Urban Substation Layout Method and System for Urban Expansion Areas Based on Artificial Intelligence Algorithms

Abstract

The present invention discloses a method and system for substation layout in urban expansion areas based on artificial intelligence algorithms. By mining the correlation law between the geospatial data and historical load data in urban expansion areas, the future power load demand data in these areas is predicted; the large language model is used to analyze the predicted data, so as to divide the urban expansion areas according to the analysis results, obtain each sub-area and its load difference data, combine it with the historical distribution network grid structure data of the urban expansion areas, and after being processed by the large language model, obtain the initial substation layout plan; by simulating the power supply capacity of each sub-area under different load growth rates, the initial substation layout plan is optimized to obtain an optimized plan and execute it. By comprehensively applying artificial intelligence algorithms such as large language models and machine learning algorithms (dynamic programming algorithms, etc.), the accurate prediction of power load demand in urban expansion areas and the effective optimization of substation layout plans are realized.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and particularly to a method and system for substation layout in urban expansion areas based on artificial intelligence algorithms. Background Art

[0002] In urban expansion areas, the grid planning of the distribution network faces complex challenges in substation site selection and capacity allocation. Currently, the substation layout plan is usually determined based on existing load prediction data. However, the load growth trend in the marginal areas of urban expansion is difficult to accurately predict, resulting in limited accuracy of the existing planning plans.

[0003] Therefore, when the dynamic change of the supply-demand balance relationship is caused by the uncertainty of load growth in the planning area, how to determine the substation layout and capacity allocation plan has become a technical problem that needs to be urgently solved by those skilled in the art. Summary of the Invention

[0004] The present invention provides a method and system for substation layout in urban expansion areas based on artificial intelligence algorithms, which solves the problem of how to determine the substation layout and capacity allocation plan when the dynamic change of the supply-demand balance relationship is caused by the uncertainty of load growth in the planning area.

[0005] To solve the above technical problems, the first aspect of the present invention provides a method for substation layout in urban expansion areas based on artificial intelligence algorithms, including:

[0006] Obtain the geospatial data and historical load data of the urban expansion area, and use a large language model to mine the association rules between the geospatial data and the historical load data to obtain key features for predicting the electricity load demand data of the urban expansion area during the target period;

[0007] Analyze the electricity load demand data through the large language model to obtain load change characteristics for zoning and clustering the urban expansion area, and obtain each sub-region and its load difference data;

[0008] Collect the historical distribution network grid structure data of the urban expansion area, and combine it with each load difference data to be processed through the large language model to obtain an initial substation layout plan;

[0009] Analyze the historical load data through a machine learning algorithm to obtain the load growth probability distribution of each sub-region during the target period to simulate the power supply capacity of each sub-region under different load growth rates;

[0010] Analyze the power supply capacity of each sub-region by using a dynamic programming algorithm to optimize the initial substation layout plan to obtain an optimized substation layout plan and execute it.

[0011] The second aspect of the present invention provides a substation layout system for urban expansion areas based on artificial intelligence algorithms, including:

[0012] A demand forecasting module, configured to obtain the geospatial data and historical load data of the urban expansion area, and use a large language model to mine the correlation rules between the geospatial data and the historical load data, so as to obtain key features to predict the electricity load demand data of the urban expansion area during the target period;

[0013] A zoning clustering module, configured to analyze the electricity load demand data through the large language model to obtain load change characteristics, perform zoning clustering on the urban expansion area, and obtain each sub-region and its load difference data;

[0014] A scheme generation module, configured to collect the historical distribution network grid structure data of the urban expansion area, combine it with each load difference data, and process it through the large language model to obtain an initial substation layout scheme;

[0015] A capacity simulation module, configured to analyze the historical load data through a machine learning algorithm to obtain the load growth probability distribution of each sub-region during the target period, so as to simulate the power supply capacity of each sub-region under different load growth rates;

[0016] A scheme optimization module, configured to analyze the power supply capacity of each sub-region by using a dynamic programming algorithm, optimize the initial substation layout scheme, obtain an optimized substation layout scheme and execute it.

[0017] Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following:

[0018] (1) By using a large language model to mine the correlation rules between geospatial data and historical load data, it is possible to more accurately predict the electricity load demand of urban expansion areas; according to the load change characteristics, perform zoning clustering on the urban expansion area to obtain the load difference data of each sub-region, thereby optimizing the substation layout and capacity configuration and improving resource utilization efficiency;

[0019] (2) By analyzing historical load data through a machine learning algorithm, simulate the power supply capacity of each sub-region under different load growth rates to ensure the stability and reliability of the power supply system; use a dynamic programming algorithm to analyze the power supply capacity to optimize the initial substation layout scheme, provide a scientific basis for urban planners and decision-makers, and improve decision-making efficiency;

[0020] (3) By comprehensively applying artificial intelligence algorithms such as large language models and machine learning algorithms, it is possible to accurately predict the electricity load demand in urban expansion areas and optimize the substation layout and capacity configuration plan, which helps improve resource utilization efficiency and ensure the stability and reliability of the power supply system, providing a scientific basis for urban planners and decision-makers. Brief Description of the Drawings

[0021] To more clearly illustrate the technical solutions of the present invention, the drawings required for the implementation will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0022] Figure 1 is a flowchart of a method for laying out substations in urban expansion areas based on artificial intelligence algorithms provided by an embodiment of the present invention;

[0023] Figure 2 is a structural diagram of a system for laying out substations in urban expansion areas based on artificial intelligence algorithms provided by an embodiment of the present invention. Detailed Embodiments

[0024] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings and embodiments. Obviously, the described embodiments are only some embodiments of the present invention, not all of them. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0025] In the description of this application, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. In the description of this application, unless otherwise specified, the meaning of "a plurality" is two or more.

[0026] In the description of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two components. The terms "vertical", "horizontal", "left", "right", "upper", "lower" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the system or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0027] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0028] At present, with the expansion of the city, the load in the area will also show a rapid growth trend, but the uncertainty of load growth will make it difficult to stabilize the supply and demand balance in the area. Therefore, in the early stage of planning, the substation layout scheme determined based on the existing load forecast data is often difficult to adapt to the long-term load growth demand in the area: on the one hand, when the actual load growth exceeds expectations, the power supply capacity of the existing substation will face bottlenecks, and the new substations are restricted by land resources and long construction periods; on the other hand, if the substation capacity is over-reserved, it will lead to excessive initial investment and low equipment utilization, which will affect the economic efficiency of the power grid. This contradiction is particularly prominent in the marginal areas of urban expansion. The load growth trend in this area is difficult to accurately predict, but it is an important bearing area for future urban development. Therefore, planners need to consider moderate advanced layout when selecting substation sites, but how to determine the reserved capacity and construction sequence so that it can meet the future load growth needs without causing resource waste has become a technical difficulty. Under the influence of load growth uncertainty, how to dynamically adjust the substation layout plan so that it can adapt to different load growth scenarios and maintain a reasonable investment scale is a key issue that needs to be solved in the planning of distribution network in urban expansion areas.

[0029] Based on this, in one embodiment, if Figure 1As shown in the figure, the first aspect of the present invention provides a method for laying out substations in urban expansion areas based on artificial intelligence algorithms, including:

[0030] S1. Obtain the geospatial data and historical load data of the urban expansion area, and use a large language model to mine the correlation rules between the geospatial data and the historical load data to obtain key features for predicting the electricity load demand data of the urban expansion area during the target period;

[0031] S2. Analyze the electricity load demand data through the large language model to obtain load change characteristics for zoning and clustering the urban expansion area, and obtain each sub-region and its load difference data;

[0032] S3. Collect the historical distribution network grid structure data of the urban expansion area, and combine it with each load difference data to be processed through the large language model to obtain the initial substation layout plan;

[0033] S4. Analyze the historical load data through a machine learning algorithm to obtain the load growth probability distribution of each sub-region during the target period to simulate the power supply capacity of each sub-region under different load growth rates;

[0034] S5. Analyze the power supply capacity of each sub-region by using a dynamic programming algorithm to optimize the initial substation layout plan, and obtain and execute the optimized substation layout plan.

[0035] Specifically, the present invention obtains geographical features such as land use types, building density, transportation networks, and population heat distribution in the urban expansion area and urban planning texts (such as urban development plans, etc.) through satellite remote sensing images, GIS systems, etc. as geospatial features, and collects time-series load data (regional electricity consumption, load curves, peak loads, etc.) and user type distributions (residential / commercial / industrial) at the substation / feeder level from the SCADA system, etc. as historical load data, and then unifies these data in a spatio-temporal coordinate system for data alignment processing and matches them to geospatial units (such as grids); then converts the structured data into natural language descriptions, and uses artificial intelligence algorithms, such as advanced large language models like deepseek, for association rule mining, enabling advanced large language models like deepseek to use various algorithms integrated in it to extract features from the input data, and after obtaining key features, combine with the Prompt design of the deepseek large language model or the Transformer model to generate the load curve of the urban expansion area during the future preset target period, that is, the electricity load demand data.

