A power grid facility layout planning method and system based on multi-source perception data

Abstract

The application discloses a kind of power grid facility layout planning method and system based on multi-source perception data, belong to big data analysis and mining technical field, this method includes following steps: acquisition multi-source data and pre-processing, grid region after generating preliminary candidate solution, calculate preliminary candidate solution fitness, by constraint check hierarchical processing, determine to be optimized solution, particle swarm algorithm iteration optimization, according to fitness variation and maximum iteration number termination, form final output, a kind of power grid facility layout planning method and system based on multi-source perception data provided in the application are coupled by space grid pretreatment and GIS technology, scattered data is converted into structured information with spatial correlation, realize multi-objective balance, avoid the limitation of traditional planning heavy single index, light comprehensive performance, rely on particle swarm algorithm hybrid coding strategy, jump out of local optimum trap, provide technical support for efficient and reliable power grid planning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data analysis and mining, and particularly discloses a power grid facility layout planning method and system based on multi-source perception data. BACKGROUND

[0002] In the prior art, a typical power grid facility layout planning method is a GIS (Geographic Information System) and load data assisted planning method. This method collects the load distribution, power supply capacity and geographic environment information (such as roads, water bodies, terrain, etc.) in the region, uses GIS software to perform spatial analysis and constraint evaluation on the planning region, and combines historical load data and power flow calculation results to preliminarily screen and capacity match the candidate power grid facility site selection. By simulating and evaluating the operation loss, reliability and environmental impact of different schemes, the planning personnel can determine a reasonable power grid facility layout under the premise of meeting the power supply reliability and economic requirements, and realize scientific planning and optimized management of the power system.

[0003] For example, the power grid planning scheme risk assessment method taking into account the operation mode of the power system disclosed in Chinese patent CN104599189B includes: using a multi-dimensional K-means clustering algorithm with low time complexity and good robustness to accurately classify load and power generation data, establishing a multi-level horizontal load and power generation level clustering model, and combining the power grid maintenance scheme to obtain the annual typical operation mode and its probability; based on the annual typical operation mode, the transient security risk of the system is evaluated by using an analytical method based on time domain simulation.

[0004] For example, the power communication network routing planning method based on reliability prediction and particle swarm optimization disclosed in Chinese patent CN111210361B includes: predicting the reliability of network nodes and links at the current time through a trained reliability prediction model of the power communication network, and obtaining the optimal power communication network service routing scheme through the reliability of nodes and links and a particle swarm optimization algorithm; the reliability prediction model of the power communication network is constructed based on a recurrent neural network. The inputs of the reliability prediction model of the power communication network include: the occupation degree of node buffer, fiber communication error rate and optical power; the outputs are: node reliability prediction results on the communication network, and link reliability prediction results are the average values of the reliability of two nodes connected thereto.

[0005] The above-mentioned technology at least has the following technical problems:

[0006] In the power grid facility layout planning based on multi-source sensing data, the main problem is that the power flow calculation has strong nonlinearity and large-scale characteristics, and even if real-time sensing data is introduced, it is difficult to verify the electrical feasibility of each candidate solution within a short time; meanwhile, the multi-source information such as terrain undulation, road direction and dynamic load distribution makes the feasible solution space present high irregularity and dynamic change, thereby further increasing the uncertainty and complexity of the planning. SUMMARY

[0007] In one aspect, a power grid facility layout planning method based on multi-source sensing data is provided, the method comprising:

[0008] Collecting and preprocessing multi-source sensing data, uniformly dividing the planning area into a spatial grid, screening preliminary buildable sites and generating preliminary candidate solutions, performing preliminary candidate solution fitness calculation to obtain a candidate solution set.

[0009] Performing hierarchical power flow calculation to obtain a preliminary constraint checking strategy, and when the preliminary constraint checking strategy is a local repair adjustment, performing local repair adjustment.

[0010] After local repair adjustment, determining a solution to be optimized, executing a hybrid optimization strategy and iterative optimization, and after iteration ends, performing current executable optimal solution executability confirmation, and outputting strategy executable optimal solution generation confirmation information.

[0011] After receiving the strategy executable optimal solution generation confirmation information, taking the solution corresponding to the strategy executable optimal solution generation confirmation information as the power grid facility layout planning strategy.

[0012] In another aspect, the application embodiments provide a power grid facility layout planning system based on multi-source sensing data, comprising:

[0013] A candidate set selection module configured to collect and preprocess multi-source sensing data, uniformly divide the planning area into a spatial grid, screen preliminary buildable sites and generate preliminary candidate solutions, perform preliminary candidate solution fitness calculation, and obtain a candidate solution set.

[0014] A constraint checking and repair module configured to perform hierarchical power flow calculation to obtain a preliminary constraint checking strategy, and when the preliminary constraint checking strategy is a local repair adjustment, perform local repair adjustment.

[0015] An iterative optimization module configured to, after local repair adjustment, determine a solution to be optimized, execute a hybrid optimization strategy and iterative optimization, and after iteration ends, perform current executable optimal solution executability confirmation, and output strategy executable optimal solution generation confirmation information.

[0016] An optimal solution generation confirmation module configured to, after receiving the strategy executable optimal solution generation confirmation information, take the solution corresponding to the strategy executable optimal solution generation confirmation information as the power grid facility layout planning strategy.

[0017] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0018] 1. This invention achieves a balance between exploration space and convergence speed through deep fusion of multi-source data and multi-objective optimization, collaborative design of hierarchical screening and intelligent iteration, and a dynamic mechanism for precise balance between exploration space and convergence speed. The system couples spatial gridding preprocessing with GIS technology to transform scattered data into structured information with spatial correlation, achieving a balance between multiple objectives and avoiding the limitations of traditional planning that emphasizes a single indicator while neglecting comprehensive performance. Hierarchical screening uses fitness thresholds and linearization constraints for pre-checking, eliminating inferior solutions and focusing on high-potential solutions. Intelligent iteration relies on a hybrid coding strategy of particle swarm optimization, using discrete PSO for qualitative decision-making of discrete variables and continuous PSO for fine-tuning quantitative parameters of continuous variables. It combines dynamic updates of individual and global optimal values ​​to escape the trap of local optima, avoiding ineffective iteration and preventing non-convergence termination through early warning and supplementation. The final output scheme is scientifically guaranteed by quantitative analysis and algorithm verification, providing technical support for efficient and reliable power grid planning.

[0019] 2. This invention provides accurate basic data support for planning through multi-source data acquisition and gridded preprocessing. Spatial grid statistics of load and power supply data, along with GIS calculations of terrain constraint coefficients, achieve data spatialization and standardization, avoiding planning deviations caused by information silos. The initial screening of feasible sites and generation of attribute tables, combined with a greedy algorithm to randomly generate candidate solutions, ensures solution diversity while reducing the computational complexity of subsequent evaluations through gridding. This step ensures the comprehensiveness and accuracy of the original data and lays an efficient foundation for subsequent optimization through structured processing, avoiding redundant calculations of invalid solutions.

[0020] 3. This invention achieves quantitative evaluation of preliminary candidate solutions through rapid simulation of DC power flow models and calculation of multi-dimensional indicators. Fitness stratification and constraint pre-checking, while ensuring efficiency, initially screen out high-quality solutions, reducing the computational load of subsequent deep optimization. The stratified processing strategy reflects the optimization logic of prioritizing major solutions while avoiding minor ones, preserving potentially repairable solutions and preventing resource waste. Local repair adjustments target specific violations precisely, improving the efficiency of transforming minor violation solutions into feasible solutions, and providing high-quality input for iterative optimization.

[0021] 4、The application adapts different optimization requirements of qualitative decision and quantitative parameters in power grid planning by adopting hybrid coding and type updating rules of particle swarm algorithm, ensures the scientificity of optimization direction, dynamically updates individual and global optimal value, combines the double termination conditions of fitness change value and maximum iteration number, avoids premature convergence into local optimum, and prevents invalid iteration from wasting resources. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0023] Figure 1 is a flow chart of a power grid facility layout planning method based on multi-source perception data provided by the embodiments of the present application;

[0024] Figure 2 is a structural schematic diagram of a power grid facility layout planning system based on multi-source perception data provided by the embodiments of the present application;

[0025] Figure 3 is a mind map of the power grid facility layout planning based on multi-source perception data provided by the embodiments of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the present application will be described below in combination with the drawings.

[0027] In the embodiments of the present application, the words such as example, for example, and the like are used to represent as an example, illustration or description. Any embodiment or design scheme described as an example in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word example is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, and / or the meanings expressed can be both, or can be either one of the two.

