Power supply path dynamic generation and evaluation method based on historical load analysis

Abstract

The invention provides a historical load analysis-based power supply path dynamic generation and evaluation method, which comprises the following steps of: acquiring historical load data of a commercial district and topological structure data of a power distribution network, extracting load space distribution characteristics, and determining a dynamic change trend of load space distribution by adopting a clustering algorithm; performing time sequence analysis on the offset distance of the load center and the change rate of the power utilization density, and identifying the change period of the load in the commercial district and the sudden characteristics of sudden increase or sudden decrease of the load; key section distribution is identified according to the section identification standard, the load matching degree of the topological structure of the power distribution network is evaluated according to the key section distribution, and the load distribution balance and the line bearing capacity between topological nodes are calculated; and identifying weak nodes and areas with unbalanced load distribution according to the load distribution balance among the topological nodes and the line bearing capacity, and adjusting the line layout according to the identified weak nodes.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, and in particular to a power supply path dynamic generation and evaluation method based on historical load analysis. BACKGROUND

[0002] The adaptability evaluation of distribution network topology is a core research field of power system planning and operation optimization, which is of great significance to guarantee the reliability of urban power supply and improve the efficiency of power grid operation. With the acceleration of urbanization, the dynamic changes of load distribution put higher requirements on the distribution network, and the reasonable design of the topology directly affects the carrying capacity and adaptability of the power grid. At present, the evaluation method based on historical load analysis is relatively mature in the static scene, but in the face of the changes in the spatial distribution of load brought by urban development, the limitations of existing methods gradually appear. The main defects are that these methods often rely on fixed section identification standards, and are difficult to adapt to the dynamic changes such as load center shift, electricity density reconstruction and load growth pole in the scenes of emerging business district formation, transformation and expansion, resulting in the evaluation results deviating from the actual demand, affecting the scientificity of medium and long-term planning and the flexibility of short-term operation optimization. It needs to be clear that the network path topology, i.e. the power supply path, is the physical implementation of power transmission in the distribution network, which uniformly defines the actual flow channel of power from the source to the load, and is the core object of evaluation and optimization in this research. The core challenge is that the correlation mechanism between historical load growth pattern and network topology characteristics cannot be quantitatively characterized, making it difficult to accurately identify the influence of load changes on the adaptability of the topology structure. The lack of this quantitative mechanism further makes it impossible to establish a dynamic adjustment mechanism for the identification of key sections in response to changes in the spatial distribution of load, and the identification results of the sections lag behind the actual load changes. The static nature of the section identification standard eventually leads to a lack of effective connection between medium and long-term planning models and short-term operation optimization models, forming a coordination difficulty in the "planning and operation" two-stage decision-making. These unresolved technical factors make it difficult for the topology structure evaluation to adapt to the complex scenarios of urban development, and further cause the problems of planning deviation and reduced operation efficiency. Therefore, how to establish a topology structure evaluation model that can dynamically adjust the identification standard of key sections based on historical load analysis, and coordinate the needs of medium and long-term network path planning and short-term operation optimization, has become a key problem that needs to be solved in this research. SUMMARY

[0003] The present application provides a power supply path dynamic generation and evaluation method based on historical load analysis, mainly including:

[0004] (1) Obtain historical load data and distribution network topology data of the business district, extract the spatial distribution characteristics of the load, and determine the dynamic change trend of the spatial distribution of the load using a spatio-temporal clustering algorithm;

[0005] (2) Determine the position change of the load center of the business district at different time points by the dynamic change trend of the load space distribution to obtain the offset trajectory of the load center of the business district, obtain the electricity load per unit area in the business district, and combine the offset trajectory of the load center of the business district to reconstruct the electricity density, and obtain the offset distance of the load center of the business district and the change rate of the electricity density;

[0006] (3) Time series analysis is performed on the offset distance of the load center of the business district and the change rate of the electricity density, and the change period of the load of the business district and the burstiness characteristics of the sudden increase or decrease of the load are identified;

[0007] (4) According to the change period and burstiness characteristics of the load, the load change rate at different time points is calculated, and if the load change rate exceeds the preset change threshold, the position of the key section is recalculated based on historical load analysis to obtain a dynamically adjusted section identification standard; and a section identification standard document containing a power supply guarantee section and priority is generated in combination with the business operation characteristics;

[0008] (5) According to the section identification standard, the distribution of the key section is identified, the load matching degree of the power distribution network topology is evaluated according to the distribution of the key section, and the load distribution balance and line carrying capacity between the nodes of the topology are calculated;

[0009] (6) According to the load distribution balance and line carrying capacity between the nodes of the topology, the weak nodes and the areas with uneven load distribution are identified, and the line layout is adjusted according to the identified weak nodes;

[0010] (7) Real-time collection of load data corresponding to the adjusted line layout, if a load mutation is monitored, the section identification standard and operating parameters are dynamically adjusted, the power supply path is dynamically generated, and a short-term operation optimization scheme is formed.

[0011] Preferably, the step (1) comprises:

[0012] Obtain the transformer node topology structure database and load value from the power distribution automation master station, wherein the load value is collected by the power distribution network data collection device within a preset sampling time interval;

[0013] Obtain the load density matrix by using Gaussian kernel density estimation according to the transformer node load value and geographic coordinate information, wherein the load density matrix is associated with the land use type identifier in the geographic information library;

[0014] If the transformer node load value exceeds the load fluctuation threshold value calculated according to the historical load data, a multi-dimensional feature vector containing the load value, fluctuation amplitude and duration is generated;

[0015] The multi-dimensional feature vectors are classified by using a spectral clustering method to obtain a node group with similar power consumption characteristics, and a dynamic change trend of load space distribution in the node group is calculated by using a sliding time window.

[0016] Preferably, the step (2) comprises:

[0017] A polygon region boundary is constructed according to a set of coordinate points of the business district, a fixed edge length division method is used to generate a grid cell, and an initial load density distribution map is obtained by acquiring power consumption load data in the grid cell from the power distribution device;

[0018] For the initial load density distribution map, a weighted average operation is performed with the power consumption load value of the grid cell as the weight to obtain a load center coordinate sequence of the business district;

[0019] The load center coordinate sequence is smoothed by using a cubic spline interpolation method, and a time-series load center position point set is obtained by sampling the smoothed curve at a fixed time interval;

[0020] The unit area power consumption load is calculated according to the grid cell area and the power consumption load value, and the unit area power consumption load is spatially reconstructed by using a bilinear interpolation method to obtain a power density distribution map;

[0021] The Euclidean distance between two points is calculated by using the load center positions of adjacent time points to obtain an original offset distance, and the original offset distance is normalized according to the diagonal length of the business district boundary to obtain a standardized offset distance;

[0022] The power density change rate is calculated according to the standardized offset distance and the difference between the power density distribution maps of adjacent time points.

[0023] Preferably, it further comprises: calculating the load center position at different time points according to the load space distribution trend, extracting the coordinate change sequence of the load center position at different time points, constructing a load center offset trajectory according to the coordinate change sequence, obtaining power consumption load data within a unit area of the business district, spatiotemporally matching the power consumption load data with the offset trajectory, reconstructing the power density distribution of each region according to the spatiotemporal matching result, and extracting the gradient change rate of the power density distribution at the corresponding time point, specifically including:

[0024] Load sampling data and geographic position coordinate information are obtained, and the load sampling data is collected by a power distribution automation master station;

[0025] A load weighted average center coordinate sequence is calculated according to the load sampling data and the geographic position coordinate information;

[0026] A load center offset trajectory curve is obtained by curve fitting the load weighted average center coordinate sequence by using a cubic spline interpolation method.