[0036] Analyze the electricity load demand data through the DeepSeek large language model, extract its peak-valley difference, load density, growth slope, seasonal volatility, etc., and form load change characteristics represented by high-dimensional vectors. Through clustering algorithms and semantic rules of the DeepSeek large language model, partition and cluster the urban expansion area to obtain each sub-region (such as high-load industrial areas, low-growth residential areas, etc.) and their load difference data.

[0037] Associate the historical grid data (substation coordinates, transformer capacity, line impedance, power supply radius, line impedance, etc., that is, the historical planning scheme for the urban expansion area) with the load difference data of the sub-regions, and use the DeepSeek large language model to call the power grid design rule library (such as IEEE standards) to generate the candidate site locations and capacities, that is, the initial layout scheme of the substation.

[0038] Analyze the historical load data of the urban expansion area through machine learning algorithms to predict the load growth probability distribution of each sub-region within a preset target time period in the future, and use Monte Carlo to generate samples to simulate the equipment utilization rate of each sub-region under various load growth rates, and then obtain the power supply capabilities of each sub-region under various different load growth rates.

[0039] Finally, use the dynamic programming algorithm to analyze the power supply capabilities of each sub-region to optimize the initial layout scheme of the substation. By constructing the objective function, state variables, and decision variables and performing recursive optimization, obtain the optimized layout scheme of the substation and execute it.

[0040] After learning and understanding a large amount of data through large language models such as DeepSeek, the present invention can master the internal laws of load changes and make intelligent predictions about future trends. Combining this cutting-edge artificial intelligence technology with a variety of artificial intelligence algorithms to generate a more dynamic and flexible substation layout optimization strategy, while meeting the future load growth requirements, it can also take into account investment benefits, achieving a balance between the advancement and economy of the plan. This solution improves the scientificity and adaptability of power grid planning by deeply integrating the semantic understanding of large language models and the numerical calculation capabilities of traditional algorithms, realizing the full-chain intelligence from data to decision-making, and is applicable to urban expansion scenarios with high uncertainty.

[0041] In one embodiment, step S1 includes:

[0042] Divide the urban expansion area into several grid cells, obtain the geospatial data and historical load data of each grid cell, and perform similarity grouping on the historical load data of each grid cell through a clustering algorithm based on dynamic time warping to obtain the grid cell load grouping result;

[0043] Extract industrial land distribution data from the geospatial data, and use the association rule mining algorithm in the large language model to quantify the support and confidence between the industrial land distribution data and the load grouping results of each grid cell, so as to obtain an association rule set characterizing the relationship between the industrial land types and the electricity load fluctuation patterns in the urban expansion area;

[0044] Extract the time series data of economic indicators corresponding to the industrial land types from the geospatial data according to the association rule set, combine it with the industrial land distribution data and the historical load data, and process it through a third-order tensor decomposition algorithm to obtain the key features characterizing the electricity load and industrial economic development;

[0045] Based on the key features, use a recursive neural network to predict and model the electricity load in the urban expansion area, and obtain the electricity load demand data in the target period in the urban expansion area.

[0046] Specifically, the present invention divides the urban expansion area into several grid cells based on the population density data of the urban expansion area. For example, the population density of a grid cell in the core business district of a certain urban area reaches 25,000 people per square kilometer during the day on weekdays and drops to 8,000 people per square kilometer at night, while the residential grid cell shows the opposite population density change characteristics; or uses GIS tools to divide the urban expansion area into regular grid cells of 1km×1km to ensure that each grid covers a continuous geographical space; through remote sensing image analysis or network public data, extract geographical space data such as industrial land types, building densities, and population heat distributions of each grid, and spatially interpolate and distribute the hourly load data provided by the power company according to the grid geographical coordinates to form a time series load data set for each grid; then calculate the DTW distance matrix for the load curve of each grid to solve the phase shift problem of load fluctuations on the time axis (such as the peak-valley difference between weekdays and holidays), and use a hierarchical clustering algorithm combined with DTW distance to generate groups with similar load patterns to obtain the grid cell load grouping results. By using the dynamic time warping algorithm to measure the similarity of these electricity load time series curves, the grid cells can be clustered into different electricity consumption characteristic types such as commercial-dominated, residential-dominated, and mixed types.

[0047] Screen the industrial land types from geospatial data (for example, as shown in the industrial land distribution data, the proportion of commercial land in the commercial-dominated grid cells exceeds 60%, the proportion of office land is about 25%, while the proportion of residential land in the residential-dominated grid cells reaches 75%). Statistically analyze the area proportion of different land types in each grid, and input the obtained grid cell load grouping results and the industrial land data converted into natural language descriptions into the DeepSeek large language model for Prompt design and calculation of support and confidence. Through association rule mining, it is found that when the proportion of commercial land exceeds 50%, the confidence of the bimodal characteristics of the weekday electricity load is 0.85 and the support is 0.72, so as to generate an association rule set representing the relationship between industrial land types and electricity load fluctuation patterns in urban expansion areas. The grid labels can also be encoded as transaction items (such as "grid ID_industrial type_load group"), set the minimum support threshold, and then use the FP-Growth algorithm to mine frequent item sets, and screen strong association rules according to the lift (such as lift>1.5) and confidence (such as conf>0.7), so as to obtain an association rule set representing the relationship between industrial land types and electricity load fluctuation patterns in urban expansion areas.

[0048] Based on the grid location and industrial land type labels in the association rule set, obtain the time series data of economic indicators in this area (such as GDP growth rate, industrial added value, etc., and the time series data of economic indicators reflects that the quarterly-on-quarter growth rate of commercial retail sales in commercial-dominated grid cells has a significant positive correlation with the electricity load growth rate). Then use the third-order tensor decomposition algorithm to construct a third-order tensor with dimensions of "grid×time×feature", and its features include: economic indicators, industrial distribution (land type ratio), and load data (weekly average load, etc.). And use the alternating least squares method to decompose the tensor to extract multiple latent factors, and use them as key features representing the electricity load and industrial economic development after verifying the decomposition stability through core consistency diagnosis. At the same time, the eigenvectors obtained after the third-order tensor decomposition show that there is a stable co-evolutionary relationship among the expansion of commercial land, the growth of retail sales, and the growth of electricity load. It is also possible to construct a third-order tensor including time (monthly granularity), space (grid cells, such as 100 grids), and characteristic indicators (proportion of industrial land, GDP growth rate, electricity load value). For missing economic indicators (such as some grids have no GDP data), use KNN interpolation to complete them, and then use CP decomposition to decompose the original tensor into three factor matrices and a core tensor, and extract the factor combination with the highest weight in the core tensor (such as the correlation coefficient between the night load in the commercial land intensive area and the regional service industry GDP reaches 0.73), and use it as the key feature to input into the prediction model.

[0049] Construct a multi-dimensional input vector by integrating tensor decomposition factors, historical load time series data (the previous 24 months), and economic indicators (GDP, population), etc. Generate training samples using the sliding window method. Use a recurrent neural network with two hidden layers (neuron numbers 128 and 64) and a Dropout layer (ratio 0.2) to prevent overfitting for predicting and modeling the electricity load in the urban expansion area. The loss function is Huber Loss, and hyperparameters are selected through Bayesian optimization. Finally, the obtained key features are input into the trained recurrent neural network for processing, and the electricity load demand data including the electricity demand growth trend and the peak-valley situation of the electricity demand in the future preset target period in the urban expansion area is output.

[0050] The present invention captures the heterogeneity of local electricity consumption patterns (such as the load difference between commercial areas and residential areas) through grid cell division to improve prediction accuracy; uses DTW clustering to identify non-linear time series similarities (such as sudden load changes during holidays), avoiding the sensitivity of the Euclidean distance to phase shifts; uses explicit modeling of the causal relationship between industry and load to provide physical interpretability for association rule mining, supporting the transformation of power grid planning from "statistical correlation" to "mechanism-driven", thereby improving the confidence of rules; uses third-order tensor decomposition to capture the non-linear coupling relationship of "economic indicators - industrial distribution - load fluctuations", and crosses multi-dimensional features to enhance prediction robustness.

[0051] In one embodiment, step S2 includes:

[0052] Parse the semantic text in the electricity load demand data through the semantic tokenizer in the large language model to obtain load change features including load numerical features and text features;

[0053] Extract the regional spatial layout data of the urban expansion area based on the load change features, quantify the Euclidean distance similarity of the urban expansion area through the hierarchical clustering algorithm, and perform regional division based on the similarity calculation result to obtain several sub-regions;

[0054] Process the historical load data of each sub-region through a conditional random field to construct a time series state transition probability matrix including load density and load peak-valley ratio, and perform probability inference on the load growth pattern of each sub-region according to each time series state transition probability matrix to obtain the load growth probability distribution of each sub-region;

[0055] Extract the periodic and trend features of the load peak-valley difference of each sub-region from each load growth probability distribution, and process them through the time series decomposition algorithm to obtain the load difference data of each sub-region.