[0028] In the embodiments of the present application, sometimes the subscript such as W1 may be written in the form of non-subscript such as W1, and the meanings expressed are consistent when the difference is not emphasized.

[0029] In order to make the technical problems, technical solutions and advantages of the present application more clear, the following will be described in detail in combination with the drawings and specific embodiments.

[0030] As Figure 1 shown in a grid of a power grid facility layout planning method based on multi-source perception data, the processing flow of the method can include the following steps: first, collect and preprocess load, power, remote sensing data, divide the planning area grid and determine the buildable site combined with GIS; then use the greedy algorithm to generate a preliminary candidate solution, calculate the relevant indicators of the preliminary candidate solution through power flow simulation and weighting, and select the high fitness candidate solution set; then, the candidate solution set is checked for constraints, and is divided into pending solutions, slightly illegal solutions and serious illegal solutions; then, the feasible solution and the slightly illegal solution are used as the optimization solution, and the particle swarm optimization algorithm is used for iterative optimization, and the iteration is stopped according to the fitness change and the number of iterations; finally, the iteration result is verified.

[0031] Collect and preprocess multi-source perception data, divide the planning area into spatial grids, screen preliminary buildable sites and generate preliminary candidate solutions, calculate the fitness of the preliminary candidate solutions, and obtain the candidate solution set.

[0032] Further, the preliminary buildable sites are screened and the preliminary candidate solutions are generated, and the specific analysis method is as follows:

[0033] Collect and preprocess multi-source perception data, including load data, power data and remote sensing data.

[0034] It should be noted that the multi-source perception data includes load data, power data and remote sensing data, wherein the load data includes total power load, load density and time dynamic characteristics of the planning area, the power data includes power output, power capacity and power voltage and current, and the remote sensing data includes topographic data, land cover data and ground attachment distribution data.

[0035] The planning area is uniformly divided into spatial grids, the total load, load density and power capacity of each grid are counted, the terrain constraint coefficient is calculated according to the GIS software, and the grids in the planning area that meet the station building requirements are marked as preliminary buildable sites, and an attribute table is generated for the preliminary buildable sites.

[0036] It should be explained that GIS software (Geographic Information System) is a tool for acquiring, storing, managing, analyzing and visualizing spatial data. Its core function is to combine geographic location with attribute information, support spatial data processing and spatial decision analysis.

[0037] It should be noted that according to the range size and accuracy requirement of the planning area, the area is uniformly divided into several spatial grids, and the total load and load density are counted: relying on the basic load data in the load type data, the position and load value of each load point in the planning area are extracted first, and then the load value is summarized to the corresponding grid according to the grid belonging of the load point, and the total load of each grid is obtained; then combined with the area of the grid, the load density of each grid is calculated by dividing the total load by the area of the grid. Based on the power supply type data, the spatial position and capacity parameters of the power supply basic parameters are first located in the grid of the existing and newly built power supply, and then the rated capacity of each power supply is summarized to the corresponding grid to obtain the power supply capacity of each grid. This index directly reflects the power supply capacity of the grid, which needs to be compared with the total load in the future to judge whether the supply and demand are balanced.

[0038] It should be noted that the accuracy target of the power grid planning is converted into the length of the grid side, and in this embodiment, the system pre-sets the corresponding rules between the accuracy requirement and the grid side length, and stores them in the configuration file or parameter table for unified calling. In the running process, the system first calculates the spatial resolution required according to the load distribution, terrain characteristics and data sampling accuracy in the planning area, which reflects the degree of detail in depicting the station site selection and power grid layout, and then finds or calculates the matching grid side length according to the pre-set rules. The rules can establish the correlation between the accuracy requirement and the grid side length through interval mapping table, function fitting or interpolation method. When the accuracy requirement is high, it means that the load distribution and constraint conditions need to be described in more detail, and the system needs to use smaller grid side length to improve the fineness of spatial division; therefore, the higher the accuracy requirement, the smaller the corresponding grid side length.

[0039] The elevation data and terrain type data of the terrain data in the remote sensing type data are imported into the GIS software; then the terrain analysis tool of GIS is used to quantitatively score the terrain conditions of each grid, and finally the terrain constraint coefficient of each grid is output.

[0040] The preliminary buildable sites are screened and the attribute table of the preliminary buildable sites is extracted, and the greedy algorithm is used to randomly generate each preliminary candidate solution based on the attribute table of the preliminary buildable sites.

[0041] It should be noted that the station building core conditions are first selected: the terrain constraint coefficient, the grid land type is construction land, there is no built important infrastructure in the grid, and the difference between the total load and the power supply capacity of the grid.

[0042] Then in the GIS software, the above conditions are converted into spatial screening rules, and all grids are batch screened. The grids meeting the conditions are marked as preliminary buildable sites, and the attribute table of each preliminary buildable site is extracted to generate the attribute table of the integrated data. The core attributes include: basic spatial attributes, load attributes, power supply attributes, and terrain attributes.

[0043] It should be noted that when generating a preliminary candidate solution based on the preliminary buildable site attribute table using a greedy algorithm, first the core objective of the algorithm to improve power supply stability needs to be clarified, and the load attributes including the load total of the corresponding grid of each site, the density, the power supply attributes including the estimated capacity of the proposed site, the existing power supply capacity in the surrounding area, the terrain attributes including the terrain constraint coefficient and the site coordinates, and other key data are extracted from the attribute table; then the basic parameters and constraint boundaries are initialized; then the algorithm enters the iteration generation stage, the first step is to calculate the load coverage ratio (proposed capacity divided by the total load of the corresponding grid) of each site from all preliminary buildable sites according to the load coverage benefit priority principle, mark its coverage range and record the covered load; the second step is to exclude sites with overlapping coverage ranges and sites with proposed capacities that cannot cover the load of their own grids, calculate the marginal benefit (newly covered load divided by the cost of new construction) of the remaining sites, select the site with the highest marginal benefit to join the set and update the relevant data, repeat this step until the covered load meets the standard, the cost reaches the threshold, or there is no site that meets the constraints; the third step is to configure the supporting power supply equipment for the sites in the candidate solution, plan the line path and determine the impedance parameters, and form a complete preliminary candidate solution; finally, to avoid local optimization, introduce random disturbance in the site sorting step of generating the candidate solution each time.

[0044] Further, the fitness of the preliminary candidate solution is calculated, and the specific analysis method is as follows:

[0045] For a preliminary candidate solution, each electrical connection point of each electrical device in it is recorded as each node.

[0046] It should be noted that the electrical connection point of the electrical device refers to the physical connection and electrical energy exchange interface between all electrical devices in the plan and between the device and the line, which is defined as a node.

[0047] The preliminary candidate solution is simulated quickly using the DC power flow model, the current variation amplitude of the connection line of each node in the preliminary candidate solution, the voltage variation amplitude of each node, and the total operating loss of the whole network are calculated and recorded, the actual electricity demand of the planning area in the planning period and the maximum power supply capacity that the power grid facilities can provide under operating conditions are obtained, which are recorded as load demand and power supply capacity respectively.

[0048] It should be noted that the process of using the direct current flow model to simulate the preliminary candidate solution quickly, first, the network topology parameters, typical load data and power output information are prepared, and the direct current flow model is selected to establish the linearized power flow equation; then, the station access scheme corresponding to the preliminary candidate solution is mapped into the network structure, and the power flow calculation is carried out under different typical working conditions, and the voltage of each node and the current distribution of each branch are obtained. Based on the calculation results, the change amplitude of each node voltage under different working conditions, the change amplitude of each branch current and the total operating loss of the whole network are calculated to measure the stability and energy efficiency level of the power grid operation; at the same time, the actual electricity demand in the planning period is calculated through the load time sequence data, and combined with the maximum output capacity of the network under the conditions of voltage constraint and line thermal limit, the maximum power supply level that the power grid facilities can provide is determined, which is recorded as load demand and power supply capacity respectively.

[0049] The robustness coefficient is calculated by tracking the change amplitude of node voltage and line current.

[0050] The robustness coefficient is calculated by tracking the change amplitude of node voltage and line current. Wherein, Q represents robustness, W represents the change amplitude of each node voltage, a represents the characteristic distribution factor of the change amplitude of each node voltage, R represents the change amplitude of the line current connected to each node, and b represents the characteristic distribution factor of the change amplitude of the line current between each node.

[0051] The operation loss degree is obtained according to the ratio of the total active loss obtained by power flow calculation to the load power.

[0052] The operation loss degree is obtained by dividing the total active loss of the whole network obtained by power flow calculation by the load demand in the planning period.

