Power distribution network supply chain balanced configuration method and device based on multi-target collaborative optimization and dynamic weight distribution

Abstract

The invention discloses a power distribution network supply chain balanced configuration method and device based on multi-objective collaborative optimization and dynamic weight allocation, and relates to the field of power distribution network planning and supply chain management.The method comprises the steps that firstly, a multi-objective optimization model including economical efficiency, reliability and environmental influence is constructed; then, through a dynamic weight distribution mechanism based on real-time operation data of each link of the supply chain, carrying out collaborative optimization on multiple targets; then, solving the optimization model by adopting an evolutionary algorithm fused with a self-adaptive mechanism to obtain an optimal configuration scheme; and finally, putting the scheme into operation and continuously collecting new data so as to realize closed-loop dynamic adjustment of the weight and the scheme. Through the dynamic weight distribution and closed-loop feedback mechanism, the problem that a traditional static configuration method is difficult to adapt to the real-time operation state change of the power distribution network is effectively solved, and the timeliness, balance and comprehensive benefits of a supply chain configuration scheme are remarkably improved.

Description

Technical Field

[0001] This application relates to the fields of distribution network planning and supply chain management, and in particular to a method and apparatus for balanced configuration of distribution network supply chain based on multi-objective collaborative optimization and dynamic weight allocation. Background Technology

[0002] In the process of building new power systems and energy transition, the distribution network supply chain is facing multiple uncertainties. On the one hand, frequent extreme natural disasters have led to a sharp increase in the risk of material transportation disruptions, and sudden accidents have caused damage to key distribution equipment assets and the spread of faults. On the other hand, the drastic load fluctuations caused by the grid connection of new energy sources have resulted in short-term, highly volatile demand for distribution network materials. Most existing optimization methods take cost as the sole objective function, overemphasizing cost control and leading to problems such as delays in the delivery of important materials and insufficient reserves. At the same time, the static allocation mechanism with fixed weights is difficult to adapt to different scenarios such as the rapid spread of faults in smart distribution terminals and the high requirements for emergency repair, resulting in a serious mismatch between resource allocation and actual demand. Summary of the Invention

[0003] The purpose of this application is to provide a distribution network supply chain balancing configuration method and device based on multi-objective collaborative optimization and dynamic weight allocation. It can collaboratively process multiple objectives and dynamically adjust the importance of each objective according to real-time operating data, thereby realizing balanced, efficient and adaptive distribution network supply chain management.

[0004] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation, including: A multi-objective optimization model is established with economic indicators, reliability indicators, low-carbon indicators, and safety indicators as optimization objectives. The multi-objective optimization model includes a first objective function, a second objective function, a third objective function, and a fourth objective function. The economic indicators include equipment procurement cost, operation and maintenance cost, and loss cost. The reliability indicators include average user outage time and average system outage frequency. The low-carbon indicator is the total carbon emissions during the operation of the distribution network. The safety indicators include equipment load rate margin and short-circuit current margin. Real-time collection of operational data from the power distribution network supply chain, including load fluctuation data, equipment health status data, and environmental meteorological data; Based on the operational data, the dynamic weight values ​​of economic indicators, reliability indicators and environmental impact indicators are calculated using a preset dynamic weight calculation rule. The dynamic weight values ​​are used to characterize the priority of each target in the current operational state. Based on the aforementioned multi-objective optimization model and dynamic weight values, an equilibrium optimization model considering spatiotemporal dynamic constraints is established. The optimal distribution network supply chain balance configuration scheme is obtained by solving the equilibrium optimization model that considers spatiotemporal dynamic constraints.

[0005] Optionally, the expression for the first objective function is as follows: ; In the formula, , and These are equipment transportation costs, inventory costs, and power outage loss costs; The expression for equipment transportation costs is as follows: ; In the formula, For equipment To the fault point The transportation distance In order to allocate the amount of supplies, For unit distance transportation cost, For the number of devices, This represents the number of fault points. The constraints on equipment transportation costs are as follows: ; In the formula, This represents the upper limit of transportation capacity. The expression for inventory cost is as follows: ; In the formula, For the first Safety stock levels of such materials Unit inventory holding cost For excess inventory, Excessive storage incurs penalty costs. Category of inventory materials; The expression for the cost of power outage losses is as follows: ; In the formula, For users Unit power outage losses, for User power outage status during the specified time period =1 indicates a power outage. =0 is normal For users Load weight, For the number of time periods, For the number of users.

