Intelligent line planning and fault early warning method suitable for electric power engineering design

Abstract

The invention discloses a line intelligent planning and fault early warning method suitable for electric power engineering design, and relates to the technical field of electric power engineering. The method comprises the following steps: quantifying geographic information, geological conditions and construction cost into a continuous decision curved surface through spatial semantic segmentation and a cost curved surface model, and identifying a large-area low-cost region as a primary feasible region in combination with a dynamic clustering algorithm; a multi-dimensional decision index is constructed based on spatial features, geological stability and facility association data, fault regions such as terrain abrupt change or facility conflict are identified through real-time association coefficients, and an exception handling mechanism is triggered; and selecting a seed region with an optimal construction condition as an anchor point, constructing a joint probability model of spatial features and cost, and generating an optimal path considering feasibility probability and cost effectiveness in combination with an expectation maximization algorithm and a graph optimization technology. According to the invention, scientificity and economical efficiency of cable planning under complex geological conditions are improved, and key technical support is provided for intelligentization of electric power engineering.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric power engineering, and in particular relates to a line intelligent planning and fault early warning method applicable to electric power engineering design. Background Art

[0002] In the field of power engineering, underground cable line planning and fault warning have long faced the following technical bottlenecks:

[0003] Traditional planning methods are crude and inefficient: They rely on manual experience or single-source GIS analysis, making it difficult to comprehensively consider multiple factors such as topography, geology, and underground facility conflicts. This leads to highly subjective path selection and makes it difficult to quantify construction risks and costs. Static zoning strategies cannot dynamically adapt to changing project conditions. For example, when new underground facilities are added during construction or geological survey data is updated, re-surveying and planning are required, resulting in low efficiency.

[0004] Insufficient multi-source data fusion capabilities: Geographic information, geological exploration, facility distribution, and other data are typically stored independently, lacking semantic associations and in-depth mining. For example, traditional methods may only use geological data to avoid undesirable areas without establishing a quantitative correlation with construction costs and maintenance difficulties. Existing semantic segmentation technologies (such as remote sensing image classification) can only achieve coarse regional divisions and cannot accurately identify mixed "partially suitable / partially unsuitable" areas, resulting in conservative or aggressive classification of feasible areas.

[0005] Lack of a dynamic risk warning mechanism: Route planning is disconnected from fault warning. Existing technologies primarily focus on post-construction monitoring, making it difficult to predict construction risks in areas with sudden changes in terrain or densely populated facilities during the planning phase. This leads to high costs for rerouting later. Traditional decision-making models (such as cost-benefit analysis) consider only a single metric and ignore the coupled effects of multiple factors (such as the interaction between geological stability and facility distance), easily overlooking hidden risks.

[0006] Poor adaptability of optimization algorithms: Path search algorithms lack real-time feedback mechanisms. Changing construction conditions require recalculating the global path, preventing local optimization. Probabilistic model parameters (such as feasibility weights) are often preset with fixed values, making them difficult to adapt to regional characteristics. This results in optimization results that deviate from actual requirements.

[0007] To address these challenges, the present invention proposes the following solutions: Intelligent Multimodal Data Fusion: Through spatial semantic segmentation and cost surface models, discrete multi-source data (GIS, remote sensing, geological reports) is transformed into continuous decision surfaces, achieving a transition from "data silos" to "collaborative analysis." Dynamic Clustering and Weight Adaptation: An iterative clustering weight adjustment mechanism is introduced to automatically identify large, low-cost areas. Regional coordination is quantified using correlation coefficients, addressing the problem of traditional clustering algorithms easily falling into local optimality. Real-time Risk Warning Embedded Planning: The difference in decision indicators between adjacent sub-regions is converted into correlation coefficients, identifying fault areas during the planning phase and triggering exception handling processes (such as data review or model adjustment) to achieve risk feedforward control. Dual Probability Model Coupling with Graph Optimization: A joint probability model of spatial features and costs is constructed, and the correlation coefficient is integrated into the graph search algorithm through edge weight design. This ensures global optimality while improving local adaptability. Through a data-driven intelligent decision-making system, this solution significantly improves the scientific and economic efficiency of cable route planning under complex geological conditions, providing key technical support for the intelligent transformation of power engineering. Summary of the Invention