[0053] The capacity matching degree is obtained by dividing the power supply capacity by the load demand.

[0054] At the same time, the matching degree of line and terrain constraint is calculated through remote sensing and GIS data, which is recorded as terrain constraint degree.

[0055] The terrain constraint degree is obtained by multiplying the constraint coefficient of a certain terrain type and the length proportion of the line of this terrain, and then weightedly averaging the terrain constraint degrees of all terrain types.

[0056] It should be noted that the amplitude of the line current change between the nodes and the characteristic distribution factor thereof can be obtained through historical operation data and simulation results statistics, and combined with stability and sensitivity analysis to determine typical values and upper and lower limits, and then normalized and stored in the database. The parameters can be directly extracted during system operation to characterize the line current fluctuation characteristics and support planning and regulation.

[0057] After the capacity matching degree, the operation loss degree, the robustness coefficient and the terrain constraint degree are normalized, they are combined in a weighted manner to obtain the preliminary candidate solution fitness.

[0058] It should be noted that when calculating the preliminary candidate solution fitness, the capacity matching degree, the operation loss degree, the robustness coefficient and the terrain constraint degree need to be normalized to eliminate dimensional differences: the robustness coefficient, the capacity matching degree and the terrain constraint degree are positive indicators, the actual value of the indicator is subtracted from the minimum value of the indicator to obtain the actual difference value of the indicator, the maximum value of the indicator is subtracted from the minimum value of the indicator to obtain the maximum difference value of the indicator, and the actual difference value of the indicator is divided by the maximum difference value of the indicator to calculate the positive normalization value; the operation loss degree is a reverse indicator, which needs to be obtained by subtracting the actual value of the indicator from the maximum value of the indicator to obtain the actual difference value of the indicator, and the maximum value of the indicator is subtracted from the minimum value of the indicator to obtain the maximum difference value of the reverse indicator, and the actual difference value of the indicator is divided by the maximum difference value of the reverse indicator to calculate the reverse normalization value. After normalization, according to the core target of power grid planning, the normalized distribution coefficients of each indicator are set, and then the preliminary candidate solution fitness is combined and calculated according to the weighted formula. The closer the value is to 1, the better the comprehensive performance of the preliminary candidate solution in supply and demand balance, operation efficiency, power grid stability and terrain adaptability.

[0059] It should be noted that the capacity matching degree reflects the power supply and demand balance ability, the larger the value is, the more reasonable the power supply capacity and load demand matching is, and the stronger the contribution to the preliminary candidate solution fitness is; the operation loss degree measures the energy utilization efficiency, the larger the value is, the higher the total active power loss occupies the load proportion, and the more serious the energy waste is, and the weaker the contribution to the fitness is; the robustness coefficient represents the stability of the power grid against fluctuations, the larger the value is, the smaller the amplitude of the node voltage and the line current change is, and the stronger the ability to cope with fluctuations is, thereby improving the fitness; the terrain constraint degree reflects the line and terrain adaptability, the larger the value is, the flatter the terrain the line passes through, the lower the construction difficulty and cost, and the positively supports the preliminary candidate solution fitness. In summary, the four types of indicators jointly affect the fitness evaluation of the candidate solution through positive or negative effects, and provide quantitative basis for optimization decision-making.

[0060] It should be noted that the weight is preliminarily allocated according to the correlation strength of each index and the planning demand, then the preset index-weight mapping rule is extracted from the database, the weight range in different planning scenarios is indicated through interval mapping, the importance parameters of each index in real-time statistics are substituted into the mapping rule, the coefficient value is preliminarily determined, then the preliminary coefficient is normalized and checked, if the sum is not 1, the coefficient is modified by dividing the preliminary coefficient by the sum of the coefficients to ensure that the final four normalized distribution coefficients are in the interval of 0-1 and the sum is 1, after the setting is completed, it can be used for the weighted fusion calculation of the four indexes, providing quantitative basis for the preliminary candidate solution preliminary candidate solution fitness evaluation, supporting the optimization decision of power grid planning scheme.

[0061] It should be noted that the greater the capacity matching degree, the more reasonable the load distribution, the less the redundant load of line transmission, the smaller the operation loss degree, and the lower the fault risk caused by supply and demand imbalance, and the robustness coefficient is also improved accordingly; but the improvement of the capacity matching degree may be limited by the terrain constraint degree, if the terrain constraint degree is greater, the distance between the transformer substation and the load center may be forced to lengthen, and the line path needs to bypass the complex terrain, at this time, even if the matching degree is improved by increasing the equipment capacity, the operation loss degree may increase due to the increase of line impedance, and the maintenance difficulty of equipment in complex terrain is increased, and the robustness coefficient may decrease with the increase of terrain constraint degree.

[0062]

[0063] wherein F s represents the fitness of the s th preliminary candidate solution, s represents the preliminary candidate solution number of the planning area, s = 1, 2, 3 … o, M s represents the capacity matching degree of the s th preliminary candidate solution, a represents the capacity matching degree normalization distribution coefficient, L s represents the operation loss degree of the s th preliminary candidate solution, b represents the operation loss degree normalization distribution coefficient, K s represents the robustness coefficient of the s th preliminary candidate solution, g represents the robustness coefficient normalization distribution coefficient, T s represents the terrain constraint degree of the s th preliminary candidate solution, g represents the terrain constraint degree normalization distribution coefficient.

[0064] It should be noted that in power grid planning, indexes such as capacity matching degree and operation loss degree have spatial and temporal distribution unevenness in the planning area or period. By using the discrete grid summation method, the index values of all grids can be accumulated, the local index is converted into the overall cumulative effect, thereby avoiding the locality and one-sidedness of single-point sampling. Through summation calculation, the fitness value can comprehensively reflect the performance characteristics and change trend of the scheme in the entire evaluation range, so that the evaluation of the candidate solution is closer to the overall comprehensive performance, rather than being limited to local optimization.​

[0065] The fitness of each preliminary candidate solution is obtained by traversing each preliminary candidate solution.

[0066] Further, a candidate solution set is obtained, and the specific analysis method is as follows:

[0067] After obtaining the fitness of each preliminary candidate solution, a preset candidate solution fitness threshold in the database is extracted.

[0068] If the fitness of a preliminary candidate solution is greater than or equal to the candidate solution fitness threshold, the preliminary candidate solution is recorded as a high-fitness candidate solution.

[0069] It should be noted that if the fitness of a preliminary candidate solution is greater than or equal to the candidate solution fitness threshold, it means that the candidate solution has excellent comprehensive performance in the core evaluation dimension of power grid planning, and can be included in subsequent adjustment and further compared and screened with other high-fitness candidate solutions to reach the qualified standard for entering the next optimization link, and therefore is recorded as a high-fitness candidate solution.

[0070] If the fitness of a preliminary candidate solution is less than the candidate solution fitness threshold, the preliminary candidate solution is recorded as a low-fitness candidate solution.

[0071] It should be noted that if the fitness of a preliminary candidate solution is less than the candidate solution fitness threshold, it means that the candidate solution does not reach the qualified line for entering the next link in the comprehensive evaluation dimension of power grid planning, and this result usually reflects that the candidate solution has at least one or more obvious performance short board, and continuing to retain will significantly reduce the subsequent optimization efficiency and even affect the feasibility of the final planning scheme, and therefore is recorded as a low-fitness candidate solution.

[0072] All high-fitness candidate solutions are counted to obtain a candidate solution set.

[0073] Layered power flow calculation is performed to obtain a preliminary constraint checking strategy, and when the preliminary constraint checking strategy is local repair adjustment, local repair adjustment is performed.

[0074] Further, layered power flow calculation is performed to obtain a preliminary constraint checking strategy, and the specific analysis method is as follows:

[0075] The system first linearizes the candidate solution set, preliminarily checks the constraints such as voltage, line current-carrying capacity and power supply radius coverage on the premise of ensuring calculation efficiency, and the specific process includes: constructing linearized power flow equations according to node load and line impedance, and quickly solving node voltage amplitude and line current.

[0076] It needs to be explained that the premise of ensuring the efficiency of calculation refers to the linearization process, which converts the complex nonlinear problem into a linear equation set by ignoring the nonlinear terms (voltage amplitude square term, phase angle product term) in the power flow equation, greatly reduces the calculation steps and variable coupling degree when solving, realizes the rapid batch checking of a large number of candidate solutions, and balances between efficiency and accuracy, saving computing resources for subsequent depth optimization.

