Electric power material efficient scheduling and logistics optimization method for emergency scene

Abstract

The invention discloses an electric power material efficient dispatching and logistics optimization method for an emergency scene, and belongs to the field of power grid material dispatching and transportation. According to the material dispatching requirements of the power system under different disaster levels, a hierarchical response mechanism is provided, and the hierarchical response mechanism comprises a single-bin quick response mode, a multi-bin collaborative supplement mode and a whole-network centralized support mode. A mathematical optimization model including supply, transfer and demand nodes and transportation paths is established, and improved optimization algorithms such as Sine mapping population initialization, differential variation disturbance and elitist retention, dynamic reverse learning and the like are introduced to improve the solving efficiency and the diversity of solutions. Under the assumption that node positions, material requirements, vehicle parameters and the like are known, the model realizes full-process optimization of supply distribution, transportation paths and resource scheduling. The method can effectively improve the emergency material dispatching efficiency, reduces the transportation cost and the material shortage risk, and has a good practical application value.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of power grid material transportation, and more particularly relates to a power material efficient scheduling and logistics optimization method for emergency scenarios. BACKGROUND

[0002] With the continuous improvement of the emergency response mechanism of China's power system, higher requirements are put forward for the efficiency of power repair and the ability of material support in emergencies. Especially in Yunnan Province with complex terrain and large regional span, the power grid covers a wide area, and the transportation is not convenient and the communication is limited in some remote areas, so the rapid response and efficient deployment of emergency repair materials become the key factors affecting the emergency disposal capacity of the power system. In the emergency scenario of simultaneous occurrence of multiple distributed faults, the traditional method of relying on manual experience for path judgment and material scheduling has been difficult to adapt to the precise support demand of multi-source and multi-type power repair materials in a complex dynamic environment.

[0003] In recent years, in order to deal with large-scale material scheduling and path optimization problems, heuristic intelligent algorithms such as ant colony algorithm, particle swarm algorithm and genetic algorithm have been widely used in logistics path optimization and vehicle scheduling problems, and to a certain extent, the scheduling efficiency has been improved.

[0004] However, in the face of time-varying demand characteristics in the disaster evolution process, road state uncertainty, repair task priority difference and multi-warehouse collaborative supply, the traditional heuristic algorithm still has significant limitations in global optimization ability and model adaptability. Especially in the background of variable road conditions in the disaster area and multiple concurrent emergency tasks, the path planning is easy to fall into local optimum, and it is difficult to effectively adapt to the dynamic disturbance and resource conflict in the actual material transportation process.

[0005] Under this background, how to build an intelligent logistics scheduling method for power materials in emergency repair scenarios, realize the rapid matching of repair tasks and material types, the linkage scheduling of multi-source warehouse system, and the real-time optimization of emergency repair path, has become the core problem to be solved in the current power material support system. SUMMARY

[0006] Therefore, the present application proposes an emergency material scheduling method for power supply that integrates multi-warehouse material collaboration and intelligent path planning, and constructs an emergency logistics optimization model for practical application, aiming at the key pain points such as dynamic material demand, complex path planning and limited supply resources in the power repair scenario. The model is solved by state optimization algorithm (SBO) and multi-objective scheduling strategy, which improves the intelligence, refinement and real-time of emergency response, and provides technical support for efficient repair in power system emergencies.

[0007] In order to achieve the above purpose, the present application adopts the following technical solutions: the method comprises:

[0008] The material scheduling system problem in the emergency response process of the power system is described in mathematical form;

[0009] An efficient power material scheduling and logistics optimization model is established, including assumption conditions, constraint conditions and objective functions;

[0010] An efficient power material scheduling and logistics optimization model solving algorithm is constructed, and an improved state optimization algorithm is used to solve the efficient power material scheduling and logistics optimization model.

[0011] In one scheme, the mathematical description of the material scheduling system problem is as follows:

[0012] (1) Light disaster scenario: single warehouse rapid response; when the disaster scenario is relatively local, the repair demand scale is small, and the response time requirement is medium, the first level transfer warehouse near the disaster area is preferentially started to respond to the task;

[0013] (2) Moderate disaster scenario: multi-warehouse collaborative replenishment; when a single transfer warehouse cannot meet the demand for types or quantities of disaster area repair materials, the system will automatically start the nearby multi-warehouse replenishment mechanism;

[0014] (3) Major disaster scenario: whole network centralized support. When the disaster spreads widely, the repair task is intensive, the material demand is large or the category is complex, the system will trigger the whole network cooperation + central warehouse support mechanism;

[0015] The entire optimization decision includes three aspects: 1) determining the transfer node that meets the demand node,

[0016] 2) establishing the transportation path between the three types of nodes, and 3) determining the supply quantity received by the supply node and the quantity allocated to the demand node.