[0056] Specifically, the present invention constructs a feature semantic parsing corpus by extracting semantic features such as total load, load density, load peak-valley ratio, and load growth rate from the prediction results, and then uses the semantic tokenizer in the DeepSeek large language model to perform structured parsing on the load prediction description text through named entity recognition, obtaining load change features composed of load numerical features (load values, growth rates, time, etc., such as peak load increment: 80 MW, time: 2024, etc.) and text features (industry keywords, etc., such as industry type: food street, region: commercial area, grid cell).

[0057] Based on the regional labels in the load change features, the regional spatial layout data of the urban expansion area is extracted and used as input data together with the load change features. Then, the Euclidean distance similarity of the urban expansion area is quantified using the hierarchical clustering algorithm based on the input features (such as load density, industry type, etc.), and regional division is performed based on the similarity calculation results to obtain several sub-regions. In the expansion planning of the urban core area, new regions often exhibit different spatial clustering characteristics. For example, the planned land area of the eastern new area of a certain city reaches 35 square kilometers, of which commercial land accounts for 35%, industrial land accounts for 25%, and residential land accounts for 40%. The hierarchical clustering based on the load density and growth rate features shows that this area can be divided into 3 load feature sub-regions, where the predicted load density of the commercial area is 6000 kW / km², the industrial area is 4500 kW / km², and the residential area is 2500 kW / km².

[0058] The historical load data of each sub-region is discretely state-divided through a conditional random field, and a feature function is designed. The state transition frequency between states in history is statistically analyzed using the state transition feature, the observed values such as load density and peak-valley ratio are associated based on the observed features, and the weight parameters of the feature function are solved using the maximum likelihood estimation method and optimized using the L-BFGS algorithm. Furthermore, a time-series state transition probability matrix including load density and load peak-valley ratio is constructed, and the probability distribution of the state sequence in the next k steps (such as the next 12 months) is predicted using the Viterbi algorithm based on the time-series state transition probability matrix of each sub-region, obtaining the load growth probability distribution of each sub-region. Through the analysis of the historical load data of these sub-regions, it is found that the average annual growth rate of the load density in the commercial area is 12%, and the daily load peak-valley ratio is 2.1; the average annual growth rate of the load density in the industrial area is 8%, and the daily load peak-valley ratio is 1.5; the average annual growth rate of the load density in the residential area is 6%, and the daily load peak-valley ratio is 1.8. The state transition probability constructed by the conditional random field shows that the load density growth probability in the commercial area remains above 0.85 within the next 3 years.

[0059] Perform STL decomposition on the load growth probability distribution of each sub-region to extract the periodic and trend characteristics of the load peak-valley difference in each sub-region, and quantify the slope of the trend term, the difference between the maximum and minimum values of the periodic term, and the standard deviation of the residual term through the time series decomposition algorithm to calculate the characteristic difference index of each sub-region, and obtain the load difference data of each sub-region including load density, load peak-valley difference, and load growth rate, etc.

[0060] The present invention parses the text and numerical load data through a semantic tokenizer to fully explore the implicit impact of unstructured information on load changes; uses hierarchical clustering combined with Euclidean distance and semantic features to achieve regional segmentation that is more in line with the actual load distribution; captures the load state transition law through a conditional random field to quantify the growth risk under uncertain scenarios; uses time series decomposition to separate the periodicity and trend of the load peak-valley difference, providing a basis for differential power grid planning.

[0061] In one embodiment, step S2 further includes:

[0062] Establish a characteristic text dictionary including total load, load density, load growth rate, and load peak-valley ratio according to the electricity load demand data, and extract numerical features from the characteristic text dictionary through the tokenization rules in the large language model to obtain a regional load prediction numerical matrix;

[0063] Obtain the land use type data of the urban expansion area based on the regional load prediction numerical matrix, and calculate the Euclidean distance similarity of load density and growth rate through the spectral clustering algorithm based on the land use type data to perform unsupervised classification on the urban expansion area to obtain several sub-regions;

[0064] Use the principal component analysis method to process the historical load data of each sub-region to obtain the load density principal component feature vectors of each sub-region to obtain the load growth data of each sub-region, and perform regression fitting on each load growth data to obtain the load growth potential coefficient of each sub-region;

[0065] Use the Fourier transform to process the load peak-valley data of each sub-region to obtain the load peak-valley characteristic sequences of each sub-region, combine them with the load growth potential coefficients of each sub-region, and perform processing using the contour line drawing method based on kernel density estimation to quantify the load density gradient, load growth trend line, and load peak-valley change band, and obtain the load difference data of each sub-region.

[0066] Specifically, based on the electricity load demand data, the present invention establishes a characteristic text dictionary including the total load, load density, load growth rate, and load peak-valley ratio associated with grid cells (such as the peak load density reaching 5,500 kW / km², a 12.5% increase compared to the same period last year, and the daily peak-valley difference being 2,800 kW), stores the characteristic names, dimensions, calculation rules, and examples in JSON format, and then performs numerical extraction through a large language model: identifies the keywords (such as total load, peak-valley ratio, etc.) and associated numerical values in the characteristic text dictionary, and parses the numerical values based on the formulas in the dictionary to extract numerical characteristics to generate a region-characteristic matrix, obtaining a regional load prediction numerical matrix.

[0067] Based on the regional characteristics or grid cells in the regional load prediction numerical matrix, extract the land use type data of the urban expansion area (such as land use nature, plot area, floor area ratio, building density, etc.), encode the land use types as categorical variables (such as industry = 1, commerce = 2, etc.) and associate them with the load matrix according to the regional ID; and calculate the Euclidean distance similarity of feature vectors such as load density and load growth rate for these encoded data through spectral clustering algorithm, perform Gaussian kernel function conversion on them to construct a normalized Laplacian matrix, and perform eigen-decomposition on it, take the first k eigenvectors (k is the preset number of clusters) for K-means clustering to obtain several sub-regions.

[0068] Take the historical load data matrix of each sub-region as input to calculate the covariance matrix, obtain the load density principal component feature vectors of each sub-region as input, and after being processed by the ridge regression model, output the load growth potential data of each sub-region; among them, the load density distribution data shows after principal component analysis that the first principal component reflects the overall load level of the region, accounting for 65% of the total variance contribution rate, and the second principal component reflects the unevenness of load distribution, accounting for 25% of the total variance contribution rate. 90% of the load density spatial distribution characteristics can be explained by these two principal components; and the historical load growth data shows that the average annual load growth rate in the commercial area has remained above 13% in the past 5 years, about 7.5% in the industrial area, and about 5.5% in the residential area. The growth potential coefficients obtained by regression fitting are 1.8, 1.3, and 1.1 respectively, reflecting the development potential differences in different regions.

[0069] The load peak-valley data of each sub-region is processed using Fourier transform to obtain the load peak-valley characteristic sequences of each sub-region, and a multi-dimensional kernel function is designed in combination with the load growth potential coefficients of each sub-region: load density gradient (spatial kernel (Gaussian kernel) + load density value), load growth trend line (temporal kernel + growth rate), and load peak-valley change band (frequency domain kernel + Fourier amplitude). The contour drawing method based on kernel density estimation is used for processing, and the kernel density estimation values are calculated for the grid spatial points, and the contour points are connected to form gradient lines, trend lines, and change bands to obtain the load difference data of each sub-region. Among them, load peak-valley characteristic analysis shows that the peak-valley ratio in the commercial area on weekdays is 2.2, which drops to 1.6 on holidays. The peak-valley ratio in the industrial area remains stable at about 1.4 throughout the year. The peak-valley ratio in the residential area reaches 1.9 in summer and about 1.5 in other seasons. Obvious daily and seasonal cycle characteristics are extracted through Fourier transform. The contour map generated based on kernel density estimation shows that the load density decreases from the city center to the periphery. The contour interval is the densest in the commercial area, reflecting the largest load density gradient. The load growth trend line points to the main urban development axis, and the load peak-valley change band shows an obvious layered structure, with the largest peak-valley difference in the core area and gradually decreasing outward.

[0070] The present invention unifies the format of multi-source load data by constructing a characteristic text dictionary to enhance data fusion and model generalization capabilities; uses spectral clustering combined with land use types and load characteristics for spatial-load collaborative classification to achieve more accurate regional division; extracts core growth driving factors through principal component analysis and quantifies development potential using a regression model; fuses Fourier transform and kernel density estimation to visualize spatio-temporal dynamics and generate a multi-dimensional load difference map to support refined power grid planning.

[0071] In one embodiment, step S3 includes:

[0072] Extract the spatial distribution characteristics of the electricity load of each sub-region from the load difference data through a convolutional neural network, and process the spatial distribution characteristics of the electricity load of each sub-region using a spatial coordinate mapper to generate a load density heat distribution matrix for each sub-region.

[0073] Obtain the historical distribution network grid structure data of the urban expansion area, and use the support vector regression algorithm to optimize the calculation of the power supply radius and load coverage rate in the historical distribution network grid structure data to obtain a candidate substation siting sequence.