[0008] In order to overcome the shortcomings and deficiencies of the above-mentioned prior art, the present invention adopts the following technical solutions:

[0009] The intelligent line planning and fault warning method applicable to power engineering design has the following process:

[0010] Step 1: Preliminary division of feasible areas: S11. Obtain the underground pipeline network area and divide it into multiple adjacent sub-areas, and collect spatial feature data, geological feature data and facility-related data of each sub-area; S12. Construct a spatial semantic segmentation model and a cost surface model, wherein the spatial semantic segmentation model divides the area into areas suitable for laying cables, areas not suitable for laying cables and areas with existing underground facilities and encodes them, and the cost surface model calculates the cost weight according to the terrain, geology and facility distance; S13. Fusion of spatial semantic segmentation results, cost weights and spatial position coordinates forms a comprehensive feature vector, clusters and adjusts it through Gaussian mixture model to obtain a primary feasible area; S14. Calculate the spatial feature decision index, geological feature decision index and facility-related decision index of each sub-area; S15. Calculate the real-time correlation coefficient based on the decision index of adjacent sub-areas, generate a fusion decision index and perform exception processing;

[0011] Step 2: Finely determine feasible cable channels: S21: Select seed areas with good construction conditions, low costs, and low risks from the primary feasible areas;

[0012] S31. Construct a feasibility probability model based on spatial features and a cost probability model based on cost; S41-S42. Combine the two probability models to obtain a comprehensive probability value, use the expectation maximization algorithm to optimize, and determine the final feasible cable channel.

[0013] Preferably, the spatial semantic segmentation model in S12 adopts a deep learning model to encode the sub-region semantic labels as: area suitable for laying cables 1, area not suitable for laying cables 0, and area with existing underground facilities 2; the cost weight formula is: Among them, S slope is the normalized slope value; S geology S is the geological stability score, geology ∈[1,10]; D facility is the facility distance; (x, y) is the spatial location coordinate; α, β, γ are the preset weights.

[0014] Preferably, the cluster adjustment step in S13 includes:

[0015] Calculate the average cost weight of all sub-regions in the cluster, Among them, λ1 and λ2 are adjustment coefficients used to balance the importance of area and cost;

[0016] Set the weight threshold to divide the large clusters and small clusters, assign the small cluster sub-areas to the large cluster with the smallest distance and iterate again.

[0017] Preferably, the decision indicator calculation in S14 includes:

[0018] Spatial feature decision calculation: The first decision indicator corresponding to the sub-region is obtained based on the spatial feature data and the spatial feature decision model. The spatial feature decision model is: Among them, I1 is the first decision indicator; S t is the spatial eigenvalue at time t; S targe is the target space eigenvalue;

[0019] Geological feature decision index: The second decision index corresponding to the sub-region is obtained based on the geological feature data and the geological feature decision model. The geological feature decision model is: Among them, I2 is the second decision indicator; G t is the geological characteristic value at time t; G targe is the target geological characteristic value;

[0020] Facility association decision index: The third decision index corresponding to the sub-region is obtained based on the facility association data and the facility association decision model. The facility association decision model is: Among them, I3 is the third decision indicator; F t is the facility association value at time t; F targe Associate a value with the target facility.

[0021] Preferably, the real-time correlation coefficient in S15 is calculated as:

[0022] A first difference ΔI1 is obtained according to a first decision metric of the adjacent sub-regions, a second difference ΔI2 is obtained according to a second decision metric of the adjacent sub-regions, and a third difference ΔI3 is obtained according to a third decision metric of the adjacent sub-regions. The first difference ΔI1, the second difference ΔI2, and the third difference ΔI3 are all normalized, ΔI1∈[0,1], ΔI2∈0,1], and ΔI3∈[0,1]. ΔI includes ΔI1, ΔI2 and ΔI3;

[0023] The real-time correlation coefficient R is obtained based on the correlation model R=γ1ΔI1+γ2ΔI2+γ3ΔI3, where γ1, γ2, and γ3 are adjustment coefficients. The closer the correlation coefficient R is to 1, the more coordinated the engineering properties of adjacent sub-areas are.