[0017] In one scheme, the assumption conditions include:

[0018] (1) The locations of the supply, transfer and demand nodes are known, and the supply quantity of the supply node is known;

[0019] (2) Each demand node can only be accessed once by a vehicle from a transfer node;

[0020] (3) There is no supply-demand relationship between each demand level node;

[0021] (4) All vehicles provided by the same type of node are of the same size and constant speed;

[0022] (5) Only after all supplies arrive at the transfer node, can loading and distribution begin, and the loading order depends on the urgency of the demand node.

[0023] In one scheme, the improved state optimization algorithm has been improved in three aspects:

[0024] First, a Sine mapping population initialization strategy is adopted, and an elite retention strategy is used to improve the distribution balance and coverage of the initial solution, thereby enhancing the quality of the algorithm's global search starting point.

[0025] Secondly, a differential mutation perturbation mechanism is introduced. By constructing the difference vector between individuals and superimposing the perturbation, the search is effectively guided to escape the local optimum trap.

[0026] Finally, by combining a dynamic reverse learning strategy, adaptive reverse updates are implemented based on the individual evolutionary state, guiding the algorithm out of stagnation areas and improving population diversity and search robustness.

[0027] In one approach, the objective function includes:

[0028] To minimize transportation costs, including time and vehicle costs, and to allocate emergency supplies appropriately;

[0029] Minimize total transit time;

[0030] Minimize the degree of shortage.

[0031] In one embodiment, the constraints include:

[0032] Vehicle transport capacity constraints that represent two transport processes at any given time;

[0033] Limits on the number of vehicles a node can dispatch at any given time;

[0034] Vehicles carrying a certain type of supply can visit transit nodes at most once at any given time; the amount of supply arriving at transit and demand nodes at any given time.

[0035] The demand at any given time will not exceed the actual demand.

[0036] The total amount of supply sent from the transport node to the demand node in any cycle is equal to the total amount at the transport node;

[0037] At any given time, vehicles dispatched by transit nodes only visit each demand node once; when a vehicle arrives at a node at any time, the amount of supplies it carries will not exceed its transport capacity.

[0038] At any given time, all types of supplies required by a demand node are transported by the same vehicle.

[0039] In one scheme, the Sine mapping population initialization strategy is specifically as follows: based on the sine function, by selecting initial values ​​and parameters, the Sine mapping effectively avoids the problem of the initial population being too concentrated in local areas, thereby improving the optimization algorithm's coverage and diversity of the solution space.

[0040] In one scheme, the introduction of the differential mutation perturbation mechanism specifically involves: introducing an elite retention strategy to improve the algorithm; in each iteration, the individual with the best fitness is directly retained to the next generation to avoid being replaced in crossover or mutation operations, thereby effectively avoiding the loss of high-quality solutions.

[0041] In one approach, the dynamic reverse learning strategy specifically involves constructing a reverse solution that maps the current individual to the boundary of the search space, and then adaptively adjusting the frequency and magnitude of the reverse solution by combining the current iteration count with the maximum iteration count.

[0042] In the early stages of the search, the reversed solution participates in the population update with a high probability, guiding the search to expand towards potentially high-quality regions; in the later stages of the search, the role of reverse learning gradually weakens to ensure the accuracy and stability of the local search.