[0074] Based on the land use restriction conditions of each sub-region, preprocess the candidate substation siting sequence through the large language model, and use the genetic algorithm to perform multi-objective optimization operations on the preprocessed candidate substation siting sequence to obtain a substation siting coordinate set.

[0075] Quantify the load density thermal distribution matrix of each sub-region through the hierarchical clustering algorithm to obtain the maximum load capacity of the power supply area, obtain the standard transformer capacity sequence from the equipment specification library, and calculate the transformer combination plan based on the standard transformer capacity sequence using the dynamic programming algorithm to obtain the main transformer capacity configuration table;

[0076] Under the constraints of the power supply reliability level requirements and the load transfer requirements, determine the number of outgoing lines based on the main transformer capacity configuration table through the graph theory algorithm to obtain the substation outgoing line plan, and combine it with the substation site selection coordinate set and the main transformer capacity configuration table to obtain the initial substation layout plan.

[0077] Specifically, encode the load difference data (load density, growth rate, peak-valley ratio, etc.) of each sub-region into a multi-channel matrix by grid as input data and input it into the convolutional neural network for processing, so that it extracts the spatial features of the regional power consumption load through a sliding window of 500m×500m, outputs the load spatial feature vector of each grid (such as industrial area feature = high density, low fluctuation, residential area feature = medium density, high fluctuation), and uses a coordinate mapper to associate the grid longitude and latitude coordinates (the boundary coordinates of each sub-region can also be directly used) with the feature vector to construct a spatial index, so as to perform Gaussian kernel smoothing on the load density of each grid to generate a load density thermal distribution matrix with a resolution of 100m×100m.

[0078] Obtain the historical distribution network grid structure data of the urban expansion area (such as the power supply radius, load coverage rate, line impedance, transformer load rate, etc. of the historical planning scheme), perform multi-objective optimization of minimizing the power supply radius and maximizing the load coverage rate, use the radial basis function (RBF) kernel, and optimize the parameters through grid search to output the candidate substation site sequence; among them, the substation site selection optimization also needs to consider geographical condition constraints, such as the nature of construction land, traffic accessibility, topography and landform, etc. A certain candidate site is located near the intersection of the urban main road, covers an area of 0.8 hectares, has a flat terrain, convenient transportation, a power supply radius coverage rate of 95%, and the lowest operating cost, and is determined as the optimal site by the genetic algorithm.

[0079] Taking the candidate site coordinates and land use restrictions (such as ecological red lines and requirements for avoiding residential areas) as inputs, the policy text (such as "The substation should be at least 200m away from the residential area") is converted into a geofence constraint through rule parsing by the DeepSeek large language model, and the candidate points that violate the constraints (such as points within the ecological protection area) are removed to process the candidate substation site sequence. Binary coding is used to represent whether the substation is selected (e.g., 1 = selected, 0 = not selected). Using the land acquisition, line laying costs, etc. of the processed candidate substation site sequence as costs, a fitness function is constructed for evolutionary operations to obtain the Pareto optimal solution set, and the substation site selection coordinate set is selected through the entropy weight method.

[0080] Through the hierarchical clustering algorithm, the load density values of the load density thermal distribution matrix base of each sub-region are aggregated into several capacity levels according to the Euclidean distance, and the 95% quantile load value of each cluster is taken as the upper limit of the power supply area capacity to obtain the maximum load capacity of the power supply area. Then, based on the maximum load capacity of the power supply area, the standard transformer capacity sequence is obtained from the equipment specification library. Taking the sub-region as a stage, the cumulative configured capacity and remaining budget as states, and the selection of the transformer model as a decision to define the state. Then, based on these states, transformer capacity, and regional load demand, a state transition equation is constructed, and the minimization of the total cost (equipment purchase + operation and maintenance) is used as the optimization goal to solve using the dynamic programming algorithm to obtain the main transformer capacity configuration table. Among them, based on the load characteristics, the main transformer capacity configuration plan given by the dynamic programming algorithm is 2 transformers with a capacity of 63 MVA. Considering the load growth expectation and spare capacity requirements, the rated capacity of the transformer meets 125% of the maximum load, and it also has the ability to transfer the load when a single main transformer fails.

[0081] Taking the requirement that the outgoing line circuit should meet the N-1 criterion (when any line fails, the remaining lines can carry all the loads) and the load transfer path should be connected in the topology as the reliability constraint, the substation and load points are modeled as graph nodes, the lines are edges, and the weight = line capacity / length. An initial radial grid is constructed through the minimum spanning tree to ensure the lowest cost, and redundant lines are added until the N-1 check is satisfied for the outgoing line circuit calculation. The number of outgoing lines and the path are output, that is, the substation outgoing line plan, which is combined with the substation site selection coordinate set and the main transformer capacity configuration table to obtain the initial substation layout plan. Among them, in terms of the outgoing line circuit configuration, considering that the proportion of first-class load for power supply reliability reaches 40%, the minimum spanning tree obtained by the graph theory algorithm contains 8 backbone nodes, and 10 outgoing lines are configured in combination with the load transfer requirements. Among them, 6 lines are used as normal operating circuits, and 4 lines are used as standby connection circuits to quickly realize load transfer when the main transformer fails or the line is under maintenance.

[0082] The present invention extracts the thermal distribution characteristics of load density through a convolutional neural network to accurately depict the spatial heterogeneity of regional power demand; integrates support vector regression and genetic algorithm for collaborative optimization considering multiple dimensions such as power supply radius, coverage rate, and land use constraints; generates a transformer capacity configuration table based on hierarchical clustering and dynamic programming to achieve dynamic matching between equipment selection and load growth; adopts a graph theory algorithm to generate a line-out plan to balance reliability and economy to reduce power outage risks; this solution fully considers multiple dimensions such as load distribution characteristics, geographical location constraints, and power supply reliability requirements, and ensures the scientificity and economy of power grid planning through multi-level optimization, forming a complete site planning solution.

[0083] In one embodiment, step S4 includes:

[0084] Extract the trend term, periodic term, and random term in the historical load data of each sub-region through a time series decomposition algorithm, and process the extraction results using the kernel density estimation method to obtain the load growth probability distribution matrix of each sub-region;

[0085] Based on each load growth probability distribution matrix, use a long short-term memory network to predict the load growth data at monthly, quarterly, and annual time scales to obtain a multi-time scale load growth trend vector;

[0086] According to the multi-time scale load growth trend vector, use the Monte Carlo method to perform probability sampling on the load levels of each sub-region under multiple growth rate scenarios to obtain the main transformer load probability distribution to quantify the average utilization hours, maximum load rate, and load duration of the equipment, and obtain the equipment utilization rate characteristic matrix;

[0087] Fit the equipment utilization rate characteristic matrix through polynomial regression to quantify the change trend of the power supply capacity of each sub-region at multiple time scales, and obtain the power supply capacity of each sub-region at different load growth rates.

[0088] Specifically, the present invention takes the historical load time series of each sub-region as input, extracts the trend term (reflecting long-term growth), periodic term (seasonal fluctuations), and random term (impact of noise or emergencies) of each sub-region through a time series decomposition algorithm, and takes the growth rate sequence of the extracted trend term data as input, processes it using a Gaussian kernel function, and outputs a load growth probability density function to construct a load growth probability distribution matrix for each sub-region; among them, the time series decomposition of the load data shows obvious multi-level characteristics. For example, in the historical load data of a certain commercial area, the trend term shows an annual growth rate of 8.5%, the periodic term contains obvious double-peak characteristics within a day, the peak-to-valley difference reaches 45%, and the random term fluctuates greatly during holidays, with a standard deviation of 15%. The probability distribution matrix obtained through kernel density estimation shows that the cumulative probability of the load growth probability in this region between 6% and 10% reaches 0.75.

[0089] Taking the load growth probability distribution matrix, historical periodic term, and external variables (such as economic indicators, population growth rate, etc.) of each sub-region as input, a long short-term memory network is used to predict the load growth data at multiple time scales such as monthly, quarterly, and annual, and the prediction results at different time scales are spliced to output a multi-time scale load growth trend vector; among them, the prediction results at multiple time scales show growth characteristics of different cycles. The monthly data is significantly affected by seasonal factors, with an average monthly growth rate of 12% in summer and dropping to about 5% in winter. The quarterly data shows a stepped growth related to the industrial production cycle, and the annual data shows a steady upward trend as a whole, with an average annual growth rate maintained at 8% level.

