulate surface runoff, soil infiltration, and precipitation, forming a 3D hydrological model. Spatiotemporal alignment is achieved through data collection timestamps and geographic coordinates. The 3D topographic and surface models are integrated with the 3D hydrological model, ultimately forming an engineering area model containing topographic, surface, and hydrological parameters on a 3D visualization platform.

[0020] Based on the engineering area model, non-standard runoff zones are identified and divided to obtain the first runoff zone distribution model.

[0021] Furthermore, based on the engineering area model, non-standard runoff plots are identified and divided to obtain a first runoff plot distribution model, including: identifying multi-scale soil erosion area features based on the engineering area model to obtain a first-scale regional feature sequence, a second-scale regional feature sequence, and a third-scale regional feature sequence; dividing the engineering area model into non-standard runoff plots based on the first-scale regional feature sequence to obtain a first runoff plot division model; dividing the engineering area model into non-standard runoff plots based on the second-scale regional feature sequence to obtain a second runoff plot division model; dividing the engineering area model into non-standard runoff plots based on the third-scale regional feature sequence to obtain a third runoff plot division model; performing multi-scale comparative analysis based on the first, second, and third runoff plot division models to obtain multi-scale division comparative analysis results; and globally collaboratively reconstructing the first, second, and third runoff plot division models based on the multi-scale division comparative analysis results to generate the first runoff plot distribution model.

[0022] Specifically, runoff plots are testing facilities used for quantitative research on the patterns of soil erosion on slopes and in small watersheds. They typically consist of a embankment, a plot enclosed by the embankment, a catchment channel, runoff and sediment collection and storage facilities, a protective belt, and a drainage system. Non-standard runoff plots refer to multiple runoff division units dynamically generated based on multi-dimensional characteristics such as actual topography, surface, and hydrological parameters, rather than relying on traditional division rules or fixed grid methods. The boundaries and sizes are dynamically adjusted based on environmental characteristics to more closely reflect actual soil erosion patterns. The identification and division of non-standard runoff plots involves dividing the power transmission and transformation project area into several regions with different soil erosion characteristics and runoff responses for refined management.

[0023] The specific process of identifying and dividing non-standard runoff plots based on the engineering area model first involves multi-scale soil erosion area feature identification based on the engineering area model. This multi-scale feature identification involves extracting and identifying soil erosion features at different spatial resolutions, including first, second, and third scales. The first scale is used to analyze the overall topography, water system, and soil type of the entire engineering area. The second scale is used to analyze medium-scale soil erosion features such as local slopes, gullies, and vegetation cover areas. The third scale is used to analyze micro-topographic changes and small-scale runoff paths. Using GIS spatial analysis tools, the slope, water system, soil type, and vegetation cover of the engineering area model are analyzed to obtain first-scale, second-scale, and third-scale regional feature sequences. The first-scale regional feature sequence identifies soil erosion features at a larger macro-spatial scale based on the engineering area model, analyzing the overall topographic slope distribution, water system direction, and overall soil type of the engineering area. The second-scale regional feature sequence identifies soil erosion characteristics at a medium-to-medium scale based on engineering regional models. For example, it analyzes the impact of mesoscopic features such as local slope, soil moisture, secondary gullies, and vegetation cover on runoff formation. The third-scale regional feature sequence identifies soil erosion characteristics at a smaller microscopic spatial scale. For example, it analyzes changes in micro-topography and local runoff paths. Each of the first, second, and third-scale regional feature sequences includes information such as spatial location, slope, hydrological properties, soil type, and vegetation cover.

[0024] Based on the first-scale, second-scale, and third-scale regional feature sequences, GIS spatial segmentation algorithms are used, such as the watershed algorithm based on slope and watershed, and clustering algorithms such as K-means. Segmentation conditions are also employed, such as separating high-slope and soil erosion-sensitive areas into separate sub-areas, forming sub-area boundaries based on the intersection of primary and secondary channels, and further subdividing sub-areas based on small-scale features. Non-standard runoff sub-areas are then divided into three models for the engineering area model, resulting in a first, second, and third runoff sub-area model. The first model generates large-scale sub-areas based on the first-scale feature sequence, possessing overall connectivity and watershed structure characteristics. The second model divides medium-scale sub-areas based on the second-scale feature sequence, using local slopes, secondary channels, and hydrological differences. The third model further subdivides sub-areas based on the third-scale sequence.

[0025] Cross-validation of three runoff plot division models (model 1, model 2, and model 3) was conducted through multi-scale comparative analysis. Plot boundaries, areas, and slopes were compared, boundary conflict zones were marked, and plots with inconsistent characteristics were identified. The differences in boundary locations and soil erosion sensitivity among the three models were analyzed to obtain multi-scale division comparison analysis results. Based on these results, the three runoff plot division models were globally and collaboratively reconstructed, adjusting plot boundaries to generate a first runoff plot distribution model containing multiple runoff plots. For example, a multi-objective optimization method was used to construct multi-objective optimization indices, including macroscopic connectivity, local feature preservation, and small-scale sensitivity. Macroscopic connectivity ensures hydrological watershed connectivity between plots, while local feature preservation preserves actual slope, soil type, gully, and vegetation cover differences. Constraints were set based on the multi-objective optimization indices, such as continuous and non-intersecting plot boundaries, separate division of highly sensitive areas, and consistency between the total area of ​​multiple plots and the actual engineering area. In the optimization process, the three-scale superimposed boundaries are used as the initial solution. The positions of the cell boundary points are fine-tuned through mutation operations, and the shape and area of ​​the cells are adjusted. The boundary features of the three scales are fused through cross operations to generate candidate schemes. The multi-objective fitness of each candidate scheme is calculated, the Pareto optimal solution is selected, and the optimization is continuously iterated until the boundary converges or the preset number of iterations is reached. This enables the global collaborative reconstruction of the first, second, and third runoff cell division models.

