Power distribution network optimal reconstruction path generation method based on investment return rate constraint

Abstract

The application discloses a power distribution network framework optimal reconstruction path generation method based on investment yield rate constraints, realizes multidimensional deep fusion of technical indexes, full life cycle costs and quantitative benefits by constructing a sensitivity-cost-yield three-dimensional coupling tensor and a marginal investment efficiency matrix, effectively solves the economically unfeasible problems caused by one-sided analysis of traditional methods, thereby significantly reducing total reconstruction costs and shortening the investment recovery period; the investment yield rate and reliability improvement gradient constraints are introduced through a multi-constraint graph sequence decision algorithm, the limitations of static planning ignoring time sequence dynamic correlation are overcome, and power supply reliability is greatly improved; in combination with distributed robust optimization processing of uncertainties such as loads and equipment failures, the adaptability and robustness of the reconstruction scheme in the worst scenario are enhanced; and finally, the structured path directly guiding engineering implementation is output, and the scientificity, timeliness and operability of planning and decision are comprehensively improved.

Description

Technical Field

[0001] This invention belongs to the technical field of power system distribution networks, and particularly relates to a method for generating the optimal transformation path of distribution network structure based on investment return constraints. Background Technology

[0002] Distribution network upgrades are a core means to improve power supply reliability, adapt to distributed power source integration, and meet the demands of load growth. The rationality of the upgrade path directly affects investment efficiency, power supply security, and operational efficiency. With the expansion of distribution network scale, the increasing complexity of the operating environment, and tightening investment constraints, traditional methods for generating network upgrade paths face numerous challenges: Traditional methods only base their renovation plans on technical sensitivity (such as voltage sensitivity and load transfer sensitivity) without fully integrating economic indicators such as cost and benefits, resulting in technical feasibility but economic infeasibility, or redundant investment problems. Traditional methods often employ static programming models, failing to consider the dynamic correlation of the transformation sequence. Furthermore, the constraints only focus on the cost ceiling, lacking rigid constraints on the rate of return on investment and the reliability improvement gradient, making it difficult to achieve a balance between multiple objectives. Traditional methods lack effective handling of uncertain parameters such as load growth and equipment failure rate, and plan based solely on deterministic prediction data. This leads to problems such as lower-than-expected benefits and insufficient reliability in actual operation of the transformation path.

[0003] Therefore, there is an urgent need for a transformation path generation method that integrates multi-dimensional coupled analysis, multi-constraint dynamic optimization, and uncertainty handling to solve the core problems of traditional methods such as one-sided analysis, single optimization, and weak robustness, and to improve the scientificity and practicality of power distribution network transformation decisions. Summary of the Invention

[0004] The purpose of this invention is to provide a method for generating the optimal transformation path of a distribution network based on the return on investment constraint. By constructing a three-dimensional coupled tensor of sensitivity, cost, and benefit, designing a multi-constraint graph sequence decision algorithm, introducing sub-Bruker optimization and a closed-loop evaluation mechanism, the optimal transformation path that meets the return on investment requirement and takes into account both reliability and economy is generated, thereby achieving multi-objective optimization of reducing the total transformation cost, shortening the investment payback period, and improving power supply reliability.

[0005] To solve the above problems, the technical solution of the present invention is as follows: A method for generating optimal transformation paths for distribution network structures based on investment return constraints includes the following steps: Based on the basic data, dynamic data, and constraint data of the distribution network, a three-dimensional coupled tensor of sensitivity, cost, and benefit is constructed to characterize the correlation between core elements related to the transformation; a marginal investment efficiency matrix is ​​constructed to quantify the marginal investment efficiency of implementing different transformation measures in different regions; a dynamic graph model of the distribution network transformation process is established, and edge weights are determined based on the aforementioned three-dimensional coupled tensor of sensitivity, cost, and benefit and the marginal investment efficiency matrix; an optimal transformation path is generated through a multi-constraint graph sequence decision algorithm; based on the uncertainty of load growth rate and equipment failure rate, the transformation path is optimized and corrected using bibliometric optimization, and a structured optimal transformation path that directly guides the implementation of the project is output.

[0006] According to an embodiment of the present invention, the construction of the sensitivity-cost-benefit three-dimensional coupled tensor T∈R M×N×P The system includes: a first dimension M, a set of sensitivity indicators, including voltage quality sensitivity, line load rate sensitivity, N-1 safe pass rate sensitivity, and distributed power generation absorption sensitivity; a second dimension N, a set of upgrade cost indicators, including equipment procurement cost, construction and installation cost, power outage loss cost, and operation and maintenance cost; and a third dimension P, a set of upgrade benefit indicators, including increased electricity sales revenue, reduced network loss revenue, improved reliability revenue, and ancillary service revenue; and a tensor element T. m,n,p The coupling strength of the m-th type sensitivity index, the n-th type cost index, and the p-th type benefit index is represented by a normalized weighted calculation.

