Electric power natural disaster emergency resource allocation method and system based on multi-data fusion

Abstract

The invention relates to the technical field of electric power, and discloses an electric power natural disaster emergency resource allocation method and system based on multi-data fusion, and the key points of the technical scheme are as follows: multi-source data acquisition and fusion, resource demand prediction, dynamic allocation scheme generation, and real-time monitoring and adjustment. By constructing a multi-source data fusion system, establishing a high-precision demand prediction model and dynamically generating an optimal allocation scheme in combination with traffic accessibility, accurate and efficient overall planning of emergency resources is realized, the resource utilization rate is improved, and the power recovery time after disasters is shortened.

Description

Technical Field

[0001] This invention relates to the field of power technology, and more specifically, to a method and system for emergency resource allocation in power natural disasters based on multi-data fusion. Background Technology

[0002] Existing emergency resource allocation schemes generally suffer from the following significant drawbacks: First, multi-source data is fragmented and scattered, with data on weather warnings, power grid damage, resource inventory, and traffic conditions not forming effective linkages, easily leading to "data silos" and distorted supply and demand assessments. Second, resource demand forecasting lacks scientific model support and relies heavily on empirical estimations, making it difficult to accurately quantify the shortage of repair teams, equipment, and materials in disaster-stricken areas. Third, allocation schemes are static and do not dynamically adjust in real time based on traffic accessibility, easily leading to inefficiency due to road blockages or resource misallocation. Fourth, resource utilization is insufficient, with surplus areas experiencing idle resources and shortage areas experiencing resource shortages, further prolonging the power restoration cycle.

[0003] While existing research has made some progress in data collection and demand forecasting, it still has limitations: some studies focus on a single data dimension (such as using only power grid data) and have not achieved multi-source data fusion; some studies use traditional algorithms to predict demand, and the accuracy is difficult to meet the requirements of complex disaster scenarios; and some studies ignore the dynamic coupling relationship between traffic and resource allocation, resulting in poor practicality of the solutions.

[0004] Therefore, there is an urgent need for an emergency resource management method and system that integrates multi-data fusion, scientific prediction, and dynamic allocation to solve the problems of data fragmentation, inaccurate prediction, and inefficient allocation in existing schemes. Summary of the Invention

[0005] This invention aims to address the shortcomings of existing technologies by providing a method and system for emergency resource allocation in power natural disasters based on multi-data fusion. By constructing a multi-source data fusion system, establishing a high-precision demand prediction model, and dynamically generating the optimal allocation scheme in conjunction with traffic accessibility, this invention achieves precise and efficient coordination of emergency resources, improves resource utilization, and shortens the power restoration time after a disaster.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: a method and system for emergency resource allocation in power natural disasters based on multi-data fusion, comprising the following steps:

[0007] S1. Multi-source data collection and fusion: Collect data on weather warnings, power grid damage, resource inventory, and traffic conditions, and then clean, standardize, and integrate the data to form a unified data mart;

[0008] S2. Resource Demand Forecasting: Based on fused data, a BP neural network + disaster impact coefficient coupled model is used to quantify resource demand and prioritize demand areas.

[0009] S3. Dynamic dispatching scheme generation: Construct a MILP model and generate the optimal dispatching scheme by combining traffic accessibility;

[0010] S4. Real-time monitoring and adjustment: Track resource transportation, power grid damage, and traffic changes, and dynamically update the allocation plan.

[0011] As a preferred embodiment of the present invention, the line fault probability mapping formula in step 1 is:

[0012]

[0013] Where α and β are fitting parameters, v norm This is the standardized value for disaster intensity.

[0014] As a preferred embodiment of the present invention, the activation function of the hidden layer of the BP neural network in step 2 is:

[0015]

[0016] The weight update uses gradient descent, and the iterative formula is as follows:

[0017]

[0018] Where η is the learning rate and E is the prediction error.

[0019] As a preferred embodiment of the present invention, the emergency power generation vehicle demand correction formula in step 2 is:

[0020]

[0021] in, The BP output is the basic requirement, γ is the disaster correction factor, and P loss P represents the load loss power. gen,rate This refers to the rated power of the generator car.