[0077] It needs to be explained that when the system carries out linearization processing on the candidate solution set and preliminarily checks the constraints, the specific process is as follows: First, from the attribute table of the candidate solution and the previous simulation data, the load power of each node, the impedance parameters of each line, and the constraint standards such as voltage upper / lower limit, line carrying capacity upper limit, and power supply coverage range are extracted; Then, based on the linearization principle of ignoring voltage phase angle change and simplifying power balance relationship, the linearized power flow equation is extracted, the linear equation set is quickly solved through matrix operation, and the voltage amplitude calculation value of each node, the current calculation value of each line and the power supply radius of each node are obtained.

[0078] It needs to be supplemented that the construction of the linearized power flow equation needs to take the power grid topology and basic parameters as the core and be completed in three steps: First, define the basic parameters and variables, determine the node set and line set based on the connection relationship of electrical equipment in the candidate solution, extract the load power of each node and the impedance parameters of each line, and define the to-be-solved variables as the voltage amplitude of each node and the current of each line. Second, simplify the nonlinear power flow equation. The traditional nonlinear power flow equation contains nonlinear product terms of voltage amplitude and phase angle. When linearized, small perturbation assumption is used to ignore high-order small terms: assuming that the node voltage amplitude fluctuates around the rated voltage, the square term of the voltage amplitude can be approximated as the square of the rated voltage plus twice the rated voltage multiplied by the deviation, and the sine and cosine terms of the phase angle can be approximated as the angle value and 1, thereby converting the power balance equation into a linear equation about the voltage deviation and the phase angle; At the same time, the relationship between line current and voltage is linearized through Ohm's law, and the current is directly represented by the ratio of node voltage difference to line impedance. Third, construct the equation set and arrange the form. With the active power and reactive power balance of each load node as the constraint, combined with the linear relationship between line current and voltage, a linear equation set with node voltage amplitude deviation and phase angle as variables is established, which is arranged in matrix form, and finally a linearized power flow equation that can be quickly solved by matrix inversion method is formed, laying a foundation for the subsequent rapid calculation of node voltage and line current.

[0079] The candidate solution in the candidate solution set that meets the three constraint conditions is recorded as the to-be-determined solution, and the to-be-determined solution set is obtained by statistics.

[0080] The to-be-determined solution with the highest fitness in the to-be-determined solution set is determined as the executable optimal solution, and the strategy executable optimal solution generation confirmation information is output.

[0081] It should be noted that the pending solution with the highest fitness of the preliminary candidate solution set is determined as the executable optimal solution, which not only meets the load demand of the planning area and ensures power supply stability, but also realizes the optimal balance of operation efficiency and construction cost. It can be directly used as the core basis for subsequent power grid construction, equipment configuration, and operation and maintenance strategy development, while avoiding resource waste and power supply risks caused by selecting a suboptimal solution, thereby ensuring the practicality of the power grid planning scheme. Therefore, this candidate solution is recorded as a pending solution.

[0082] The three constraints are that the voltages of each node are within the allowable range of the rated voltage, the load flow of each line does not exceed the rated capacity, and the power supply coverage meets the preset area and load requirements.

[0083] If there is no pending solution in the candidate solution set, the preliminary constraint check is used to determine the fine evaluation layer and the non-fine evaluation layer. One or two candidate solutions that do not meet the constraints are recorded as slightly defective solutions and stored in the fine evaluation layer candidate solution set. The preliminary constraint check strategy for the fine evaluation layer candidate solution set is recorded as local repair adjustment.

[0084] It should be noted that if there is no pending solution in the candidate solution set, the system will perform hierarchical processing on the candidate solutions based on the preliminary constraint check results. First, the number and severity of constraint violations of each candidate solution are determined. Candidate solutions with only one or two constraint violations are recorded as slightly defective solutions and are stored in the fine evaluation layer candidate solution set. Candidate solutions with three or more constraint violations or a single severe violation are classified as non-fine evaluation layer solutions and are directly eliminated. For the fine evaluation layer candidate solution set, the preliminary constraint check strategy is recorded as local repair adjustment. The core of this strategy is that the slightly defective solutions are not reconstructed as a whole, but are locally modified based on the specific violation items. Through minimal adjustment cost, the slightly defective solutions can meet the constraint requirements as much as possible, thereby qualifying for the subsequent optimal solution screening process. This avoids wasting candidate solutions with potential optimization value due to slight defects, thereby ensuring the diversity and practicality of the power grid planning scheme.

[0085] The three full non-compliance candidate solutions are recorded as severely defective solutions and are stored in the non-fine evaluation layer candidate solution set. The preliminary constraint check strategy for the non-fine evaluation layer candidate solution set is recorded as direct discard.

[0086] It should be noted that for candidate solutions with three full non-compliance constraints, the number of violations is large and the defects are severe, which far exceeds the range that can be modified by local repair. Not only is the repair cost extremely high, but it also leads to the failure of the original candidate solution design logic. Moreover, it is difficult to ensure the comprehensive performance after repair. At the same time, retaining such solutions occupies a large amount of system computing resources and reduces the efficiency of subsequent planning processes. Therefore, these solutions are recorded as severely defective solutions and are stored in the non-fine evaluation layer candidate solution set. The preliminary constraint check strategy for the non-fine evaluation layer candidate solution set is recorded as direct discard.

[0087] Further, when the preliminary constraint check strategy is local repair adjustment, local repair adjustment is performed, and the specific analysis method is as follows:

[0088] For a slight violation solution, key position identification is performed, that is, node voltage, line current-carrying capacity and load distribution are analyzed to locate the nodes and lines that cause constraint violation.

[0089] It should be noted that, first, for node voltage constraint violation, the calculated value of the voltage amplitude of each node obtained by solving the linearized power flow equation is compared with the preset voltage allowed range in the database, and the nodes whose voltage amplitudes exceed the interval are screened out. At the same time, the connection line parameters of the node and the voltage of the adjacent node are combined to determine whether the voltage drop of the node itself caused by the excessive load is too large or the voltage transmission caused by the abnormal impedance of the upstream power supply line is insufficient, to preliminarily locate the nodes and lines related to the voltage violation; secondly, for line current-carrying capacity constraint violation, the current calculation value of each line is compared with the upper limit of the current-carrying capacity to find out the line whose current exceeds the limit. Further, the load power of the nodes at both ends of the line and the line impedance are combined to analyze whether the current is increased due to overload or the current is abnormal due to mismatching of the line parameters, to determine the line and related nodes corresponding to the current-carrying capacity violation; finally, for load distribution constraint violation, the load nodes outside the coverage range are identified by superimposing the power supply coverage radius of the power supply site and the load node position on the GIS map, and whether the connection line of the load node has the problem of excessive power supply loss due to long path or not connected to the corresponding power supply branch line is checked to locate the nodes and lines of load distribution violation. Through the above three-dimensional analysis, the core nodes and core lines that cause constraint violation are finally integrated.

[0090] If there is a node load exceeding the configured substation capacity, the node is marked as a key node with insufficient capacity, and the output distribution of adjacent variable facilities is adjusted according to the load gap size and the remaining available capacity of each facility to meet the load demand.

[0091] It should be noted that when the system finds that the load power of a certain node exceeds the rated power supply capacity of its corresponding configured substation in the key position identification, the node will be directly marked as a key node with insufficient capacity. This marking clearly identifies the core position of the imbalance between load supply and demand, and the imbalance is focused on the fact that the single substation power supply capacity cannot cover the node load. For the load gap of such key nodes, the solution needs to focus on adjusting the output distribution of adjacent substation facilities: first, the system accurately calculates the size of the load gap of the insufficient capacity node, and at the same time retrieves the real-time operation data of the adjacent substation facilities around the node through the database to count the remaining available capacity of each facility; then, according to the principle of near distribution and load balance, an output adjustment strategy is developed, which preferentially selects the adjacent substation with the closest distance to the insufficient capacity node and sufficient remaining available capacity, and increases the output power of the substation by adjusting the line switch state between the substation and the node, optimizing the transformer tap or adding the line loop, etc., and distributes the newly added output power to the node; finally, through real-time monitoring of the load power of the node and the output power of the substation, it is confirmed that the load gap is completely filled.

[0092] If there is a node whose voltage deviates from the allowable range of the rated voltage, the node is marked as a voltage abnormal node, and the impedance distribution range is determined by the node voltage deviation to improve the voltage level.