[0027] The tangent direction and curvature value of each sampling time point are calculated according to the load center offset trajectory curve, and trajectory feature data is obtained;

[0028] The trajectory feature data is subjected to spatial grid node interpolation operation by using the inverse distance weighted interpolation method, and an electricity density distribution map is obtained;

[0029] The electricity density difference value between adjacent time points is calculated according to the electricity density distribution map, and density change gradient data is obtained.

[0030] Preferably, the step (3) comprises:

[0031] The distance difference value sequence between adjacent sampling points is calculated according to the load center coordinate time series data, and the frequency spectrum signal is obtained by using discrete Fourier transform on the distance difference value sequence;

[0032] The electricity density change rate sequence is subjected to sliding time window smoothing processing, and the residual sequence is obtained by calculating the difference between the smoothed curve and the original curve at each sampling point;

[0033] The mutation position is marked according to the sampling point whose residual sequence exceeds the three times threshold of standard deviation, and the data segment is obtained by extending the mutation position forward and backward for a fixed time length, and the change amplitude is obtained by calculating the difference between the initial value and the terminal value of the electricity density in the data segment;

[0034] The change rate is calculated according to the change amplitude and time interval, the mutation feature vector is obtained by symbol judgment on the change rate and change amplitude, and the periodic pattern recognition of the time interval of adjacent mutation events in the mutation feature vector sequence is carried out by using autocorrelation analysis.

[0035] Preferably, the step (4) comprises:

[0036] The load change rate of adjacent sampling points is calculated by using a fixed length time window, and the standardized load change rate sequence is obtained by subtracting the mean value from the load change rate sequence and dividing by the standard deviation;

[0037] The load mutation marker is obtained according to the comparison result of the standardized load change rate sequence and the early warning reference value, and the smoothed historical load curve is obtained by weighted average on the historical load data before the load mutation marker time by using a decreasing coefficient sequence;

[0038] The load density distribution is calculated according to the smoothed historical load curve, and the key section initial position is obtained by processing the load density distribution by using the maximum gradient method;

[0039] According to the key section initial position setting search interval, the density estimation operation is performed on the load data in the search interval to obtain a density peak position, the density peak position is determined as the key section accurate position, and a standardized section identification parameter is generated as a new section identification standard.

[0040] Obtain load anomaly records during commercial holidays and promotion activities, analyze commercial district new large-scale power equipment commissioning, property expansion and tenant format adjustment plans, locate key line points of peak period power flow changes, interface commercial center regional power supply transformer load rate data, and generate section identification standard documents containing commercial peak period power supply guarantee sections, core merchant uninterrupted power supply priority, and adjustable load capacity, including:

[0041] Obtain holiday and promotion activity calendar in the commercial district operation database, extract load data corresponding to the period according to the activity calendar, and obtain load anomaly records by using three standard deviations for screening;

[0042] According to the load anomaly records and tenant format adjustment information, the unit area power load benchmark value is calculated, and the load change sequence is generated for the planning area of the commercial district according to the benchmark value;

[0043] The Monte Carlo method is used to sample the load change sequence, and the line current density value is obtained by power flow calculation. If the current density change amount and the benchmark value ratio exceed the threshold value, the line node coordinates are marked;

[0044] Obtain transformer load rate data in the power supply monitoring device, and calculate the transformer load rate threshold value according to the load rate data;

[0045] Calculate the total power capacity and interruptible load capacity of each load unit group, and generate a substation power supply scheme containing guarantee sections, priority sequences, and adjustable capacities.

[0046] Preferably, the step (5) comprises:

[0047] According to the section coordinate points, a cubic spline interpolation method is used to generate a section distribution curve, and the section distribution curve is used to calculate the equidistant sampling point coordinate set;

[0048] The equidistant sampling point coordinate set is subjected to a Thiessen polygon division operation to obtain a set of partition boundary points in the section coverage range;

[0049] According to the set of partition boundary points, a power distribution network topology graph is constructed, and a maximum flow algorithm is used to calculate the maximum allowable load flow value between nodes in the power distribution network topology graph;

[0050] A power flow calculation is performed on the power distribution network topology to obtain line power values, and if the line power values are divided by line rated capacity values, line load rate indicators are obtained.

[0051] Preferably, the step (6) comprises:

[0052] According to the load balancing degree index, the nodes are meshed and divided, the load density values in the grid cells are obtained by using a density calculation method, and a node distribution density map is obtained by mean standardization processing.

[0053] A local spatial statistical method is used to calculate the load density difference values of adjacent grid cells for the node distribution density map, and if the load density difference values exceed twice the overall standard deviation, the region is marked as a load imbalance region.

[0054] By calculating the load rate of each node in the load imbalance region, if the node load rate exceeds the transformer rated capacity threshold, the node is marked as a weak node.

[0055] According to the load difference value and the distance between the weak node and the adjacent node, a load transfer index is calculated, the node with the highest load transfer index and the load rate lower than the preset percentage is selected as a tie node, and the path coordinate sequence of the new line is generated based on the shortest channel distance principle.

[0056] Preferably, the step (7) comprises:

[0057] Load time series data is generated according to the acquisition time tag of the power distribution automation master station, and a smoothed load sequence is obtained by exponential smoothing processing of the load time series data.

[0058] The load change amount is calculated by calculating the difference value of adjacent data points for the smoothed load sequence, the load change rate sequence is obtained by dividing the load change amount by the acquisition time interval, and the standard deviation value is obtained according to the load change rate sequence.

[0059] The load data before and after the moment when the load change rate exceeds the standard deviation threshold is obtained, and the load mean and variance are obtained by calculating the load data as the correction reference value of the section identification parameter;

[0060] The load threshold in the section identification standard is updated using the correction reference value, and the power supply partition boundary is divided by the Voronoi polygon method according to the load threshold, and the power supply partition boundary is used to determine the power distribution network subnet division, and the rated capacity and actual load of the interconnection line between the partitions are calculated.

[0061] A plurality of power supply path alternative schemes are generated by using the shortest path algorithm, the node voltage deviation and line load rate of each scheme are calculated, and a short-term operation optimization scheme is formed.

[0062] The technical scheme provided by the embodiment of the present application can include the following beneficial effects:

[0063] The application discloses a power supply path dynamic generation and evaluation method based on historical load analysis, extracts load space distribution features by acquiring historical load and topology structure data, determines load change trend by adopting a time-space clustering algorithm, analyzes load center offset trajectory and power density change, performs time series analysis on load center offset distance and power density change rate, identifies load change period and burst features, dynamically adjusts section identification standard, identifies key section distribution according to the adjusted standard, evaluates topology structure load matching degree, identifies weak nodes and adjusts line layout, monitors load data in real time, dynamically adjusts operation parameters when load mutates, and generates a short-term optimization scheme. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 A flowchart of the power supply path dynamic generation and evaluation method based on historical load analysis of the present application.

[0065] Figure 2 A schematic diagram of the power supply path dynamic generation and evaluation method based on historical load analysis of the present application. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application is described in detail below in combination with the drawings and specific embodiments.