[0024] Preferably, the seed area selection step in S21 includes: screening sub-areas with a semantic label of 1, performing K-means clustering based on the comprehensive score of spatial features, geological features, cost weights and facility associations, extracting the cluster center as the seed area, and the comprehensive score taking into account the normalized decision index and cost weights.

[0025] Preferably, the probability model based on spatial features in S31 takes into account terrain slope, geological stability and facility distance, and the probability model based on cost takes into account construction cost and maintenance cost, and the model parameters are determined by historical data or expert weights.

[0026] Preferably, the comprehensive probability value in S41 is calculated as: P total =ω s P s i +μ e P e i Among them, P s i is the feasibility probability of sub-region i; P e i is the cost probability of subregion i, ω s is the weight of the spatial feature, μ e is the weight of the cost feature, and ω s +μ e =1.

[0027] Preferably, the expectation maximization algorithm optimization step in S42 includes:

[0028] A graph model is constructed with sub-regions as nodes. The edge weights of adjacent sub-regions are positively correlated with the association coefficient. The Dijkstra algorithm or A* algorithm is used to solve the minimum total weight path. The weights are iteratively adjusted until convergence, and the final feasible cable channel is output.

[0029] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0030] 1. This invention achieves a shift from a traditional experience-driven approach to a data-driven approach by introducing a deep learning model for spatial semantic segmentation and combining it with a cost surface model to quantitatively assess geographic, geological, and construction cost factors. This method automatically identifies areas suitable for cable laying, eliminating the subjectivity and inefficiency inherent in human judgment and significantly enhancing the intelligence and scientific nature of power engineering design.

[0031] 2. After initially demarcating feasible areas, the present invention further integrates and analyzes sub-areas using multi-dimensional decision-making indicators (including spatial characteristics, geological characteristics, and facility-related characteristics). It also introduces a real-time correlation coefficient to assess the coordination between adjacent areas, effectively identifying potential "fault zones" or high-risk connection points. This mechanism not only improves the continuity and consistency of the path, but also enhances the safety and stability of the construction process and reduces the risk of subsequent maintenance.

[0032] 3. This method constructs a probability model based on spatial features and a probability model based on cost, and employs an expectation-maximization algorithm and graphical model optimization techniques to search for the optimal connectivity path starting from a seed region. This ensures that the selected cable channel meets both project feasibility requirements and construction and maintenance costs. This method can efficiently identify the globally optimal path in complex geographical environments, achieving an organic unity of safety, feasibility, and cost-effectiveness in power engineering design. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 A flow chart showing a method for intelligent line planning and fault early warning applicable to power engineering design according to the present invention is shown;

[0035] Figure 2 A flow chart showing the preliminary division of feasible areas of the present invention is shown;

[0036] Figure 3 The flowchart of the present invention for determining the final feasible cable channel is shown. DETAILED DESCRIPTION

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0038] In addition, the described features, structures or characteristics can be combined in any suitable manner in one or more example embodiments. In the following description, many specific details are provided to provide a full understanding of the example embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure can be practiced while omitting one or more of the specific details, or other methods, components, steps, etc. can be adopted. In other cases, well-known structures, methods, implementations or operations are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0039] Example 1:

[0040] See Figure 1 As shown, the line intelligent planning and fault warning method applicable to power engineering design of this embodiment has the following process:

[0041] Step 1: Preliminary division of feasible areas.

[0042] See Figure 2 As shown in the figure, the specific process of preliminary division of feasible areas is as follows:

[0043] S11. Data collection and feature extraction.

[0044] Obtain the underground pipeline network area and divide it into multiple adjacent sub-areas. Then collect data for each sub-area. The data sources include but are not limited to geographic information system (GIS), satellite remote sensing images, geological exploration reports, topographic data, geological condition data, existing underground facility distribution data, etc.