[0007] According to one embodiment of the present invention, the marginal investment efficiency matrix E has a dimension of K×L, where K is the type of renovation measure and L is the number of distribution network zones; matrix element E k,l The marginal investment efficiency of implementing type k transformation measures in region l is expressed by the following formula: Where, ΔR k,l ΔC represents the incremental marginal revenue after the modification. k,l For the marginal cost increment, ΔI k,l Marginal investment amount; λ k,l The risk adjustment factor is dynamically calculated based on the equipment failure probability distribution, and its value range is [1.0, 1.5].

[0008] According to an embodiment of the present invention, the constraints of the multi-constraint graph sequence decision algorithm include: Return on investment constraint: the return on investment (ROI) of the renovation project throughout its entire life cycle ≥ ROI target ROI target Target return on investment; reliability improvement gradient constraint: N-1 safety pass rate improvement value ΔP in adjacent years. t≥δ, where δ is the preset gradient threshold; Cost constraint: Total cumulative renovation cost ∑C t ≤C max Cmax is the upper limit of the total investment budget; timing constraint: the implementation sequence of the transformation measures must meet the preset technical dependencies.

[0009] According to an embodiment of the present invention, the multi-constraint graph sequence decision algorithm adopts a two-stage optimal path search, including: Stage 1: A* pruning search, using the maximization of cumulative investment return rate as the heuristic function, pruning transformation paths with marginal investment efficiency below a threshold to obtain a candidate path set; Stage 2: Improved genetic algorithm optimization, using the candidate path set as the initial population, iteratively optimizing to generate the optimal transformation path.

[0010] According to one embodiment of the present invention, the fitness function optimized by the improved genetic algorithm is constructed by weighting the return on investment, the N-1 safe pass rate, and the investment payback period or cumulative cost.

[0011] According to an embodiment of the present invention, the improved genetic algorithm optimization adopts an adaptive crossover and mutation strategy: crossover probability P c And the probability of mutation P m The crossover probability decreases as fitness increases, and the mutation probability decreases as fitness increases, when the fitness of an individual is greater than or equal to the average fitness of the population. The crossover operation adopts a time-series segmented crossover mode, which is divided into multiple time segments according to the year of modification, and gene exchange is only performed on non-critical time segments.

[0012] According to an embodiment of the present invention, the uncertainty-based optimization process includes: constructing a moment uncertainty fuzzy set U of uncertainty parameters. U={ξ∈R Q ∣E[ξ]=μ,V[ξ]≤Σ,ξmin≤ξ≤ξmax} Where ξ is the uncertainty parameter vector, E[ξ] is the expectation of the parameter vector ξ, μ is the expectation vector, V[ξ] is the variance of the parameter vector ξ, Σ is the upper bound matrix of variance, and ξmin and ξmax are the parameter value boundaries; the optimization objective is to maximize the expected return in the worst case within the fuzzy set U, and the solution is: in, For the profit function, This is the cost function.

[0013] According to an embodiment of the present invention, the method further includes: closed-loop evaluation and digital twin feedback: collecting measured data after the transformation is implemented, correcting the sensitivity matrix, marginal investment efficiency matrix and three-dimensional coupling tensor based on the measured data, and dynamically adjusting the fuzzy set boundary of the uncertainty parameters; re-running the multi-constraint graph sequence decision algorithm based on the updated parameters to perform rolling correction on the remaining transformation path.

[0014] According to one embodiment of the present invention, the output structured optimal transformation path includes: a transformation timeline, including an annual list of transformation measures and implementation cycle; a specific switching operation sequence; and a list of equipment replacement or addition, including model, specifications, quantity, and installation location.

[0015] A device for generating optimal transformation paths for a distribution network based on investment return constraints includes: a three-dimensional coupled tensor construction module, used to receive basic data, dynamic data, and constraint data of the distribution network, construct a sensitivity-cost-benefit three-dimensional coupled tensor to characterize the correlation between core elements related to the transformation; a marginal investment efficiency matrix calculation module, used to construct a marginal investment efficiency matrix to quantify the marginal investment efficiency of implementing different transformation measures in different regions; an optimal path generation module, used to establish a dynamic graph model of the distribution network transformation process, determine edge weights based on the sensitivity-cost-benefit three-dimensional coupled tensor and the marginal investment efficiency matrix, and generate the optimal transformation path through a multi-constraint graph sequence decision algorithm; and an optimization module, used to optimize and correct the transformation path based on the uncertainty of load growth rate and equipment failure rate using bibliometric optimization, and output a structured optimal transformation path that directly guides the implementation of the project.