[0006] Optionally, the expression for the second objective function is as follows: ; In the formula, To ensure sufficient redundancy of emergency repair materials, To ensure a timely response to emergency repairs, This refers to the backup power capacity index. , and All are elastic coefficients; The expression for the redundancy of emergency repair materials is as follows: ; In the formula, For the first Redundancy of similar materials For standard demand, This refers to the total number of categories of emergency repair materials; The expression for the emergency repair response time is as follows: ; In the formula, For the first The time it takes for the supplies to reach the fault location To allow the maximum time limit, The total number of batches of emergency repair materials; The expression for the backup power capacity index is as follows: ; In the formula, For nodes Backup power capacity, As the importance weight of the nodes, The total number of nodes; in, .

[0007] Optionally, the expression for the third objective function is as follows: ; In the formula, To allow time for material preparation, For transportation time; The expression for the material preparation time is as follows: ; In the formula, The standard time for the release of goods of category k; The expression for transit time is as follows: ; In the formula, For equipment To the fault point The transportation distance In order to allocate the amount of supplies, This represents the average transport speed.

[0008] Optionally, the expression for the fourth objective function is as follows: ; In the formula, For the region The level of material inventory, This represents the regional average inventory level. Total number of regions; The expression for the inventory variance constraint is as follows: ; In the formula, This represents the maximum allowable inventory variance.

[0009] Optionally, the dynamic weight calculation rule is specifically as follows: When load fluctuation data exceeds a preset threshold or equipment health status data falls below a preset threshold, the dynamic weight value of reliability indicators is increased, while the dynamic weight value of economic indicators is decreased. When environmental meteorological data indicate that the region is in a high carbon emission sensitive period, the dynamic weight value of environmental impact indicators should be increased.

[0010] Optionally, the expression for the equilibrium optimization model considering spatiotemporal dynamic constraints is as follows: ; In the formula, The weights for each objective function, , , , Each objective function represents a different objective function.

[0011] Optionally, the distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation further includes the following after the last step: The optimal distribution network supply chain balancing configuration scheme is used as the final distribution network supply chain configuration decision and applied to the planning, construction, or operation and scheduling of the distribution network.

[0012] Secondly, this application provides a power distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation, comprising: A multi-objective optimization model construction module is used to establish a multi-objective optimization model with economic indicators, reliability indicators, low-carbon indicators, and safety indicators as optimization objectives. The multi-objective optimization model includes a first objective function, a second objective function, a third objective function, and a fourth objective function. The economic indicators include equipment procurement costs, operation and maintenance costs, and loss costs. The reliability indicators include average user outage time and average system outage frequency. The low-carbon indicator is the total carbon emissions during the operation of the distribution network. The safety indicators include equipment load rate margin and short-circuit current margin. The data acquisition module is used to collect real-time operational data of the power distribution network supply chain, including load fluctuation data, equipment health status data, and environmental meteorological data. The dynamic weight value calculation module is used to calculate the dynamic weight values ​​of economic indicators, reliability indicators and environmental impact indicators respectively according to the operation data and using preset dynamic weight calculation rules. The dynamic weight values ​​are used to characterize the priority of each target in the current operation state. The equilibrium optimization model construction module is used to establish an equilibrium optimization model that considers spatiotemporal dynamic constraints based on the multi-objective optimization model and dynamic weight values. The solution module is used to solve the equilibrium optimization model that considers spatiotemporal dynamic constraints to obtain the optimal distribution network supply chain equilibrium configuration scheme.

[0013] Optionally, the distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation further includes: The planning and scheduling module is used to take the optimal distribution network supply chain balance configuration scheme as the final distribution network supply chain configuration decision and apply it to the planning, construction or operation scheduling of the distribution network.

[0014] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a distribution network supply chain balancing configuration method and device based on multi-objective collaborative optimization and dynamic weight allocation, which has the following beneficial effects: Comprehensiveness: A multi-objective optimization model integrating economic, reliability, and environmental protection dimensions has been constructed, avoiding the one-sidedness of single-objective optimization and generating a balanced configuration scheme with better overall benefits.

[0015] Adaptability: Through a dynamic weight allocation mechanism, the priority of each optimization objective can be automatically adjusted according to the real-time operation data of the distribution network (such as load, equipment status, and environment), so that the decision-making process is closely integrated with the actual state of the power grid, significantly improving the timeliness and adaptability of the solution.