[0043] Beneficial effects of this invention:

[0044] This invention addresses the multi-level transportation needs and resource allocation constraints of emergency power supplies in sudden disaster scenarios, constructing a multi-node scheduling model for power emergency logistics. With the optimization objectives of minimizing global transportation costs, response time, and task delay penalties, an improved intelligent scheduling method based on the State Optimization Algorithm (ISBO) is proposed. A three-level encoding and decoding mechanism suitable for emergency logistics environments is designed to jointly model and collaboratively optimize order tasks, transportation resources, and route planning. Furthermore, by combining Sine mapping initialization, differential mutation perturbation, and dynamic back-learning strategies, the algorithm's search capability and convergence performance are enhanced, further improving the robustness and computational efficiency of emergency scheduling in large-scale disaster scenarios. Attached Figure Description

[0045] Figure 1 This is a flowchart of the method of the present invention;

[0046] Figure 2 This is a schematic diagram of power material storage according to an embodiment of the present invention;

[0047] Figure 3 This is an emergency supplies transportation network according to one embodiment of the present invention;

[0048] Figure 4 A diagram illustrating the encoding and decoding process for emergency power supply dispatching tasks;

[0049] Figure 5The flowchart for the improved ISBO algorithm. Detailed Implementation

[0050] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0051] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. To facilitate understanding, the invention will now be described more fully with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to make the disclosure of the invention more thorough and complete.

[0052] like Figure 1 As shown, the specific implementation process of a method for efficient scheduling and logistics optimization of power resources in emergency scenarios is as follows:

[0053] Step 1: Describe the material dispatching system problem in the power system emergency response process in a mathematical form.

[0054] In power system emergency response, the efficient operation of the material dispatching system is a crucial prerequisite for ensuring the smooth progress of repair operations. Faced with practical problems such as widespread power grid paralysis after a disaster, scattered repair points, and significant differences in demand, building an emergency material logistics system with rapid response capabilities and high dispatching flexibility has become a key approach to improving the efficiency of power emergency response.

[0055] Taking Yunnan Province as a typical example, the complex geography and transportation structure of the region present significant challenges in the allocation of materials during power emergency repairs, including slow response times, high uncertainty in route access, and uneven regional supply capacity. To address the high-frequency and time-sensitive dispatching needs for emergency repair materials in emergencies, the current power grid system has established a three-tiered emergency material logistics network structure, centered on the Kunming main supply node and radiating to five fixed primary transfer warehouses (distributed in different geographical areas within Yunnan Province), where "n" represents a variable number of nodes requiring emergency repair tasks.

[0056] To meet the response needs of emergency repair tasks under disaster scenarios of different scales, this invention introduces a tiered allocation mechanism. Based on Table 1 and considering the severity of the emergency scenario and the scale of material demand, the required response strategy is dynamically determined. Figure 2 This invention demonstrates the tiered dispatching process for materials in emergency scenarios within the Yunnan Province power system, as described in this invention. Based on the disaster level and the scale of the repair task, the system dynamically determines the appropriate dispatching response mode, which includes single-warehouse rapid response, multi-warehouse collaborative allocation, and centralized support across the entire network.

[0057] Table 1 Classification of Power Grid Disaster Levels

[0058]

[0059] Specifically, it includes the following:

[0060] (1) Minor Disaster Scenarios: Rapid Response from a Single Warehouse. When the disaster scenario is relatively localized, the scale of repair needs is small, and the response time requirement is moderate, the nearest primary transfer warehouse will be prioritized for task response. The system automatically matches the optimal single transfer node based on the geographical location of the disaster site, warehouse inventory, and accessibility to complete the material assembly and direct transportation tasks, achieving rapid response scheduling under the "minimum path + single point coverage" mode.

[0061] (2) Moderate Disaster Scenario: Multi-Warehouse Collaborative Replenishment. When a single transit warehouse cannot meet the needs of disaster area for the types or quantities of emergency repair materials, the system will automatically activate the nearest multi-warehouse replenishment mechanism. While ensuring timely response, other primary warehouses located near the disaster area will be selected to jointly undertake part of the material supply task, and the task will be segmented and merged from multiple sources. In this mode, route selection needs to comprehensively consider the departure windows of multiple starting points, vehicle configuration, and the timing coordination of material arrival, achieving efficient replenishment under the "regional collaboration + multi-path merging" strategy.

[0062] (3) Major Disaster Scenarios: Centralized Support Across the Entire Network. When a disaster affects a wide area, repair tasks are intensive, and the demand for materials is large or complex, the system will trigger a network-wide collaborative + central warehouse support mechanism. Not only will the five primary transfer nodes be required to fully output their stored critical emergency supplies, but strategic reserve materials will also be directly allocated from the Kunming central supply node, constructing a composite dispatch network that includes relay scheduling and direct response. In this scenario, the system needs to simultaneously optimize multiple path networks, match the coordinated execution strategies of multiple warehouse resources, different vehicles, and multiple repair point tasks, and achieve a collaborative dispatch mode of "distributed dispatch + centralized support + dynamic replenishment".