[0016] Because the present invention adopts the above technical solution, it has the following advantages and positive effects compared with the prior art: The optimal transformation path generation method for distribution network based on return on investment constraints in one embodiment of the present invention achieves multi-dimensional deep integration of technical indicators, life-cycle costs, and quantified benefits by constructing a three-dimensional coupled tensor of sensitivity, cost, and benefit and a marginal investment efficiency matrix. This effectively solves the economic infeasibility problem caused by the one-sided analysis of traditional methods, thereby significantly reducing the total transformation cost and shortening the investment payback period. By introducing gradient constraints of return on investment and reliability improvement through a multi-constraint graph sequence decision algorithm, it overcomes the limitation of static planning that ignores the dynamic correlation of time series, and greatly improves the reliability of power supply. Combined with the decomposed bar optimization to handle uncertainties such as load and equipment failure, it enhances the adaptability and robustness of the transformation scheme under worst-case scenarios. Finally, it outputs a structured path that can directly guide the implementation of the project, realizing a comprehensive improvement in the scientificity, timeliness, and operability of planning decisions. Attached Figure Description

[0017] Figure 1This is a flowchart of a method for generating the optimal transformation path of a distribution network structure based on investment return constraints in one embodiment of the present invention; Figure 2 This is a schematic diagram of the sensitivity-cost-benefit three-dimensional coupled tensor structure in one embodiment of the present invention; Figure 3 This is a block diagram of a distribution network optimal transformation path generation device based on investment return rate constraints in one embodiment of the present invention. Detailed Implementation

[0018] The following description, in conjunction with the accompanying drawings and specific embodiments, provides a more detailed explanation of the method for generating the optimal transformation path of a distribution network structure based on investment return constraints proposed in this invention.

[0019] This embodiment provides a method for generating the optimal transformation path of a distribution network based on investment return constraints, including the following steps: S1: Based on the basic data, dynamic data, and constraint data of the distribution network, construct a three-dimensional coupled tensor of sensitivity-cost-benefit to characterize the correlation between the core elements related to the transformation; S2: Construct a marginal investment efficiency matrix to quantify the marginal investment efficiency of implementing different transformation measures in different regions; S3: Establish a dynamic graph model of the distribution network transformation process, determine the edge weights based on the three-dimensional coupled tensor of sensitivity-cost-benefit and the marginal investment efficiency matrix, and generate the optimal transformation path through a multi-constraint graph sequence decision algorithm; S4: Based on the decomposed bar optimization to handle the uncertainty of load growth rate and equipment failure rate, optimize and correct the transformation path, and output a structured optimal transformation path that directly guides the implementation of the project.

[0020] This method achieves a multi-dimensional deep integration of technical indicators, life-cycle costs, and quantified benefits by constructing a three-dimensional coupled tensor of sensitivity, cost, and benefit, and effectively solves the economic infeasibility problem caused by the one-sided analysis of traditional methods. This significantly reduces the total cost of the transformation and shortens the investment payback period. By introducing investment return rate and reliability improvement gradient constraints through a multi-constraint graph sequence decision algorithm, it overcomes the limitation of static planning that ignores the dynamic correlation of time series and greatly improves power supply reliability. Combined with the handling of uncertainties such as load and equipment failure by split-blown bar optimization, it enhances the adaptability and robustness of the transformation scheme under worst-case scenarios. Finally, it outputs a structured path that can directly guide the implementation of the project, and achieves a comprehensive improvement in the scientificity, timeliness, and operability of planning and decision-making.

[0021] For details, please refer to Figure 1In step S1, basic data, dynamic data and constraint data of the distribution network to be transformed are collected. The basic data of the distribution network includes network topology data, line electrical parameters and transformer equipment parameters. The dynamic data includes regional load time series data, equipment fault statistics data and electricity market price data. The constraint data includes investment budget details, target rate of return, reliability requirements (N-1 safety pass rate ≥ 95%), etc.

[0022] Please refer to Figure 2 Based on the collected data, a three-dimensional coupled tensor T∈R of sensitivity, cost, and benefit is constructed. M×N×P Includes: Defining tensor dimensions: The first dimension M is a set of sensitivity indicators, including voltage quality sensitivity, line load rate sensitivity, N-1 safe pass rate sensitivity, and distributed power generation absorption sensitivity; the second dimension N is a set of upgrade cost indicators, including equipment procurement cost, construction and installation cost, power outage loss cost, and operation and maintenance cost; the third dimension P is a set of upgrade benefit indicators, including increased electricity sales revenue, reduced network loss revenue, improved reliability revenue, and ancillary service revenue; tensor elements are calculated as follows: Tensor element T m,n,p The coupling strength between the m-th type of sensitivity index, the n-th type of cost index, and the p-th type of revenue index is obtained through normalized weighted calculation: Among them, S m C is the normalized value of the m-th type of sensitivity index. n R is the normalized value of the nth type of cost indicator. p This is the normalized value of the p-th type of return indicator.

[0023] By constructing a three-dimensional coupled tensor of sensitivity, cost, and benefit, the load rate sensitivity, construction cost, and network loss reduction benefit are normalized and their product is calculated as a coupling strength index. Maximizing this product means that the renovation measures simultaneously possess the urgency of technical needs, economic feasibility, and significant energy-saving benefits. Tensor decomposition further extracts the principal components that are strongly positively correlated with load sensitivity and network loss benefit, and negatively correlated with construction cost. This allows for the mathematically accurate identification of the optimal renovation combinations for heavily loaded lines that achieve significant network loss reduction with lower construction investment. Based on this, measures with the highest marginal investment efficiency, such as line upgrades and the addition of tie switches, are given priority recommendation.