[0016] High efficiency: The differential evolution algorithm with integrated adaptive mechanism effectively solves the problem of solving high-dimensional and nonlinear optimization models and can obtain high-quality Pareto optimal solutions in a reasonable time.

[0017] Closed-loop: A closed-loop feedback mechanism of "configuration-execution-monitoring-re-optimization" has been established to realize dynamic management of the entire life cycle of supply chain configuration solutions, ensuring that the solutions can continuously adapt to the ever-changing internal and external environment. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating a distribution network supply chain balancing configuration method based on multi-objective collaborative optimization and dynamic weight allocation in one embodiment of this application. Figure 2 This is a schematic diagram of a dynamic weight allocation mechanism provided in an embodiment of this application; Figure 3 This is a schematic diagram of the functional modules of a power distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation, provided as an embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] This application is divided into three main parts: design of multi-objective collaborative optimization function, implementation of dynamic weight allocation mechanism, and design and solution of equilibrium optimization model. The technical solution of this application will be further described in detail below.

[0023] In one exemplary embodiment, such as Figure 1As shown, a distribution network supply chain balancing configuration method based on multi-objective collaborative optimization and dynamic weight allocation is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method includes the following steps: Step 101: Establish a multi-objective optimization model with economic indicators, reliability indicators, low-carbon indicators, and safety indicators as optimization objectives; the multi-objective optimization model includes: a first objective function, a second objective function, a third objective function, and a fourth objective function; the economic indicators include equipment procurement costs, operation and maintenance costs, and loss costs; the reliability indicators include average user outage time and average system outage frequency; the low-carbon indicator is the total carbon emissions during the operation of the distribution network; the safety indicators include equipment load rate margin and short-circuit current margin.

[0024] The multi-objective collaborative optimization function design includes four objective functions: the first objective function, the second objective function, the third objective function, and the fourth objective function.

[0025] The first objective function is to minimize the cost of power supply reliability, which includes equipment transportation costs, inventory costs, and power outage loss costs. It aims to achieve refined cost control while ensuring power supply reliability. Unlike traditional supply chains that only focus on explicit logistics costs, this objective forces a balance between the "reliability premium" and cost of the distribution network through power outage loss costs.

[0026] ; In the formula, , and These are equipment transportation costs, inventory costs, and power outage loss costs, respectively.

[0027] Equipment transportation costs: ; In the formula, Let i be the transportation distance from the fault point j. In order to allocate the amount of supplies, The cost per unit distance for transportation.

[0028] The above formula must satisfy spatial constraints, which are crucial for ensuring the normal operation of the supply chain transportation link. In the scenario of the distribution network supply chain, power equipment such as transformers and distribution cabinets is usually large in volume and heavy in weight. If the spatial constraints are not followed and materials beyond the transportation capacity are forcibly arranged for transportation, it may cause damage to transportation vehicles and failure of transportation tasks, thereby delaying the delivery of power equipment and affecting the progress of power grid construction or the efficiency of fault repair. At the same time, planning the transportation route in combination with the transportation distance can effectively reduce transportation costs and time costs. For example, preferentially selecting routes with short distances and good road conditions can improve transportation efficiency.

[0029] Spatial constraints: ; In the formula, represents the quantity of allocated materials transported from to the fault point ; is the upper limit of transportation capacity. The determination of is relatively complex and is closely related to factors such as the type of transportation vehicle (such as the load and volume of a truck), road conditions (such as narrow roads restricting the passage of large transportation vehicles), and policy restrictions (such as traffic restrictions in certain sections), and is formulated according to different scenarios.

[0030] Inventory cost: ; In the formula, is the safety inventory of the type of material, is the unit inventory holding cost, is the inventory excess quantity, is the excess storage penalty cost.

[0031] [[ID=,37]]Outage loss cost: [[ID=,38]] [[ID=,39]]; [[ID=,40]] [[ID=,41]]In the formula, [[ID=,42]] [[ID=,43]]is the unit outage loss of user [[ID=,44]] [[ID=,45]], [[ID=,46]] [[ID=,47]]is the user outage state during the [[ID=,48]] [[ID=,49]]period, [[ID=,50]] [[ID=,51]]=1 for outage, [[ID=,52]] [[ID=,53]]=0 for normal, [[ID=,54]] [[ID=,55]]is the load weight of user [[ID=,56]] [[ID=,57]]. [[ID=,58]] [[ID=,59]]

[0032] The second objective function is to maximize the fault resilience ability. Taking the standby power capacity, material redundancy, etc. as indicators, it enhances the anti-interference ability of the distribution network supply chain in multi-level fault scenarios. When cascading faults occur in the distribution network, a scheme with high fault resilience ability can reduce the fault propagation speed and gain a critical time window for emergency repair and configuration. [[ID=,{61}]] [[ID=,{62}]]

[0033] [[ID=,{63}]] [[ID=,{64}]]; [[ID=,{65}]] In the formula, To ensure sufficient redundancy of emergency repair materials, To ensure a timely response to emergency repairs, This refers to the backup power capacity index. , and It is the elastic coefficient.