[0063] Emergency supplies transportation network such as Figure 3As shown. The entire optimization decision includes three aspects: 1) determining the transfer nodes that meet the demand nodes, 2) establishing transportation paths between the three types of nodes, and 3) determining the quantity of supplies received by the supply nodes and the quantity allocated to the demand nodes. This invention is based on power material scheduling in emergency scenarios, with minimizing path costs and response time as the main optimization objectives. It fully considers multiple factors such as transportation resource availability, route accessibility, vehicle capacity limitations, and task priorities. Through dynamic path reconstruction and joint resource allocation, it effectively ensures the efficient supply and accurate delivery of repair materials under different emergency levels.

[0064] Step 2: Establish an efficient scheduling and logistics optimization model for power materials, including assumptions, constraints, and objective function.

[0065] S201, Model Assumptions

[0066] 1. The locations of supply, transit, and demand nodes are known, and the supply quantity of the supply nodes is known;

[0067] 2. Each demand node can only be accessed once by a vehicle from a transit node;

[0068] 3. There is no supply and demand relationship between each demand level node;

[0069] 4. All vehicles provided by nodes of the same type are of the same size and have a constant speed;

[0070] 5. Loading and distribution can only begin after all supplies have arrived at the transshipment nodes, and the loading order depends on the urgency of the demand nodes.

[0071] S202, Establishment of an efficient power material dispatching and logistics optimization model

[0072] Objective function:

[0073] f=minΣf1+f2+f3 (1)

[0074] Equation (1) aims to minimize transportation costs, including time and vehicle costs, and to rationally allocate emergency supplies. Wherein:

[0075]

[0076] Equation (2) minimizes the total transportation time. The first term represents the loading and unloading time of vehicles between the supply point and the transfer node, as well as the vehicle transportation time. The second term represents the queuing loading time of the supply at the transfer node. The third term represents the loading and unloading time of vehicles between the supply node and the demand node, as well as the vehicle transportation time. Equation (3) minimizes the vehicle transportation cost. The first and second terms represent the fixed transportation costs of the supply point-transfer node and transfer node-demand node paths, respectively. The third and fourth terms represent the loading, unloading, and vehicle transportation costs of the supply point-transfer node and transfer node-demand node paths, respectively. The fifth term represents the delay cost. Equation (4) minimizes the degree of shortage, that is, minimizes the product of the urgency of emergency supplies for all demand nodes and the quantity of unmet demand. Different demand nodes have different degrees of urgency for emergency supplies. Therefore, demand nodes with higher urgency should be given relatively more supply to ensure reasonable allocation. The constraints are:

[0077]

[0078] Constraints (5) and (12) represent vehicle transport capacity constraints for both transport processes in any given period. Constraint (6) states that the total supply sent from the supply point to the transit node in any given period shall not exceed the supply node's inventory. Constraints (7) and (15) represent restrictions on the number of vehicles that a node can dispatch in any given period. Equation (8) states that a vehicle carrying a certain type of supply may visit the transit node at most once in any given period. Constraints (9) and (10) represent the supply arriving at the transit and demand nodes in any given period, respectively. Constraint (11) states that when hour, This indicates that the emergency supply y carried by vehicle h is greater than the emergency supply unloaded before the vehicle arrives at node j. This indicates that before unloading the supply at node i, the amount of supply carried by vehicle h is within the vehicle capacity Q. h Within. When When the vehicle carrying goods y does not cross the path between nodes i and j, constraint (13) indicates that the demand of each demand node in any period will not exceed its actual demand. Constraint (14) indicates that the total amount of supply y sent from the transport node to the demand node in any period is equal to the total amount at the transport node. Constraint (16) indicates that in any period, the vehicle dispatched by the transit node visits each demand node only once. Constraint (17) indicates that the amount of goods in the vehicle when it arrives at node i in any period will not exceed the vehicle's transport capacity. Constraint (18) indicates that the amount of goods in the vehicle is greater than or equal to the actual demand of node j when the vehicle places node i at node j in any period. Constraint (19) indicates that in any period, all types of goods required by a demand node are transported by the same vehicle. Constraints (20) and (21) respectively represent the relationship between two variables in the two transport processes. Constraint (22) indicates that the routing variable and the location variable are 0-1 decision variables. Constraint (23) indicates that the flow variable is a real variable.