[0026] By identifying regional characteristics of soil erosion at multiple scales and dividing non-standard runoff plots, the accuracy of the division is improved. Then, by generating a first runoff plot distribution model through global collaborative reconstruction, the pertinence and reliability of soil erosion prediction and construction safety assessment are improved.

[0027] Furthermore, based on the first runoff plot distribution model, the water and soil erosion coupling characteristics are optimized to obtain the second runoff plot distribution model.

[0028] Furthermore, based on the first runoff plot distribution model, water and soil erosion coupling features are optimized to obtain a second runoff plot distribution model. This includes: identifying water and soil erosion features for each runoff plot within the first runoff plot distribution model to obtain multiple plot water and soil erosion features; performing pairwise coupling evaluations based on the multiple plot water and soil erosion features to obtain a water and soil erosion coupling evaluation sequence; based on the water and soil erosion coupling evaluation threshold, performing hypercoupling correlation analysis on the multiple plot water and soil erosion features according to the water and soil erosion coupling evaluation sequence to obtain a water and soil erosion hypercoupling relationship network; and performing boundary correction on the first runoff plot distribution model based on the water and soil erosion hypercoupling relationship network to generate the second runoff plot distribution model.

[0029] Specifically, soil erosion characteristics are extracted for each runoff plot in the first runoff plot distribution model, resulting in multiple plot soil erosion characteristics. These characteristics include, but are not limited to, slope, aspect, soil moisture, runoff volume, rainfall, and vegetation cover. Based on these multiple plot soil erosion characteristics, pairwise coupling evaluations are performed. Coupling evaluation refers to the correlation between different plots in the soil erosion process, such as the enhanced erosion caused by upstream runoff flowing into downstream areas, or the synergistic relationship between vegetation restoration and soil stabilization between neighboring plots. For example, correlation analysis methods such as Pearson correlation coefficient and cosine similarity are used to assess the coupling relationship of soil erosion characteristics between different runoff plots. Based on the results of multiple coupling evaluations, a soil erosion coupling evaluation sequence is formed, which includes the correlation strength between each runoff plot and other plots.

[0030] Based on actual needs and expert experience, a water and soil erosion coupling evaluation threshold is set to screen out water and soil erosion characteristics of multiple communities with high correlation and coupling relationships. According to the water and soil erosion coupling evaluation threshold, the water and soil erosion coupling evaluation sequence is subjected to hypercoupling correlation analysis to screen out coupling relationships higher than the water and soil erosion coupling evaluation threshold. A water and soil erosion hypercoupling relationship network is constructed based on the screened hypercoupling relationships. This network can be represented using graph theory methods, where nodes represent runoff communities and edge strength represents the strength of the coupling relationship. Based on the water and soil erosion hypercoupling relationship network, spatial analysis tools in GIS, such as buffer analysis and spatial overlay analysis, are used to correct the boundaries of the first runoff community distribution model, adjust the boundaries of multiple runoff communities, adjust or reconstruct the boundaries of highly coupled communities, and smooth the boundaries of weakly coupled communities, making the community division more consistent with the actual distribution of water and soil erosion characteristics. By introducing a water and soil erosion coupling mechanism to correct the boundary of the first runoff plot distribution model, a second runoff plot distribution model is formed. This corrected second runoff plot distribution model not only considers the geometric features of topography and geomorphology, but also more accurately reflects the actual changes in water and soil erosion characteristics, thereby improving the accuracy and reliability of runoff plot division.

[0031] Based on the power transmission and transformation project construction plan and future meteorological dataset for the power transmission and transformation project area, the second runoff distribution model is used to perform direct soil and water loss accident simulation under multiple nodes to obtain the first soil and water loss simulation map.

[0032] Furthermore, based on the power transmission and transformation project construction plan and future meteorological dataset for the power transmission and transformation project area, the second runoff area distribution model is used to perform multi-node direct soil erosion accident simulation to obtain a first soil erosion simulation map. This includes: dividing the power transmission and transformation project into multiple nodes based on the power transmission and transformation project construction plan to obtain multiple node project construction plans; sorting out the multi-node construction scene features based on the multiple node project construction plans and the future meteorological dataset to establish multiple node construction scene vectors; extracting the Sth runoff area model based on the second runoff area distribution model, where S is a positive integer; simulating construction on the Sth runoff area model based on the multiple node construction scene vectors to obtain multiple node area simulation data; performing iterative learning of distillation loss under equilibrium adversarial conditions based on the direct soil erosion event set to establish a direct soil erosion simulation channel; inputting the multiple node area simulation data into the direct soil erosion simulation channel to obtain multiple direct soil erosion simulation paths; organizing the multiple direct soil erosion simulation paths to obtain the direct soil erosion accident distribution of the Sth area, and adding the direct soil erosion accident distribution of the Sth area to the first soil erosion simulation map.