[0034] Redundancy of emergency repair materials: ; In the formula, For the redundancy of the w-th type of materials, This is the standard demand quantity.

[0035] Emergency repair response time: ; In the formula, Let m be the time it takes for the m-th material to reach the fault point. The maximum allowed time limit.

[0036] Backup power capacity index: ; In the formula, Let n be the backup power capacity of node n. This represents the importance weight of the nodes.

[0037] ; , and Adjustments are made based on the risk level of the power distribution network.

[0038] The third objective function is to minimize the response time. Targeting the "time-sensitive" characteristics of the distribution network, the material pre-configuration strategy is optimized by improving the material preparation time, and the optimal route is planned by combining the transportation time with real-time traffic conditions, thereby achieving a "minute-level" response improvement.

[0039] ; In the formula, To allow time for material preparation, For delivery time.

[0040] Material preparation time: ; In the formula, This refers to the standard time for the outbound shipment of the kth type of materials.

[0041] Delivery time:

[0042] In the formula, Let i be the transportation distance from the fault point j. In order to allocate the amount of supplies, This represents the average transport speed.

[0043] Certain time constraints, or thresholds, need to be met. These time constraints directly affect the service quality and processing capacity of the supply chain. In scenarios with surging demand, such as the significant increase in demand for power equipment during the summer peak electricity consumption period, failure to meet time constraints will result in equipment delivery not being possible, severely impacting business production and residents' lives. Furthermore, time constraints define the boundaries for the response time minimization objective in multi-objective optimization, ensuring that the optimized resource allocation scheme is practically feasible and effective in the time dimension.

[0044] In different scenarios, The thresholds differ, such as those for important power users like hospitals. The threshold needs to be appropriately reduced, especially under extreme weather conditions. The threshold can be appropriately relaxed.

[0045] The fourth objective function is to maximize the regional resource balance. This objective function aims to solve the problem of "regional supply and demand imbalance" in the power distribution network supply chain and avoid the coexistence of local stockpiling and shortages in neighboring areas.

[0046] ; In the formula, The inventory level of materials in region g. This represents the regional average inventory level.

[0047] Inventory variance constraint: ; In the formula, This represents the maximum allowable inventory variance.

[0048] Step 102: Collect real-time operational data of the power distribution network supply chain, including load fluctuation data, equipment health status data, and environmental meteorological data.

[0049] Step 103: Based on the operational data, using a preset dynamic weight calculation rule, calculate the dynamic weight values ​​of the economic indicators, reliability indicators, and environmental impact indicators respectively. The dynamic weight values ​​are used to characterize the priority of each target in the current operational state.

[0050] Specifically, it includes: Based on the different scenarios and changing business needs of the distribution network supply chain, a dynamic weight allocation mechanism is designed (see details). Figure 2 ), and adjust the weights of each objective in the multi-objective optimization function in real time.

[0051] Preset the possible scenario types and their corresponding initial weight vectors, and assign initial weight values ​​to them.

[0052] Scene recognition: An intelligent scene recognition method combining multi-source heterogeneous data fusion and deep learning is adopted.

[0053] Establish a data acquisition framework to obtain multi-dimensional status parameters of supply chain nodes in real time, including node equipment operating load, failure rate, demand fluctuation, inventory level, and transportation delay rate.

[0054] After preprocessing, the collected data is input into a multimodal fusion model based on the Transformer architecture. This model automatically learns the correlation features between different types of data and uncovers hidden scene information. Cross-modal feature fusion and scene information mining are achieved through the following innovative mechanisms.

[0055] Construct a Transformer encoder that includes a spatiotemporal attention mechanism, the structure of which includes: In the multimodal feature embedding layer, numerical data (such as equipment load and failure rate) is processed by "linear mapping + batch" normalization, text data is represented by word vectors extracted through a pre-trained BERT model, image data is extracted by ResNet-50 to extract feature maps, and geospatial data is encoded into two-dimensional tensors through latitude and longitude.