[0024] In step S2, a marginal investment efficiency matrix is ​​constructed to quantify the marginal investment efficiency of implementing different renovation measures in different regions. The marginal investment efficiency matrix E has a dimension of K×L, where K represents the type of renovation measure and L represents the number of distribution network zones; matrix elements E k,lThe marginal investment efficiency of implementing type k transformation measures in region l is expressed by the following formula: Where, ΔR k,l ΔC represents the incremental marginal revenue after the modification. k,l For the marginal cost increment, ΔI k,l Marginal investment amount; λ k,l As a risk adjustment factor, λ is dynamically calculated based on the equipment failure probability distribution. The higher the failure probability, the higher the λ. k,l The larger the value, the range of its values ​​is [1.0, 1.5].

[0025] In step S3, a dynamic graph model of the power distribution network transformation process is established. The edge weights are determined based on the sensitivity-cost-benefit three-dimensional coupled tensor and the marginal investment efficiency matrix. The optimal transformation path is generated through a multi-constraint graph sequence decision algorithm.

[0026] In constructing the dynamic graph model, the power distribution network transformation process is modeled as a dynamic graph G. t =(V t E t W t ),in: V t The node set at time t (including device nodes, area nodes, and load nodes); E t Let t be the edge set at time t (including line connections and modification timing relationships); W t The edge weight set (obtained by weighting tensor elements and marginal investment efficiency matrix elements) is shown in the application example below for its specific calculation process, which will not be described in detail here.

[0027] The constraints of the multi-constraint graph sequence decision algorithm in this step include: Return on investment constraint: The return on investment (ROI) of the renovation project throughout its entire life cycle ≥ ROI target ROI target The target rate of return on investment is set at 8%–12% by default; the reliability improvement gradient constraint is the N-1 safety pass rate improvement value ΔP between adjacent years. t ≥δ, where δ is the preset gradient threshold; Cost constraint: Total cumulative renovation cost ∑C t ≤C max C max The total investment budget is capped; timing constraints: the implementation order of the renovation measures must meet the preset technical dependencies, such as line upgrades must precede the expansion and access of distributed power sources.

[0028] The multi-constraint graph sequence decision algorithm in this step adopts a two-stage optimal path search: Stage 1: A* pruning search, using the maximization of cumulative investment return rate as the heuristic function, prunes transformation paths with marginal investment efficiency below a threshold to obtain a candidate path set; Stage 2: Improved genetic algorithm optimization, using the candidate path set as the initial population, iteratively optimizes to generate the optimal transformation path.

[0029] In Phase 2, the fitness function is designed using the candidate path set as the initial population: in, These are the weighting coefficients. For an N-1 safe pass rate, For cumulative costs, For the investment recovery period, The target investment payback period is set. An optimal transformation path (including transformation measure type, implementation area, and timing) is generated through adaptive crossover mutation and elite retention strategies. The specific strategies are as follows: Dynamic adjustment strategy for crossover probability: crossover probability The initial maximum value is set to 0.9 and the minimum value to 0.4, which are gradually decreased as the iteration progresses. When the individual fitness... average fitness of the population At that time, according to the formula Adjustments are made to reduce the crossover probability of individuals whose fitness is closer to the optimal value; when At that time, keep To promote gene recombination. Late iteration (number of iterations) Introduce a decay factor for the total number of iterations. To further reduce the crossover probability and stabilize the optimal solution, a time-series crossover operation is adopted, dividing the process into 10 time segments based on the year of modification. Gene exchange is only performed on non-critical time segments (without irreplaceable technical dependencies). After crossover, the path validity is verified; if the time constraints are violated, the crossover period is backtracked and reselected. The mutation probability is dynamically adjusted: the mutation probability... The initial maximum value is 0.2 and the minimum value is 0.05, decreasing as the iteration progresses. At that time, according to the formula Adjustments are made, and individuals with lower fitness have a higher probability of variation; when At that time, keep To increase genetic diversity. Later stages of iteration. Introducing convergence factor To balance local exploration and convergence speed, mutation operations are performed with a focus on the type of modification measure, implementation area, and timing. After mutation, the marginal investment efficiency threshold must be met (not less than 70% of the current population's average marginal investment efficiency).

[0030] Elite retention strategy: Each generation selects the top 5% of individuals by fitness (2-3 individuals are retained when the population size is 50) as elite candidates, which must simultaneously meet all constraints and robustness requirements (the worst-case return on investment should not be less than 95% of the target value). Elite individuals are directly copied into the next generation without participating in crossover and mutation operations; 50% of the new individuals in the next generation are generated by adjusting non-core elements based on the elite individuals as templates; during crossover, priority is given to retaining the key time segments and core modification measures of the elite individuals; when the fitness improvement of elites is ≤0.5% for three consecutive generations, they are replaced with high-quality individuals with ≤60% genetic similarity to the existing elites to avoid the algorithm getting trapped in local optima.

[0031] In step S4, based on the uncertainty of load growth rate and equipment failure rate in the split-bar optimization process, the transformation path is optimized and corrected, and a structured optimal transformation path that directly guides the implementation of the project is output.