[0045] S12. Build spatial semantic segmentation and cost surface model.

[0046] Spatial semantic segmentation: A deep learning model is used to perform spatial semantic segmentation on underground pipeline network areas. By training on a large amount of annotated geospatial data, the model can automatically classify areas into different semantic categories, including areas suitable for cable laying, areas unsuitable for cable laying, and areas with existing underground facilities. Each sub-area is assigned a semantic label: areas suitable for cable laying are coded as 1, areas unsuitable for cable laying are coded as 0, and areas with existing underground facilities are coded as 2.

[0047] Cost Surface Model: A cost surface model is constructed based on a geographic information system. This model takes into account factors such as topography, geological conditions, and construction difficulty, assigning different cost weights to different locations. For example, areas with steep terrain and complex geology have high construction costs and correspondingly high cost weights; whereas areas with flat terrain and stable geology have low construction costs and correspondingly low cost weights.

[0048] Define the cost weight formula:

[0049] Among them, S slope is the normalized slope value; S geology S is the geological stability score, geology ∈[1,10]; D facility is the facility distance; (x, y) is the spatial location coordinate; α, β, γ are the preset weights.

[0050] Slope refers to the degree of inclination of the terrain, typically expressed as an angle or percentage. In power engineering design, steeper terrain increases construction difficulty and cost, requiring more civil engineering work to ensure the safety and stability of the cables. Therefore, in the cost weight formula, a greater slope value corresponds to a higher cost weight.

[0051] The geological stability score is a quantitative assessment of the geological conditions of a specific area, taking into account factors such as soil type, rock structure, and groundwater conditions. A high score indicates good geological conditions and is suitable for infrastructure construction such as cable laying; a low score, on the other hand, may indicate the risk of geological hazards such as landslides and subsidence, making such construction activities unsuitable. Therefore, the lower the geological stability score, the higher the corresponding cost weight, reflecting higher construction costs and risks.

[0052] Facility distance refers to the distance from the planned cable installation location to the nearest existing underground utility. This distance influences the number and complexity of obstacles encountered during construction. Shorter distances can increase construction difficulty by requiring more precise maneuvers to avoid damaging existing underground infrastructure, such as water pipes, gas lines, or other cables. Therefore, shorter facility distances are associated with higher cost weights.

[0053] S13. Clustering primary feasible regions.

[0054] Feature fusion: The results of spatial semantic segmentation and the cost surface model are fused to obtain a comprehensive feature vector. This feature vector contains spatial position, semantic category, and cost information. Specifically, the spatial position coordinates (x, y), semantic category code (0 / 1 / 2), and cost weight C(x, y) of each sub-region are concatenated into a comprehensive feature vector.

[0055] Clustering algorithm: Clusters the comprehensive feature vectors. During the clustering process, the weight of each cluster is calculated after a certain number of iterations. The weight is calculated based on the area of ​​the cluster and the construction cost. For example, clusters with large areas and low construction costs have larger weights.

[0056] Specifically, the initial number of clusters K is set, each cluster obeys a multidimensional Gaussian distribution, and the mean μ of each Gaussian distribution is randomly initialized. k , covariance matrix ∑k and prior probability π k ;

[0057] Calculate the posterior probability P(k|i) that each subregion belongs to each cluster, Update the mean, covariance, and prior probability based on the posterior probability to minimize the distribution difference between the data and the model.

[0058] Weight calculation: After every fixed number of iterations, clustering is paused and the weights of all current clusters are calculated. The total area of ​​each sub-region contained in each cluster is counted (geometric area can be calculated by GIS overlay), and the average cost weight of all sub-regions in the cluster is calculated. Among them, λ1 and λ2 are adjustment coefficients used to balance the importance of area and cost, such as giving priority to large areas with low cost.

[0059] Cluster k area refers to the sum of the areas of all subregions within the kth cluster in cluster analysis. This can be calculated by overlaying geometric areas using a GIS (Geographic Information System). It reflects the spatial extent of the cluster and is an important indicator of cluster size. It is used in weight calculations to reflect the impact of cluster size on weights.