[0056] In the spatiotemporal location coding module, a location coding function incorporating temporal periodicity and spatial topological relationships is designed: ; In the formula, Used to capture the absolute order information of positions in a sequence. For location index, For dimensional indexing, For encoding dimensions. In the power distribution network scenario, In the time dimension, it can correspond to the time of data collection. Through the periodic changes of the sine curve, the model can perceive the sequential relationship of the time series (such as the alternating pattern of load peaks and valleys). In the spatial dimension, it can correspond to the physical location of the node in the distribution network topology. As a time-periodic encoding item, it is specifically designed to capture the periodic characteristics of data acquisition time. In power distribution networks, it can be combined with equipment aging cycles to predict the peak failure rate of equipment within a specific time interval. Adjust the weight of time encoding in total location encoding. The larger the value, the more the model focuses on the impact of time periodicity on the scene (such as the increase in equipment failure rate during typhoon season). This spatial topology coding term, based on graph embedding technology, represents the topological connections between nodes. In a distribution network, it can identify electrical coupling relationships between nodes, predict fault propagation, and, based on the topological distance between nodes, prioritize the allocation of emergency repair materials from nearby warehouses, thus shortening response time. The weights that control spatial coding are larger, and the more the model depends on spatial topological information (such as prioritizing the resource status of geographically neighboring nodes under natural disasters). and The value can be optimized through training with historical data.

[0057] To achieve deep fusion of multi-source heterogeneous data, a three-stage attention fusion architecture is designed, which gradually transitions from single-modal feature mining to cross-modal correlation analysis in a progressive manner.

[0058] Intramodal self-attention architecture: Self-attention is calculated separately for each modal feature, aiming to capture the correlation within a single data source.

[0059] Self-attention computation: First, through a learnable weight matrix... Input features Projected as query vectors respectively Key vector Value vector : ; Calculate the first The position is the first Attention score at each position To measure the strength of the association between the two: ; In the formula, It is the feature dimension.

[0060] The attention score is normalized using the Softmax function to obtain the first... Attention weights of each position to all positions : ; Use attention weights on the value vector Perform a weighted summation to obtain the first... The output after self-attention at each position : ; By integrating the outputs from all positions, the final result of the modality feature after self-attention is obtained, thereby capturing the correlation between features within a single modality.

[0061] ; Cross-modal cross-attention: After completing single-modal feature mining, cross-modal cross-attention is used to calculate the association between different modalities and establish semantic connections across data sources. This mechanism can associate the potential link between increased equipment failure rates and transportation route congestion, and identify the scenario logic where transportation obstruction leads to the inability to deliver emergency repair materials in a timely manner, thereby exacerbating the scope of the failure's impact.

[0062] Feature sequence of mode A: ( (where n is the sequence length of A); Modal B feature sequence: ( (where m is the sequence length of B); Use modality A for "query" and modality B for "key" and "value". Query vector (from A): ( (This is the query weight matrix for mode A). Key vector (from B): ( (This is the bond weight matrix for mode B). Value vector (from B): ( (The weight matrix for mode B). Calculate the cross-modal attention score: This measures the association strength between the i-th feature in modality A and the j-th feature in modality B. ; Normalized attention weights: Convert scores into weights using Softmax. ; In the formula, It is the location index of the mode B feature in cross-modal computation.

[0063] We use the weights to sum the value vector of mode B, and obtain the cross-attention output of mode A based on B.

[0064] Spatiotemporal global attention: Based on the single-modal internal features and cross-modal correlation information obtained in the first two stages, the global attention layer combines temporal and spatial dimension information to construct a spatiotemporal dependency model of resource flow between regions, thereby achieving a comprehensive understanding of the global state of the power distribution network supply chain and optimizing resource allocation schemes.

[0065] Based on the multimodal fusion features obtained from the Transformer architecture, considering the significant time-series characteristics of distribution network data, it is necessary to further extract time-dimensional features to capture the changing patterns of the data over time.

[0066] A time-series convolutional network is introduced to extract features from time-series data and capture the changing trends of state parameters over time.

[0067] A temporal feature extraction network integrating multi-scale dilated convolution and gating mechanisms is constructed, as follows: Constructing multi-scale dilated convolutional layers: A parallel dilated convolutional structure is adopted, with four convolutional branches set with dilation rates of {1,2,4,8} to capture features at different time granularities. Feature extraction is performed on time-series data from multiple time scales, enriching the temporal dimension feature expression of the data.