[0067] As Figures 1-2 , the power supply path dynamic generation and evaluation method based on historical load analysis of the present embodiment can specifically include:

[0068] In the embodiment of the present application, the power distribution network system includes transformer nodes, power supply lines and load points, the power supply path is defined as a physical transmission channel of electric power from a source point to a load. Load data is divided into historical load data and real-time load data, the historical load data includes node load value, geographic coordinates and time series change, and the real-time load data is used to monitor load mutation. Load space distribution features reflect the time-space change rule of power demand in a business district, and a key section refers to a key line point position affecting load matching degree of the power supply path. The embodiment of the present application dynamically adjusts section identification standard by analyzing historical load data, optimizes power supply path generation, and improves adaptability and operation efficiency of the power distribution network.

[0069] S101, acquire historical load data and power distribution network topology structure data of a business district, extract load space distribution features, and determine dynamic change trend of the load space distribution by adopting a time-space clustering algorithm.

[0070] In the embodiment of the present application, the dynamic evaluation function of the power distribution network can be started by the power distribution automation master station or a special application program, and the specific starting mode can be determined according to the actual scene. The user can start the evaluation function through the master station interface operation or through a remote instruction trigger. After the function is started, the historical load data and the topological structure data of the commercial area are obtained from the power distribution automation master station, the historical load data includes the load value, geographic coordinates and time sequence record of the transformer node, and the topological structure data includes the connection relationship between nodes and the line parameters.

[0071] S1011, the load value and geographic coordinates of the transformer node are obtained from the power distribution automation master station, and a load basis matrix is generated in combination with the sampling time interval.

[0072] In the embodiment of the present application, the transformer node load value and latitude and longitude coordinates are collected by the power distribution network data collection device at a preset sampling interval, such as 10 minutes. The load value reflects the node power demand, and the geographic coordinates are used to locate the node position. According to the collected data, a load basis matrix is generated, and the matrix elements include the load value and coordinate information of each node at different time points.

[0073] S1012, a load density matrix is generated based on the load basis matrix using Gaussian kernel density estimation, and the land use type in the geographic information library is associated.

[0074] In the embodiment of the present application, Gaussian kernel density estimation is performed on the load basis matrix, a bandwidth such as 500 meters is selected, and a load density matrix reflecting the regional power consumption density is generated. The land use type of the area where the node is located, such as commercial office, retail or catering entertainment, is obtained from the geographic information library, the load density is associated with the land use type, and the power consumption characteristics of different areas are identified.

[0075] S1013, if the node load value exceeds the fluctuation threshold value calculated based on the historical load data, a multi-dimensional feature vector is generated, and a spectral clustering method is used to divide the node group.

[0076] In the embodiment of the present application, the load peak-to-valley ratio is calculated based on the historical load data, and the fluctuation threshold value is set to 2.8. If the node load value exceeds the threshold value, it is marked as a peak period, and a multi-dimensional feature vector containing the load value, fluctuation amplitude and duration is generated. Based on the feature vector, a similarity matrix is constructed, and a spectral clustering method is used to divide the nodes into groups with similar power consumption characteristics.

[0077] S1014, the load capacity of the node group is analyzed in time sequence, and a density-based spatial clustering algorithm is used to divide the load change similar area in combination with the geographic position.

[0078] In the embodiment of the present application, the load capacity in each group is analyzed in time sequence by using a sliding time window such as 6 hours, and the load change rate is calculated. In combination with the geographical position of the node, the nodes with similar load changes and similar geographical positions are divided into the same area by using a density-based spatial clustering algorithm, and the dynamic change trend of the load space distribution in the business district is identified.

[0079] In the embodiment of the present application, by extracting the load space distribution features and performing spatio-temporal clustering analysis, the dynamic change law of the power demand in the business district can be accurately identified, and reliable data basis can be provided for the load center offset analysis and section adjustment. Compared with the traditional static evaluation method, the embodiment of the present application can better adapt to the complex changes of the urban load distribution, and improve the accuracy and practicability of the evaluation results.

[0080] S102, based on the dynamic change trend of the load space distribution, the offset trajectory of the load center in the business district and the power density change rate are calculated.

[0081] In the embodiment of the present application, by analyzing the dynamic change trend of the load space distribution, the position change of the load center in the business district at different time points is determined, and a smooth offset trajectory curve is generated. At the same time, in combination with the power load data per unit area in the business district, an interpolation method is used to reconstruct the power density distribution, and the power density change rate is extracted, which provides a key basis for load change period analysis and section adjustment. Through accurate quantification of the load center offset and the power density change, the embodiment of the present application can effectively reflect the spatio-temporal dynamic characteristics of the power demand in the business district.

[0082] As Figure 2 S1021, the load sampling data and geographical position coordinates of the power distribution terminal in the business district are obtained from the power distribution automation master station, a load base matrix is generated, and a load weighted average center coordinate sequence is calculated.

[0083] In the embodiment of the present application, the power distribution automation master station collects the load value and geographical coordinate information of each node in the business district through the power distribution terminal at a fixed time interval such as 10 minutes. The load value reflects the power demand of the node at a specific time point, and the geographical coordinates are used to locate the spatial position of the node in the business district. According to the collected load value and coordinate information, a load base matrix is generated, and the matrix elements include the load value and latitude and longitude coordinates of each node at different time points. For the load base matrix, a weighted average method is used, and the node load value is used as the weight to calculate the coordinates of the load center in the business district, and a load center coordinate sequence sorted by time sequence is generated. For example, in a business district including 20 power distribution terminals, the load value of a terminal at 8 o'clock in the morning is 350 kW, and the geographical coordinates are (500, 600). The load center coordinates are calculated by weighted average, which reflect the position of the regional power concentration.

[0084] S1022, the load center coordinate sequence is smoothed by using a cubic spline interpolation method, a load center offset trajectory curve is generated, and trajectory feature data is extracted.

[0085] In the embodiment of the application, the discrete load center coordinate sequence is processed by cubic spline interpolation to generate a continuous and smooth offset trajectory curve, so as to eliminate the influence of noise in the coordinate sequence. The curve is sampled at a fixed time interval, for example, 10 minutes, to obtain a time-series load center position point set. For the offset trajectory curve, the tangent direction angle and the curvature value of each sampling time point are calculated to generate trajectory feature data, including time stamp, position coordinate, offset direction and curvature. For example, between 8 am and 9 am, the load center offsets from the office building area to the retail area, the tangent direction angle changes from 20 degrees to 50 degrees, and the curvature value is 0.015, indicating that the offset direction changes relatively gently. The trajectory feature data can directly reflect the dynamic migration law of the load center in the commercial area.

[0086] S1023, based on the trajectory feature data, the offset trajectory is fitted by using the least square method to calculate the standardized offset distance.

[0087] In the embodiment of the application, based on the time-series load center position point set, the offset trajectory is fitted by using the least square method to generate a smooth trajectory function. The Euclidean distance of the load center positions of adjacent time points is calculated to obtain the original offset distance, and the distance is normalized based on the diagonal length of the commercial area boundary to generate the standardized offset distance. For example, in a commercial area with a side length of 2 kilometers, the load center offset distance in a certain time period is 400 meters, and the standardized offset distance after normalization is 0.2. The standardized offset distance can quantify the moving amplitude of the load center, and provide a basis for subsequent power density change analysis.

[0088] S1024, the power load data of each area in the commercial area is collected from the power distribution terminal, the unit area power load is calculated, and the power density distribution is reconstructed by using the bilinear interpolation method.