[0060] The average cost weight of cluster k is the average of the cost weights C(x,y) of all sub-regions within the kth cluster. The cost weight C(x,y) is related to factors such as the construction cost and resource consumption of the sub-region. This average reflects the average cost level of the sub-regions within the kth cluster, reflecting the cost characteristics of the cluster. It is used to measure the impact of the cluster's cost on the weight in the weight calculation.

[0061] The total area represents the area of ​​the entire study region (i.e., the sum of the areas covered by all clusters). It serves as a reference quantity and is used to compare with the area of ​​cluster k when calculating cluster weights to determine the proportion of cluster k's area in the total area. This factor then participates in the weight calculation to measure the size of the cluster within the overall region.

[0062] Cluster adjustment: Based on a weight threshold, clusters with weights greater than or equal to the threshold are marked as large clusters, and those with weights less than the threshold are marked as small clusters. All sub-regions within a small cluster are traversed, and for each sub-region belonging to a small cluster, its distance from all large clusters is calculated. The sub-region is assigned to the large cluster with the smallest distance, and the re-combined large cluster is iterated again until the clustering results converge. Cluster adjustment eliminates unreasonable small-scale area divisions.

[0063] S14. Calculate the decision index for each sub-region after clustering.

[0064] The following three types of data are obtained for each sub-region (which has been divided into primary feasible areas after clustering):

[0065] Spatial feature data: terrain slope, elevation, standard deviation of terrain relief, etc.

[0066] Geological characteristic data: soil type (such as clay, sand, which can be coded as a numerical value), geological stability score (1-10 points, 10 points is the most stable), groundwater level depth, etc.

[0067] Facility-related data: horizontal distance to the nearest underground facility, intersection angle, type of existing facilities (such as gas pipelines, drainage pipelines, coded as conflict risk level), etc.

[0068] Spatial feature decision calculation: Obtain the first decision indicator corresponding to the sub-region based on the spatial feature data and the spatial feature decision model. Assume that the spatial feature decision model is: Among them, I1 is the first decision indicator; S t is the spatial characteristic value at time t (such as terrain slope); S targe is the target space characteristic value (such as the slope suitable for laying cables).

[0069] Geological feature decision index: Obtain the second decision index corresponding to the sub-region based on the geological feature data and the geological feature decision model. Assume that the geological feature decision model is: Among them, I2 is the second decision indicator; G t is the geological characteristic value at time t (such as the stability index of soil type); G targe is the target geological characteristic value.

[0070] Facility association decision index: Obtain the third decision index corresponding to the sub-region based on the facility association data and the facility association decision model. Assume that the facility association decision model is: Among them, I3 is the third decision indicator; F t is the facility association value at time t (such as the distance to other underground facilities); F targe Associate a value with the target facility.

[0071] S15. Judgment and integration.

[0072] Real-time correlation coefficient calculation: The real-time correlation coefficient of adjacent sub-regions is obtained based on the first decision indicator, the second decision indicator, and the third decision indicator of the adjacent sub-regions. Specifically:

[0073] A first difference ΔI1 is obtained according to a first decision indicator of adjacent sub-regions, a second difference ΔI2 is obtained according to a second decision indicator of adjacent sub-regions, and a third difference ΔI3 is obtained according to a third decision indicator of adjacent sub-regions.

[0074] The first difference ΔI1, the second difference ΔI2 and the third difference ΔI3 are all normalized. ΔI includes ΔI1, ΔI2, and ΔI3; ΔI1∈[0,1], ΔI2∈[0,1], and ΔI3∈[0,1]. Based on the correlation model R=γ1ΔI1+γ2ΔI2+γ3ΔI3, a real-time correlation coefficient R is obtained, where γ1, γ2, and γ3 are adjustment coefficients, and γ1+γ2+γ3=1. The closer the correlation coefficient R is to 1, the more coordinated the engineering properties of adjacent sub-areas (e.g., consistent slopes, similar geological stability, and consistent distances from existing facilities). The correlation coefficient R is used to identify "fault zones": If the correlation coefficient between adjacent sub-areas is too low (e.g., sudden changes in terrain or significant differences in geological conditions), the connection between the two sub-areas may present construction risks (e.g., uneven settlement or sudden changes in excavation difficulty), requiring focused inspection.