[0068] Constructing gated recurrent units: Introducing a gating mechanism to dynamically weight and fuse multi-scale convolutional features, avoiding simple information superposition.

[0069] Construct a temporal attention enhancement module: further explore the dependencies of data in the time dimension, highlight the time points that are of great value to scene analysis, so that the model can more accurately capture key change patterns in time series data, thereby improving the accuracy and reliability of scene recognition.

[0070] After extracting data features, these features need to be transformed into specific scenario recognition results so that weights can be adjusted according to the scenario. Then, a Softmax classifier is used to output the probability distribution of the current scenario of the supply chain, thereby accurately identifying scenario types such as routine operation, fault propagation, demand surge, and impact of natural disasters.

[0071] The original multimodal fusion features suffer from high dimensionality, information redundancy, and nonlinear distribution. Directly inputting them into a classifier increases computational burden and reduces accuracy. First, Principal Component Analysis (PCA) is used to process the data. Its core principle is to project the original features onto the direction of maximum variance by eigenvalue decomposition of the covariance matrix. Building upon PCA dimensionality reduction, Locally Linear Embedding (LLE) algorithm further mines the nonlinear structure of the data. These two steps improve the quality of the features, making them more suitable for subsequent classification decisions and enhancing the accuracy and efficiency of scene recognition.

[0072] When faced with complex and ever-changing power distribution network supply chain scenarios, the generalization ability and classification accuracy of a single classifier still have limitations. To address this issue, this application integrates three base classifiers: Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting Tree (XGBoost), and introduces a dynamic weighted voting strategy, which is a key step in making classification decisions based on optimized data.

[0073] The classification results obtained by the multi-classifier integration module only preliminarily determine the scene category. However, the traditional Softmax function cannot adjust the output scene probability according to the characteristics of different scenes, thus failing to meet the requirements of accurate recognition. To solve this problem, this application introduces a temperature parameter adjustment mechanism and designs a dynamic adjustment strategy to optimize the probability of the classification results, thereby achieving accurate scene recognition.

[0074] The improved Softmax formula: ; In the formula, This indicates that the sample belongs to the scene category. The probability is the final output result, and its value ranges from 0 to 1. In the distribution network supply chain scenario identification, it represents the likelihood that the current data belongs to a specific scenario (such as routine operation, fault propagation, etc.), and is a key basis for determining the scenario type. To enable the ensemble classifier to determine which scene category a sample belongs to The output score reflects the model's tendency to predict whether a sample belongs to a certain category based on the input features; the higher the score, the greater the likelihood that the model believes the sample belongs to that category. The temperature coefficient directly affects the shape of the probability distribution. For scene categories.

[0075] In normal operating scenarios, data fluctuations are relatively small, at which point larger... A value that allows the model to output a more uniform probability distribution avoids misjudgments caused by slight data variations. When When the value is small, the effect of the exponential term strengthens, the probability distribution becomes sharper, and the model's confidence in the prediction results increases. In failure scenarios, such as sudden equipment failure or supply chain disruptions, a smaller exponential value... Values ​​can make the model more focused on key features, highlight the probability of real-world scenarios, thereby accurately identifying abnormal situations and improving the sensitivity and accuracy of fault scenarios.

[0076] Data uncertainty is measured by calculating the entropy value of the current scene features. When the entropy value is high, indicating that the data distribution is relatively scattered and the uncertainty is large, it should be appropriately increased. The value makes the model output more robust; when the entropy value is low, the data distribution is concentrated, and the uncertainty is small, reducing the entropy value makes the model output more robust. This value enhances the model's ability to distinguish between different scenarios.

[0077] Dynamic weight adjustment: Based on real-time collected business data and changes in the scenario, an adaptive weight adjustment algorithm, such as weight adjustment based on fuzzy logic or weight optimization based on reinforcement learning, is used to dynamically update the weights of each objective.

[0078] Step 104: Based on the multi-objective optimization model and dynamic weight values, establish an equilibrium optimization model that considers spatiotemporal dynamic constraints.

[0079] By combining multi-objective collaborative optimization functions, the temporal evolution characteristics of business volume at supply chain nodes, and spatiotemporal dynamic constraints, an equilibrium optimization model considering spatiotemporal dynamic constraints is established.

[0080] Its mathematical expression is: ; In the formula, The weights of the objective function, , , .

[0081] Step 105: Solve the equilibrium optimization model considering spatiotemporal dynamic constraints to obtain the optimal distribution network supply chain equilibrium configuration scheme.