[0079] Step 3: Construct a solution algorithm for the efficient scheduling and logistics optimization model of power materials. Use the improved state optimization algorithm to solve the efficient scheduling and logistics optimization model of power materials.

[0080] The core of the power emergency material dispatching problem lies in how to achieve global coordination of graded material response, transportation route optimization, and vehicle dispatching plans under the complex background of multi-source supply, multi-level warehousing, and multi-target emergency repair tasks. The dispatching system must not only balance task response time and route operating costs, but also fully consider multiple factors such as the collaborative relationships between multiple supply nodes, the load-bearing capacity of transfer nodes, and the dynamic evolution of task priorities, constituting a typical high-dimensional coupled, multi-stage linkage optimization problem. Due to the large scale of variables and diverse constraints, designing an intelligent optimization algorithm with global optimization capabilities, dynamic route adjustment capabilities, and resource collaborative control mechanisms becomes crucial for efficiently solving such problems.

[0081] State-based optimization (SBO) algorithms, by simulating the human behavior of climbing a social ladder, abstract the solution process of complex optimization problems into a dynamic process evolving in a multi-dimensional state space. In this state space, each candidate scheduling scheme is considered a "state body," whose internal structure includes an ordered combination of key variables such as supply route selection, transfer node allocation, vehicle scheduling sequence, and material allocation strategies for each disaster-stricken point. In the initial stage, the algorithm uses a heuristic strategy to generate a representative and diverse initial state population. Each state body, after decoding, can be mapped to a complete power emergency logistics scheduling solution. The core mechanism of SBO is that, just as individuals improve their status by learning from and imitating successful role models in society, each state body in the algorithm continuously absorbs information from high-quality solutions, undergoing directional evolution and self-updating in the state space. During the fitness evaluation process, the algorithm constructs a comprehensive evaluation function, fully considering multi-dimensional objectives such as total transportation cost, path time delay, task response penalty, and load balancing among storage nodes, ensuring an accurate characterization of the global merits of the scheduling scheme. During state evolution, SBO introduces a state transition matrix mechanism and integrates three evolutionary strategies: local fine-tuning, global jumps, and perturbation reconstruction, simulating various behavioral patterns of individuals in the process of "social ascent" in reality. Local exploration achieves neighborhood-optimal search through fine-grained variable adjustments; global jumps leverage historical high-quality states to guide direction and enhance the ability to escape local optima; while the perturbation reconstruction mechanism breaks the stagnation of the population, reconstructing diverse search paths to maintain population activity. To further enhance the effectiveness of the search strategy, SBO employs a state memory mechanism to store and track high-fitness states, continuously assigning them higher search weights in subsequent iterations to achieve repeated use and evolutionary reinforcement of high-quality solutions. Simultaneously, the algorithm introduces a "reward"-based strategy control mechanism, providing quantitative feedback on the fitness changes brought about by each state transition. If a state transition operation brings significant performance improvement, the probability of applying the corresponding strategy in subsequent iterations will be increased; conversely, it will be suppressed, thus dynamically adjusting the execution ratio of the strategy set. This mechanism endows the algorithm with excellent adaptive and directional control capabilities, enabling flexible adjustment of search behavior for different problem structures and avoiding blind iteration and premature convergence.

[0082] The SBO algorithm was applied to the power emergency logistics dispatching scenario in Yunnan Province, demonstrating its adaptability and robustness. By modeling the unified state of the three-layer dispatching structure—supply nodes, transfer nodes, and repair demand nodes—the algorithm effectively addresses the problems of multi-stage path combinations and uneven resource distribution. Furthermore, its feedback-driven and disturbance adjustment mechanisms enable the algorithm to possess stronger recovery and adaptability when dealing with uncertain disturbances such as path interruptions, task fluctuations, and insufficient inventory. Experimental verification shows that SBO can quickly converge to a high-quality solution set within a finite number of iterations and outperforms traditional heuristic algorithms in terms of search stability, optimization accuracy, and global optimization capability, demonstrating good engineering feasibility and application promotion value.