[0032] Among them, uncertainty is handled based on the pluripotent bar optimization, including: constructing a moment uncertain fuzzy set U for uncertain parameters such as load growth rate and equipment failure rate. U={ξ∈R Q ∣E[ξ]=μ,V[ξ]≤Σ,ξmin≤ξ≤ξmax} Where ξ is the uncertainty parameter vector, E[ξ] is the expectation of the parameter vector ξ, μ is the expectation vector, V[ξ] is the variance of the parameter vector ξ, Σ is the upper limit matrix of variance, and ξmin and ξmax are the parameter value boundaries; Robust optimization objective: Maximize the expected return in the worst case within the fuzzy set U to ensure the adaptability of the modification path to uncertainty. Solution: in, For the profit function, This is the cost function.

[0033] Through the above optimization, the optimal modification path is obtained. The output is a structured optimal modification path that can directly guide project implementation, including: The renovation schedule includes an annual list of renovation measures and their implementation cycle; a specific sequence of switch operations; and a list of equipment to be replaced or added, including model, specifications, quantity, and installation location.

[0034] Furthermore, the above method also includes: closed-loop evaluation and digital twin feedback steps: collecting measured data after the transformation is implemented, correcting the sensitivity matrix, marginal investment efficiency matrix and three-dimensional coupling tensor based on the measured data, and dynamically adjusting the fuzzy set boundary of the uncertainty parameters; re-running the multi-constraint graph sequence decision algorithm based on the updated parameters to perform rolling corrections on the remaining transformation paths.

[0035] For example, after the renovation is implemented, measured data such as equipment operating status, power supply reliability, network loss, and actual benefits are collected and transmitted back to the decision-making center through the SCADA system and digital twin platform. Based on the measured data, the sensitivity matrix, marginal investment efficiency matrix, and three-dimensional coupling tensor are corrected, and the fuzzy set boundary and moment information of the uncertainty parameters are dynamically adjusted. Each year, based on the updated parameters, the multi-constraint graph sequence decision algorithm is re-run to optimize and adjust the remaining renovation paths, forming a closed-loop optimization of planning-implementation-evaluation-correction.

[0036] The following application example will be used to specifically introduce the method for generating the optimal transformation path of the distribution network based on the investment return rate constraint.

[0037] Taking a large 10kV distribution network as an example, the effectiveness of this method is verified. The specific parameters are as follows: Distribution network parameters: voltage level 110kV / 10kV, total line length 1500km, 1200 transformers, existing N-1 safety pass rate 78%; Load characteristics: The current peak load is 2000MW, and the load growth rate is predicted to be 5%~8% / year (uncertainty parameter); Investment constraints: Total investment budget of 500 million yuan, target return on investment (ROI) ≥ 10%, target investment payback period ≤ 8 years; Comparison methods: traditional static programming (based on technical sensitivity and cost ceiling constraints) and greedy algorithm (selecting the optimal modification measures year by year).

[0038] 1. Data Acquisition and Preprocessing Collect 5-year baseline data: Static data: distribution network topology, line resistance / reactance, equipment model and service life, regional load distribution; Dynamic data: load time series data, equipment failure statistics (number of failures and causes of failures in the past 3 years), and electricity market price data; Constraints include: detailed investment budget, target rate of return, and reliability requirements (N-1 safety pass rate ≥ 95%). The data is cleaned and normalized to construct the model input dataset.

[0039] 2. Construction of 3D Coupled Tensor and Marginal Investment Efficiency Matrix 1) Three-dimensional coupling tensor: The dimension is 4×4×4 (4 types of sensitivity indicators, 4 types of cost indicators, and 4 types of benefit indicators). Through coupling analysis, the core coupling relationship of high line load rate sensitivity, low construction cost, and high benefit from reduced network loss is identified, and line upgrades and the addition of tie switches are determined to be the highest priority renovation measures.

[0040] 2) Marginal investment efficiency matrix: It includes 8 types of renovation measures and 10 distribution network zones. The risk adjustment factor is calculated based on the probability of equipment failure. The old line area has λ=1.4 and the new area has λ=1.0. The matrix results show that the marginal investment efficiency of "East Zone line upgrade" and "West Zone interconnection switch addition" is the highest ≥1.8.

[0041] 3. Execution of the multi-constraint graph sequence decision algorithm 1) Dynamic graph modeling: Construct a 10-year time-series dynamic graph, and fuse the edge weights with the coupling tensor and marginal investment efficiency matrix results; 2) A* Pruned Search: Using "maximizing cumulative ROI" as the heuristic function, 25 candidate paths are obtained after pruning; 3) Improved genetic algorithm optimization: Set the population size to 50, the number of iterations to 100, the crossover probability to 0.7, the mutation probability to 0.1, and the fitness function weights ω1=0.4, ω2=0.3, ω3=0.3 to finally generate the optimal modification path.

[0042] In dynamic graph modeling, a power distribution network is divided into three renovation zones: East Zone, West Zone, and South Zone (L=3). Considering two types of renovation measures—line upgrades and the addition of tie switches (K=2)—the renovation is planned to be implemented in year t, including the following steps: Step 1: Extract the core elements of the 3D coupled tensor Key coupling relationships were identified through tensor decomposition, yielding normalized values ​​for the dimensions of line load rate sensitivity, construction cost, and network loss reduction benefits: Sensitivity normalization values: The highest line load rate sensitivity is set at 0.85 for the East Zone; 0.60 for the West Zone; and 0.45 for the South Zone.