[0022] As a preferred embodiment of the present invention, the priority formula for the demand area in step 2 is:

[0023]

[0024] Among them, P imp,r P ind,r P nor,r These represent the residential, industrial, and general load power (P) of region r. imp,total ω represents the corresponding total power. imp =100, ω ind =50, ω nor =10 represents the load weight.

[0025] As a preferred embodiment of the present invention, the formula for road travel time in step 3 is:

[0026]

[0027] Among them, L S v is the length of the road. s For real-time speed, k s This represents the congestion coefficient.

[0028] As a preferred embodiment of the present invention, the objective function of MILP in step 3 is:

[0029]

[0030] Where ω1 = 0.3, ω2 = 0.4, and ω3 = 0.3 are weighting coefficients; N res,surplus,ware N represents the surplus of warehouse resources. res,total,ware C represents the total resources of the warehouse. ware→r T represents the unit transportation cost from the warehouse to region r; avg This represents the average transportation time for resources.

[0031] The power natural disaster emergency resource allocation system based on multi-data fusion includes: a multi-source data acquisition and fusion module, a resource demand prediction module, a dynamic allocation scheme generation module, a real-time monitoring and adjustment module, and a data interaction module. Each module works together to achieve the above functions.

[0032] As a preferred technical solution of the present invention, an edge-cloud collaborative architecture is adopted, in which the edge layer collects real-time data, the cloud layer runs an optimization model, and the terminal layer supports visual interaction.

[0033] As a preferred technical solution of the present invention, by adjusting the disaster correction coefficient γ and traffic parameters to adapt to different disaster scenarios such as typhoons, earthquakes, and rainstorms, there is no need to reconstruct the model.

[0034] In summary, this invention, by constructing a comprehensive system encompassing "multi-source data fusion - scientific demand forecasting - dynamic allocation - real-time adjustment," effectively addresses the core issues of data fragmentation, inaccurate demand, low efficiency, and poor adaptability in the allocation of emergency resources for power natural disasters. It possesses significant advantages across multiple dimensions.

[0035] First, it achieves deep integration of multi-source data such as weather warnings, power grid damage, resource inventory, and traffic conditions. By cleaning data to supplement missing values, standardizing to eliminate differences in magnitude, and constructing a data linkage matrix through correlation and fusion, it breaks down "data silos," significantly improves the accuracy of supply and demand judgment, and avoids the idleness of surplus areas and the shortage of shortage areas caused by resource misallocation.

[0036] Secondly, it innovatively adopts a "BP neural network + disaster impact coefficient" coupled model, which combines fused data to accurately quantify resource demand, introduces differentiated disaster correction coefficients to adapt to different disaster types, reduces demand forecasting errors, and reduces the cost of ineffective resource reserves;

[0037] Third, a dynamic allocation model is constructed with the goal of maximizing resource utilization and minimizing transportation costs. It incorporates traffic accessibility constraints in real time, prioritizes efficient routes, and updates the plan every 30 minutes in case of abnormal situations such as road blockages or new faults, thereby shortening the average transportation time of resources.

[0038] Fourth, through configurable disaster correction coefficients, traffic parameters, and standardized edge-cloud architecture, it can adapt to various disasters such as typhoons, earthquakes, and rainstorms, as well as the characteristics of different regions, without the need to reconstruct the model, thereby reducing implementation costs;

[0039] Fifth, the technological advantages of each link in the whole process are optimized through collaborative efforts: satellite data ensures information sharing, demand forecasting ensures accurate resource allocation, dynamic allocation improves transportation efficiency, and real-time adjustments can be made to deal with emergencies, thus shortening the repair time of power grid fault lines. Attached Figure Description

[0040] Figure 1 A flowchart of a power natural disaster emergency resource allocation method based on multi-data fusion provided in an embodiment of the present invention. Detailed Implementation

[0041] The present application will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any way. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application. These all fall within the protection scope of the present application.

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0043] Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items.

[0044] Furthermore, the technical features involved in the various embodiments of this application described below can be combined with each other as long as they do not conflict with each other.