[0033] Specifically, based on the construction drawings and schedule of the power transmission and transformation project, the construction plan is divided into multiple construction nodes. Each node corresponds to a specific construction stage or area, such as foundation pit excavation, tower foundation construction, road hardening, and backfilling. This results in multiple node construction plans. Simultaneously, future meteorological data for the area is obtained from a meteorological service platform, forming a future meteorological dataset. This dataset includes rainfall, rainfall intensity, and wind speed. The construction plans for each node and the future meteorological data are then analyzed to identify multi-node construction scenario characteristics. This involves spatiotemporally overlaying the construction plans and corresponding future meteorological data to create multiple node construction scenario vectors. Each node construction scenario vector represents the construction scenario of a construction node under corresponding meteorological conditions.

[0034] From the second runoff cell distribution model, which includes multiple runoff cell models, the S-th runoff cell model (where S is a positive integer) is extracted. Using construction scenario vectors from multiple nodes as input, construction simulations are performed on the extracted S-th runoff cell model in a digital twin environment. These simulations simulate dynamic changes such as surface disturbance, runoff path alteration, soil loosening, and slope exposure, while also recording soil erosion during the simulation. Through physical modeling and data-driven approaches, cell response data for each node is obtained, generating multi-node cell simulation data.

[0035] Historical data on direct soil erosion events are collected, including a set of events directly caused by construction disturbances or local meteorological factors under similar landforms and construction conditions in the past. For example, during the construction of power transmission and transformation projects, phenomena such as erosion, increased runoff, and sediment transport caused by the excavation of tower foundations, earthwork filling, and the construction of construction roads, or by localized heavy rainfall directly affecting exposed surfaces, are recorded. The direct soil erosion event set records the time, intensity, location, and extent of soil erosion under direct disturbance conditions. Through sample balance adversarial learning, the direct soil erosion event set is balanced to ensure the balance of the training data. Iterative learning using a distillation loss function optimizes the performance of the inference model, establishing a direct soil erosion inference channel that can accurately and quickly output the soil erosion response path of a small community. Simulation data from multiple node communities are input into the direct soil erosion inference channel to generate multiple direct soil erosion inference paths. Each direct soil erosion inference path represents the spatial evolution process of runoff convergence, erosion intensification, and sediment migration within a community under specific construction nodes and corresponding meteorological conditions. Multiple direct soil erosion projection paths are spatiotemporally organized. For example, construction stages at different nodes may have continuous relationships. These projection paths are aligned according to construction sequence to form the direct soil erosion accident distribution in the Sth sub-area. This Sth sub-area direct soil erosion accident distribution is then overlaid onto the overall regional time-series map to generate the first soil erosion projection map. The second soil erosion projection map not only displays the predicted paths and extent of direct soil erosion but also reflects the intensity of soil erosion through color depth. For example, dark areas indicate very severe soil erosion, while light areas indicate relatively mild soil erosion. In other words, the first soil erosion projection map can intuitively display the potential soil erosion risk areas, impact range, and intensity of each sub-area under different construction nodes.

[0036] By simulating the construction process and combining it with future weather conditions, the timing of potential direct soil erosion is predicted, generating a first-stage soil erosion projection map. This improves the accuracy, comprehensiveness, and intuitiveness of soil erosion risk prediction. Furthermore, by predicting soil erosion incidents in advance, effective preventative measures can be taken to reduce the environmental impact of soil erosion and ensure the smooth progress of power transmission and transformation projects.

[0037] Furthermore, based on the direct soil erosion event set, an iterative learning process for distillation loss under equilibrium adversarial conditions is performed to establish a direct soil erosion projection channel. This includes: performing fault tree tracing based on the direct soil erosion event set to obtain a first soil erosion tracing distribution; training a first direct soil erosion projection model based on the first soil erosion tracing distribution; performing sample equilibrium evaluation based on the first soil erosion tracing distribution to obtain a first sample equilibrium coefficient; injecting equilibrium adversarial samples into the first soil erosion tracing distribution based on the first sample equilibrium coefficient and a sample equilibrium threshold to obtain a second soil erosion tracing distribution; training a second direct soil erosion projection model based on the second soil erosion tracing distribution; and performing iterative learning for distillation loss based on the first and second direct soil erosion projection models to generate the direct soil erosion projection channel.

[0038] Specifically, fault tree analysis is employed to conduct causal tracing analysis on a set of direct soil erosion events. Fault tree analysis is a top-down event logic reasoning method that starts from the outcome of a soil erosion event and traces its causes and processes. Specifically, the direct soil erosion event is taken as the top event and decomposed into several bottom-level contributing factors, such as heavy rainfall, construction disturbance, vegetation destruction, and excessive surface slope. A fault tree structure is constructed based on logical relationships. For each historical direct soil erosion event sample, based on meteorological data, construction progress records, and topographic data, the satisfaction status and contribution probability of each soil erosion event in the period leading up to the event are calculated. By matching the fault tree model, the causes and occurrence paths of each direct soil erosion event are obtained, generating the first soil erosion tracing distribution.

[0039] Based on the first soil erosion tracing distribution, a suitable machine learning model is selected, for example, a Long Short-Term Memory (LSTM) network is used to learn from the time-series data. The input features from the first soil erosion tracing distribution are used as input sample sequences in chronological order, including meteorological data, construction scenarios, and topographic surface data, to train the LTM network model. The LTM network model analyzes the causes and paths of the first soil erosion tracing distribution and inputs the predicted soil erosion path results. During the training phase, mean squared error is used as the loss function for optimization, and the trained LTM network model is validated using a validation set. Finally, a first direct soil erosion inference model is obtained, which can output the corresponding soil erosion response trend based on newly input construction and meteorological data.