[0090] Based on the multi-time-scale load growth trend vectors and the load growth probability distribution matrices of each sub-region, 1000 sets of random growth rate sequences (such as pessimistic / neutral / optimistic scenarios) are generated through the Monte Carlo method, and the load rates of the main transformers are calculated for each scenario to obtain the load probability distribution of the main transformers, so as to quantify the average utilization hours, maximum load rate, and load duration of each main transformer device, and obtain the device utilization rate characteristic matrix (each row corresponds to a sub-region, and each column is an index value), that is, the utilization rate of the main transformer in each sub-region; among them, the load data of the main transformer shows significant differences under different growth scenarios. The maximum load rate reaches 85% within 5 years in the high-growth scenario, 75% in the medium-growth scenario, and 65% in the low-growth scenario; and the Monte Carlo sampling results show that the probability that the load rate exceeds 80% in the high-growth scenario is 0.35, and this probability drops to 0.15 in the medium-growth scenario; as for the evaluation of the device utilization rate, a multi-dimensional index system is adopted. The annual average utilization hours of the main transformer reach 5500 hours in the high-load region, the maximum load rate is 85%, and the proportion of full-load operation time is 25%. The matrix composed of these characteristic indicators reflects the trend that the operation pressure of the device gradually increases with the load growth, and the change trend of the power supply capacity shows an obvious non-linear relationship with the device utilization rate. When the device utilization rate exceeds 75%, the speed of power supply capacity improvement significantly slows down.

[0091] Taking the device utilization rate characteristic matrix of each sub-region as the input and the power supply capacity (defined as the maximum load growth rate that can be carried) as the target variable, the optimal order (usually 2 - 3 orders) is determined through cross-validation to construct a regression model to fit the power supply capacity curve under different load growth rates, and the power supply capacity matrix of each sub-region under different load growth rates is output; among them, the polynomial regression fitting results show that in the high-load growth scenario, the power supply capacity improvement amplitude after 3 years is only 85% of the load growth amplitude; when the device utilization rate reaches 80%, the power supply reliability rate drops to 99.95%, and the voltage qualification rate drops to 98%. If the high-speed growth trend is maintained, the power supply capacity constraint will become the key factor restricting regional development within 5 years.

[0092] The present invention separates the trend, cycle, and random components of load changes through time series decomposition to improve the interpretability and accuracy of prediction; uses the long short-term memory network (LSTM) to capture the multi-level laws of monthly, quarterly, and annual load growth to adapt to the phased characteristics of urban expansion; adopts Monte Carlo simulation to quantify the extreme scenario probability of device load (such as overload risk) to support the grid resilience planning; reveals the dynamic relationship between power supply capacity and load growth based on polynomial regression, which can guide the dynamic adjustment of transformer capacity and operation and maintenance strategies; this solution realizes a complete chain from data analysis to decision support through a technical closed-loop of "decomposition - prediction - simulation - quantification", and is especially suitable for the grid adaptability planning scenario under the background of rapidly increasing load growth rate.

[0093] In one embodiment, step S5 includes:

[0094] Based on the power supply capabilities of each sub-region, solve a multi-stage optimization function with the goal of minimizing investment costs and the constraint that the equipment utilization rate is not lower than a preset threshold through a dynamic programming algorithm, and obtain a substation construction time schedule;

[0095] According to the transformer capacity specification constraints, optimize the main transformer capacity configuration table in the initial substation layout scheme using a genetic algorithm to obtain a main transformer capacity configuration scheme;

[0096] Based on the main transformer capacity configuration scheme, obtain equipment load rate data, and calculate the annual average utilization hours and maximum load rate of the equipment through a multi-objective optimization algorithm to obtain a substation capacity expansion scheme;

[0097] Extract grid structure data from the historical distribution network grid structure data according to the substation capacity expansion scheme, and calculate the substation power supply radius and load coverage rate through a graph theory algorithm to obtain a substation site selection optimization scheme;

[0098] Combine the substation site selection optimization scheme, the substation outgoing line scheme, the substation construction time schedule, the main transformer capacity configuration scheme, and the substation capacity expansion scheme to obtain a substation optimized layout scheme and execute it.

[0099] Specifically, the present invention obtains load prediction data and its corresponding power supply capacity data under three growth scenarios of high, medium, and low based on the historical load data of each sub-region, and divides the planning period into several stages through a dynamic programming algorithm. Using the capacity of the existing substations, the cumulative investment cost, and the equipment utilization rate in each period as state variables, and the capacity and location of the newly built / expanded substations in each period as decision variables, solve a multi-stage optimization function with the goal of minimizing investment costs and the constraint that the equipment utilization rate is not lower than a preset threshold. By recursively solving the optimal decisions in each stage, output a substation construction time schedule; among them, in the substation construction plan, significant differences are shown in different load growth scenarios. The average annual growth rate reaches 12% in the high-growth scenario, 8% in the medium-growth scenario, and 5% in the low-growth scenario. The dynamic programming algorithm calculates that 2 substations need to be built within 3 years in the high-growth scenario and 1 substation needs to be built within 5 years in the medium-growth scenario with the constraint that the equipment utilization rate is not lower than 65%.

[0100] The transformer capacity selection is encoded with real numbers, and the capacity matching standard specifications (such as multiples of 50 MVA) and the total capacity not less than 1.2 times the maximum load in the area (N - 1 criterion redundancy) are injected as constraints. Based on the load coverage rate (covered load / total load) and the capacity utilization rate (actual load / total capacity), a fitness function is constructed to perform evolutionary operations to optimize the main transformer capacity configuration table and obtain the main transformer capacity configuration plan. Among them, the transformer capacity configuration plan directly affects the investment benefit. The load forecast of a substation's power supply area shows that the maximum load will reach 80 MVA within 5 years. The optimal configuration plan given by the genetic algorithm is to install 2 main transformers of 50 MVA initially and increase the capacity to 3 main transformers of 50 MVA in the 4th year, and the average equipment utilization rate is maintained above 75%.

[0101] According to the main transformer capacity configuration plan, obtain the historical load rate data of each transformer device in the plan. Taking maximizing the annual average utilization hours of the device (reflecting economy) and minimizing the maximum load rate (reducing overload risk) as the objective function, perform non - dominated sorting by the NSGA - II algorithm and stratify according to the objective values, retain the Pareto - front solutions, and calculate the crowding degree to ensure that the solution set is evenly distributed in the objective space, and finally output the set of capacity expansion plans. Among them, the device operation data shows that the annual average utilization hours of the main transformers in the existing substation is 5200 hours, and the maximum load rate reaches 82%. The capacity expansion timing determined by multi - objective optimization shows that the expansion is started when the annual average utilization hours exceed 5500 hours or the maximum load rate exceeds 85%.

[0102] Extract the power grid structure data (i.e., substation configuration locations, quantities, transmission lines, etc.) from the historical distribution network grid structure data according to the substation capacity expansion plan. Taking the substations and load centers in the power grid structure data as nodes and the transmission lines and their weights = line length × unit cost as edges, use the shortest path algorithm to calculate the shortest distance from the substation to each load point. If the distance is not greater than the power supply radius threshold (such as 5 km), it is determined to be covered, and iterate to optimize the distance calculation and coverage determination process to adjust the substation location until the coverage rate ≥ 95%, and output the coordinates of the new substations and the optimized power supply radius to obtain the substation site layout optimization plan. Among them, in the substation site layout optimization, considering that the typical power supply radius of a 110 - kV substation is 3 km and the load coverage rate requirement is not less than 95%, the calculation results of the graph theory algorithm show that 2 substation sites need to be set in the commercial area with a distance of 2.5 km between stations, 1 substation site in the industrial area with a power supply radius of 3.5 km, and 1 substation site in the residential area with a power supply radius of 4 km.

[0103] Finally, check the timing consistency between the time schedule and the capacity expansion plan (such as whether the capacity expansion is carried out after the construction is completed) for conflict monitoring, and accumulate the investments of each sub - plan to ensure that the total budget does not exceed the limit. After passing the verification, combine the substation site selection optimization plan, the substation outgoing line plan, the substation construction time schedule, the main transformer capacity configuration plan, and the substation capacity expansion plan to obtain the substation optimized layout plan (for example, build 2 substations with a capacity of 100 MVA each in the first stage, and the power supply reliability reaches 99.95%. In the second stage, expand 1 substation and increase the capacity of the original substation, and the power supply reliability remains above 99.93%, and the voltage qualification rate is always higher than 98.5%) and execute it.

[0104] The present invention collaborates dynamic programming and genetic algorithms to cover the full - cycle decision - making of substation construction timing, capacity configuration, and expansion planning; adopts a multi - objective optimization algorithm to quantify equipment utilization rate and load rate, taking into account investment costs and grid resilience; fuses geographical space constraints and time - stage objectives through graph theory algorithms to achieve dynamic adaptive layout; combines the optimization results of multiple stages to ensure the executability of the plan under complex constraints; through algorithm collaboration and multi - objective optimization, the scheme realizes the intelligent upgrade of the power grid from micro - equipment configuration to macro - space - time planning, providing a scientific, flexible, and implementable planning tool for high - dynamic urban power grids.