[0083] S301, State-Based Optimization (SBO) Algorithm

[0084] The SBO algorithm simulates the fundamental motivation for humans to climb the social ladder—a behavior rooted in our need for self-improvement. This ambition reflects the core objective of optimization: iterative optimization. Just as people gain an advantage by connecting with successful peers, SBO agents learn from high-performance solutions to improve search efficiency. SBO first generates two distinct agent populations, representing individuals from different social backgrounds, which then evolve through a process modeled after seeking guidance from social elites. Therefore, random initialization is employed, with the initialization phase laying the foundation for the SBO algorithm by generating two populations, which can be mathematically modeled using the following formula:

[0085]

[0086] x i,j =U(lb) j ,ub j (25)

[0087] Among them, X 1 and X 2 This dual-population design represents family members with varying levels of knowledge and social status. It ensures that each family is represented by at least two individuals, thus capturing diversity within the family and dynamically updating the elite membership as the algorithm iterates. i,j Let lb be the j-th decision variable for the i-th individual, and D be the number of decision variables. j and ub j These are the lower and upper bounds, respectively. This uniform initialization on the N×D matrix of both populations establishes the dimensionality of the problem and ensures a diverse starting point. s

[0088] In the second phase of SBO, the elite participation phase, the SBO algorithm replicates the complex dynamics of human social status structures to enhance the search for optimal solutions. This phase simulates individuals seeking guidance from high-status mentors (elite agents in the algorithm) to accelerate their progress. Unlike the isolated family framework, this development transcends independent groups by establishing interconnections between different social units, creating a more adaptive and robust search mechanism.

[0089]

[0090] Where, x i Let x represent the i-th individual in the population. i′ Indicates the next iteration. It is through roulette wheel betting method from X e Elite individuals selected from the population, x b This is the best solution found so far. Parameters w1 and w2 are generated using rand, while w3 and w4 are design parameters used to dynamically adjust the impact of the high-level geocentric circle on exploration and development. The calculation methods for w3 and w4 are as follows:

[0091]

[0092] w4 = unifrnd(-w3, w3) (28)

[0093] The resource acquisition phase is crucial for the transition from exploration to exploitation by acquiring and utilizing valuable insights, such as social capital within human networks. Resource acquisition mechanisms vary depending on state-related success; for socially successful individuals, resources are selectively acquired through inputs from two sources on average: one from elite individuals within the same household unit, and the other from the best individuals in the overall population. This process can be modeled mathematically as follows:

[0094] To indicate:

[0095]

[0096] Where j idx =randi(D) for idx = 1, 2, 3 reflects a mixture of familial and external elite influences. Conversely, socially unsuccessful individuals rely entirely on familial resources. Their resource renewal follows...

[0097] The following rules apply:

[0098]

[0099] The row vector m is initially zero and is updated before social interaction in the following way:

[0100] m(u(1:ceil(rand×D)))=1 (31)

[0101] Where u = randperm(D) provides a random permutation of the decision variable indices. During the resource evaluation phase, the algorithm assesses whether the acquired resources enhance the individual's fitness. A previously established flag vector is used to track progress: 1 = fitness improvement (success); 0 = no improvement (failure).

[0102] S302. Improved Status-based Optimization (ISBO) Algorithm

[0103] Although the State Optimization (SBO) algorithm performs well in power emergency material dispatching, it still has limitations in terms of population diversity, global search, and local refinement when faced with the complex characteristics of tasks such as high-dimensional coupling, dynamic disturbances, and multi-stage path selection.

[0104] To improve its convergence efficiency and solution quality, this invention makes three improvements to SBO:

[0105] First, a Sine mapping population initialization strategy is adopted, which improves the distribution balance and coverage breadth of the initial solution through elite retention, enhancing the quality of the algorithm's global search starting point. Second, a differential mutation perturbation mechanism is introduced, which effectively guides the search away from local optima by constructing difference vectors between individuals and superimposing perturbations. Finally, a dynamic back-learning strategy is combined, which implements adaptive back-updates based on the individual evolutionary state, guiding the algorithm out of stagnant regions and improving population diversity and search robustness. The synergistic effect of these three strategies enables ISBO to have stronger search capabilities and adaptability when facing multi-objective and multi-constraint emergency scheduling problems.