[0040] The samples in the first soil erosion tracing distribution are classified and statistically analyzed. For example, they can be classified according to the intensity of soil erosion, including mild, moderate, and severe, or grouped according to event type, including rainfall-induced and construction-disrupted types. The number of samples in each category is counted, and the sample proportion and variance of each category are calculated. A sample equilibrium average is then calculated to form the first sample equilibrium coefficient, which reflects the degree of difference in the number of samples of different categories in the direct soil erosion event set. A larger equilibrium coefficient indicates a larger sample distribution deviation. A sample equilibrium threshold is set to determine whether the sample types in the soil erosion event set have reached a balanced state. When the first sample equilibrium coefficient exceeds the sample equilibrium threshold, a sample augmentation process is triggered to supplement minority samples. For example, new samples are generated by using generative adversarial networks or time series data augmentation methods. Specifically, the target data of minority class samples is calculated using the first sample equilibrium coefficient. The generative adversarial network is trained on the original time series samples and generates time series data that is consistent with the real distribution but has diverse sample attributes through adversarial learning. The rationality of the newly added samples is analyzed through sample consistency verification. The newly added minority class samples that meet the rationality criteria are injected into the first soil erosion tracing distribution to generate the second soil erosion tracing distribution. Similarly, based on the second distribution of soil erosion, a second direct soil erosion inference model is trained and generated, enabling the second direct soil erosion inference model to better learn the characteristics of minority events and further improve the prediction accuracy of the direct soil erosion inference model.

[0041] Finally, based on the first and second direct soil erosion inference models, iterative learning of distillation loss is performed. The second direct soil erosion inference model is used as the teacher model, and the first direct soil erosion inference model is used as the student model, trained using a knowledge distillation strategy. The second direct soil erosion inference model provides high-precision soft labels for events, i.e., the probability distribution of event occurrence. The first direct soil erosion inference model is optimized by minimizing a weighted combination of hard label loss and soft label distillation loss. Through multiple rounds of iterative training, the first direct soil erosion inference model gradually absorbs the higher-order feature representations of the second direct soil erosion inference model while maintaining its own efficiency, achieving a balance between model performance and direct soil erosion inference efficiency, and generating the direct soil erosion inference channel. By generating the direct soil erosion inference channel, the occurrence path and impact range of direct soil erosion accidents can be accurately and quickly predicted based on different construction scenarios and future weather conditions, providing accurate and reliable data support for the dynamic assessment of soil erosion risks in power transmission and transformation engineering construction.

[0042] Based on the power transmission and transformation project construction plan and the future meteorological dataset, the indirect soil erosion accident simulation under multiple nodes of the second runoff area distribution model is performed to obtain the second soil erosion simulation map.

[0043] Furthermore, based on the power transmission and transformation project construction plan and the future meteorological dataset, the second runoff distribution model is used to perform multi-node indirect soil erosion accident simulation to obtain a second soil erosion simulation map. This includes: performing fault tree tracing based on the indirect soil erosion event set to obtain a third soil erosion tracing distribution; performing iterative learning of distillation loss under equilibrium adversarial conditions based on the third soil erosion tracing distribution to establish an indirect soil erosion simulation channel; inputting the simulation data of the multiple node communities into the indirect soil erosion simulation channel to obtain multiple indirect soil erosion simulation paths; organizing the multiple indirect soil erosion simulation paths to obtain the indirect soil erosion accident distribution of the Sth community, and adding the indirect soil erosion accident distribution of the Sth community to the second soil erosion simulation map.

[0044] Specifically, historical data on indirect soil erosion events are collected to form an indirect soil erosion event set. This set refers to the collection of secondary or cascading soil erosion events triggered by direct disturbance events through hydrological, topographical, or soil conduction. For example, upstream construction leading to increased runoff or sediment output can affect downstream areas, causing erosion or slope instability, thus creating indirect soil erosion events. Using fault tree analysis, starting from the outcomes of these indirect soil erosion events, the causes and processes of these events are traced to generate a third soil erosion tracing distribution, including the causal relationships and occurrence paths of these events. Similarly, based on this third soil erosion tracing distribution, a first indirect soil erosion inference model is trained. A sample balance evaluation is then performed based on the third soil erosion tracing distribution. Based on the evaluation results and the sample balance threshold, adversarial sample injection is applied to the third soil erosion tracing distribution to increase minority sample data, forming a fourth soil erosion tracing distribution. Finally, a second indirect soil erosion inference model is trained based on this fourth distribution, using iterative learning through distillation loss to construct an indirect soil erosion inference pathway.

[0045] Simulated data from multiple node communities are input into an indirect soil erosion projection channel. This channel analyzes the input data and outputs multiple indirect soil erosion projection paths. These paths are then processed to obtain the distribution of indirect soil erosion events in the S-th community, including the extent, intensity, and potential impact areas of soil erosion. This distribution is then added to a second soil erosion projection map. The second map not only displays the path range of indirect soil erosion but also uses color depth to reflect the intensity of soil erosion; for example, dark areas indicate very severe erosion, while light areas indicate relatively mild erosion. By analyzing the set of indirect soil erosion events and constructing the second soil erosion projection map, the comprehensiveness and reliability of soil erosion risk prediction are further improved. This optimizes the construction plan and protective measures for power transmission and transformation projects, reduces the environmental impact of soil erosion, and ensures the safety and ecological stability of power transmission and transformation projects.

[0046] Based on the first and second soil erosion projection maps, the second runoff plot distribution model is optimized with global perturbation under multiple nodes to obtain the third runoff plot distribution model.