[0043] Cost reciprocal: The construction cost in the East District is lower, so we take 0.70; the West District takes 0.50; and the South District takes 0.40 (the lower the cost, the higher the weight).

[0044] Normalized revenue values: The East region has the greatest potential for reducing network losses, so we take 0.90; the West region takes 0.65; and the South region takes 0.55.

[0045] Coupling strength calculation (simplified formula): T 分区 = East District: 0.58 × 0.7 × 0.9 = 0.5355 West District: 0.6 × 0.5 × 0.65 = 0.195 South area: 0.45×0.4×0.55=0.099.

[0046] Step 2: Extract elements of the marginal investment efficiency matrix Calculate the risk adjustment factor λ based on the equipment failure rate. k,l : The lines in the eastern area are old, with λ=1.4; the western and southern areas are newly built, with λ=1.0. Marginal investment efficiency E k,l Calculation results: East District Line Upgrade: E 1,l =1.85 (High returns, high sensitivity) New switch added in the East Zone: E 2,l =1.6 West District Line Upgrade: E 1,2 =1.2 New switch added in the west area: E 2,2 =1.45 (The tie switch has higher marginal efficiency) South District Line Upgrade: E 1,3 =0.95 New switch added in the South District: E 2,3 =1.1.

[0047] Step 3: Generate a weighted set of edge weights W t Animated Graph G t It contains two types of edges: Spatial connection edge: indicates electrical connection between areas (such as East Zone - West Zone); Temporal decision edge: Represents the modification action of a certain partition and a certain modification measure; The time-series decision edge weights (core) are fused using a linear weighting method: W t (Partition l, Measure k) = α·T 分区l +β·E k,l Where α = 0.4 and β = 0.6 (adjusted based on engineering experience, emphasizing investment efficiency).

[0048] The calculation results are as follows: Step 4: Spatial Connectivity Edge Weights (Auxiliary) The weight of the inter-regional electrical coupling strength is determined by the spatial correlation of the sensitivity index: East Zone - West Zone: The two zones have strong load complementarity, and the weight is set to 0.75; East-South region: The coupling is relatively weak, and the weight is set to 0.45; West Zone - South Zone: Connected via the main line, with a weight set to 0.60.

[0049] Final edge weight set W t constitute: A* search uses the cumulative weight maximization heuristic function, prioritizing the East Zone-Line upgrade as the primary renovation measure in year t; the genetic algorithm is based on the complete W... t The set is used for chromosome coding and fitness assessment.

[0050] The A* pruning search, using the maximization of cumulative return on investment as its heuristic function, prunes transformation paths whose marginal investment efficiency is below a threshold, thus obtaining a set of candidate paths. Specifically: Step 1: Calculate the marginal efficiency of investment E for each improvement measure. k,l Contribution to ROI: After the first round of pruning: only three effective renovation measures will be retained: upgrading the lines in the east area, adding new switches in the east area, and adding new switches in the west area.

[0051] Step 2: Define A* search nodes and heuristic functions Node status representation: [Modified sequence, cumulative cost, cumulative ROI, current year]; Heuristic function design: g(n) = Cumulative ROI (actual value) of confirmed renovations; h(n) = Estimated maximum future ROI = MIN(remaining budget, revenue at threshold 1.30) / cumulative cost; f(n) = g(n) + h(n) (Prioritize expanding the node with the largest f value).

[0052] Step 3: Hierarchical Search and Dynamic Pruning Year 1 Decision-Making Layer (Root Node Expansion) Root: [∅, Cost=0, ROI=0%, Year=0] East Zone Line Upgrade: [EL, Cost=8000, ROI=15.0%, Year=1] (E=1.50) New switch added in the East Zone: [ES, Cost=3000, ROI=14.0%, Year=1] (E=1.40) New switch in the West Zone: [WS, Cost=2500, ROI=15.2%, Year=1] (E=1.52); Without pruning, all 3 nodes are retained, and the newly added switches in the western region are expanded first in order of f(n) (highest ROI).

[0053] The decision-making level in the second year (expanded from the newly added switch nodes in the western area) Current status: [WS, Cost=2500, ROI=15.2%, Year=1]; Optional actions (technical dependencies must be considered: if the East Zone line upgrade is selected, switches can be added to that zone later; if the East Zone new switch is selected first, then the upgrade cannot be performed): Key pruning: If the instantaneous marginal efficiency (new ROI in this step / new cost in this step) of a certain expansion path is less than 1.30, then prune it directly. All expansions in this layer meet the criteria.

[0054] Year 3 Decision-making level (from WS → expansion of new switch nodes in the eastern region) Current status: [WS→ES, Cost=5500, ROI=14.8%, Year=2] Remaining budget: 245 million yuan; technical constraints: new switches have been added in the East Zone and cannot be upgraded further (timing dependency conflict).