[0089] In the embodiment of the present application, the real-time power consumption load value of each area in the commercial area is collected by the power distribution terminal, and the unit area power consumption load is calculated in combination with the area of the terminal coverage area. For example, the load value of a power distribution terminal covering an area of 2000 square meters is 500 kilowatts, and the unit area power consumption load is 0.25 kilowatts per square meter. Based on the set of boundary coordinate points of the commercial area, a polygonal area boundary is constructed, and a square grid unit is divided with a fixed side length, such as 500 meters. For the power consumption load data of each grid unit, a bilinear interpolation method is used for spatial reconstruction to generate a continuous power density distribution map. The power density distribution map can intuitively show the differences in power intensity in different areas in the commercial area, such as the power density of the retail area in the afternoon period reaching 0.3 kilowatts per square meter, and the power density of the catering area at night reaching 0.28 kilowatts per square meter.

[0090] S1025, according to the power density distribution map and the offset trajectory, the power density difference value of adjacent time points is calculated, and the power density change rate is extracted.

[0091] In the embodiment of the present application, the power density distribution map of each time point on the load center offset trajectory is time-sequentially aligned, the power density difference value of adjacent time points is calculated, and the density change gradient data is generated. Based on the gradient data, the projection component of the power density along the offset trajectory direction is calculated, and the power density change rate sequence is generated. For example, from 10 o'clock to 10 o'clock and 10 minutes, the load center offsets to the commercial retail area, the standardized offset distance is 0.15, the power density difference value is 0.06 kilowatts per square meter, and the change rate is 0.36 kilowatts per square meter per hour. The high and low of the change rate reflects the dynamic adjustment speed of the power intensity of the commercial area, especially during the morning and evening peak periods, the change rate can reach 0.5 kilowatts per square meter per hour, indicating that the commercial activity migration leads to rapid changes in power demand.

[0092] In the embodiment of the present application, by constructing the load center offset trajectory and reconstructing the power density distribution, the spatio-temporal variation characteristics of the power demand of the commercial area can be accurately quantified. Compared with the traditional method relying on static load analysis, the embodiment of the present application uses dynamic interpolation and trajectory fitting technology, which can more accurately capture the movement law of the load center and the change trend of the power density. For example, in a commercial complex of 4 square kilometers, the load center is located in the office building area at 8 o'clock in the morning on weekdays, offsets 500 meters to the retail area at 12 o'clock noon, and offsets 300 meters to the catering area at 18 o'clock at night, and the offset angle changes by 80 degrees. The power density change rate is closely related to the load center offset, and when the standardized offset distance exceeds 0.1, the power density change rate usually exceeds 0.04 kilowatts per square meter per hour. This dynamic analysis method can provide a reliable basis for key section adjustment, significantly improve the adaptability of the power distribution network to the complex load changes of the commercial area, and optimize the generation efficiency of the power supply path.

[0093] S103, time series analysis is performed based on the business district load center offset distance and the power density change rate to identify the load change period and the burst characteristics.

[0094] In the embodiment of the present application, by analyzing the time series of the load center offset distance and the power density change rate, the periodicity of the business district load change and the burst characteristics of the sudden increase or decrease are extracted. The burst characteristics include the change rate, the change amplitude and the duration, which are used to describe the dramatic fluctuation behavior of the load. The embodiment of the present application adopts the frequency domain analysis and residual analysis method combined with autocorrelation analysis to accurately identify the periodic pattern and abnormal events of the load change, and provides a reliable basis for dynamically adjusting the key section identification standard.

[0095] S1031, according to the load center coordinate time series data, the offset distance difference sequence between adjacent time points is calculated, and the discrete Fourier transform is used to extract the load change period.

[0096] In the embodiment of the present application, the spatial coordinates of adjacent time points are extracted from the load center coordinate time series data, the Euclidean distance between the two points is calculated, and the offset distance difference sequence is generated. For the sequence, the discrete Fourier transform is used to convert it to the frequency domain to generate the frequency spectrum signal. The frequency component with the maximum amplitude is identified from the frequency spectrum, and its reciprocal is taken as the basic period of the load center movement. For example, in the 24-hour monitoring data of a business district, the frequency spectrum shows that the main peak value is located at 0.0417Hz, and the corresponding period is about 24 hours, which reflects the daily periodic change of the load center with business activities. The frequency domain analysis can effectively capture the long-term regularity of the load change and provide a basis for the quantification of the periodic characteristics.

[0097] S1032, the power density change rate sequence is smoothed by using a sliding time window to generate a residual sequence and mark the mutation points.

[0098] In the embodiment of the present application, for the power density change rate sequence, a fixed-size sliding time window such as 30 minutes is used for smoothing to generate a smooth curve. The difference between the smooth curve and the original curve at each time point is calculated to generate a residual sequence. The standard deviation of the power density change rate is determined based on the historical data, and the mutation judgment threshold is set to 3 times the standard deviation, such as 0.15 kilowatts per square meter per hour. If the value of a time point in the residual sequence exceeds the threshold, it is marked as a mutation point. For example, at 8:30 am, the residual value reaches 0.18 kilowatts per square meter per hour, which exceeds the threshold, and is marked as a mutation point, indicating that the power density has changed significantly. The smoothing and residual analysis can effectively filter out noise and highlight the characteristics of the mutation event.

[0099] S1033, data segments are extracted for the mutation points, the change amplitude and the change rate are calculated, and a mutation feature vector is generated.

[0100] In the embodiment of the application, a data segment is obtained by extending forward and backward from the mutation point for a fixed time length, such as 20 minutes. The difference between the initial value and the terminal value of the power consumption density in the segment is calculated to obtain the change amplitude. The change rate is obtained by dividing the change amplitude by the time interval. The mutation type is determined by the sign judgment, such as positive sudden increase or negative sudden decrease, and a mutation feature vector is generated, including the mutation type, the change rate, the change amplitude, and the duration. For example, at 8 o'clock in the morning, the data segment shows that the power consumption density rises from 0.12 kW / m2 to 0.35 kW / m2, the change amplitude is 0.23 kW / m2, the duration is 15 minutes, the change rate is 0.92 kW / m2 / h, and the feature vector records a sudden increase event. The mutation feature vector provides a structured description of the quantitative burstiness feature.

[0101] S1034, the time interval of the mutation feature vector sequence is identified for periodic pattern by using autocorrelation analysis.

[0102] In the embodiment of the application, the time interval of adjacent mutation events is extracted from the mutation feature vector sequence to generate a time interval sequence. The autocorrelation function of the time interval sequence is calculated by using the autocorrelation analysis method to identify the periodic pattern. For example, in the working day data, the autocorrelation function shows a significant peak at 14 hours, indicating that the mutation events occur periodically at 14 hours, reflecting the morning and evening peak power consumption rules. In the weekend data, the autocorrelation function has no obvious periodicity, and the mutation events are randomly distributed. Autocorrelation analysis can effectively reveal the timing rules of mutation events and provide a basis for load behavior prediction.

[0103] In the embodiment of the application, the periodicity and burstiness features of the commercial area load change are accurately identified by time series analysis, which can effectively depict the dynamic rules of power consumption behavior. For example, in a typical commercial complex, the load center increases in the morning from 8 o'clock to 9 o'clock due to the increase in power consumption demand of the office area, triggering a sudden increase event, and the power consumption density change rate reaches 0.9 kW / m2 / h, lasting for 12 minutes. At 22 o'clock in the evening, a sudden decrease event occurs due to the closing of the catering area, with a change amplitude of 0.2 kW / m2 and a duration of 18 minutes. Periodic analysis shows that the mutation events in the working day have a basic period of 24 hours, which is consistent with the business operation rules. Compared with the traditional static analysis method, the embodiment of the application can more accurately capture the periodicity and abnormal fluctuation features of the load by frequency domain analysis and residual analysis, providing key data support for the dynamic optimization of the power distribution network, and significantly improving the response speed of key section adjustment and the adaptability of the power supply path.