[0105] In one embodiment, step S5 further includes:

[0106] Based on the historical load data of each sub - region, construct a load prediction function through a recurrent neural network to calculate the load growth probability under multiple scenarios, and obtain a multi - scenario load prediction matrix;

[0107] Taking the load data at different times in each scenario in the multi - scenario load prediction matrix as state variables, and taking the construction time of the new substation and the transformer capacity as decision variables, with the weighted combination of equipment utilization rate and the power supply capacity of each sub - region as the objective function, solve through the dynamic programming algorithm to obtain the substation construction decision space;

[0108] Taking the substation construction decision space as the input, establish an optimization function with the combination of construction time and transformer capacity as the decision vector through a heuristic algorithm, perform weighted calculation on the average equipment utilization rate and power supply reliability rate, and output the substation construction time schedule;

[0109] Use a sorting algorithm to prioritize the construction scale of each stage in the substation construction time schedule, and generate a construction scale allocation sequence under the constraints of voltage qualification rate and power supply reliability rate to obtain a phased construction plan to optimize the initial substation layout plan, and obtain the substation optimized layout plan and execute it.

[0110] Specifically, the present invention takes the historical load data (daily / monthly / yearly) of each sub-region, economic indicators (GDP, population), etc. as inputs and inputs them into a recurrent neural network for training. With the goal of minimizing the cross-entropy loss between the predicted load and the actual load, the trained recurrent neural network model takes the historical load data of each sub-region as input and outputs the load growth probabilities under multiple scenarios. For the medium growth scenario: the load grows linearly according to the historical trend; for the high growth scenario: the economy accelerates to drive a 12% annual increase in load; for the low growth scenario: policy restrictions result in an annual increase of ≤5%. A three-dimensional matrix (scenario × time × sub-region) is obtained, that is, a multi-scenario load prediction matrix. Among them, when constructing the load prediction function, various influencing factors need to be considered. For example, the planning of a new area shows that 350,000 square meters of commercial complexes, 500,000 square meters of high-tech industrial parks, and 250,000 square meters of residential communities will be built within the next 5 years. Based on this, the predicted average annual growth rate in the high growth scenario reaches 15%, in the medium growth scenario is 10%, and in the low growth scenario is 6%. The occurrence probabilities of each scenario are 0.3, 0.5, and 0.2 respectively.

[0111] Based on the multi-scenario load prediction matrix, the load prediction data at different times under each scenario, the capacity and location distribution of existing substations, and the equipment utilization rate are used as state variables, and the construction time of new substations and the transformer capacity are used as decision variables. With the goal of maximizing the comprehensive benefit of the weighted sum of the equipment utilization rate and the power supply capacity, a state transition equation is constructed and solved by dynamic programming recursion, that is, the optimal value function of each state is calculated backward from the end of the planning period, and the optimal decision path is recorded, and the substation construction decision space (the set of all feasible decisions) is output. Among them, the substation construction decision space includes two dimensions: the capacity of newly built substations and the expansion capacity.

[0112] Taking the combination of construction time and transformer capacity as decision variables, and the weighted equipment utilization rate and power supply reliability rate as the objective function, the construction time and capacity in the substation construction decision space are encoded. Roulette wheel selection is used to retain individuals with high fitness, arithmetic crossover is used to generate the offspring time and capacity, and Gaussian mutation is used to perturb the time points, and a Pareto optimal solution set is output to manually select the final substation construction schedule. Among them, the establishment of the objective function comprehensively considers the average equipment utilization rate and the power supply reliability rate. The weight of the average equipment utilization rate is 0.6, and the weight of the power supply reliability rate is 0.4. The substation construction schedule shows that 2 110 kV substations will be built in the first stage, the original substations will be upgraded and 1 new substation will be built in the second stage, and the construction of all 4 substations will be completed in the third stage.

[0113] Taking the load gap (demand - existing capacity), voltage qualification rate (qualified when > 95%), and power supply reliability rate (high - priority when > 99%) as sorting indicators, using a sorting algorithm to prioritize the construction scale of each stage in the substation construction schedule, calculating the comprehensive score using the TOPSIS method, taking the mandatory priority construction of areas with unqualified voltage qualification rate and the single - stage investment not exceeding 30% of the total budget as hard constraints to generate a construction scale allocation sequence, outputting a phased construction plan to be added to the initial substation layout plan for optimization, obtaining an optimized substation layout plan and implementing it; among them, the load transfer capacity analysis in the grid topology structure shows that 25% of the capacity needs to be reserved for mutual supply between adjacent substations, and capacity expansion is started when the main transformer load rate exceeds 75%. The maximum load - transfer capacity of the existing 220 - kV substation to the 110 - kV substation is 40 MVA; the optimized capacity configuration plan after synthesis is that in the first stage, both substations adopt 2 main transformers of 63 MVA, the annual utilization hours of equipment are controlled within 5000 hours, and the peak - valley difference rate does not exceed 0.4. In the second stage, the newly built substation adopts a combination of 3 main transformers of 50 MVA, and the original substation is increased in capacity to 3 main transformers of 63 MVA. In the third stage, the newly built substation adopts 2 main transformers of 63 MVA configuration. This phased construction plan not only meets the load growth demand but also ensures the economic operation of the equipment.

[0114] The present invention generates a multi - scenario load - growth probability matrix through an RNN, covering multiple possibilities under uncertainty and enhancing the planning resilience; combines dynamic programming with a heuristic algorithm to achieve the collaborative optimization of long - term construction decisions and short - term capacity configuration; ensures the priority implementation of key projects through the fusion of sorting algorithms and technical constraints; the phased construction plan directly guides the project implementation, reducing the deviation risk between planning and execution; through algorithmic chain collaboration and multi - objective decision - making, this solution realizes the full - process optimization from load forecasting to construction implementation, providing reliable technical support for the urban power grid planning with high uncertainty.

[0115] In one embodiment, after step S5, it further includes:

[0116] Obtaining the load change data and equipment operation data when implementing the optimized substation layout plan;

[0117] Performing sequence modeling on the load change data through a recurrent neural network to obtain a load - growth dynamic prediction sequence, and using a clustering algorithm to process the equipment operation data to obtain an equipment utilization state vector;

[0118] Quantifying the remaining life of the equipment based on the equipment utilization state vector using a fuzzy inference algorithm to obtain a grading result of equipment utilization rate to adjust the optimized substation layout plan, obtaining a construction plan adjustment plan and implementing it.

[0119] Specifically, the present invention collects in real time the load change data during the simulation execution of the substation optimal layout scheme in the urban expansion area, such as hourly load, peak-valley difference, growth rate and its external correlation data, such as meteorological data, holiday marks, economic indicators, etc., and equipment operation data, such as transformer load rate, oil temperature, winding temperature, insulation aging index, switch cabinet operation times, fault alarm records, line current, voltage deviation and active power loss, etc., and cleans and aligns these data.

[0120] Taking historical load data and external variables (temperature, holidays) as covariates to generate a training sample sequence to train a recurrent neural network. Using the bidirectional LSTM layer of the network to capture the bidirectional temporal dependence of load changes (such as morning and evening peaks), using its attention mechanism to determine the influence of weighted key time points (such as extreme weather days), and predicting the load curve for the next 72 hours through its output layer in multiple steps. And using MSE and peak-valley difference penalty term as the loss function during the training process, and using early stopping method to prevent overfitting to train and optimize the recurrent neural network. And taking the real-time obtained load change data as input and inputting it into the trained recurrent neural network model to output a load growth dynamic prediction sequence; taking the load rate mean / variance, temperature rise rate, cumulative operation time, etc. in the equipment operation data as numerical features, and using the fault history label (such as encoding "overload times > 3" as 1) as a categorical feature to construct a feature engineering, and clustering the equipment based on the feature engineering using the K-means algorithm to obtain the equipment utilization state vector; among them, the load dynamic prediction can also be continuously modeled using a sliding time window in units of weeks. The real-time monitoring data shows that the load growth rates of a commercial area substation in the past 4 weeks are 2.8%, 3.2%, 3.5%, and 3.8% respectively, showing a gradually accelerating trend. Through the recurrent neural network, it is predicted that the growth rate in the next 4 weeks will remain above 3.5%, and the load peak is expected to exceed 85 MVA; the equipment operation status monitoring covers multiple key parameters. The temperature of the A-phase winding of the main transformer is maintained at 55 °C, which is 5 °C higher than the same period last year. The vibration value is 0.8 mm / s, which is within the normal range, but the load rate has reached 78%. The clustering analysis results show that the transformer is in a high-load and high-temperature operation state.

[0121] Taking the device utilization status vector (such as the load rate, temperature, aging index of transformer equipment, etc.) as the input variable, and the remaining life level (short / medium / long) and recommended operations (maintenance / expansion / replacement) of the device as the output variables, using triangular / trapezoidal functions to quantify fuzzy concepts (such as "high load rate" defined as >85%), mapping the device status vector to a fuzzy set (such as load rate 78% → membership degree of "high" is 0.6, membership degree of "medium" is 0.4), calculating the activation strength of each rule (taking the minimum value of the input membership degree), and using the weighted average method to generate the quantified remaining life value (such as remaining life = 5 years) to obtain the device utilization rate grading result for dynamically adjusting the substation optimization layout plan and executing it. For example, for the short-life devices in the substation optimization layout plan, they are preferentially included in the expansion plan or arranged for replacement, etc.; for the medium-life devices, the load distribution is optimized (such as transferring part of the load to adjacent substations, etc.); for the long-life devices, the current capacity is maintained, and the operating status is monitored; after such adjustment, a construction plan adjustment plan (such as advancing the original expansion plan in the 5th year to the 3rd year, etc.) can be obtained for execution.