[0093] It should be noted that if there is a node whose voltage deviates from the allowable range of the rated voltage, the node is marked as a voltage abnormal node. Subsequent improvement needs to first calculate the difference between the actual voltage and the rated voltage to obtain the node voltage deviation value, and then determine the adjustment distribution range of the line impedance according to the deviation size. By reducing the impedance of the upstream line of the abnormal node, the local line impedance is appropriately increased to consume excess voltage, and finally the node voltage returns to the allowable range, improving the overall voltage level.

[0094] It should be noted that through the DC power flow model, the power flow sensitivity of the voltage abnormal node is analyzed to calculate which load nodes are most sensitive to changes in voltage or line power, thereby determining the adjustable nodes and their adjustable range, including the active and reactive output of the generator and the flexible adjustable load. Subsequently, the system reasonably allocates generator output and load changes to each adjustable node according to the power flow sensitivity to form an adjustment vector, causing the power flow to redistribute along the target line, and ensuring that the node voltage, line power, and system total power balance meet various constraint conditions during the adjustment process, thereby changing the voltage drop and power distribution on the line, and achieving the perception of changing the equivalent impedance of the line.

[0095] In the embodiment, the system pre-sets the corresponding rules of the node voltage deviation value and the line impedance adjustment range, and stores them in the configuration file or parameter table for unified calling. In the running process, the system first calculates the voltage deviation value of each node to reflect the deviation degree of the node voltage in the power grid operation, and then finds or calculates the matching line impedance adjustment range according to the pre-set rules. The rules can establish the correlation between the node voltage deviation value and the line impedance adjustment range through the interval mapping table, function fitting or interpolation method. When the node voltage deviation value is large, it means that the voltage of some nodes deviates from the rated value too much, which may lead to line overload or uneven power distribution. At this time, the system needs to adjust the related line impedance to improve the power flow distribution and relieve the voltage deviation. Therefore, the higher the deviation value is, the larger the corresponding line impedance adjustment range is.

[0096] If there is a line with a load current exceeding the rated capacity, the line is marked as an overload critical line, and the load gap is distributed to each node output proportion in proportion to the remaining capacity to reduce the current carrying pressure.

[0097] It should be noted that, first, the line overload is accurately calculated, that is, the difference between the actual load current and the rated capacity, and it is converted into the corresponding load gap according to the line power transmission characteristics; then all the nodes connected to the overload line are locked, and the remaining power supply capacity of each node is counted through real-time monitoring data; then the proportion of the remaining capacity of each node to the total remaining capacity is calculated; then the load gap is distributed in proportion to this proportion, and finally the node output equipment is regulated to reduce the transmission load of the original overload line, thereby reducing the line current carrying pressure and restoring it to the rated capacity range.

[0098] If the above three conditions do not exist, a warning information is generated.

[0099] It should be noted that if the above three conditions do not exist, it means that the minor violation is not caused by known abnormal reasons, but may be caused by unknown factors, such as measurement error, transient disturbance, environmental factor or small drift of equipment, etc. Unknown reasons may be accumulated or intensified in the future, thereby affecting the stability of the power grid facility operation, so a warning information is generated.

[0100] In the embodiment, the warning information can be: "Attention! There is an unknown reason causing a minor violation."

[0101] After the local repair adjustment, the to-be-optimized solution is determined, the hybrid optimization strategy and iterative optimization are executed, the executability of the current executable optimal solution is confirmed after the iteration is completed, and the strategy executable optimal solution generation confirmation information is output.

[0102] Further, the to-be-optimized solution is determined, and the specific analysis method is as follows:

[0103] After the local repair adjustment, the constraint check is performed on the nodes and lines affected by the adjustment, and the adjusted candidate solution that meets the three constraint conditions is recorded as a feasible solution, and the system iterative optimization is directly performed.

[0104] It should be noted that the system needs to prioritize the constraint check on the nodes and lines affected by the adjustment, because the repair operation is only for specific problem nodes / lines, which may indirectly change the load, voltage of associated nodes or the current capacity of lines. For example, adjusting the output distribution of a substation may affect the voltage of other nodes and the current of lines connected to it. Therefore, the repair needs to be checked to confirm that it does not cause new constraint violations. If the affected nodes / lines and the overall candidate solution all meet the three constraints, it means that the local repair has completely solved the original minor violation problem, and the adjusted candidate solution meets the standard of no constraint violation and has the potential to be executed. Therefore, it is recorded as a feasible solution, and the feasible solution is directly included in the system iterative optimization process. The essence is to allow high-quality solutions to quickly enter the next stage of fine evaluation.

[0105] For the minor violation solution, the system still allows it to enter the iterative optimization, but adds a constraint penalty to the preliminary candidate solution fitness.

[0106] It should be noted that for the candidate solution that still has a minor constraint violation after local repair adjustment, the system does not directly eliminate it but allows it to enter the iterative optimization. The core is to preserve solutions with potential optimization value. Such solutions may perform well in other dimensions such as capacity matching degree and terrain constraint degree, and are only not up to standard due to minor defects. Direct elimination will reduce the diversity of candidate solutions and may miss the opportunity to become a high-quality solution after subsequent optimization. However, to reflect the impact of constraint compliance on the preliminary candidate solution fitness, the system adds a constraint penalty when calculating the preliminary candidate solution fitness of the solution: the penalty degree is positively correlated with the violation degree, and by reducing the preliminary candidate solution fitness score of the violation solution, its competitiveness in the iterative optimization is lower than that of the feasible solution without violation.

[0107] In this embodiment, the system predefines the corresponding rules of violation degree and penalty degree and stores them in a configuration file or parameter table for unified calling. During operation, the system first calculates the violation degree of each node or device to reflect the degree of deviation from the design or safety requirements, and then finds or calculates the matching penalty degree according to the pre-set rules. The rules can establish the correlation between the violation degree and the penalty degree through interval mapping table, function fitting or interpolation method. When the violation degree is high, it means that the operation deviates seriously or the risk is large, and the system needs to increase the penalty degree to strengthen the management or constraint measures. Therefore, the higher the violation degree, the greater the corresponding penalty degree.

[0108] For the serious violation solution, the optimization priority is reduced and it does not participate in the iteration temporarily.

[0109] It should be noted that for the serious violation solution stored in the non-precision evaluation layer, the system processes it through the adaptive penalty function: first, the penalty is dynamically generated according to the number of violations, the more violations, the greater the penalty coefficient, and finally the adaptive penalty function is used to calculate the preliminary candidate solution fitness value of the solution, so that its optimization priority is significantly lower than that of the feasible solution and the slight violation solution. At the same time, the system does not allow the serious violation solution to participate in the current iteration optimization process, and the core reason is that the defects of the solution need to be radically adjusted. If it is directly put into iteration, not only is it difficult to transform into a high-quality solution through fine-tuning, but also it will occupy a large amount of computing resources and slow down the optimization efficiency.

[0110] When the feasible solution and the slight violation solution are detected, the feasible solution and the slight violation solution are recorded as the solution to be optimized, and the iteration optimization is started.

[0111] Further, the hybrid optimization strategy and the iteration optimization are performed, and the specific analysis method is as follows:

[0112] The maximum number of iterations preset in the database is extracted.

[0113] The input solution to be optimized is subjected to iteration effect judgment after each iteration optimization.

[0114] After each iteration optimization, the average value of the fitness of all solutions to be optimized is calculated, and the change value of the average value from the last time is observed, which is recorded as the fitness change value. If the fitness change value is less than or equal to the fitness change value threshold, the current global optimal solution generated by the iteration optimization is recorded as the optimal solution. If the maximum number of iterations has not been reached at this time, if the fitness change value is greater than the fitness change value threshold, the iteration optimization is continued.

[0115] It should be noted that in the iteration optimization process of power grid planning, after each iteration, the system first calculates the average value of the fitness of all solutions to be optimized, and then subtracts the average value from the fitness average value of the last iteration. The difference obtained is the fitness change value, which reflects the improvement amplitude of the comprehensive performance of the candidate solution in this round of iteration. The greater the change value, the more significant the iteration optimization effect; the smaller the change value, the closer the performance of the candidate solution to stability, and the limited optimization space.

[0116] Subsequently, the system compares the fitness change value with a preset fitness change value threshold in the database: if the fitness change value is less than or equal to the threshold, it indicates that the performance of the candidate solution generated in the current iteration has been basically stable, and it is difficult to further improve the comprehensive quality by continuing iteration, at this time, whether the preset maximum iteration number is reached or not, the system will determine the global optimal solution in the current iteration, that is, the solution with the highest fitness among all the solutions to be optimized, as the final executable optimal solution, and end the iteration process; if the fitness change value is greater than the threshold, it indicates that the candidate solution still has a large optimization space, and the maximum iteration number has not been reached, and the next round of iteration optimization needs to be continued until the fitness change value meets the threshold requirement or the maximum iteration number is reached, to ensure that the final output optimal solution has sufficient comprehensive performance.