[0047] Furthermore, based on the first and second soil erosion projection maps, the second runoff plot distribution model is subjected to multi-node global perturbation optimization to obtain a third runoff plot distribution model. This includes: performing multi-node perturbation impact analysis on each runoff plot model within the second runoff plot distribution model based on the first soil erosion projection map to obtain a first plot of plot perturbation analysis; performing multi-node perturbation impact analysis on each runoff plot model based on the second soil erosion projection map to obtain a second plot of plot perturbation analysis; performing multi-node perturbation impact coupling analysis on each runoff plot model based on the first and second plots of plot perturbation analysis to obtain a perturbation impact coupling distribution; performing multi-node collaborative optimization on the first and second plots of plot perturbation analysis based on the perturbation impact coupling distribution to obtain a third plot of plot perturbation analysis; and performing multi-node global optimization on each runoff plot model based on the third plot of plot perturbation analysis to generate the third runoff plot distribution model.

[0048] Specifically, using overlay analysis and calculation with a geographic information system, the spatial data of multiple runoff plot models within the second runoff plot distribution model are spatially registered with the first soil erosion projection map. Multi-node disturbance impact analysis is performed on multiple runoff plot models. Using a digital twin model, simulations are conducted to obtain the sensitivity and response intensity of external disturbances, including excavation, backfilling, and changes in drainage, to the soil erosion state within each runoff plot under different construction nodes and meteorological conditions. This includes erosion intensity, loss range, response time, and cumulative effects, generating a first plot disturbance analysis map reflecting the disturbance distribution under direct soil erosion conditions. Similarly, based on the second soil erosion projection map, multi-node disturbance impact analysis is performed on each runoff plot model within the second runoff plot distribution model, generating a second plot disturbance analysis map reflecting the disturbance distribution under indirect soil erosion conditions.

[0049] Based on the first and second disturbance analysis maps of each cell, a multi-node disturbance influence coupling analysis is performed on the runoff cell model. For example, using an additive algorithm, weights are set according to factors such as the intensity, range, and duration of the disturbance. Through weighted calculation, the temporal and spatial interactions of direct and indirect disturbances in each cell are identified, obtaining the disturbance influence coupling distribution. This distribution reflects which cells exhibit synergistic enhancement or mutual weakening under multi-node conditions. Based on the disturbance influence coupling distribution, an optimization algorithm, such as a genetic algorithm or particle swarm optimization algorithm, is used to perform multi-node collaborative optimization on the first and second disturbance analysis maps of the cells, generating an optimized third disturbance analysis map of the cells. For example, optimization objectives are set, such as minimizing the intensity of inter-cell disturbance risk, balancing the upstream and downstream disturbance transmission effects, and improving the stability and connectivity of cell division. The boundary position, area ratio, and disturbance response of the cells are used as adjustable variables. The optimization algorithm is used to search and iteratively update under multi-node disturbance conditions. The particle swarm optimization algorithm is used to simulate the position movement of the population in multi-dimensional space. A global search is performed based on the gradient of the disturbance risk distribution. After multiple iterations, the solution with the minimum overall disturbance risk and stable spatial structure is output, forming the third analytical map of cell disturbance.

[0050] Finally, based on the third analysis map of cell disturbance, a global optimization algorithm, such as the simulated degradation algorithm, is used to perform multi-node optimization on each runoff cell model. Taking the third analysis map of cell disturbance as the initial state, new candidate partitioning schemes are generated by randomly disturbing the cell boundary positions or shapes. The overall disturbance intensity change after adjustment is calculated, and candidate partitioning schemes are selected to be retained or discarded according to the acceptance criteria of simulated annealing. Through multiple iterations, the global optimal solution is gradually approached, generating the third runoff cell distribution model. The third runoff cell distribution model comprehensively considers the impact of direct and indirect soil erosion, and can more accurately reflect the distribution of soil erosion risk in the power transmission and transformation project area. It provides comprehensive and reliable data support for the construction of power transmission and transformation stations, improves the effectiveness and rationality of risk prevention and control in the construction area of ​​power transmission and transformation stations, reduces soil erosion, and ensures project safety and ecological stability.

[0051] Furthermore, twin modeling is performed based on the real-time monitoring dataset of the power transmission and transformation project area, including: real-time monitoring of the power transmission and transformation project area to obtain a regional dataset; and data cleaning based on the regional dataset to obtain the real-time monitoring dataset.

[0052] Specifically, multiple monitoring devices deployed within the project area are used to monitor the power transmission and transformation project in real time, obtaining multi-dimensional data including topographic, surface, and hydrological parameters to form a regional dataset. The collected regional data then undergoes data cleaning, and coordinate transformation and projection are used to unify the geographic coordinates of the regional data, unifying data from different monitoring devices into a unified geographic coordinate system to ensure the spatial accuracy and consistency of the multi-dimensional data. Statistical analysis is then used to identify and remove noise and outliers from the regional dataset. Missing values ​​are corrected using interpolation methods such as linear interpolation and Kriging interpolation to fill data gaps, thus obtaining a real-time monitoring dataset. Data cleaning improves the accuracy and completeness of the real-time monitoring dataset, thereby ensuring the accuracy and reliability of the constructed project area model.

[0053] Furthermore, constructing an engineering area model includes: performing deviation detection on the engineering area model based on the real-time monitoring dataset to obtain modeling deviation detection results; and performing feedback optimization on the engineering area model based on the modeling deviation detection results.