[0045] This disclosure aims to address the problems of fragmented and scattered multi-source data (weather warnings, power grid damage, resource inventory, traffic conditions) in the emergency resource allocation for power natural disasters, resulting in low demand forecast accuracy and insufficient resource utilization due to static allocation schemes. Therefore, this disclosure proposes a method and system for emergency resource allocation for power natural disasters based on multi-data fusion, to achieve precise coordination and dynamic scheduling of power emergency resources and ensure rapid power grid restoration after disasters. This method integrates satellite communication data acquisition technology with a BP neural network prediction model to construct a multi-dimensional data fusion system. It combines configurable disaster correction coefficients (e.g., typhoon γ=1.2, earthquake γ=1.5) and differentiated traffic accessibility constraints (road carrying capacity, congestion coefficient) to adapt to the resource allocation needs of different disaster types and regions, thereby breaking down data silos, improving demand forecast accuracy, and reducing system implementation complexity and emergency operation costs.

[0046] Please refer to Figure 1 , Figure 1 A flowchart of the power natural disaster emergency resource allocation method based on multi-data fusion, as described in an embodiment of this disclosure, is shown. The overall process mainly includes the following four steps:

[0047] Step 1: Multi-source data acquisition and fusion.

[0048] Step 1.1: Multi-source data acquisition. Four types of core data are collected through edge nodes (deployed in substations, traffic monitoring stations, and resource warehouses), with a time resolution of 5 minutes to ensure real-time performance.

[0049] Meteorological warning data: disaster type (typhoon / earthquake / rainstorm), intensity parameters (typhoon wind speed v, earthquake magnitude M, rainstorm precipitation R), and impact range (latitude and longitude boundary (x1,y1)-(x2,y2)). The data are sourced from meteorological API and satellite remote sensing.

[0050] Damaged power grid data: Number of faulty lines N fault Location of the faulty line (x) fault ,y fault ), Number of substations out of service N sub,out Load loss rate δ load (δ load =P loss / P normal P loss For power loss, P normal (Normal power); based on fault waveform data and UAV inspection data, the fault location accuracy is ≤ ±50m.

[0051] Emergency resource data: Resource warehouse location (x ware ,y ware ), Inventory Quantity (Repair Team N) crew,ware Emergency power generation vehicle N gen,ware Cable length L cable,ware Resource type parameters (generator vehicle rated power P) gen,rate Cable Type cable ).

[0052] Traffic data: Road number s, traffic status λ s (1 indicates passage, 0 indicates obstruction), real-time speed v s Congestion coefficient k s (k s =v design / v s ,v design (Design speed), road carrying capacity C s (Unit: vehicles / hour), data sourced from traffic monitoring equipment and navigation platforms.

[0053] Step 1.2: Multi-data fusion processing

[0054] Data cleaning: Remove outliers (such as negative inventory or wind speeds exceeding reasonable limits), and use interpolation to supplement missing data (e.g., when traffic data for a certain area is missing, it is generated based on data from adjacent areas). The formula is:

[0055]

[0056] Where, d s-1 d s This represents the distance between the adjacent road and the target road.

[0057] Data standardization: Converting data of different magnitudes into a unified standard (value range [0,1]), such as wind speed standardization:

[0058]

[0059] Among them, v min v max These represent the minimum and maximum wind speeds for this type of disaster (e.g., v in a typhoon scenario). min =10m / s, v max =40m / s).

[0060] Data correlation and fusion: Constructing a correlation matrix of "disaster-power grid-resources-transportation" to quantify the coupling relationships between data, such as the mapping between typhoon wind speed and line fault probability.

[0061]

[0062] Wherein, α=2.3 and β=1.8 are fitting parameters, which are obtained through training with historical disaster data.

[0063] Step 2: Resource demand forecasting based on fused data.

[0064] Specifically, a coupled model of "BP neural network + disaster impact coefficient" is used to predict the resource demand of the disaster-stricken area:

[0065] Step 2.1: BP Neural Network Modeling

[0066] Network structure: The input layer consists of 5 fused data features (standardized disaster intensity value I). norm Fault line density ρ fault Load loss rate δ load Area S area Number of users N user The hidden layer has two layers, each with 12 neurons; the output layer has three types of resource requirements (repair team N). crew,demand Emergency power generation vehicle N gen,demand Cable length L cable,demand ).

[0067] Activation function and weight update: The hidden layer uses the Sigmoid function.

[0068]

[0069] The weight update uses gradient descent, and the iterative formula is as follows:

[0070]

[0071] Where η = 0.01 is the learning rate, and E is the prediction error.