[0122] The present invention monitors the load and device data in real time to perform data-driven dynamic optimization, thereby improving the adaptability and real-time performance of the planning scheme; quantifies the remaining life of the device through fuzzy inference to support preventive maintenance and dynamic capacity adjustment; adopts prediction and device status-based dynamic correction of the construction plan to reduce investment waste and operation risks; through real-time data perception, intelligent algorithm reasoning and closed-loop feedback control, the scheme realizes the full-link closed-loop from power grid operation monitoring to planning dynamic optimization, providing an intelligent guarantee for the reliability and economy of the urban power grid.

[0123] In the embodiments of the present application, based on the problem of how to determine the substation layout and capacity configuration scheme when the load growth uncertainty in the planned area leads to dynamic changes in the supply-demand balance relationship, a substation layout method for urban expansion areas based on artificial intelligence algorithms is designed. It combines the geographical spatial data and historical load data of the urban expansion area, uses large language models to mine the correlation laws between the two to predict the electricity load demand in the urban expansion area during a specific future period; based on the predicted electricity load demand and the load change characteristics of the urban expansion area, combines its historical distribution network grid structure data and uses large language models to formulate the substation layout and capacity configuration scheme; simulates the power supply capacity of the urban expansion area under different load growth rates through machine learning algorithms, and uses dynamic programming algorithms to optimize the initially formulated plan to obtain the optimized plan and execute it; by comprehensively applying large language models and various algorithms, a substation layout and capacity configuration scheme for the planned area with load growth uncertainty is constructed, providing a scientific and efficient method for the power network planning of urban expansion areas, and helping to improve the power supply capacity and electricity consumption efficiency of the city.

[0124] It should be noted that although the steps in the above flowcharts are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders.

[0125] In another embodiment, as Figure 2 shown, the second aspect of the present invention provides a substation layout system for urban expansion areas based on artificial intelligence algorithms, including:

[0126] A demand prediction module 10, configured to obtain the geospatial data and historical load data of the urban expansion area, and use a large language model to mine the association rules between the geospatial data and the historical load data, obtain key features to predict the power consumption load demand data of the urban expansion area during the target period;

[0127] A zoning clustering module 20, configured to analyze the power consumption load demand data through the large language model, obtain load change characteristics to perform zoning clustering on the urban expansion area, and obtain each sub-region and its load difference data;

[0128] A scheme generation module 30, configured to collect the historical distribution network grid structure data of the urban expansion area, and combine it with each load difference data, and process it through the large language model to obtain an initial substation layout scheme;

[0129] A capacity simulation module 40, configured to analyze the historical load data through a machine learning algorithm, obtain the load growth probability distribution of each sub-region during the target period, and simulate the power supply capacity of each sub-region under different load growth rates;

[0130] A scheme optimization module 50, configured to analyze the power supply capacity of each sub-region by using a dynamic programming algorithm, optimize the initial substation layout scheme, obtain an optimized substation layout scheme and execute it.

[0131] It should be noted that each module in the above substation layout system for urban expansion areas based on artificial intelligence algorithms can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules. For the specific limitations of a substation layout system for urban expansion areas based on artificial intelligence algorithms, refer to the limitations of a substation layout method for urban expansion areas based on artificial intelligence algorithms in the above text. The two have the same functions and effects and will not be elaborated here.

[0132] In summary, the present invention relates to the field of information technology, and discloses a substation layout method and system for urban expansion areas based on artificial intelligence algorithms. By mining the correlation law between the geospatial data and historical load data of urban expansion areas, the future power load demand data of these areas is predicted; the large language model is used to analyze the predicted data, so as to divide the urban expansion areas according to the analysis results, obtain each sub-region and its load difference data, combine it with the historical distribution network grid structure data of the urban expansion areas, and after being processed by the large language model, obtain the initial substation layout plan; by simulating the power supply capacity of each sub-region under different load growth rates, the initial substation layout plan is optimized to obtain an optimized plan and execute it. By comprehensively applying artificial intelligence algorithms such as large language models and machine learning algorithms and dynamic programming algorithms, the accurate prediction of the power load demand in urban expansion areas and the effective optimization of the substation layout plan are realized. At the same time, the intelligent and dynamic optimization of urban power planning is realized, and the scientificity and adaptability of power grid planning are improved.

[0133] Each embodiment in this specification is described in a progressive manner. For the parts that are the same or similar in each embodiment, reference can be made to each other. The key point of each embodiment is to illustrate the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. It should be noted that the above technical features of the embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the above technical features in the embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0134] The above embodiments only represent several preferred implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those of ordinary skill in the art in this technical field, without departing from the technical principle of the present invention, several improvements and substitutions can still be made, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the protection scope of the claimed rights.

Claims

1. A method for laying out substations in urban expansion areas based on artificial intelligence algorithms, characterized in that Including: Obtain the geospatial data and historical load data of the urban expansion area, and use a large language model to mine the correlation rules between the geospatial data and the historical load data, so as to obtain key features to predict the electricity load demand data of the urban expansion area during the target period; Analyze the electricity load demand data through the large language model to obtain load change characteristics for zoning clustering of the urban expansion area, and obtain each sub-region and its load difference data; Collect the historical distribution network grid structure data of the urban expansion area, and combine it with each load difference data to be processed through the large language model to obtain the initial substation layout plan; Analyze the historical load data through a machine learning algorithm to obtain the load growth probability distribution of each sub-region during the target period, so as to simulate the power supply capacity of each sub-region under different load growth rates; Use a dynamic programming algorithm to analyze the power supply capacity of each sub-region to optimize the initial substation layout plan, obtain the optimized substation layout plan and execute it; The obtaining of the geospatial data and historical load data of the urban expansion area, and using a large language model to mine the correlation rules between the geospatial data and the historical load data, so as to obtain key features to predict the electricity load demand data of the urban expansion area during the target period, includes: Divide the urban expansion area into several grid cells, obtain the geospatial data and historical load data of each grid cell, and perform similarity grouping on the historical load data of each grid cell through a clustering algorithm based on dynamic time warping to obtain the grid cell load grouping result; Extract the industrial land distribution data from the geospatial data, and use the association rule mining algorithm in the large language model to quantify the support and confidence between the industrial land distribution data and the load grouping results of each grid cell, so as to obtain an association rule set representing the relationship between the industrial land types and the electricity load fluctuation patterns in the urban expansion area; Extract the time series data of economic indicators corresponding to the industrial land types from the geospatial data according to the association rule set, combine it with the industrial land distribution data and the historical load data, and process it through a third-order tensor decomposition algorithm to obtain key features representing the relationship between electricity load and industrial economic development; Based on the key features, use a recurrent neural network to predict and model the electricity load of the urban expansion area, so as to obtain the electricity load demand data of the urban expansion area during the target period.

2. The substation layout method for urban expansion areas based on artificial intelligence algorithms according to claim 1, characterized in that, The analyzing of the electricity load demand data through the large language model to obtain load change characteristics for zoning clustering of the urban expansion area, and obtaining each sub-region and its load difference data, includes: Parse the semantic text in the electricity load demand data through the semantic tokenizer in the large language model to obtain load change characteristics including load numerical characteristics and text characteristics; Extract the regional spatial layout data of the urban expansion area based on the load change characteristics, quantify the Euclidean distance similarity of the urban expansion area through the hierarchical clustering algorithm, and perform regional division based on the similarity calculation result to obtain several sub-regions; Process the historical load data of each sub-region through a conditional random field to construct a time-series state transition probability matrix including load density and load peak-valley ratio, and perform probability inference on the load growth pattern of each sub-region according to each time-series state transition probability matrix to obtain the load growth probability distribution of each sub-region; Extract the periodic and trend characteristics of the load peak-valley difference of each sub-region from each load growth probability distribution, and perform processing through the time-series decomposition algorithm to obtain the load difference data of each sub-region.