[0117] If the maximum iteration number is reached at this time, if the fitness change value is greater than the fitness change value threshold, a warning information is generated.

[0118] It should be noted that when the iteration optimization reaches the preset maximum iteration number, if the calculated fitness change value is still greater than the fitness change value threshold, it indicates that although the iteration has reached the upper limit of the number of times, the comprehensive performance of the candidate solution is still in a significant fluctuation state, and the optimization has not been stabilized, which may be that the parameter setting of the current optimization algorithm cannot adapt to the adjustment requirement of the candidate solution, resulting in that the performance improvement cannot converge all the time, so the warning information is generated.

[0119] In this embodiment, the warning information can be: "Attention! The iteration optimization has not converged."

[0120] The specific iteration process is as follows: each input solution to be optimized is first mapped to a particle, the position vector of which is represented by a hybrid coding network planning scheme, including discrete decision variables and continuous variables, and the velocity vector of the particle represents the change trend of the current position in the search space, including discrete variable speed and continuous variable speed.

[0121] The system initializes the position of each particle as the coding of the solution to be optimized, empirically sets the speed, calculates the initial fitness, and records the individual optimal value and the global optimal value, sets the current position as the individual optimal value, and the solution with the highest fitness in the group and meeting the constraints as the global optimal value.

[0122] It should be noted that the system adopts particle swarm optimization algorithm, each input to be optimized solution is converted into a particle in the algorithm, and the scheme optimization is realized by simulating the motion of particles in the search space. The specific process is as follows: firstly, particle mapping and coding, each to be optimized solution is mapped to a particle, and the position vector adopts hybrid coding mode, which contains discrete decision variables corresponding to qualitative selection in power grid planning, such as substation location, line path direction, transformer type selection, and also contains continuous variables corresponding to quantitative parameters, such as substation rated capacity, line cross-sectional area, and node load distribution ratio. This coding mode can completely restore all key parameters of the planning scheme, ensuring one-to-one correspondence between particle position and actual scheme.

[0123] Then, parameter initialization, the system initializes two core parameters for each particle: one is the position vector, which directly takes the hybrid coding result of the to-be-optimized solution as the initial position of the particle, ensuring that the particle starts searching from a potential scheme starting point; the other is the velocity vector, which is empirically set, and the historical optimization data is referred to, the velocity of discrete variables is set as the switching probability of adjacent schemes, and the velocity of continuous variables is set as the step range of parameter adjustment. The direction and size of the velocity vector determine the trend of the position change in the next iteration of the particle. The velocity of discrete variables controls the adjustment direction of qualitative selection of the scheme (such as whether to switch the line path), and the velocity of continuous variables controls the adjustment range of quantitative value of the parameter (such as the increase or decrease of capacity).

[0124] Then, fitness calculation and optimal value recording, after initialization, the system calculates the initial fitness of each particle according to the fitness evaluation standard mentioned in the foregoing; then records the individual optimal value and the global optimal value: the initial position and fitness of each particle are set as the individual optimal value of the particle; among the group composed of all particles, the particle with the highest fitness and meeting the three constraints of node load, voltage and line carrying capacity is selected, and its position and fitness are set as the global optimal value. Each particle will adjust the motion direction according to the individual optimal value and the global optimal value, gradually approach a better planning scheme, and promote the continuous iteration optimization.

[0125] In each iteration, the mixed optimization strategy of discrete variables is called discrete iteration optimization, and the position updating is performed through the discrete PSO rule.

[0126] It should be noted that when the position updating rule of discrete PSO is executed, a random number (such as a uniform random number between 0 and 1) is generated, which is compared with each switching probability in the velocity vector of discrete variables: if the random number is less than the switching probability to the individual optimal value, the discrete position of the particle is updated to the option corresponding to the individual optimal value; if the random number is between the individual optimal value switching probability and the global optimal value switching probability, the global optimal value corresponding option is updated; if the random number is greater than all switching probabilities, the current position remains unchanged.

[0127] Finally, the rationality of the updated discrete variables is verified. After the position update is completed, the new selected discrete options are simply checked for compliance with the basic logic of power grid planning. If they are in compliance, the updated results are retained as the new positions of the discrete variables for this round of iteration. If they are not in compliance, the speed calculation and position update are triggered again to ensure that the update of the discrete variables is always within the range of feasible options.

[0128] The discrete PSO formula is:

[0129]

[0130]

[0131] where V i t+1 represents the new speed vector of the particle after updating, i = 1, 2, 3,..., n, e represents a natural constant, v i represents the current speed of particle i, X represents the position vector of particle i at the tth generation, and n represents the total number of particles participating in iteration.

[0132] The mixed optimization strategy for continuous variables is denoted as continuous iterative optimization, which is updated through the speed and position update formula of continuous PSO.

[0133] It should be noted that for the continuous variables in the particle position vector, the mixed optimization strategy adopted by the system is denoted as continuous iterative optimization, and its core is to complete parameter iteration through the classic continuous PSO (Particle Swarm Optimization) speed and position update formula, to ensure that the continuous variables can dynamically adjust the values according to the optimal solution of the particle itself and the group, and gradually optimize the quantitative parameters of the power grid planning scheme. The specific process is as follows:

[0134] First, the core update logic of continuous PSO is determined. The optimization of continuous variables follows the principle of approaching the individual optimal value and the global optimal value: the speed of the continuous variable of each particle is adjusted according to the individual optimal value and the global optimal value, and the current position is corrected based on the updated speed, so that the continuous variable value continuously approaches the better direction.

[0135] ​​It should be noted that the inertia weight, the individual learning factor and the group learning factor are key parameters for adjusting the convergence characteristics of the particle swarm optimization algorithm, and their values can be pre-set in the system to balance the global search and local search capabilities, thereby ensuring the stability and convergence efficiency of the optimization process. In specific implementation, a large amount of historical optimization running data can be collected to record the value distribution of the inertia weight, the individual learning factor and the group learning factor under different problem sizes and complexities, and long-term statistical analysis can be performed on these data to obtain the inertia weight mean value range, the typical lower limit and upper limit of the individual learning factor and the group learning factor suitable for different application scenarios. Subsequently, these statistical values can be comprehensively processed, such as taking the average value, weighted average or adding a safety margin based on experience values, to obtain the pre-set inertia weight, individual learning factor and group learning factor in the database, which can be directly extracted from the database when actually called to guide the iterative update of the particle swarm.

[0136] The continuous PSO formula is:

[0137]

[0138]

[0139] where V i t+1 represents the new speed vector of the particle after updating, ω represents the inertia weight, which controls the inertia of the particle moving in the current speed direction, t represents the iteration number, c1 represents the individual learning factor, P represents the historical optimal position of the particle itself, X i t +1 represents the new position vector of the particle after updating, c2 represents the group learning factor, r1 and r2 represent random numbers for increasing search diversity, G represents the global optimal position, i represents the particle number, i = 1, 2, 3,..., n, and n represents the total number of particles participating in iteration.

[0140] During iteration, the update direction is determined according to the individual optimal value and the global optimal value.

[0141] After updating, the system re-performs the fitness calculation of the particle, and dynamically updates the individual optimal value and the global optimal value of the particle according to the new fitness value, gradually approaches the optimal solution using historical optimal experience and global information, and generates the current executable optimal solution.

[0142] It should be noted that after completing the iterative update of the discrete variable and the continuous variable, the system will start a new round of particle fitness calculation to determine whether the updated scheme is better, obtain the new fitness value of each particle, and ensure that the evaluation result can truly reflect the comprehensive performance of the updated scheme.

[0143] ​​Subsequently, the system dynamically updates the individual optimal value and the global optimal value based on the new fitness value: for the individual optimal value, the new fitness value of each particle is compared with its historical individual optimal fitness value, if the new value is higher, the individual optimal value of the particle is updated to the current new position and the new fitness value, retaining the best historical experience of the particle; if the new value is lower, the original individual optimal value is maintained unchanged, avoiding the loss of high-quality historical scheme due to single improper update. For the global optimal value, the system will traverse the new fitness value and the individual optimal value of all particles, filter out the particle position with the highest fitness value and meeting the three constraints, if the new fitness value of a particle exceeds the current global optimal value, the global optimal value is updated to the new position and the new fitness value of the particle; if the new values of all particles do not exceed the existing global optimal value, the principle of whether there is less constraint violation or larger potential optimization space is combined to judge whether the original global optimal value needs to be retained, to ensure that the global optimal value always represents the most reliable scheme in terms of comprehensive performance in the current population.