[0075] Judgment and generation of fusion decision indicators: determine whether multiple real-time correlation coefficients are all greater than a preset threshold. If so, obtain the fusion decision indicator corresponding to the sub-region based on the fusion model, the first decision indicator, the second decision indicator and the third decision indicator.

[0076] The fusion model operates as follows: the first, second, and third decision indicators are normalized to generate the first, second, and third character strings, respectively; the characteristic character of the first character string is extracted as the first character, the characteristic character of the second character string is extracted as the second character, and the characteristic character of the third character string is extracted as the third character; the first, second, and third characters are concatenated in sequence to form the fused decision indicator. After normalizing the three decision indicators (I1, I2, and I3) for spatial, geological, and facility association, the fused indicator is generated through string feature extraction and concatenation (for example, the normalized values ​​are rounded to integers and concatenated to form "8-6-9").

[0077] Abnormal handling: If multiple real-time correlation coefficients are not all greater than the preset threshold, the fusion abnormality strategy is triggered, such as rechecking the accuracy of data collection, adjusting the decision model, etc. By judging whether the real-time correlation coefficients of adjacent sub-regions are all greater than the threshold, it is possible to identify whether there is a "fault area". For example: if the geological stability scores of two adjacent sub-regions differ too much, their real-time correlation coefficients may be lower than the threshold, indicating that there is a risk of uneven settlement at the junction. If such faults are not excluded, they may cause the cable to break due to sudden changes in geological conditions after laying; only when all adjacent sub-regions meet the real-time correlation coefficient greater than the preset threshold, the fusion decision indicator is generated. This ensures that the final path consists of continuous, low-risk areas, avoiding the failure of the overall plan due to local incoordination.

[0078] Step 2: Carefully determine the feasible cable channel.

[0079] Seed Area Selection: Select several seed areas from each primary feasible area. These seed areas are typically locations within the area with relatively good construction conditions and low costs. Representative locations can be selected as seed areas by evaluating the comprehensive characteristics of the primary feasible area.

[0080] The core area with the best construction conditions, lowest cost and lowest risk is selected from the primary feasible area as the starting point for subsequent detailed planning (similar to the "anchor point" to reduce the search space).

[0081] Step 3: Build a probability model.

[0082] Probabilistic model based on spatial characteristics: Considering factors such as terrain, geology, and distance from other underground facilities, a probability model is constructed to calculate the feasibility probability of laying cables at different locations. For example, locations with flat terrain, stable geology, and a long distance from other underground facilities have a higher feasibility probability. The model output is a probability value between [0, 1], which represents the confidence level P that the sub-region meets the spatial feasibility requirements. s i .

[0083] Cost-based probability model: Considering factors such as construction cost and maintenance cost, a probability model is constructed to calculate the cost probability of laying cables in different locations. Locations with low construction cost and low maintenance cost have a higher cost probability. The model output is a probability value between [0,1], which represents the feasibility probability P of the sub-region in the cost dimension. e i .

[0084] Step 4: Determine the final feasible cable channel based on the probability model.

[0085] See Figure 3 As shown in the figure, the specific process of determining the final feasible cable channel by combining the probability model is as follows:

[0086] S41, the feasibility probability P obtained by the seed region using the probability model based on spatial features s i and the cost probability P obtained based on the cost probability model e i Combine them to get a comprehensive probability value P total , P total =ω s P s i +μ e P e i Among them, P s i is the feasibility probability of sub-region i; P e i is the cost probability of subregion i, ω s is the weight of the spatial feature, μ e is the weight of the cost feature, and ω s +μ e =1.

[0087] S42. Optimize the comprehensive probability value using the expectation maximization algorithm to determine the final feasible cable channel.