[0055] Step 4: Generate a candidate path set After a three-level search, seven complete paths were ultimately retained, for example: 1. Path 1: West Zone Switch → East Zone Line Upgrade → West Zone Switch Total cost: 2500 + 8000 + 2500 = 130 million, cumulative ROI = 15.1% 2. Path 2: East Zone Line Upgrade → West Zone Switch → East Zone Switch Total cost: 8000 + 2500 + 3000 = 135 million, cumulative ROI = 14.9% 3. Path 7: East Zone Line Upgrade → East Zone Switch → West Zone Switch Total cost: 8000 + 3000 + 2500 = 135 million, cumulative ROI = 14.7% Pruning effect: The original combination space has possibilities → after efficiency threshold pruning → 7 high-quality candidate paths, which meets the "20~30" control range described in the patent (this example is simplified to fewer).

[0056] Step 5: Enter the genetic algorithm optimization. Using the above 7 paths as the initial population, input them into the improved genetic algorithm with the following fitness function: The solution ultimately converges to Path 2 as the optimal solution because it has the most balanced N-1 pass rate improvement (+5% after the upgrade of the East District, +3% after the switch of the West District, and +2% after further optimization of the switch of the East District), which satisfies the gradient constraint that the improvement in adjacent years is ≥3%.

[0057] 4. Optimization and verification of the Blue Bar. A fuzzy set was constructed with load growth of 5%~8% and equipment failure rate baseline value ±30%. In the worst-case scenario with load growth of 8%+ and equipment failure rate increase of 30%, the expected ROI of the transformation path was 10.2% (≥ target value 10%), and the N-1 safety pass rate was 98.1% (≥ required value 95%), which verified the robustness.

[0058] 5. Comparison of Experimental Results Therefore, the distribution network upgrade path generated by this method achieves a 18% reduction in total upgrade cost, a 2.3-year shortening of the investment payback period, and an N-1 safety pass rate of 98.5%, while meeting the target return on investment of 10%. Furthermore, the path generation time is only 4.7 hours, supporting annual rolling revisions. Compared to traditional methods, this method has significant advantages in economy, reliability, decision-making efficiency, and robustness. The output structured upgrade scheme can directly guide project implementation, validating its practical value in large-scale distribution network upgrades.

[0059] Based on the same concept, this application also provides a device for generating the optimal transformation path of a distribution network structure based on investment return constraints. Please refer to [link / reference]. Figure 3 The device includes: a three-dimensional coupled tensor construction module, used to receive basic data, dynamic data, and constraint data of the distribution network, construct a sensitivity-cost-benefit three-dimensional coupled tensor, and characterize the correlation between core elements related to the transformation; a marginal investment efficiency matrix calculation module, used to construct a marginal investment efficiency matrix, and quantify the marginal investment efficiency of implementing different transformation measures in different regions; an optimal path generation module, used to establish a dynamic graph model of the distribution network transformation process, determine edge weights based on the sensitivity-cost-benefit three-dimensional coupled tensor and the marginal investment efficiency matrix, and generate the optimal transformation path through a multi-constraint graph sequence decision algorithm; and an optimization module, used to optimize and correct the transformation path based on the uncertainty of load growth rate and equipment failure rate using bibliometric optimization, and output a structured optimal transformation path that directly guides the implementation of the project.

[0060] This device is used to implement the above-mentioned method for generating the optimal transformation path of the distribution network structure based on the investment return rate constraint. Its implementation method is similar and will not be described in detail here.

[0061] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments. Even if various changes are made to the present invention, if these changes fall within the scope of the claims of the present invention and their equivalents, they shall still fall within the protection scope of the present invention.

Claims

1. A method for generating the optimal transformation path of a distribution network structure based on investment return constraints, characterized in that, Includes the following steps: Based on the basic, dynamic, and constraint data of the power distribution network, a three-dimensional coupled tensor of sensitivity, cost, and benefit is constructed to characterize the correlation between core elements related to the transformation; a marginal investment efficiency matrix is ​​constructed to quantify the marginal investment efficiency of implementing different transformation measures in different regions. A dynamic graph model of the power distribution network transformation process is established. The edge weights are determined based on the sensitivity-cost-benefit three-dimensional coupled tensor and the marginal investment efficiency matrix. The optimal transformation path is generated through a multi-constraint graph sequence decision algorithm. Based on the uncertainty of load growth rate and equipment failure rate in the optimization of the distributed bar, the transformation path is optimized and corrected, and a structured optimal transformation path is output to directly guide the implementation of the project.

2. The method for generating the optimal transformation path of a distribution network based on investment return constraints according to claim 1, characterized in that, The constructed sensitivity-cost-benefit three-dimensional coupled tensor T∈R M×N×P The system includes: a first dimension M, a set of sensitivity indicators, including voltage quality sensitivity, line load rate sensitivity, N-1 safe pass rate sensitivity, and distributed power generation absorption sensitivity; a second dimension N, a set of upgrade cost indicators, including equipment procurement cost, construction and installation cost, power outage loss cost, and operation and maintenance cost; and a third dimension P, a set of upgrade benefit indicators, including increased electricity sales revenue, reduced network loss revenue, improved reliability revenue, and ancillary service revenue; and a tensor element T. m,n,p The coupling strength of the m-th type sensitivity index, the n-th type cost index, and the p-th type benefit index is represented by a normalized weighted calculation.