[0104] S104, calculate the load change rate according to the commercial area load change period and burstiness characteristics, if the change rate exceeds the preset threshold, dynamically adjust the key section identification standard based on historical load analysis, and generate a section identification standard document containing power supply guarantee section and priority based on commercial operation characteristics.

[0105] In the embodiment of the application, by analyzing the load change period and burstiness characteristics, the load change rate at different time points is calculated and compared with the preset threshold to identify significant changes. Based on historical load data, the key section position is dynamically adjusted, and the section identification standard suitable for dynamic load change is generated. In addition, combined with holiday, promotion activity and format adjustment and other operation characteristics, the section identification standard document containing power supply guarantee section, merchant priority and adjustable load capacity is generated. The embodiment of the application can improve the accuracy of key section identification and the adaptability of power distribution network to complex electricity demand of commercial area by combining dynamic analysis and operation data.

[0106] S1041, calculate the load change rate using a fixed length time window and perform standardization processing, and mark the load mutation time.

[0107] In the embodiment of the application, a fixed 60-minute sliding time window is used to calculate the load change rate between adjacent sampling points, and a load change rate sequence is generated. The sequence is standardized by subtracting the mean and dividing by the standard deviation to obtain a dimensionless standardized load change rate sequence. The 95% quantile of the load change rate is extracted from the historical data as a warning reference value, for example, 0.3 kilowatt per hour. If the standardized load change rate exceeds the reference value, it is marked as a load mutation time. For example, at 8 o'clock in the morning in a certain commercial area, the load rises from 250 kilowatts to 400 kilowatts, and the standardized load change rate is 2.5, which exceeds the reference value, so the system records this time as a mutation point. Standardization processing eliminates the difference in load base of different time periods, making the change rate comparable across time periods, and providing a reliable trigger condition for section adjustment.

[0108] S1042, generate a smooth curve based on the historical load data of the mutation time, and determine the key section position using the maximum gradient method and density estimation.

[0109] In the embodiment of the present application, for the load mutation marking moment, the historical load data of the previous 24 hours is traced back, a decreasing coefficient sequence such as 0.95 is used for weighted average, a smooth historical load curve is generated, and data close to the mutation moment is given higher weight. Based on the smooth curve, the load density distribution is calculated, and the maximum gradient method is used to identify the density jump point as the initial position of the key section. For example, at the junction of the office area and the retail area, the load density gradient reaches 0.12 kilowatts per square meter per meter, which is marked as the initial section position. A 100-meter search interval is set on both sides of the initial position, and the load data in the interval is subjected to kernel density estimation, and the density peak position such as the offset of 20 meters and the density of 0.45 kilowatts per square meter is selected as the accurate position of the key section. The smooth section reference curve is generated by using the cubic spline interpolation to connect multiple section positions, and the normalized section identification parameter is formed after normalization. This method can accurately capture the spatial variation of load distribution and improve the accuracy of section identification.

[0110] S1043, extract holiday and promotion activity data from the commercial area operation database, filter load anomaly records and predict load changes.

[0111] In the embodiment of the present application, the holiday and promotion activity calendar is obtained from the commercial area operation database, the load data of the corresponding period is extracted, and the three-sigma method is used to filter the load anomaly records. For example, during the Spring Festival promotion period, the load suddenly increases from an average of 350 kilowatts to 800 kilowatts, exceeding the threshold of 600 kilowatts of mean value plus three times the standard deviation, and is marked as an anomaly record. Combined with the information of newly added large power equipment such as a 400 kilowatt central air conditioner, property expansion area such as 2000 square meters, and tenant format adjustment, the unit area power load benchmark value of each format is calculated, such as 0.22 kilowatts per square meter for the catering area and 0.28 kilowatts per square meter for the cinema. The planning area of the commercial area is predicted according to the format, and a change sequence containing time stamp and load capacity is generated. This prediction method can identify potential high-load scenarios in advance and provide forward-looking basis for section adjustment.

[0112] S1044, mark key line nodes using the Monte Carlo method and power flow calculation to determine the power supply guarantee section.

[0113] In the embodiment of the present application, 1000 times of Monte Carlo probability sampling is performed on the load change sequence, and the line current density is obtained by combining the power flow calculation. If the ratio of the current density change amount to the reference value exceeds 80%, the corresponding line node coordinates are marked. For example, during the peak period of the retail area, the current density change amount of a certain line reaches 85%, and its node is marked as a key point. The transformer load rate data is obtained from the power supply monitoring device, and the load rate threshold is calculated, such as 90% of the rated capacity of 800 kW, i.e. 720 kW. If the load rate exceeds the threshold, such as 93% in the supermarket area, it is marked as a power supply guarantee section. Marking the key nodes and guarantee sections can clearly identify the key areas of power flow change and optimize the pertinence of section identification.

[0114] In the embodiment of the present application, by analyzing the power consumption contribution value of the merchant and the characteristics of the load unit, a substation power supply scheme is generated. The annual power consumption and turnover of the merchant are collected, the power consumption contribution value per unit turnover is calculated, such as 0.018 kWh per yuan for supermarkets, and the highest power supply priority is divided. The load unit is divided for the power supply guarantee section, a similarity matrix is constructed based on the physical distance between units, such as 50 meters, and the electrical connection, and a hierarchical clustering method is used to group units with a distance of less than 100 meters and electrical connection. The total power consumption capacity and interruptible load capacity of each group are calculated, such as the total capacity of 1800 kW in the core area, of which 500 kW of lighting load is interruptible. The generated section identification standard document includes guarantee sections, priority sequences and adjustable capacity, providing comprehensive guidance for dynamic optimization of distribution networks. For example, during the holiday peak, the system adjusts the section location according to the abnormal records to ensure uninterrupted power supply in the supermarket area, while optimizing the load distribution of the line through the adjustable capacity to improve the power supply reliability and operating efficiency.

[0115] S105, according to the dynamically adjusted section identification standard, identify the key section distribution of the commercial area, and evaluate the load matching degree and line carrying capacity of the distribution network topology.

[0116] In the embodiment of the present application, by applying the dynamically adjusted section identification standard, the key section distribution of the commercial area is identified, and a section coverage network reflecting the load space characteristics is generated. Based on the section distribution, a distribution network topology graph is constructed, and the load distribution balance and line carrying capacity between the nodes of the topology are evaluated by using the maximum flow algorithm and power flow calculation. The embodiment of the present application can accurately evaluate the adaptation degree of the distribution network and the dynamic load of the commercial area through multi-dimensional quantitative analysis, and provide a scientific basis for optimizing the power supply path.

[0117] S1051, based on the section identification standard, extract the key section coordinates, generate the section distribution curve by using cubic spline interpolation, and perform grid division.