[0082] Intelligent optimization algorithms, such as genetic algorithms, particle swarm optimization algorithms, and simulated annealing algorithms, are used to solve the elastic equilibrium optimization model to obtain the optimal resource allocation scheme for the distribution network supply chain.

[0083] Step 106: The optimal distribution network supply chain balancing configuration scheme is used as the final distribution network supply chain configuration decision and applied to the planning, construction, or operation and scheduling of the distribution network.

[0084] Based on the same inventive concept, this application also provides a distribution network supply chain balancing configuration device for implementing the above-mentioned distribution network supply chain balancing configuration method based on multi-objective collaborative optimization and dynamic weight allocation. The solution provided by this device is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more embodiments of the distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation provided below can be found in the limitations of the distribution network supply chain balancing configuration method based on multi-objective collaborative optimization and dynamic weight allocation described above, and will not be repeated here.

[0085] In one exemplary embodiment, such as Figure 3 As shown, a power distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation is provided, comprising: The multi-objective optimization model construction module 201 is used to establish a multi-objective optimization model with economic indicators, reliability indicators, low-carbon indicators, and safety indicators as optimization objectives. The multi-objective optimization model includes a first objective function, a second objective function, a third objective function, and a fourth objective function. The economic indicators include equipment procurement costs, operation and maintenance costs, and loss costs. The reliability indicators include average user outage time and average system outage frequency. The low-carbon indicator is the total carbon emissions during the operation of the distribution network. The safety indicators include equipment load rate margin and short-circuit current margin. The data acquisition module 202 is used to collect real-time operational data of the power distribution network supply chain, including load fluctuation data, equipment health status data, and environmental meteorological data. The dynamic weight value calculation module 203 is used to calculate the dynamic weight values ​​of economic indicators, reliability indicators and environmental impact indicators respectively according to the operation data and using preset dynamic weight calculation rules. The dynamic weight values ​​are used to characterize the priority of each target in the current operation state. The equilibrium optimization model construction module 204 is used to establish an equilibrium optimization model that considers spatiotemporal dynamic constraints based on the multi-objective optimization model and dynamic weight values. The solution module 205 is used to solve the equilibrium optimization model considering spatiotemporal dynamic constraints to obtain the optimal distribution network supply chain equilibrium configuration scheme. The planning and scheduling module 206 is used to take the optimal distribution network supply chain balance configuration scheme as the final distribution network supply chain configuration decision and apply it to the planning, construction or operation scheduling of the distribution network.

[0086] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0087] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation, characterized in that, The distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation includes: A multi-objective optimization model is established with economic indicators, reliability indicators, low-carbon indicators, and safety indicators as optimization objectives. The multi-objective optimization model includes a first objective function, a second objective function, a third objective function, and a fourth objective function. The economic indicators include equipment procurement cost, operation and maintenance cost, and loss cost. The reliability indicators include average user outage time and average system outage frequency. The low-carbon indicator is the total carbon emissions during the operation of the distribution network. The safety indicators include equipment load rate margin and short-circuit current margin. Real-time collection of operational data from the power distribution network supply chain, including load fluctuation data, equipment health status data, and environmental meteorological data; Based on the operational data, the dynamic weight values ​​of economic indicators, reliability indicators and environmental impact indicators are calculated using a preset dynamic weight calculation rule. The dynamic weight values ​​are used to characterize the priority of each target in the current operational state. Based on the aforementioned multi-objective optimization model and dynamic weight values, an equilibrium optimization model considering spatiotemporal dynamic constraints is established. The optimal distribution network supply chain balance configuration scheme is obtained by solving the equilibrium optimization model that considers spatiotemporal dynamic constraints.

2. The distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation according to claim 1, characterized in that, The expression for the first objective function is as follows: ； In the formula, , and These are equipment transportation costs, inventory costs, and power outage loss costs; The expression for equipment transportation costs is as follows: ; In the formula, For equipment To the fault point The transportation distance In order to allocate the amount of supplies, For unit distance transportation cost, For the number of devices, This represents the number of fault points. The constraints on equipment transportation costs are as follows: ; In the formula, This represents the upper limit of transportation capacity. The expression for inventory cost is as follows: ; In the formula, For the first Safety stock levels of such materials Unit inventory holding cost For excess inventory, Excessive storage incurs penalty costs. Category of inventory materials; The expression for the cost of power outage losses is as follows: ; In the formula, For users Unit power outage losses, for User power outage status during the specified time period =1 indicates a power outage. =0 is normal For users Load weight, For the number of time periods, For the number of users.