3. A method for laying out a substation in an urban expansion area based on an artificial intelligence algorithm according to claim 1, characterized in that, The analysis of the electricity load demand data by the large language model to obtain load change characteristics for zoning clustering of the urban expansion area to obtain each sub-region and its load difference data further includes: Establish a feature text dictionary including total load, load density, load growth rate, and load peak-valley ratio according to the electricity load demand data, and extract numerical features from the feature text dictionary through the word segmentation rules in the large language model to obtain a regional load prediction numerical matrix; Obtain the land use type data of the urban expansion area based on the regional load prediction numerical matrix, and calculate the Euclidean distance similarity of load density and growth rate through the spectral clustering algorithm based on the land use type data to perform unsupervised classification on the urban expansion area to obtain several sub-regions; Use the principal component analysis method to process the historical load data of each sub-region to obtain the load density principal component feature vector of each sub-region to obtain the load growth data of each sub-region, and perform regression fitting on each load growth data to obtain the load growth potential coefficient of each sub-region; Use Fourier transform to process the load peak-valley data of each sub-region to obtain the load peak-valley feature sequence of each sub-region, combine it with each load growth potential coefficient, and perform processing using the contour line drawing method based on kernel density estimation to quantify the load density gradient, load growth trend line, and load peak-valley change band to obtain the load difference data of each sub-region.

4. A substation layout method for urban expansion areas based on artificial intelligence algorithms according to claim 1, characterized in that Collect the historical distribution network grid structure data of the urban expansion area, combine it with each load difference data, and perform processing through the large language model to obtain the initial substation layout plan, including: Extract the spatial distribution characteristics of electricity load of each sub-region from each load difference data through a convolutional neural network, and use a spatial coordinate mapper to process each spatial distribution characteristic of electricity load to generate a load density heat distribution matrix of each sub-region; Obtain the historical distribution network grid structure data of the urban expansion area, and use the support vector regression algorithm to optimize the calculation of the power supply radius and load coverage rate in the historical distribution network grid structure data to obtain a candidate substation layout sequence; Based on the land use restriction conditions of each sub-region, the large language model is used to preprocess the candidate substation siting sequence, and the genetic algorithm is used to perform multi-objective optimization operations on the preprocessed candidate substation siting sequence to obtain the substation siting coordinate set; The load density heat distribution matrix of each sub-region is quantified by the hierarchical clustering algorithm to obtain the maximum load capacity of the power supply area, and the standard transformer capacity sequence is obtained from the equipment specification library. Based on the standard transformer capacity sequence, the dynamic programming algorithm is used to calculate the transformer combination scheme to obtain the main transformer capacity configuration table; Under the constraints of the power supply reliability level requirements and the load transfer requirements, based on the main transformer capacity configuration table, the graph theory algorithm is used to determine the number of outgoing lines to obtain the substation outgoing line scheme, and the substation initial layout scheme is obtained by combining with the substation siting coordinate set and the main transformer capacity configuration table.

5. A method for laying out substations in urban expansion areas based on artificial intelligence algorithms according to claim 1, characterized in that, The historical load data is analyzed by the machine learning algorithm to obtain the load growth probability distribution of each sub-region during the target period to simulate the power supply capacity of each sub-region under different load growth rates, including: The trend term, periodic term, and random term in the historical load data of each sub-region are extracted by the time series decomposition algorithm, and the extraction results are processed by the kernel density estimation method to obtain the load growth probability distribution matrix of each sub-region; Based on each load growth probability distribution matrix, the long short-term memory network is used to predict the load growth data at monthly, quarterly, and annual time scales to obtain the multi-time scale load growth trend vector; According to the multi-time scale load growth trend vector, the Monte Carlo method is used to perform probability sampling on the load levels of each sub-region under multiple growth rate scenarios to obtain the main transformer load probability distribution to quantify the average utilization hours, maximum load rate, and load duration of the equipment, and obtain the equipment utilization rate characteristic matrix; The equipment utilization rate characteristic matrix is fitted by polynomial regression to quantify the change trend of the power supply capacity of each sub-region at multiple time scales to obtain the power supply capacity of each sub-region under different load growth rates.

6. The substation layout method for urban expansion areas based on artificial intelligence algorithms according to claim 4, characterized in that The dynamic programming algorithm is used to analyze the power supply capacity of each sub-region to optimize the substation initial layout scheme, obtain the substation optimized layout scheme and execute it, including: Based on the power supply capacity of each sub-region, the dynamic programming algorithm is used to solve the multi-stage optimization function with the minimum investment cost as the goal and the equipment utilization rate not less than the preset threshold as the constraint to obtain the substation construction time schedule; According to the transformer capacity specification constraints, the genetic algorithm is used to optimize the main transformer capacity configuration table in the substation initial layout scheme to obtain the main transformer capacity configuration scheme; Based on the main transformer capacity configuration scheme, the equipment load rate data is obtained to calculate the annual average utilization hours and maximum load rate of the equipment through the multi-objective optimization algorithm to obtain the substation capacity expansion scheme; Extract power grid structure data from the historical power distribution network grid structure data according to the substation capacity expansion plan, and calculate the power supply radius and load coverage rate of the substation through graph theory algorithms to obtain an optimized substation siting plan; Combine the optimized substation siting plan, the substation outgoing line plan, the substation construction time schedule, the main transformer capacity configuration plan, and the substation capacity expansion plan to obtain an optimized substation layout plan and execute it.

7. The layout method of a substation in an urban expansion area based on an artificial intelligence algorithm according to claim 4, characterized in that The use of dynamic programming algorithm to analyze the power supply capacity of each sub-region to optimize the initial substation layout plan, obtain an optimized substation layout plan and execute it, further includes: Based on the historical load data of each sub-region, construct a load prediction function through a recurrent neural network to calculate the load growth probability under multiple scenarios, and obtain a multi-scenario load prediction matrix; According to the multi-scenario load prediction matrix, use the load data at different times in each scenario as state variables, and use the construction time and transformer capacity of the newly added substation as decision variables, with the weighted combination of equipment utilization rate and the power supply capacity of each sub-region as the objective function, and solve through the dynamic programming algorithm to obtain the substation construction decision space; Taking the substation construction decision space as the input, establish an optimization function with the combination of construction time and transformer capacity as the decision vector through a heuristic algorithm, perform weighted calculation on the average equipment utilization rate and power supply reliability rate, and output the substation construction time sequence table; Use a sorting algorithm to sort the construction scales of each stage in the substation construction time sequence table by priority, and generate a construction scale allocation sequence under the constraints of voltage qualification rate and power supply reliability rate to obtain a phased construction plan to optimize the initial substation layout plan, obtain an optimized substation layout plan and execute it.

8. The substation layout method for urban expansion areas based on artificial intelligence algorithms according to claim 1, characterized in that After using the dynamic programming algorithm to analyze the power supply capacity of each sub-region to optimize the initial substation layout plan, obtain an optimized substation layout plan and execute it, further includes: Obtain the load change data and equipment operation data when executing the optimized substation layout plan; Perform sequence modeling on the load change data through a recurrent neural network to obtain a load growth dynamic prediction sequence, and use a clustering algorithm to process the equipment operation data to obtain an equipment utilization state vector; Quantify the remaining life of the equipment based on the equipment utilization state vector using a fuzzy inference algorithm to obtain the equipment utilization rate classification result to adjust the optimized substation layout plan, obtain a construction plan adjustment plan and execute it.

9. An urban expansion area substation layout system based on artificial intelligence algorithms, characterized in that, Including: A demand prediction module, used to obtain the geospatial data and historical load data of the urban expansion area, and use a large language model to mine the correlation law between the geospatial data and the historical load data to obtain key features to predict the electricity load demand data of the urban expansion area during the target period; A partition clustering module, used to analyze the electricity load demand data through the large language model to obtain load change characteristics to perform partition clustering on the urban expansion area to obtain each sub-region and its load difference data; A scheme generation module, which is used to collect the historical distribution network grid structure data of the urban expansion area, combine it with each load difference data, and process it through the large language model to obtain the initial substation layout scheme; A capacity simulation module, which is used to analyze the historical load data through machine learning algorithms to obtain the load growth probability distribution of each sub-region during the target period, so as to simulate the power supply capacity of each sub-region under different load growth rates; A scheme optimization module, which is used to analyze the power supply capacity of each sub-region by using the dynamic programming algorithm, optimize the initial substation layout scheme, obtain the optimized substation layout scheme and execute it; Obtaining the geographical spatial data and historical load data of the urban expansion area, and using the large language model to mine the association rules between the geographical spatial data and the historical load data, obtaining key features to predict the electricity load demand data of the urban expansion area during the target period, including: Dividing the urban expansion area into several grid units, obtaining the geographical spatial data and historical load data of each grid unit, and grouping the historical load data of each grid unit by similarity through a clustering algorithm based on dynamic time warping to obtain the grid unit load grouping result; Extracting the industrial land distribution data from the geographical spatial data, and using the association rule mining algorithm in the large language model to quantify the support and confidence between the industrial land distribution data and the load grouping results of each grid unit, obtaining an association rule set representing the relationship between the industrial land types and the electricity load fluctuation patterns in the urban expansion area; Extracting the time series data of economic indicators corresponding to the industrial land types from the geographical spatial data according to the association rule set, combining it with the industrial land distribution data and the historical load data, and processing it through a third-order tensor decomposition algorithm to obtain the key features representing the electricity load and industrial economic development; Based on the key features, using a recursive neural network to predict and model the electricity load of the urban expansion area, obtaining the electricity load demand data of the urban expansion area during the target period.

Images