[0118] In the embodiment of the present application, the key section coordinate point set of the commercial area is extracted from the standard document of the section, for example, a set of 25 coordinate points, covering the boundary of the main functional division. The cubic spline interpolation method is used to connect these coordinate points to generate a smooth section distribution curve, ensuring the continuity and spatial representativeness of the curve. Along the curve, sampling is performed at a fixed interval, such as 50 meters, to generate an equidistant sampling point coordinate set. Based on the sampling point coordinates, a section grid layer is constructed, and the Thiessen polygon division method is used to divide the commercial area into multiple subareas, with each subarea centered on a section point and the boundary being the perpendicular bisector of adjacent section points. For example, in a 6000 square meter commercial area, 5 Thiessen polygon subareas are divided, each with different load distribution characteristics. This division method can naturally reflect the spatial characteristics of the section coverage range and provide an accurate spatial framework for subsequent topological analysis.

[0119] S1052, constructing a power distribution network topology map according to the subarea boundaries, and calculating the load matching coefficient between nodes using the maximum flow algorithm.

[0120] In the embodiment of the present application, based on the boundary point set of the Thiessen polygon subarea, the nodes of the power distribution network within the subarea are numbered, the number and capacity of the power supply channels between nodes are recorded, and a power distribution network topology map is constructed. The topology map represents transformers or load points as nodes and power supply lines as edges, with the weight of the edge being the rated capacity of the line. The maximum flow algorithm is used to calculate the maximum allowed load flow value between adjacent nodes in the topology map. For example, a transformer node is connected to 3 lines with a total rated capacity of 1200 kW, and the maximum flow algorithm calculates the maximum allowed load as 1000 kW. The actual load value of 700 kW divided by the maximum allowed load gives a load matching coefficient of 0.7, indicating that the load utilization of this node is moderate. The load matching coefficient can quantify the adaptation degree of node carrying capacity and actual load, providing a key indicator for evaluating network performance.

[0121] S1053, obtaining the line load rate through power flow calculation and generating a carrying capacity evaluation index combined with the load distribution dispersion.

[0122] In the embodiment of the present application, the power flow calculation is performed based on the power distribution network topology graph to obtain the power flow direction and amplitude of each line. The line load rate is calculated by dividing the actual load value by the rated capacity. For example, for a line with a rated capacity of 1000 kW, the actual power flow is 650 kW, and the load rate is 0.65. The load distribution proportion of all nodes in the topology graph is analyzed, the standard deviation of the proportion sequence is calculated, and the load distribution dispersion degree is obtained, such as 0.08, which reflects the balance of load distribution. The line load rate is weighted and calculated in combination with the load distribution dispersion degree to generate a comprehensive load bearing capacity evaluation index. For example, when the dispersion degree is 0.1, the comprehensive index is 0.68, indicating that the line has strong load bearing capacity. This multi-dimensional evaluation method can comprehensively reflect the load balance and line operation state of the network.

[0123] In the embodiment of the present application, through key section distribution and topology analysis, the load matching degree and load bearing capacity of the power distribution network can be effectively evaluated. For example, in a typical commercial complex, the load matching coefficient of the core retail area is 0.75, and the load distribution dispersion degree is 0.05, indicating that the power supply capacity and power demand are highly matched. The load matching coefficient of the newly developed area is only 0.55, and the dispersion degree is 0.12, indicating that the line capacity needs to be optimized. The power flow calculation shows that the load rate of the main line in the retail area is 0.7 during the peak period, but it is still lower than the rated capacity, and the operation is stable. Compared with the traditional static evaluation, the embodiment of the present application can more accurately capture the spatial heterogeneity of the load in the commercial area through dynamic section identification and multi-dimensional quantitative analysis, and significantly improve the scientificity of the power distribution network optimization decision and the guarantee ability of the power supply reliability.

[0124] S106, according to the load distribution balance and line load bearing capacity between the nodes of the power distribution network topology, identify weak nodes and load imbalance areas, and adjust the power supply network structure by optimizing the line layout.

[0125] In the embodiment of the present application, by analyzing the load distribution balance and line load bearing capacity between the nodes of the topology, weak nodes and areas with high load or uneven distribution are identified. Based on the identification result, the load transfer index is calculated, the line layout is optimized, the new line path is generated, and the operation stability and resource utilization efficiency of the power distribution network are improved through network reconstruction. The embodiment of the present application can effectively alleviate the load imbalance problem and enhance the adaptability of the commercial area power distribution network through fine load analysis and dynamic planning.

[0126] S1061, based on the load balance degree index, the nodes are divided into grids to generate a node distribution density map and identify load imbalance areas.

[0127] In an embodiment of the present application, according to the load balancing index, the business district is divided into 100m by 100m grid units, and the load density value in each unit is calculated, i.e. the transformer load value divided by the unit area. The load density value is subjected to mean standardization processing to generate a node distribution density map reflecting the load space distribution characteristics. The local spatial statistical method, such as the local Moran index, is used to calculate the load density difference of adjacent grid units. If the difference value exceeds twice the overall standard deviation, for example, 0.1 kW per square meter, the area is marked as a load imbalance area. For example, in a certain commercial center, the load density difference between the retail area and the surrounding area reaches 0.25 kW per square meter, far exceeding twice the standard deviation of 0.09 kW per square meter, and is marked as an imbalance area. This method can accurately locate the spatial heterogeneity of load distribution and provide target areas for subsequent optimization.

[0128] S1062, calculate the node load rate in the load imbalance area, mark the weak nodes and determine the contact nodes.

[0129] In an embodiment of the present application, for the transformer nodes in the load imbalance area, the load rate is calculated, i.e. the actual load value divided by the rated capacity. If the load rate exceeds 80%, for example, the load rate of a certain node is 88%, it is marked as a weak node. For each weak node, the load difference value with adjacent nodes is calculated, and the load transfer index is generated according to the ratio of the distance between nodes and the difference value. For example, a weak node with a load rate of 90% and an adjacent node with a load rate of 40% are 150 meters apart, and the load transfer index is 0.33. Select the node with the highest load transfer index and a load rate below 50% as the contact node, such as a node with a load rate of 38% and an index of 0.35. The selection of the contact node can maximize the load transfer effect and relieve the pressure of the weak node.

[0130] S1063, based on the shortest channel distance principle and dynamic programming method, generate the path of the new line and verify the network performance.

[0131] In an embodiment of the present application, the geographic coordinate data of the existing line channel is extracted, and the dynamic programming method is used to generate the path coordinate sequence of the new line with the constraints of minimizing the channel excavation length and maximizing the load transfer amount. For example, in a 2.5km long cable channel, a 300m long new line path is planned. After the new line is put into operation, the network performance is verified through the power flow calculation to ensure that the node voltage deviation is less than 5% of the rated value and the line capacity margin is greater than 25%. For example, after the operation, the load rate of the weak node decreases from 87% to 72%, and the voltage deviation is 2.8%. The dynamic programming method can optimize the line layout, reduce the construction cost and improve the network stability.

[0132] In the embodiment of the present application, the minimum spanning tree algorithm is used to reconstruct the power supply circuit to generate a power supply network topology scheme containing tie lines and backup power sources. For example, in a commercial complex, the weak nodes in the core area are connected to low-load nodes by newly added tie lines to form new backup channels. The power flow calculation shows that when a single device fails, the network can still maintain stable operation, and the highest load rate node is reduced to 75%. Compared with the traditional static optimization method, the embodiment of the present application can dynamically respond to the load changes in the commercial area through the load transfer index and dynamic programming, significantly reduce the overload risk of weak nodes, and improve the operation efficiency and power supply reliability of the distribution network.

[0133] S107, real-time collection of load data of the adjusted line layout, if a load mutation is detected, dynamically adjusting the cross-section identification standard and operation parameters, generating an optimized power supply path and forming a short-term operation optimization scheme.