[0054] Specifically, after the engineering area model is constructed, it is run, and a deviation detection mechanism is used to monitor deviations in the model based on real-time monitoring datasets. The real-time monitoring datasets and the running data of the engineering area model are compared and analyzed to calculate the deviations between multiple real-time monitoring data and their corresponding running data. Based on these deviations, a modeling deviation detection result is generated. When deviations are found, the engineering area model is optimized based on the modeling deviation detection result, adjusting its parameters. Deviation detection is then performed again after optimization until the engineering area model accurately reflects the actual data information of the power transmission and transformation engineering area. By detecting and providing feedback on the engineering area model, its accuracy and reliability are improved, thus enabling it to accurately reflect the actual situation of the power transmission and transformation engineering area.

[0055] Example 2, based on the same inventive concept as the intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects in the previous examples, such as... Figure 2 As shown, this application provides an intelligent identification and modeling system for non-standard runoff communities in power transmission and transformation projects. The system includes: The engineering area model construction module 11 is used to construct an engineering area model by performing twin modeling based on the real-time monitoring dataset of the power transmission and transformation engineering area; the first runoff cell distribution model acquisition module 12 is used to identify and classify non-standard runoff cells based on the engineering area model to obtain a first runoff cell distribution model; the second runoff cell distribution model acquisition module 13 is used to optimize the water and soil erosion coupling characteristics based on the first runoff cell distribution model to obtain a second runoff cell distribution model; the first water and soil erosion projection map acquisition module 14 is used to obtain a first water and soil erosion projection map based on the power transmission and transformation engineering construction plan and future meteorological dataset of the power transmission and transformation engineering area. The second runoff plot distribution model is used to perform direct soil erosion accident simulation under multiple nodes to obtain a first soil erosion simulation map; the second soil erosion simulation map acquisition module 15 is used to perform indirect soil erosion accident simulation under multiple nodes on the second runoff plot distribution model based on the power transmission and transformation project construction plan and the future meteorological dataset to obtain a second soil erosion simulation map; the third runoff plot distribution model acquisition module 16 is used to perform global disturbance optimization under multiple nodes on the second runoff plot distribution model based on the first soil erosion simulation map and the second soil erosion simulation map to obtain a third runoff plot distribution model.

[0056] Furthermore, the first runoff plot distribution model acquisition module 12 is also used for: identifying multi-scale soil erosion area features based on the engineering area model to obtain a first-scale regional feature sequence, a second-scale regional feature sequence, and a third-scale regional feature sequence; dividing the engineering area model into non-standard runoff plots based on the first-scale regional feature sequence to obtain a first runoff plot division model; dividing the engineering area model into non-standard runoff plots based on the second-scale regional feature sequence to obtain a second runoff plot division model; dividing the engineering area model into non-standard runoff plots based on the third-scale regional feature sequence to obtain a third runoff plot division model; performing multi-scale comparative analysis based on the first runoff plot division model, the second runoff plot division model, and the third runoff plot division model to obtain multi-scale division comparative analysis results; and performing global collaborative reconstruction of the first runoff plot division model, the second runoff plot division model, and the third runoff plot division model based on the multi-scale division comparative analysis results to generate the first runoff plot distribution model.

[0057] Furthermore, the first runoff plot distribution model acquisition module 12 is also used to: identify soil erosion characteristics for each runoff plot in the first runoff plot distribution model to obtain soil erosion characteristics for multiple plots; perform pairwise coupling evaluation based on the soil erosion characteristics of the multiple plots to obtain a soil erosion coupling evaluation sequence; based on the soil erosion coupling evaluation threshold, perform hypercoupling correlation analysis on the soil erosion characteristics of the multiple plots according to the soil erosion coupling evaluation sequence to obtain a soil erosion hypercoupling relationship network; and perform boundary correction on the first runoff plot distribution model according to the soil erosion hypercoupling relationship network to generate the second runoff plot distribution model.

[0058] Furthermore, the first soil erosion projection map acquisition module 14 is also used for: dividing the power transmission and transformation project into multiple nodes according to the construction plan, and obtaining multiple node project construction plans; sorting out the features of multiple node construction scenarios according to the multiple node project construction plans and the future meteorological dataset, and establishing multiple node construction scenario vectors; extracting the Sth runoff area model according to the second runoff area distribution model, where S is a positive integer; simulating construction on the Sth runoff area model based on the multiple node construction scenario vectors, and obtaining multiple node area simulation data; performing iterative learning of distillation loss under equilibrium adversarial conditions according to the direct soil erosion event set, and establishing a direct soil erosion projection channel; inputting the multiple node area simulation data into the direct soil erosion projection channel, and obtaining multiple direct soil erosion projection paths; organizing the multiple direct soil erosion projection paths, obtaining the direct soil erosion accident distribution of the Sth area, and adding the direct soil erosion accident distribution of the Sth area to the first soil erosion projection map.

[0059] Furthermore, the first soil erosion projection map acquisition module 14 is also used for: performing fault tree tracing based on the direct soil erosion event set to obtain a first soil erosion tracing distribution; training a first direct soil erosion projection model based on the first soil erosion tracing distribution; performing sample balance evaluation based on the first soil erosion tracing distribution to obtain a first sample balance coefficient; injecting balanced adversarial samples into the first soil erosion tracing distribution based on the first sample balance coefficient and a sample balance threshold to obtain a second soil erosion tracing distribution; training a second direct soil erosion projection model based on the second soil erosion tracing distribution; and performing distillation loss iterative learning based on the first direct soil erosion projection model and the second direct soil erosion projection model to generate the direct soil erosion projection channel.