[0144] Further, the current executable optimal solution executability is confirmed, and the specific analysis method is as follows:

[0145] After iteration, the generated current executable optimal solution is checked for constraints.

[0146] If the current executable optimal solution meets the three constraint conditions, it is recorded as an optimization executable solution, and the strategy executable optimal solution generation confirmation information is output.

[0147] It should be noted that if the current executable optimal solution completely meets the above three constraints, it means that the scheme has reached the target in terms of fitness and constraint violation, and the system will officially record it as an optimization executable solution, and output the strategy executable optimal solution generation confirmation information, which includes the core parameters of the scheme, the fitness score, and the specific satisfaction of each constraint, providing clear scheme confirmation basis for planning personnel.

[0148] If the current executable optimal solution still does not completely meet the constraints, the strategy executable optimal solution generation failure information is output, if the current executable optimal solution is a slight violation solution, the constraint deviation can be recorded for the reference of planning personnel, if the current executable optimal solution is a serious violation solution, the violation amount needs to be recorded, and a warning information is generated.

[0149] It should be noted that if the current executable optimal solution still does not fully satisfy the constraint, the output strategy executable optimal solution generation fails information, and further according to the degree of violation subdivision record: if the solution belongs to the slight violation solution, the system will record the deviation value of each constraint, mark the position and influence range corresponding to the deviation, for the reference of the planning personnel, and provide the direction for the subsequent improvement of the scheme; if the solution belongs to the serious violation solution, the system will record the specific violation amount, and simultaneously generate a warning information, prompting that the current optimal solution has a serious compliance risk and cannot be directly applied, providing a problem basis for parameter correction in the next round of optimization process.

[0150] After receiving the strategy executable optimal solution generation confirmation information, the strategy executable optimal solution generation confirmation information corresponding to the solution is taken as the power grid facility layout planning strategy.

[0151] It should be noted that when the system receives the optimal solution generation confirmation information, it indicates that the solution has passed all core constraint checks and has the highest comprehensive fitness, and is formally determined as the power grid facility layout planning strategy. The strategy covers the core elements such as substation site selection and capacity, line path and cross section, and node load distribution, and can directly guide the actual construction. Taking the solution as the final planning strategy, the theoretical results of iterative optimization are essentially converted into an implementable engineering scheme, which provides a basis for implementation for planning personnel and a benchmark framework for subsequent operation and expansion, ensuring that the power grid not only meets the current load but also has flexibility for future development, realizing the closed-loop landing from the optimal theoretical solution to the actual engineering strategy.

[0152] The embodiment of the application provides a structure diagram of a power grid facility layout planning system based on multi-source perception data as shown in the figure. Figure 2 The processing flow of the system can include the following steps: a candidate set selection module, a constraint checking and repairing module, an iterative optimization module and an optimal solution generation confirmation.

[0153] The candidate set selection module is used for collecting and preprocessing multi-source perception data, uniformly dividing the planning area into a spatial grid, screening preliminary buildable sites and generating a preliminary candidate solution, performing preliminary candidate solution fitness calculation, and obtaining a candidate solution set.

[0154] The constraint checking and repairing module is used for performing layered power flow calculation to obtain a preliminary constraint checking strategy, and performing local repair adjustment when the preliminary constraint checking strategy is a local repair adjustment.

[0155] The iterative optimization module is used for determining a to-be-optimized solution after local repair adjustment, executing a hybrid optimization strategy and iterative optimization, performing current executable optimal solution executability confirmation after iteration, and outputting strategy executable optimal solution generation confirmation information.

[0156] The optimal solution generation confirmation module receives the strategy executable optimal solution generation confirmation information, and takes the solution corresponding to the strategy executable optimal solution generation confirmation information as the power grid facility layout planning strategy.

[0157] Referring to Figure 3 As shown in the mind map of the power grid facility layout planning system based on multi-source perception data provided by the embodiment of the application. The system collects load, power and remote sensing data and preprocesses, and statistically calculates key parameters of regional grid, calculates terrain constraints and selects preliminary buildable sites; uses a greedy algorithm to generate a preliminary candidate solution and simulate calculation fitness, selects high fitness solutions, and determines the to-be-optimized solution after constraint pre-check and repair; finally, based on the particle swarm optimization algorithm, iteratively optimizes, terminates according to the fitness change and the maximum number of iterations, and outputs the optimized executable solution after the constraint check.

[0158] The above embodiments can be realized wholly or partially by software, hardware (such as a circuit), firmware or any other combination. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the flow or function according to the embodiments of the present application is wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another, for example, the computer instructions can be transferred from one website, computer, server or data center to another by wired (for example, infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server, data center and the like containing one or more available medium collections. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD) or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0159] It should be understood that the terms and / or herein only describe the association relationship between the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the characters / in this paper generally represent that the associated objects before and after are one or a relationship, but it can also represent a and / or relationship, which can be understood according to the context.

[0160] In the present application, at least one means one or more, and multiple means two or more. At least one or the like means any combination of the items, including any combination of single item or multiple items. For example, at least one of a, b, or c can mean a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0161] It should be understood that the size of the sequence number of each process described above does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0162] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0163] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the above-described device, apparatus and unit can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0164] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for power grid facility layout planning based on multi-source perception data, characterized in that, The method comprises: Collect and pre-process multi-source perception data in the planning area, uniformly divide the planning area into spatial grids, screen preliminary buildable sites and generate preliminary candidate solutions, calculate the fitness of the preliminary candidate solutions, and obtain a candidate solution set; Perform hierarchical power flow calculation to obtain a preliminary constraint check strategy, and perform local repair adjustment when the preliminary constraint check strategy is local repair adjustment; After local repair adjustment, determine the to-be-optimized solution, execute the hybrid optimization strategy and iterative optimization, perform current executable optimal solution executability confirmation after iteration ends, and output strategy executable optimal solution generation confirmation information; After receiving the strategy executable optimal solution generation confirmation information, take the solution corresponding to the strategy executable optimal solution generation confirmation information as the power grid facility layout planning strategy; The specific analysis method of executing the hybrid optimization strategy and iterative optimization is as follows: Extract the preset maximum number of iterations; After each iteration of the input to-be-optimized solution, perform iteration effect judgment; After each iteration optimization, calculate the average fitness of all to-be-optimized solutions, observe the change value of the average value from the last time, denoted as the fitness change value, if the fitness change value is less than or equal to the fitness change value threshold, the current global optimal solution generated by the iteration optimization is recorded as the optimal solution, if the maximum number of iterations is not reached at this time, if the fitness change value is greater than the fitness change value threshold, continue to perform iteration optimization; If the maximum number of iterations is reached at this time, if the fitness change value is greater than the fitness change value threshold, generate a warning information; The specific iteration process is as follows: each input to-be-optimized solution is first mapped to a particle, and the position vector of the particle represents the network planning scheme through hybrid coding, including discrete decision variables and continuous variables, and the speed vector of the particle represents the change trend of the current position in the search space, including discrete variable speed and continuous variable speed; The system initializes the position of each particle as the to-be-optimized solution coding, empirically sets the speed, calculates the initial fitness, and records the individual optimal value and the global optimal value, and sets the current position as the individual optimal value, and the solution with the highest fitness in the group and meeting the constraints is taken as the global optimal value; In each iteration, the discrete variable records the hybrid optimization strategy as discrete iterative optimization, and the position is updated through the discrete PSO rule; The continuous variable records the hybrid optimization strategy as continuous iterative optimization, and is updated through the speed and position update formula of the continuous PSO; During the iteration process, the update direction is determined according to the individual optimal value and the global optimal value; After updating, the system re-performs the fitness calculation of the particle, and dynamically updates the individual optimal value and the global optimal value of the particle according to the new fitness value, gradually approaches the optimal solution using historical optimal experience and global information, and generates a current executable optimal solution.

2. The method of claim 1, wherein, The specific analysis method of screening preliminary buildable sites and generating preliminary candidate solutions is as follows: Collect and pre-process multi-source perception data, including load data, power supply data and remote sensing data; Divide the planning area into spatial grids evenly, count the total load, load density and power supply capacity of each grid, calculate the terrain constraint coefficient according to the GIS software, and mark the grids meeting the station building requirements in the planning area as preliminary buildable stations according to the GIS software, and generate an attribute table for the preliminary buildable stations; Screen the preliminary buildable stations and extract the attribute table of the preliminary buildable stations, and randomly generate each preliminary candidate solution based on the attribute table of the preliminary buildable stations by using the greedy algorithm.