[0134] In the embodiment of the present application, by real-time collection of load data of the adjusted line layout, monitoring the load change trend and identifying mutation events. Based on the mutation data, dynamically adjust the cross-section identification standard, redivide the power supply partition, and generate an optimized power supply path through the shortest path algorithm to form an operation optimization scheme that adapts to short-term load changes. The embodiment of the present application can significantly improve the operation stability and power supply quality of the distribution network in the commercial area by quickly responding to load mutations and dynamically optimizing network configuration.

[0135] S1071, collecting load time series data from the power distribution automation master station, generating a smoothed load sequence by using exponential smoothing processing and calculating the load change rate.

[0136] In the embodiment of the present application, the power distribution automation master station collects the load data of the adjusted line at a fixed time interval, such as 5 minutes, to generate load time series data containing time tags. Exponential smoothing method is applied to the time series data, and the smoothing coefficient is set to 0.3 to eliminate random fluctuations and generate a smoothed load sequence reflecting the load change trend. Calculate the difference between adjacent data points in the smoothed sequence to get the load change amount, and then divide by the time interval to generate a load change rate sequence. For example, the load of a line rises from 250 kW to 400 kW from 8:00 to 8:05 in the morning, with a change of 150 kW and a change rate of 1800 kW per hour. Exponential smoothing processing can effectively filter out noise and highlight the true trend of load changes, providing a reliable data foundation for mutation detection.

[0137] S1072, based on the standard deviation of the load change rate sequence, identifying the mutation time and extracting the correction reference value.

[0138] In the embodiment of the present application, based on the historical load change rate data, the standard deviation is calculated, for example, 200 kilowatts per hour, and the mutation judgment threshold is set to three times the standard deviation, that is, 600 kilowatts per hour. The threshold is compared with the load change rate sequence, and if it exceeds the threshold, it is marked as a mutation time. For example, during the opening period of a shopping mall, the change rate reaches 700 kilowatts per hour, triggering mutation detection. The load data 30 minutes before and after the mutation time is extracted, and the load mean and variance are calculated as the modified reference value of the section identification parameter. For example, the mean before mutation is 230 kilowatts, and the variance is 25 kilowatts; the mean after mutation is 350 kilowatts, and the variance is 40 kilowatts. These statistical values reflect the intensity and fluctuation characteristics of load mutation, which are used to update the section identification standard.

[0139] S1073, updating the section identification standard according to the modified reference value, and re-dividing the power supply subarea by using the Thiessen polygon method.

[0140] In the embodiment of the present application, the load threshold in the section identification standard is updated based on the modified reference value, for example, the threshold is adjusted from 300 kilowatts to 360 kilowatts. The Thiessen polygon method is used to re-divide the power supply subarea boundary with the updated section coordinate point as the center, to ensure that the load characteristics in each subarea are similar. For example, in a commercial complex, the load thresholds of the retail area and the catering area are adjusted to 400 kilowatts and 320 kilowatts respectively. The Thiessen polygon method generates subarea boundaries by constructing perpendicular bisectors, which can naturally reflect the spatial distribution characteristics of the load, providing accurate spatial basis for subarea division.

[0141] In the embodiment of the present application, based on the updated power supply subarea, the subarea division of the distribution network is carried out, and the rated capacity and actual load of the interconnection line between the subareas are calculated. For example, the rated capacity of an interconnection line is 600 kilowatts, and the actual load is 350 kilowatts, which has sufficient load adjustment space. The shortest path algorithm is used to generate multiple candidate power supply paths, and the node voltage deviation, line load rate and path length of each path are evaluated. For example, the length of candidate path 1 is 900 meters, the voltage deviation is 2.3%, and the load rate is 62%; the length of path 2 is 1100 meters, the voltage deviation is 1.9%, and the load rate is 58%. The indicators are normalized, and the path 2 with the highest comprehensive score is selected as the short-term operation scheme, which avoids the dense flow area and has excellent voltage quality and load balance performance. This dynamic optimization mechanism can quickly respond to load mutation, for example, during the promotion activity, the load increases to 450 kilowatts, and the system reduces the load rate to 60% by adjusting the path, to ensure the stability and efficiency of power supply.

[0142] It will be apparent to those skilled in the art that the application is not limited to the details of the above-exemplified embodiments and that the present application can be implemented in other particular forms without departing from the spirit or essential characteristics thereof. The embodiments should be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the above description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. No reference herein to any prior art is to be taken as an admission that the application is not entitled to antedate such prior art by virtue of prior application. Any reference to the use of a term in the singular herein shall be understood in the context to describe a particular example or embodiment of the application and should not be construed as limiting the scope of the application to that particular example or embodiment. Any reference to use of terms in the plural herein shall be understood as describing a particular example or embodiment of the application and should not be construed as limiting the scope of the application to that particular example or embodiment.

Claims

1. A method for dynamic generation and evaluation of power supply path based on historical load analysis, characterized in that, The method comprises the following steps: (1) Obtain historical load data and power distribution network topology data of the business district, extract load spatial distribution characteristics, and determine the dynamic change trend of load spatial distribution by using a space-time clustering algorithm; (2) Determine the position change of the load center of the business district at different time points according to the dynamic change trend of the load spatial distribution, obtain the power consumption load per unit area of the business district, and reconstruct the power consumption density combined with the offset trajectory of the load center of the business district to obtain the offset distance of the load center of the business district and the change rate of the power consumption density; (3) Perform time series analysis on the offset distance of the load center of the business district and the change rate of the power consumption density, identify the change period of the load of the business district and the burstiness characteristics of the load surge or reduction; (4) Calculate the load change rate at different time points according to the change period and burstiness characteristics of the load, and if the load change rate exceeds the preset change threshold, recalculate the key section position based on historical load analysis to obtain a dynamically adjusted section identification standard; At the same time, generate an end face identification standard document containing a power supply guarantee section and priority based on the business operation characteristics; (5) Identify the distribution of key sections according to the section identification standard, evaluate the load matching degree of the power distribution network topology structure according to the distribution of key sections, and calculate the load distribution balance and line carrying capacity between topology nodes; (6) Identify weak nodes and areas with uneven load distribution according to the load distribution balance and line carrying capacity between topology nodes, and adjust the line layout according to the identified weak nodes; (7) Real-time collection of load data corresponding to the adjusted line layout, dynamic adjustment of the section identification standard and operating parameters if a load mutation is monitored, dynamic generation of a power supply path, and formation of a short-term operation optimization scheme.

2. The method of claim 1, wherein, The step (1) comprises: Obtaining transformer node topology structure database and load value from the power distribution automation master station, wherein the load value is collected by the power distribution network data collection device within a preset sampling time interval; Obtaining load density matrix by using Gaussian kernel density estimation according to transformer node load value and geographic coordinate information, wherein the load density matrix is associated with land use type identifier in geographic information database; If the transformer node load value exceeds the load fluctuation threshold calculated according to historical load data, a multi-dimensional feature vector containing load value, fluctuation amplitude and duration is generated; Classifying the multi-dimensional feature vector by using spectral clustering method to obtain node groups with similar power consumption characteristics, and calculating the dynamic change trend of load spatial distribution by sliding time window within the node group.