[0060] Furthermore, the second soil erosion projection map acquisition module 15 is also used for: performing accident tree tracing based on the indirect soil erosion event set to obtain a third soil erosion tracing distribution; performing iterative learning of distillation loss under equilibrium adversarial conditions based on the third soil erosion tracing distribution to establish an indirect soil erosion projection channel; inputting the simulation data of the multiple node communities into the indirect soil erosion projection channel to obtain multiple indirect soil erosion projection paths; organizing the multiple indirect soil erosion projection paths to obtain the indirect soil erosion accident distribution of the Sth community, and adding the indirect soil erosion accident distribution of the Sth community to the second soil erosion projection map.

[0061] Furthermore, the third runoff plot distribution model acquisition module 16 is also used for: performing multi-node disturbance impact analysis on each runoff plot model within the second runoff plot distribution model based on the first soil erosion projection map to obtain a first plot of plot disturbance analysis; performing multi-node disturbance impact analysis on each runoff plot model based on the second soil erosion projection map to obtain a second plot of plot disturbance analysis; performing multi-node disturbance impact coupling analysis on each runoff plot model based on the first and second plots of plot disturbance analysis to obtain a coupled distribution of disturbance impact; performing multi-node collaborative optimization on the first and second plots of plot disturbance analysis based on the coupled distribution of disturbance impact to obtain a third plot of plot disturbance analysis; and performing multi-node global optimization on each runoff plot model based on the third plot of plot disturbance analysis to generate the third runoff plot distribution model.

[0062] Furthermore, the engineering area model construction module 11 is also used to: perform real-time monitoring of the power transmission and transformation engineering area to obtain an area dataset; and perform data cleaning based on the area dataset to obtain the real-time monitoring dataset.

[0063] Furthermore, the engineering area model construction module 11 is also used to: perform deviation detection on the engineering area model based on the real-time monitoring dataset to obtain modeling deviation detection results; and perform feedback optimization on the engineering area model based on the modeling deviation detection results.

[0064] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0065] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for intelligent identification and modeling of non-standard runoff zones in power transmission and transformation projects, characterized in that, include: Based on the real-time monitoring dataset of the power transmission and transformation project area, twin modeling is performed to construct a model of the project area; Based on the engineering area model, non-standard runoff zones are identified and divided to obtain the first runoff zone distribution model. Based on the first runoff plot distribution model, the water and soil erosion coupling characteristics are optimized to obtain the second runoff plot distribution model; Based on the power transmission and transformation project construction plan and future meteorological dataset of the power transmission and transformation project area, the second runoff area distribution model is used to perform direct soil and water loss accident simulation under multiple nodes to obtain the first soil and water loss simulation map. Based on the power transmission and transformation project construction plan and the future meteorological dataset, the indirect soil erosion accident simulation under multiple nodes is performed on the second runoff area distribution model to obtain the second soil erosion simulation map. Based on the first and second soil erosion projection maps, the second runoff plot distribution model is optimized with global perturbation under multiple nodes to obtain the third runoff plot distribution model.

2. The intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects as described in claim 1, characterized in that, Based on the engineering area model, non-standard runoff zones are identified and divided to obtain the first runoff zone distribution model, including: Based on the engineering area model, multi-scale soil erosion area feature identification is performed to obtain the first-scale area feature sequence, the second-scale area feature sequence, and the third-scale area feature sequence. Based on the first-scale regional feature sequence, the engineering area model is divided into non-standard runoff zones to obtain the first runoff zone division model. Based on the second-scale regional feature sequence, the engineering area model is divided into non-standard runoff zones to obtain a second runoff zone division model. Based on the third-scale regional feature sequence, the engineering area model is divided into non-standard runoff plots to obtain a third runoff plot division model. Multi-scale comparative analysis was performed based on the first model, the second model, and the third model of runoff plot division to obtain the results of the multi-scale division comparative analysis. Based on the results of the multi-scale partitioning comparison analysis, the first model of runoff area partitioning, the second model of runoff area partitioning, and the third model of runoff area partitioning are globally reconstructed collaboratively to generate the first runoff area distribution model.

3. The intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects as described in claim 1, characterized in that, Based on the first runoff plot distribution model, water and soil erosion coupling characteristics are optimized to obtain a second runoff plot distribution model, including: Based on the soil and water loss characteristics of each runoff plot in the first runoff plot distribution model, the soil and water loss characteristics of multiple plots are obtained. Based on the soil and water loss characteristics of the multiple communities, a pairwise coupled evaluation was conducted to obtain a soil and water loss coupled evaluation sequence. Based on the water and soil loss coupling evaluation threshold, the water and soil loss characteristics of the multiple communities are sorted out by super-coupling association according to the water and soil loss coupling evaluation sequence to obtain the water and soil loss super-coupling relationship network. The boundary of the first runoff plot distribution model is corrected based on the water and soil erosion hypercoupled relationship network to generate the second runoff plot distribution model.