[0072] The formula for mean square error is:

[0073]

[0074] Among them, y i This is the actual value. These are predicted values.

[0075] Step 2.2: Disaster Impact Coefficient Correction. Introducing disaster type correction coefficients γ (typhoon γ = 1.2, earthquake γ = 1.5, rainstorm γ = 1.1) corrects the basic requirements of the BP neural network output, yielding the final requirements:

[0076]

[0077] in, L cableAs a fundamental requirement for the output of a BP neural network, δ corr =0.2 is the cable loss correction factor.

[0078] Step 2.3 Prioritizing Demands Based on Load Importance (Residential Load Weight ω) imp =100, Industrial load ω ind =50, Normal load ω nor =10), sort the demand areas using the following formula:

[0079]

[0080] Among them, P imp,r P ind,r P nor,r These represent the residential, industrial, and general load power (P) of region r. imp,toal Equal to the corresponding total power.

[0081] Step 3: Generate a dynamic allocation plan based on traffic accessibility.

[0082] With the goal of "maximizing resource utilization and minimizing total allocation cost", a mixed-integer linear programming (MILP) model is constructed to generate dynamic allocation solutions:

[0083] Objective function:

[0084]

[0085] Where ω1 = 0.3, ω2 = 0.4, and ω3 = 0.3 are weighting coefficients; N res,surplus,ware N represents the surplus of warehouse resources. res,total,ware C represents the total resources of the warehouse. ware→r T represents the unit transportation cost from the warehouse to region r; avg This represents the average transportation time for resources.

[0086] Constraints:

[0087] Resource supply and demand balance constraint: All regional demand is met by surplus warehouse resources.

[0088] Σ ware N res,ware→r =N res,demand,r

[0089] Where, N res,ware→r N represents the amount of resources transferred from the warehouse to region r. res,demand,r For the region's resource needs.

[0090] Warehouse resource constraints: The allocation quantity cannot exceed the warehouse surplus: N res,ware→r ≤N res,surplus,ware ;

[0091] Traffic capacity constraint: The number of transport vehicles on each road shall not exceed the road's carrying capacity.

[0092] Σ ware→r N veh,ware→r,s ≤C s

[0093] Where, N veh,ware→r,s C represents the number of vehicles transported via road s. s Let s be the carrying capacity of the road.

[0094] Transportation time constraint: The time for transporting resources to region r shall not exceed the maximum allowable delay T. max (4 hours):

[0095]

[0096] Among them, Ω path The set of roads included in the transportation route. The travel time of road s (L) S (Road length).

[0097] Deployment plan output: Clearly define the resource allocation amount (e.g., deploy 3 emergency repair teams and 2 emergency power generators from warehouse A to area B), and the transportation route (prioritizing t). s Minimum distances to the road, handover points (such as temporary supply depots at regional boundaries), and estimated delivery times.

[0098] Step 4: Real-time monitoring and dynamic adjustment.

[0099] Step 4.1 Real-time data monitoring: Three types of data are collected in real time using GPS and smart sensors:

[0100] Resource transportation status: Location of transport vehicle (x veh ,y veh ), Estimated Time of Arrival (ETA), Resource Integrity δ intact (e.g., cable breakage rate);

[0101] Power grid damage update: New faulty line ΔN fault Repair completed for line ΔN repair ;

[0102] Traffic condition changes: Road traffic status updates (e.g., previously open roads become blocked), congestion coefficient updates Δk s .

[0103] Step 4.2 Dynamically adjust the scheme when an abnormal condition (ETA>T) is triggered. max ΔN fault ≥3、Δλ sWhen =1), rolling time-domain optimization is adopted (the solution is updated every 30 minutes), and the MILP model is solved again. The adjustments include:

[0104] Path replanning: An improved Dijkstra's algorithm is used to reselect the path with the shortest travel time. The formula is:

[0105]

[0106] Resource reallocation: Resources originally allocated to the repaired area are moved to the newly added faulty area. The formula is as follows:

[0107] N res,ware-→r,neuw =N res,ware→r,old +ΔN res,demand,r

[0108] Where, ΔN res,demand,r This represents an increase in resource requirements for newly added faulty areas.