3. The distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation according to claim 1, characterized in that, The expression for the second objective function is as follows: ； In the formula, To ensure sufficient redundancy of emergency repair materials, To ensure a timely response to emergency repairs, This refers to the backup power capacity index. , and All are elastic coefficients; The expression for the redundancy of emergency repair materials is as follows: ； In the formula, For the first Redundancy of similar materials For standard demand, The total number of different types of materials needed for emergency repairs; The expression for the emergency repair response time is as follows: ； In the formula, For the first The time it takes for the supplies to reach the fault location To allow the maximum time limit, The total number of batches of emergency repair materials; The expression for the backup power capacity index is as follows: ； In the formula, For nodes Backup power capacity, As the importance weight of the nodes, The total number of nodes; in, .

4. The distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation according to claim 1, characterized in that, The expression for the third objective function is as follows: ； In the formula, To allow time for material preparation, For transportation time; The expression for the material preparation time is as follows: ； In the formula, The standard time for the release of goods of category k; The expression for transit time is as follows: ； In the formula, Let i be the transportation distance from the fault point j. In order to allocate the amount of supplies, This represents the average transport speed.

5. The distribution network supply chain balancing configuration method based on multi-objective collaborative optimization and dynamic weight allocation according to claim 1, characterized in that, The expression for the fourth objective function is as follows: ； In the formula, For the region The level of material inventory, This represents the regional average inventory level. Total number of regions; The expression for the inventory variance constraint is as follows: ； In the formula, This represents the maximum allowable inventory variance.

6. The distribution network supply chain balancing configuration method based on multi-objective collaborative optimization and dynamic weight allocation according to claim 1, characterized in that, The specific rules for calculating the dynamic weights are as follows: When load fluctuation data exceeds a preset threshold or equipment health status data falls below a preset threshold, the dynamic weight value of reliability indicators is increased, while the dynamic weight value of economic indicators is decreased. When environmental meteorological data indicate that the region is in a high carbon emission sensitive period, the dynamic weight value of environmental impact indicators should be increased.

7. The distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation according to claim 1, characterized in that, The expression for the equilibrium optimization model considering spatiotemporal dynamic constraints is as follows: ； In the formula, The weights for each objective function, , , , Each objective function represents a different objective function.

8. The distribution network supply chain balancing configuration method based on multi-objective collaborative optimization and dynamic weight allocation according to claim 1, characterized in that, The distribution network supply chain equilibrium configuration method based on multi-objective collaborative optimization and dynamic weight allocation further includes the following after the last step: The optimal distribution network supply chain balancing configuration scheme is used as the final distribution network supply chain configuration decision and applied to the planning, construction, or operation and scheduling of the distribution network.

9. A power distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation, characterized in that, The distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation includes: A multi-objective optimization model construction module is used to establish a multi-objective optimization model with economic indicators, reliability indicators, low-carbon indicators, and safety indicators as optimization objectives. The multi-objective optimization model includes a first objective function, a second objective function, a third objective function, and a fourth objective function. The economic indicators include equipment procurement costs, operation and maintenance costs, and loss costs. The reliability indicators include average user outage time and average system outage frequency. The low-carbon indicator is the total carbon emissions during the operation of the distribution network. The safety indicators include equipment load rate margin and short-circuit current margin. The data acquisition module is used to collect real-time operational data of the power distribution network supply chain, including load fluctuation data, equipment health status data, and environmental meteorological data. The dynamic weight value calculation module is used to calculate the dynamic weight values ​​of economic indicators, reliability indicators and environmental impact indicators respectively according to the operation data and using preset dynamic weight calculation rules. The dynamic weight values ​​are used to characterize the priority of each target in the current operation state. The equilibrium optimization model construction module is used to establish an equilibrium optimization model that considers spatiotemporal dynamic constraints based on the multi-objective optimization model and dynamic weight values. The solution module is used to solve the equilibrium optimization model that considers spatiotemporal dynamic constraints to obtain the optimal distribution network supply chain equilibrium configuration scheme.

10. The distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation according to claim 9, characterized in that, The distribution network supply chain balancing configuration device based on multi-objective collaborative optimization and dynamic weight allocation also includes: The planning and scheduling module is used to take the optimal distribution network supply chain balance configuration scheme as the final distribution network supply chain configuration decision and apply it to the planning, construction or operation scheduling of the distribution network.

Images