[0088] Specifically, set the initial weight ω s =ω0,μ e =μ0, the sub-regions are regarded as nodes, adjacent sub-regions are connected by edges, and the edge weights are designed as: where R ij is the adjacent sub-region correlation coefficient calculated in step 1 (the larger the value, the better the coordination).

[0089] Based on the current weight ω t and μ t , calculate the comprehensive probability value P of each sub-region total,i,t ;

[0090] Constructing graph model: If the correlation coefficient of adjacent sub-regions R ij When it is less than the set threshold, the edge weight is set to infinity. Otherwise, the edge weight is Ensure that sub-regions with high probability and high coordination are more likely to be selected.

[0091] Using Dijkstra algorithm or A* algorithm, starting from the seed area, find the minimum total weight path Z connecting all target areas t ,If there are multiple seed areas (such as starting points, end points, and transfer points), a multi-source shortest path or Steiner tree algorithm is used to generate a connected path;

[0092] Use M steps to perform parameter optimization, convergence judgment, and output the final path.

[0093] The beneficial effects of this embodiment include: the fusion of spatial semantic segmentation and cost models enables intelligent initial screening of construction areas, accurately eliminating high-risk areas and reducing initial survey costs; a dynamic clustering algorithm combined with an adaptive weight adjustment mechanism automatically identifies large, low-cost areas and optimizes path topology; a multi-decision indicator fusion model quantifies regional coordination through real-time correlation coefficients, effectively avoiding the risks of sudden terrain changes and facility conflicts; a dual-probability model coupled with an expectation-maximization algorithm establishes a quantitative balance between construction feasibility and cost-effectiveness, and combined with a graph optimization algorithm to generate a globally optimal path. This solution, through a data-driven intelligent decision-making system, improves cable channel planning efficiency and reduces construction risks, providing a reliable solution for power projects under complex geological conditions.

[0094] The weights of this invention are used to measure the degree to which different factors or variables influence a particular outcome or decision. Weights are defined as numerical values ​​assigned to each factor when comparing and evaluating multiple factors, reflecting its importance or priority. These weights can be determined based on specific circumstances and needs, and are typically developed and confirmed by professionals or stakeholders. By properly setting weights, programs or systems can be made to make more accurate decisions or predictions.

[0095] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

[0096] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to specific embodiments. Obviously, many modifications and variations are possible based on the contents of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A line intelligent planning and fault warning method applicable to power engineering design, characterized by: The method, the process is as follows: Step 1: Preliminary division of feasible areas: S11. Obtain the underground pipeline network area and divide it into multiple adjacent sub-areas, and collect spatial feature data, geological feature data and facility-related data of each sub-area; S12. Construct a spatial semantic segmentation model and a cost surface model, wherein the spatial semantic segmentation model divides the area into areas suitable for laying cables, areas not suitable for laying cables and areas with existing underground facilities and encodes them, and the cost surface model calculates the cost weight according to the terrain, geology and facility distance; S13. Fusion of spatial semantic segmentation results, cost weights and spatial position coordinates forms a comprehensive feature vector, clusters and adjusts it through Gaussian mixture model to obtain a primary feasible area; S14. Calculate the spatial feature decision index, geological feature decision index and facility-related decision index of each sub-area; S15. Calculate the real-time correlation coefficient based on the decision index of adjacent sub-areas, generate a fusion decision index and perform exception processing; Step 2: Finely determine feasible cable channels: S21: Select seed areas with good construction conditions, low costs, and low risks from the primary feasible areas; Step 3: Construct a feasibility probability model based on spatial characteristics and a cost probability model based on cost; Step 4: Combine the two probability models to obtain a comprehensive probability value, and use the expectation maximization algorithm to optimize and determine the final feasible cable channel.

2. The line intelligent planning and fault warning method applicable to power engineering design according to claim 1 is characterized in that: The spatial semantic segmentation model in S12 uses a deep learning model to encode the sub-region semantic labels as: area suitable for laying cables 1, area not suitable for laying cables 0, and area with existing underground facilities 2; the cost weight formula is: Among them, S slope is the normalized slope value; S geology S is the geological stability score, geology ∈[1,10]; D facility is the facility distance; (x, y) is the spatial location coordinate; α, β, γ are the preset weights.