[0109] This application discloses a power natural disaster emergency resource allocation system based on multi-data fusion. The system adopts an "edge-cloud" collaborative architecture and mainly includes: a multi-source data acquisition and fusion module, a resource demand prediction module, a dynamic allocation scheme generation module, a real-time monitoring and adjustment module, and a data interaction module. The functions of each module are as follows:

[0110] Multi-source data acquisition and fusion module: Collects data on weather warnings, power grid damage, resource inventory, and traffic conditions. Through data cleaning, standardization, and correlation fusion, it forms a unified data mart, providing a foundation for subsequent analysis.

[0111] Resource demand forecasting module: Based on fused data, it adopts a BP neural network coupled with a disaster impact coefficient model to quantify the resource gap (quantity and type of repair teams, equipment, and materials) in the disaster-stricken area.

[0112] Dynamic allocation scheme generation module: With the goal of "maximizing resource utilization and minimizing power restoration time", it generates allocation schemes (allocation amount, route, handover nodes) by combining traffic accessibility and resource priority.

[0113] Real-time monitoring and adjustment module: Tracks resource transportation status, power grid damage changes and traffic conditions in real time, and dynamically adjusts the plan in response to abnormal events (such as road congestion, new faults).

[0114] Data interaction module: Enables data exchange with meteorological departments, power grid dispatch centers, traffic management departments, and resource warehouses, supporting cross-departmental collaborative decision-making.

[0115] This system is deployed based on an edge-cloud collaborative architecture:

[0116] Edge layer: Deployed in substations, traffic monitoring stations, and resource warehouses, responsible for real-time collection of multi-source data (such as power grid faults and road speeds), performing local monitoring (such as resource inventory checks), and responding to rapid adjustment commands (such as temporary route changes).

[0117] Cloud layer: Deployed at the provincial power emergency command center, it runs multi-data fusion algorithms, BP neural network demand prediction models, and MILP dynamic allocation models, stores historical data (recent disaster cases, resource scheduling records), and generates global allocation plans.

[0118] Terminal layer: including handheld terminals for emergency repair personnel and large monitoring screens in the command center, supporting resource trajectory viewing, demand prediction result display, and dispatch instructions issuance, with an interface visualization degree of ≥90%, ensuring cross-departmental collaboration efficiency.

[0119] Example: Taking a typhoon disaster in South China as an example, the affected area includes 3 prefecture-level cities (regions 1, 2, and 3), 20 power grid lines were damaged, 3 substations were shut down, and the load loss was 15MW; there are 2 existing resource warehouses (warehouses A and B), 8 emergency repair teams, 5 emergency power generation vehicles, and 50km of cables; in the transportation network, road S105 (warehouse A to region 2) was congested due to the typhoon (ks = 2.5).

[0120] Specific implementation steps:

[0121] Step 1: Multi-data fusion.

[0122] Data collected: Typhoon wind speed v = 30 m / s (standardized v) norm =0.67), fault line density ρ fault = 0.05 lines / km 2 Load loss rate δ load =0.3, travel time t of road S105 S105 = 2.5 hours;

[0123] Data fusion: Calculating the line fault probability P using an association model. line =1-exp(-2.3×0.671.8)=0.72, providing a basis for demand forecasting.

[0124] Step 2: Resource Demand Forecasting

[0125] Basic requirements for BP neural network output: L cable =40km;

[0126] Corrected requirements: Typhoon correction factor γ = 1.2, therefore:

[0127] N crew,demand=6 × 1.2 × 0.05 = 0.3 (Region 1) + ... = 6 branches

[0128] N gen,demand = 4 × 1.2 × (15 / 500) = 0.144 (Region 1) + ... = 4 units

[0129] L cable,demand =40 × 1.2 × 1.2 = 57.6 km;

[0130] Priority ranking: Region 2 (including 2 hospitals) priority = 100×(5 / 15)+50×(3 / 8)+10×(2 / 12)=48.3, which is higher than Region 1 (35.2) and Region 3 (30.5).

[0131] Step 3: Dynamic Allocation Scheme Generation

[0132] Objective function solution: Warehouse A has 5 surplus teams, 3 generator trucks, and 30km of cable; Warehouse B has 3 surplus teams, 2 generator trucks, and 20km of cable. The optimized solution is: Warehouse A allocates 3 teams, 2 generator trucks, and 25km of cable to area 2 (path S207, t = 1.8 hours), Warehouse B allocates 2 teams, 1 generator truck, and 20km of cable to area 1, and has 1 team and 1 generator truck remaining as reserves.