3. The method of claim 1, wherein, The preliminary candidate solution fitness calculation is specifically analyzed as follows: For a preliminary candidate solution, each electrical connection point of each electrical equipment in the preliminary candidate solution is recorded as each node; The preliminary candidate solution is simulated quickly by using the direct current flow model, the current variation amplitude of the connection line of each node, the voltage variation amplitude of each node and the total operation loss of the whole network in the preliminary candidate solution are calculated and recorded, the actual electricity demand of the planning area in the planning period and the maximum power supply capacity that can be provided by the power grid facility under the operation condition are obtained, which are recorded as load demand and power supply capacity respectively; The robustness coefficient is calculated by tracking the voltage and current variation amplitude of the node; The operation loss degree is obtained according to the ratio of the total active loss obtained by the power flow calculation to the load power; The capacity matching degree is obtained by comparing the power supply capacity and the load demand; At the same time, the matching degree of the line and the terrain constraint is calculated by remote sensing and GIS data, which is recorded as the terrain constraint degree; After the capacity matching degree, the operation loss degree, the robustness coefficient and the terrain constraint degree are normalized, they are combined in a weighted manner to obtain the preliminary candidate solution fitness; The preliminary candidate solution fitness of each preliminary candidate solution is obtained by traversing each preliminary candidate solution.

4. The method of claim 3, wherein, The candidate solution set is obtained by the following specific analysis method: After obtaining the preliminary candidate solution fitness, the candidate solution fitness threshold is extracted; If the fitness of a preliminary candidate solution is greater than or equal to the candidate solution fitness threshold, it is recorded as a high fitness candidate solution; If the fitness of a preliminary candidate solution is less than the candidate solution fitness threshold, it is recorded as a low fitness candidate solution; The candidate solution set is obtained by counting all the high fitness candidate solutions.

5. The method of claim 1, wherein, The hierarchical power flow calculation is performed to obtain the preliminary constraint checking strategy, and the specific analysis method is as follows: The candidate solution set is linearized, the voltage, line carrying capacity and power supply coverage are preliminarily checked under the premise of ensuring calculation efficiency, and the specific process includes: constructing linearized power flow equation according to node load and line impedance, quickly solving node voltage amplitude and line current; The candidate solutions in the candidate solution set that meet the three constraint conditions are recorded as the candidate solutions to be determined, and the candidate solution set to be determined is obtained by counting; The candidate solution with the highest fitness in the candidate solution set to be determined is determined as the executable optimal solution, and the strategy executable optimal solution generation confirmation information is output; The three constraint conditions are that the voltage of each node is within the allowable range of the rated voltage, the carrying capacity of each line does not exceed the rated capacity, and the power supply coverage meets the preset area and load requirements; If there is no pending solution in the candidate solution set, the preliminary constraint check is used to determine the fine evaluation layer and the non-fine evaluation layer. One or two candidate solutions that do not satisfy the constraints are recorded as minor violation solutions, stored in the fine evaluation layer candidate solution set, and the preliminary constraint check strategy of the fine evaluation layer candidate solution set is recorded as local repair adjustment. Three fully unsatisfied candidate solutions are recorded as serious violation solutions and stored in the non-fine evaluation layer candidate solution set. The preliminary constraint check strategy of the non-fine evaluation layer candidate solution set is recorded as direct discard.

6. The method of claim 5, wherein, When the preliminary constraint check strategy is local repair adjustment, local repair adjustment is performed, and the specific analysis method is as follows: For a minor violation solution, key position identification is performed, that is, node voltage, line load flow and load distribution are analyzed to locate the nodes and lines that cause constraint violation. If there is a node load exceeding the configured substation capacity, the node is marked as a key node with insufficient capacity, and the output distribution of adjacent facilities is adjusted according to the load gap size and the remaining available capacity of each facility to meet the load demand. If there is a node voltage deviating from the rated voltage allowed range, the node is marked as a voltage abnormal node, and the impedance distribution range is determined according to the node voltage deviation to improve the voltage level. If there is a line load flow exceeding the rated capacity, the line is marked as an overloaded key line, and the load gap is distributed in proportion to the remaining capacity to reduce the current-carrying pressure. If there is no above-mentioned three cases, a warning information is generated.

7. The method of claim 1, wherein, The specific analysis method for determining the solution to be optimized is as follows: After local repair adjustment, constraint check is performed on the nodes and lines affected by the adjustment. The candidate solution that satisfies the three constraint conditions in the adjusted minor violation solution is recorded as a feasible solution, and the system iterative optimization is directly performed. For a minor violation solution, the system still allows it to enter the iterative optimization, but adds a constraint penalty in the fitness. For a serious violation solution, the optimization priority is reduced by an adaptive penalty function and does not participate in iteration for the time being. When a feasible solution and a minor violation solution are detected, the feasible solution and the minor violation solution are recorded as solutions to be optimized, and the iterative optimization is started.

8. The method of claim 1, wherein, The specific analysis method for confirming the executability of the current executable optimal solution is as follows: After iteration, constraint check is performed on the generated current executable optimal solution. If the current executable optimal solution satisfies the three constraint conditions, it is recorded as an optimization executable solution, and the output strategy executable optimal solution generation confirmation information is output. If the current executable optimal solution still does not completely satisfy the constraint, the output strategy executable optimal solution generation failure information is output. If the current executable optimal solution is a minor violation solution, the constraint deviation can be recorded for reference by the planner. If the current executable optimal solution is a serious violation solution, the violation amount needs to be recorded, and a warning information is generated.

9. A power grid facility layout planning system based on multi-source perception data, applying the method of any one of claims 1-8 for power grid facility layout planning based on multi-source perception data, characterized in that, The system comprises a candidate set selection module, a constraint check and repair module, an iterative optimization module and an optimal solution generation confirmation module. The candidate set selection module is used to collect and preprocess multi-source perception data, uniformly divide the planning area into spatial grids, screen preliminary buildable sites and generate preliminary candidate solutions, perform preliminary candidate solution fitness calculation, and obtain a candidate solution set. The constraint checking and repairing module is configured to perform layered power flow calculation to obtain a preliminary constraint checking strategy, and perform local repair adjustment when the preliminary constraint checking strategy is a local repair adjustment; The iterative optimization module is configured to determine an optimized solution after the local repair adjustment, execute a hybrid optimization strategy and iterative optimization, perform current executable optimal solution executability confirmation after the iterative optimization, and output strategy executable optimal solution generation confirmation information; The optimal solution generation confirmation module is configured to receive the strategy executable optimal solution generation confirmation information, and use a solution corresponding to the strategy executable optimal solution generation confirmation information as the power grid facility layout planning strategy; The hybrid optimization strategy and iterative optimization are executed, and the specific analysis method is as follows: The maximum number of iterations is extracted; The inputted optimized solution is subjected to iterative effect judgment after each iteration; After each iteration, the average value of the fitness of all optimized solutions is calculated, the change value of the average value from the last time is observed, and is recorded as the fitness change value. If the fitness change value is less than or equal to the fitness change value threshold, the current global optimal solution generated by the iterative optimization is recorded as the optimal solution. If the maximum number of iterations is not reached at this time, if the fitness change value is greater than the fitness change value threshold, the iterative optimization is continued; If the maximum number of iterations is reached at this time, if the fitness change value is greater than the fitness change value threshold, a warning information is generated; The specific iterative process is as follows: each inputted optimized solution is first mapped to a particle, the position vector of which represents the network planning scheme through hybrid coding, including discrete decision variables and continuous variables, and the velocity vector of the particle represents the change trend of the current position in the search space, including discrete variable velocity and continuous variable velocity; The system initializes the position of each particle as the optimized solution coding, sets the speed empirically, calculates the initial fitness, and records the individual optimal value and the global optimal value. The current position is set as the individual optimal value, and the solution with the highest fitness in the group and meeting the constraints is used as the global optimal value. In each iteration, the discrete variable uses the hybrid optimization strategy as the discrete iterative optimization, and updates the position through the discrete PSO rule; The continuous variable uses the hybrid optimization strategy as the continuous iterative optimization, and updates the speed and position through the continuous PSO update formula; During the iteration process, the update direction is determined according to the individual optimal value and the global optimal value; After the update, the system re-performs the fitness calculation of the particle, and dynamically updates the individual optimal value and the global optimal value of the particle according to the new fitness value. The historical optimal experience and global information are used to gradually approach the optimal solution, and the current executable optimal solution is generated.

Images