3. The method for generating the optimal transformation path of a distribution network based on investment return constraints according to claim 1, characterized in that, The marginal investment efficiency matrix E has a dimension of K×L, where K represents the type of renovation measure and L represents the number of distribution network zones; matrix element E k,l The marginal investment efficiency of implementing type k transformation measures in region l is expressed by the following formula: Where, ΔR k,l ΔC represents the incremental marginal revenue after the modification. k,l For the marginal cost increment, ΔI k,l Marginal investment amount; λ k,l This is the risk adjustment factor, with a value range of [1.0, 1.5].

4. The method for generating the optimal transformation path of a distribution network based on investment return constraints according to claim 1, characterized in that, The constraints of the multi-constraint graph sequence decision algorithm include: Return on investment constraint: The return on investment (ROI) of the renovation project throughout its entire life cycle ≥ ROI target ROI target Target return on investment; reliability improvement gradient constraint: N-1 safety pass rate improvement value ΔP in adjacent years. t ≥δ, where δ is the preset gradient threshold; Cost constraint: Total cumulative renovation cost ∑C t ≤C max Cmax is the upper limit of the total investment budget; timing constraint: the implementation sequence of the transformation measures must meet the preset technical dependencies.

5. The method for generating the optimal transformation path of a distribution network structure based on investment return constraints according to claim 1 or 4, characterized in that, The multi-constraint graph sequence decision algorithm adopts a two-stage optimal path search, including: Stage 1: A* pruning search, using the maximization of cumulative investment return rate as the heuristic function, pruning transformation paths with marginal investment efficiency below a threshold to obtain a candidate path set; Stage 2: Improved genetic algorithm optimization, using the candidate path set as the initial population, iteratively optimizing to generate the optimal transformation path.

6. The method for generating the optimal transformation path of a distribution network based on investment return constraints according to claim 5, characterized in that, The fitness function optimized by the improved genetic algorithm is constructed by weighting the return on investment, the N-1 safe pass rate, and the payback period or cumulative cost.

7. The method for generating the optimal transformation path of a distribution network based on investment return constraints according to claim 5, characterized in that, The improved genetic algorithm optimization employs an adaptive crossover and mutation strategy: crossover probability P c And the probability of mutation P m The fitness of an individual is dynamically adjusted as the iteration progresses; when the fitness of an individual is greater than or equal to the average fitness of the population, the crossover probability decreases as fitness increases, and the mutation probability decreases as fitness increases. The crossover operation adopts a time-series segmented crossover mode, which is divided into multiple time series according to the year of modification, and gene exchange is only performed on non-critical time series.

8. The method for generating the optimal transformation path of a distribution network based on investment return constraints according to claim 1, characterized in that, The uncertainty-based optimization process includes: constructing a moment-uncertain fuzzy set U of uncertainty parameters. U={ξ∈R Q ∣E[ξ]=μ,V[ξ]≤Σ,ξmin≤ξ≤ξmax} Where ξ is the uncertainty parameter vector, E[ξ] is the expectation of the parameter vector ξ, μ is the expectation vector, V[ξ] is the variance of the parameter vector ξ, Σ is the upper bound matrix of variance, and ξmin and ξmax are the parameter value boundaries; the optimization objective is to maximize the expected return in the worst case within the fuzzy set U, and the solution is: in, For the profit function, This is the cost function.

9. The method for generating the optimal transformation path of a distribution network based on investment return constraints according to claim 1, characterized in that, Also includes: Closed-loop evaluation and digital twin feedback: Collect measured data after the transformation is implemented, correct the sensitivity matrix, marginal investment efficiency matrix and three-dimensional coupling tensor based on the measured data, and dynamically adjust the fuzzy set boundary of the uncertainty parameters; rerun the multi-constraint graph sequence decision algorithm based on the updated parameters to make rolling corrections to the remaining transformation paths.

10. A device for generating optimal transformation paths for distribution network structures based on investment return constraints, characterized in that, include: The three-dimensional coupled tensor construction module is used to receive basic data, dynamic data and constraint data of the power distribution network, construct a sensitivity-cost-benefit three-dimensional coupled tensor, and characterize the correlation of core elements related to the transformation; the marginal investment efficiency matrix calculation module is used to construct a marginal investment efficiency matrix to quantify the marginal investment efficiency of implementing different transformation measures in different regions. The optimal path generation module is used to establish a dynamic graph model of the power distribution network transformation process. Based on the sensitivity-cost-benefit three-dimensional coupled tensor and the marginal investment efficiency matrix, the edge weights are determined, and the optimal transformation path is generated through a multi-constraint graph sequence decision algorithm. The optimization module is used to optimize and correct the transformation path based on the uncertainty of load growth rate and equipment failure rate by using the BLU bar optimization, and outputs the structured optimal transformation path to directly guide the implementation of the project.

Images