3. The method of claim 1, wherein, The step (2) comprises: Constructing a polygon area boundary according to the coordinate point set of the business district, generating a grid cell by using a fixed side length division method, and obtaining the initial load density distribution map by obtaining the power consumption load data in the grid cell from the power distribution device; For the initial load density distribution map, the power consumption load value of the grid cell is used as the weight for weighted average operation to obtain a coordinate sequence of the load center of the business district; The three spline interpolation method is used for smoothing the load center coordinate sequence, and the curve after the smoothing is sampled according to the fixed time interval to obtain a time-series load center position point set; According to the grid unit area and the electric load value, the unit area electric load is calculated, and the bilinear interpolation method is used for spatial reconstruction of the unit area electric load to obtain an electric density distribution map; The Euclidean distance between two points is calculated by using the load center positions of adjacent time points to obtain the original offset distance, and the original offset distance is normalized according to the diagonal length of the commercial district boundary to obtain the standardized offset distance; The electric density change rate is calculated according to the standardized offset distance and the difference of the electric density distribution map of adjacent time points.

4. The method of claim 3, wherein, Further comprising: According to the load space distribution trend, the load center positions at different time points are calculated, the coordinate change sequence of the load center positions at different time points is extracted, the load center offset trajectory is formed according to the coordinate change sequence, the electric load data in the unit area of the commercial district is obtained, the electric load data and the offset trajectory are spatio-temporally matched, the electric density distribution of each region is reconstructed according to the spatio-temporal matching result, and the gradient change rate of the electric density distribution at the corresponding time point is extracted, specifically including: Obtaining load sampling data and geographic position coordinate information, wherein the load sampling data is collected by a power distribution automation master station; According to the load sampling data and the geographic position coordinate information, a load weighted average center coordinate sequence is calculated; The three spline interpolation method is used for curve fitting of the load weighted average center coordinate sequence to obtain a load center offset trajectory curve; The tangent direction and the curvature value of each sampling time point are calculated for the load center offset trajectory curve to obtain trajectory feature data; The inverse distance weighted interpolation method is used for spatial grid node interpolation operation of the trajectory feature data to obtain an electric density distribution map; According to the electric density distribution map, the electric density difference value of adjacent time points is calculated to obtain density change gradient data.

5. The method of claim 1, wherein, The step (3) comprises: According to the load center coordinate time-series data, a distance difference value sequence between adjacent sampling points is calculated, and the discrete Fourier transform is used for transforming the distance difference value sequence to obtain a frequency spectrum signal; The sliding time window is used for smoothing the electric density change rate sequence, and the residual sequence is obtained by calculating the difference between the smoothed curve and the original curve at each sampling point; According to the sampling points of the residual sequence exceeding the three times threshold of the standard deviation, the mutation positions are marked, the data segments are obtained by extending the mutation positions forward and backward by a fixed time length, and the change amplitude is obtained by calculating the difference between the initial value and the terminal value of the electric density in the data segment; The change rate is calculated for the change amplitude and the time interval, the mutation feature vector is obtained by symbol judgment of the change rate and the change amplitude, and the time interval of adjacent mutation events in the mutation feature vector sequence is subjected to periodic pattern recognition by autocorrelation analysis.

6. The method of claim 1, wherein, The step (4) comprises: A fixed length time window is used to calculate the load change rate of adjacent sampling points, and the standardized load change rate sequence is obtained by subtracting the mean value from the load change rate sequence and dividing by the standard deviation; According to the comparison result of the standardized load change rate sequence and the early warning reference value, a load mutation marker is obtained, and a smoothing historical load curve is obtained by using a decreasing coefficient sequence to perform weighted average on historical load data before the load mutation marker time; A load density distribution is calculated for the smoothing historical load curve, and a key section initial position is obtained by processing the load density distribution by using a maximum gradient method; According to the key section initial position, a search interval is set, density estimation operation is performed on load data in the search interval to obtain a density peak value position, the density peak value position is determined as an accurate position of the key section, a standardized section identification parameter is generated, and the standardized section identification parameter is used as a new section identification standard. Load anomaly records during commercial holidays and promotion activities are obtained, plans for adding large-scale power consumption equipment in commercial areas, property expansion, and tenant format adjustment are analyzed, key line points of power flow changes in peak periods are located, transformer load rate data of each area of a commercial center are connected, a section identification standard document including a commercial peak period power supply guarantee section, a core merchant uninterrupted power supply priority, and an adjustable load capacity is generated, and the section identification standard document specifically includes: Holiday and promotion activity calendars in a commercial area operation database are obtained, load data in corresponding periods is extracted according to the activity calendar, and load anomaly records are obtained by using three times of standard deviation screening; According to the load anomaly records and tenant format adjustment information, a unit area power load reference value is calculated, and a load change sequence is generated for a planning area of the commercial area according to the reference value; A line current density value is obtained by performing probability sampling on the load change sequence by using a Monte Carlo method, and if a current density change amount and a reference value ratio exceed a threshold value, a line node coordinate is marked; Transformer load rate data in a power supply monitoring device are obtained, and a transformer load rate threshold value is calculated according to the load rate data; Total power consumption capacity and interruptible load capacity of each load unit group are calculated, and a substation power supply scheme including a guarantee section, a priority sequence, and an adjustable capacity is generated.

7. The method of claim 1, wherein, The step (5) comprises: A section distribution curve is generated by using a cubic spline interpolation method according to section coordinate points, and the section distribution curve is used to calculate an equidistant sampling point coordinate set; A Voronoi polygon division operation is performed on the equidistant sampling point coordinate set to obtain a substation boundary point set in a section coverage range; A power distribution network topology graph is constructed according to the substation boundary point set, and a maximum allowable load flow value between nodes in the power distribution network topology graph is calculated by using a maximum flow algorithm; Line power values are obtained by performing a power flow calculation for the power distribution network topology graph, and a line load rate index is obtained by dividing the line power values by a line rated capacity value.

8. The method of claim 1, wherein, The step (6) comprises: Nodes are grid-divided according to a load balancing degree index, load density values in grid units are obtained by using a density calculation method, and a node distribution density graph is obtained by performing mean value standardization processing. A local spatial statistical method is used to calculate a load density difference value of adjacent grid cells for the node distribution density map, and if the load density difference value exceeds twice the overall standard deviation, the region is marked as a load imbalance region; By calculating the load rate of each node in the load imbalance region, if the node load rate exceeds the transformer rated capacity threshold, the node is marked as a weak node; According to the load difference value and the distance between nodes of the weak node and adjacent nodes, a load transfer index is calculated, the node with the highest load transfer index and a load rate lower than a preset percentage is selected as a tie-in node, and a path coordinate sequence of the new line is generated based on the shortest channel distance principle.

9. The method of claim 1, wherein, The step (7) comprises: According to the power distribution automation master station acquisition time tag, a load time sequence data is generated, and an exponential smoothing process is performed on the load time sequence data to obtain a smoothed load sequence; For the smoothed load sequence, a difference value of adjacent data points is calculated to obtain a load change amount, and the load change amount is divided by the acquisition time interval to obtain a load change rate sequence, and a standard deviation value is obtained according to the load change rate sequence; The load data before and after the moment when the load change rate exceeds the standard deviation threshold is obtained, and the load data is calculated to obtain a load mean and a variance as a correction reference value of the section identification parameter; The correction reference value is used to update the load threshold in the section identification standard, and the load threshold is used to divide the power supply partition boundary by the Thiessen polygon method, and the power supply partition boundary is used to determine the power distribution network subnet division and calculate the rated capacity and actual load of the interconnection line between partitions; A shortest path algorithm is used to generate multiple power supply path alternative schemes, the node voltage deviation and line load rate of each scheme are calculated, and a short-term operation optimization scheme is formed.

Images