3. The line intelligent planning and fault warning method applicable to power engineering design according to claim 1 is characterized in that: The cluster adjustment step includes: Calculate the average cost weight of all sub-regions in the cluster, Among them, λ1 and λ2 are adjustment coefficients used to balance the importance of area and cost; Set the weight threshold to divide the large clusters and small clusters, assign the small cluster sub-areas to the large cluster with the smallest distance and iterate again.

4. The line intelligent planning and fault warning method applicable to power engineering design according to claim 1 is characterized in that: The decision indicator calculation in S14 includes: Spatial feature decision calculation: The first decision indicator corresponding to the sub-region is obtained based on the spatial feature data and the spatial feature decision model. The spatial feature decision model is: Among them, I1 is the first decision indicator; S t is the spatial eigenvalue at time t; S targe is the target space eigenvalue; Geological feature decision index: The second decision index corresponding to the sub-region is obtained based on the geological feature data and the geological feature decision model. The geological feature decision model is: Among them, I2 is the second decision indicator; G t is the geological characteristic value at time t; G targe is the target geological characteristic value; Facility association decision index: The third decision index corresponding to the sub-region is obtained based on the facility association data and the facility association decision model. The facility association decision model is: Among them, I3 is the third decision indicator; F t is the facility association value at time t; F targe Associate a value with the target facility.

5. The line intelligent planning and fault warning method applicable to electric power engineering design according to claim 1 is characterized in that: The real-time correlation coefficient in S15 is calculated as: A first difference ΔI1 is obtained according to a first decision indicator of the adjacent sub-regions, a second difference ΔI2 is obtained according to a second decision indicator of the adjacent sub-regions, and a third difference ΔI3 is obtained according to a third decision indicator of the adjacent sub-regions. The first difference ΔI1, the second difference ΔI2, and the third difference ΔI3 are all normalized. ΔI includes ΔI1, ΔI2, and ΔI3; ΔI1∈[0,1], ΔI2∈[0,1], ΔI3∈[0,1]; the real-time correlation coefficient R is obtained based on the correlation model R=γ1ΔI1+γ2ΔI2+γ3ΔI3, where γ1, γ2, and γ3 are adjustment coefficients. The closer the correlation coefficient R is to 1, the more coordinated the engineering properties of adjacent sub-areas are.

6. The line intelligent planning and fault warning method applicable to electric power engineering design according to claim 1 is characterized in that: The seed area selection step in S21 includes: screening sub-areas with a semantic label of 1, performing K-means clustering based on a comprehensive score of spatial features, geological features, cost weights and facility associations, and extracting the cluster center as the seed area, wherein the comprehensive score takes into account the normalized decision index and cost weights.

7. The line intelligent planning and fault warning method applicable to electric power engineering design according to claim 1 is characterized in that: The probability model based on spatial characteristics in S31 considers terrain slope, geological stability and facility distance, and the probability model based on cost considers construction cost and maintenance cost, and the model parameters are determined by historical data or expert weights.

8. The line intelligent planning and fault warning method applicable to electric power engineering design according to claim 1 is characterized in that: The comprehensive probability value in S41 is calculated as: P total =ω s P s i +μ e P e i Among them, P s i is the feasibility probability of sub-region i; P e i is the cost probability of subregion i, ω s is the weight of the spatial feature, μ e is the weight of the cost feature, and ω s +μ e =1.

9. The line intelligent planning and fault warning method applicable to electric power engineering design according to claim 1 is characterized in that: The expectation maximization algorithm optimization step in S42 includes: A graph model is constructed with sub-regions as nodes. The edge weights of adjacent sub-regions are positively correlated with the association coefficient. The Dijkstra algorithm or A* algorithm is used to solve the minimum total weight path. The weights are iteratively adjusted until convergence, and the final feasible cable channel is output.

Images