4. The intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects as described in claim 1, characterized in that, Based on the power transmission and transformation project construction plan and future meteorological dataset for the aforementioned power transmission and transformation project area, a multi-node direct soil erosion accident simulation is performed on the second runoff distribution model to obtain a first soil erosion simulation map, including: Based on the power transmission and transformation project construction plan, multiple nodes are divided to obtain multiple node project construction plans; Based on the construction plans for the multiple nodes and the future meteorological dataset, the characteristics of the multi-node construction scenarios are sorted out, and multi-node construction scenario vectors are established. Based on the second runoff cell distribution model, the Sth runoff cell model is extracted, where S is a positive integer; Based on the construction scenario vectors of the multiple nodes, the construction of the Sth runoff community model is simulated to obtain simulation data of the multiple node communities. Based on the set of direct soil erosion events, an iterative learning of distillation loss under equilibrium adversarial conditions is conducted to establish a direct soil erosion projection channel. The simulation data of the multiple node communities are input into the direct soil and water loss inference channel to obtain multiple direct soil and water loss inference paths. The multiple direct soil erosion projection paths are organized to obtain the distribution of direct soil erosion accidents in the Sth district, and the distribution of direct soil erosion accidents in the Sth district is added to the first soil erosion projection map.

5. The intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects as described in claim 4, characterized in that, Based on the set of direct soil erosion events, iterative learning of distillation loss under equilibrium adversarial conditions is performed to establish a direct soil erosion projection channel, including: Based on the set of direct soil erosion events, fault tree tracing is performed to obtain the first soil erosion tracing distribution. Based on the first soil erosion tracing distribution, train the first direct soil erosion inference model; Based on the first soil and water loss tracing distribution, a sample balance evaluation is performed to obtain the first sample balance coefficient. Based on the first sample equilibrium coefficient, the first soil and water loss tracing distribution is subjected to equilibrium adversarial sample injection according to the sample equilibrium threshold to obtain the second soil and water loss tracing distribution. Based on the second soil erosion tracing distribution, train the second direct soil erosion inference model; Based on the first direct soil and water loss projection model and the second direct soil and water loss projection model, distillation loss iterative learning is performed to generate the direct soil and water loss projection channel.

6. The intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects as described in claim 4, characterized in that, Based on the power transmission and transformation project construction plan and the future meteorological dataset, the indirect soil erosion accident simulation under multi-node conditions is performed on the second runoff distribution model to obtain the second soil erosion simulation map, including: Based on the indirect soil erosion event set, fault tree tracing is performed to obtain the third soil erosion tracing distribution; Based on the third soil and water loss tracing distribution, an iterative learning of distillation loss under equilibrium countermeasure is performed to establish an indirect soil and water loss inference channel. The simulation data of the multiple node communities are input into the indirect soil erosion inference channel to obtain multiple indirect soil erosion inference paths; The multiple indirect soil erosion projection paths are organized to obtain the distribution of indirect soil erosion accidents in the Sth sub-area, and the distribution of indirect soil erosion accidents in the Sth sub-area is added to the second soil erosion projection map.

7. The intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects as described in claim 1, characterized in that, Based on the first and second soil erosion projection maps, the second runoff plot distribution model is optimized using multi-node global perturbation to obtain the third runoff plot distribution model, including: Based on the first soil and water loss projection map, the multi-node disturbance impact analysis of each runoff plot model in the second runoff plot distribution model is performed to obtain the first plot of plot disturbance analysis. Based on the second soil and water loss projection map, the multi-node disturbance impact analysis of each runoff plot model is performed to obtain the second plot of plot disturbance analysis. Based on the first and second disturbance analysis maps of the cell, multi-node disturbance influence coupling analysis is performed on the runoff cell models to obtain the disturbance influence coupling distribution; Based on the coupled distribution of the disturbance impact, multi-node collaborative optimization is performed on the first and second analytical maps of the cell disturbance to obtain the third analytical map of the cell disturbance. Based on the third spectrum of the cell disturbance analysis, the multi-node global optimization of each runoff cell model is performed to generate the third runoff cell distribution model.

8. The intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects as described in claim 1, characterized in that, Based on the real-time monitoring dataset of the power transmission and transformation project area, twin modeling was performed, including: Real-time monitoring of the power transmission and transformation project area was conducted to obtain a regional dataset; Data cleaning is performed on the regional dataset to obtain the real-time monitoring dataset.

9. The intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects as described in claim 1, characterized in that, Constructing the engineering area model includes: Based on the real-time monitoring dataset, deviation detection is performed on the engineering area model to obtain the modeling deviation detection results; The engineering area model is optimized based on the modeling deviation detection results.

10. A smart identification and modeling system for non-standard runoff zones in power transmission and transformation projects, characterized in that: The steps for implementing the intelligent identification and modeling method for non-standard runoff zones in power transmission and transformation projects according to any one of claims 1 to 9 include: The engineering area model building module is used to perform twin modeling based on the real-time monitoring dataset of the power transmission and transformation engineering area and build the engineering area model. The first runoff cell distribution model acquisition module is used to identify and divide non-standard runoff cells based on the engineering area model to obtain the first runoff cell distribution model. The second runoff plot distribution model acquisition module is used to optimize the water and soil erosion coupling characteristics based on the first runoff plot distribution model to obtain the second runoff plot distribution model. The first soil erosion projection map acquisition module is used to perform direct soil erosion accident projection on the second runoff area distribution model under multiple nodes based on the power transmission and transformation project construction plan and future meteorological dataset in the power transmission and transformation project area, and to obtain the first soil erosion projection map. The second soil erosion projection map acquisition module is used to perform indirect soil erosion accident projection under multiple nodes on the second runoff area distribution model based on the power transmission and transformation project construction plan and the future meteorological dataset, and obtain the second soil erosion projection map. The module for obtaining the third runoff plot distribution model is used to perform global perturbation optimization of the second runoff plot distribution model under multiple nodes based on the first soil erosion projection map and the second soil erosion projection map, so as to obtain the third runoff plot distribution model.

Images