[0133] Solution output: 10 transport vehicles (6 for warehouse A and 4 for warehouse B), with the handover point at the boundary power supply station of area 1 and 2, and an estimated delivery time of 1.8-2.2 hours.

[0134] Step 4: Dynamic Adjustment

[0135] Real-time monitoring: Trees suddenly fell on road S207. S207 =0 (blocked), ETA increased to 3.5 hours;

[0136] Adjustment plan: The route is replanned to S308 (t = 2.1 hours), vehicles are rerouted, and area 2 is notified to activate the backup emergency power generation vehicle to ensure uninterrupted power supply.

[0137] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for emergency resource allocation in power natural disasters based on multi-data fusion, characterized in that, The method includes the following steps: S1. Multi-source data collection and fusion: Collect data on weather warnings, power grid damage, resource inventory, and traffic conditions, and then clean, standardize, and integrate the data to form a unified data mart; S2. Resource Demand Forecasting: Based on fused data, a BP neural network + disaster impact coefficient coupled model is used to quantify resource demand and prioritize demand areas. S3. Dynamic dispatching scheme generation: Construct a MILP model and generate the optimal dispatching scheme by combining traffic accessibility; S4. Real-time monitoring and adjustment: Track resource transportation, power grid damage, and traffic changes, and dynamically update the allocation plan.

2. The method for emergency resource allocation in power natural disasters based on multi-data fusion according to claim 1, characterized in that, The line fault probability mapping formula in step 1 is: Where α and β are fitting parameters, v norm This is the standardized value for disaster intensity.

3. The method for emergency resource allocation in power natural disasters based on multi-data fusion according to claim 1, characterized in that, The activation function of the hidden layer in the BP neural network in step 2 is: The weight update uses gradient descent, and the iterative formula is as follows: Where η is the learning rate and E is the prediction error.

4. The method for emergency resource allocation in power natural disasters based on multi-data fusion according to claim 1, characterized in that, The formula for adjusting the demand of emergency power generation vehicles in step 2 is: in, The BP output is the basic requirement, γ is the disaster correction factor, and P loss P represents the load loss power. gen,rate This refers to the rated power of the generator car.

5. The method for emergency resource allocation in power natural disasters based on multi-data fusion according to claim 1, characterized in that, The formula for prioritizing demand areas in step 2 is: Among them, P imp,r P ind,r P nor,r These represent the residential, industrial, and general load power (P) of region r. imp,total ω represents the corresponding total power. imp =100, ω ind =50, ω nor =10 represents the load weight.

6. The method for emergency resource allocation in power natural disasters based on multi-data fusion according to claim 1, characterized in that, The formula for road travel time in step 3 is: Among them, L S v is the length of the road. s For real-time speed, k s This represents the congestion coefficient.

7. The method for emergency resource allocation in power natural disasters based on multi-data fusion according to claim 1, characterized in that, The objective function of MILP in step 3 is: Where ω1 = 0.3, ω2 = 0.4, and ω3 = 0.3 are weighting coefficients; N res,surplus,ware N represents the surplus of warehouse resources. res,total,ware C represents the total resources of the warehouse. ware→r T represents the unit transportation cost from the warehouse to region r; avg This represents the average transportation time for resources.

8. A power natural disaster emergency resource allocation system based on multi-data fusion, characterized in that, It includes a multi-source data acquisition and fusion module, a resource demand prediction module, a dynamic allocation scheme generation module, a real-time monitoring and adjustment module, and a data interaction module. Each module works together to realize the function of the method described in any one of claims 1-7.

9. The power natural disaster emergency resource allocation system based on multi-data fusion according to claim 8, characterized in that, It adopts an edge-cloud collaborative architecture, with the edge layer collecting real-time data, the cloud layer running and optimizing models, and the terminal layer supporting visual interaction.

10. The power natural disaster emergency resource allocation system based on multi-data fusion according to claim 8, characterized in that, By adjusting the disaster correction coefficient γ and traffic parameters to adapt to different disaster scenarios such as typhoons, earthquakes, and rainstorms, there is no need to reconstruct the model.

Images