[0106] (1) Sine mapping population initialization strategy

[0107] Sine mapping is a chaotic mapping method commonly used for population initialization. In the population initialization stage, Sine mapping is widely used to generate an initial population with nonlinear distribution characteristics. This method is based on a sine function, and its mathematical expression is simple, yet it possesses good chaotic properties and ergodicity, capable of generating widely distributed and unpredictable numerical sequences within a specified interval. By appropriately selecting initial values ​​and parameters, Sine mapping can effectively avoid the problem of the initial population being too concentrated in local regions, thereby improving the optimization algorithm's coverage and diversity of the solution space. Due to its ease of implementation and good global exploration potential, Sine mapping, as an effective initialization method, is widely used in various intelligent optimization algorithms, helping to improve the algorithm's global search performance and optimization efficiency.

[0108]

[0109] Here, a and b are parameters, which typically take values ​​between 1 and π.

[0110] (2) Elite Retention Strategy

[0111] To improve the convergence stability and optimal solution preservation capability of the algorithm, this invention introduces an Elite Retention Strategy. The core idea of ​​this strategy is to directly retain the individual with the best fitness in each iteration to the next generation, avoiding its replacement during crossover or mutation operations, thereby effectively preventing the loss of high-quality solutions.

[0112] Let the population of the th generation be:

[0113]

[0114] Where represents the population size, and () is the fitness function. First, sort all individuals in the population according to their fitness from best to worst, resulting in:

[0115]

[0116] in Let represent the th best individual in the th generation. Then, select the first th best individuals as the elite set:

[0117]

[0118] in E∈[0,1] represents the proportion of elite individuals. This elite set will be directly retained to the next generation.

[0119]

[0120] The remaining NE individuals are generated according to the crossover, mutation, or other evolutionary operators in the original algorithm, forming the complete set. Ultimately, the next complete population will be:

[0121]

[0122] This strategy can maintain the algorithm's exploration capabilities while ensuring that excellent solutions are not eliminated, thereby improving the algorithm's global convergence ability and solution accuracy.

[0123] (3) Dynamic reverse learning strategy

[0124] This invention introduces a Dynamic Opposition-Based Learning (D-OBL) strategy to enhance the algorithm's global search capability and improve convergence speed. This strategy dynamically constructs a candidate solution set by combining the current solution with its corresponding inverse solution, aiming to expand the search range and improve population quality, thereby enhancing the algorithm's ability to escape local optima. The dynamic inverse learning strategy constructs the mapping inverse solution of the current individual on the search space boundary and adaptively adjusts the frequency and magnitude of inverse solution introduction by combining the current iteration number and the maximum iteration number. In the early stages of the search, inverse solutions participate in population updates with a high probability, guiding the search towards potentially high-quality regions; in the later stages of the search, the inverse learning effect gradually weakens to ensure local search accuracy and stability. The inverse solution is generated in the following form:

[0125]

[0126] in, Let L be the position of the i-th individual in the i-th iteration in the i-th dimension. j with U j Here, δ(t) represents the lower and upper bounds of this dimension, respectively, and δ(t) is a perturbation term dynamically adjusted with iteration to control the degree of deviation of the inverse solution. Its form can be set as a decreasing function or a perturbation decay factor depending on the specific problem. In the metaheuristic optimization process, inverse learning, as a strategy that combines development and exploration, can effectively supplement the blind spots of the current individual search and avoid prematurely falling into local optima. In this study, by comparing different inverse learning intensities and action periods, the D-OBL method, which combines an adaptive adjustment strategy and a perturbation guidance mechanism, was finally selected and embedded into the main algorithm framework. This strategy enables the algorithm to have stronger search jump capabilities in the early stages and gradually converge to the vicinity of a better solution in the later stages, effectively improving the globality and accuracy of the solution.

[0127] S303, Encoding / Decoding Strategy

[0128] In the power emergency material dispatch optimization problem, task allocation, vehicle scheduling, and path node matching are the core elements for constructing the scheduling solution. These elements are highly coupled and directly affect the overall scheduling efficiency and response time. To improve the algorithm's solution efficiency and enhance the model's adaptability to real-world task structures, this invention designs a three-level encoding and decoding strategy. This strategy integrates supply node scheduling, vehicle path planning, and material allocation at repair points into a unified state representation, thereby achieving effective mapping of complex decision structures. Specifically, the first layer of encoding represents the correspondence between emergency tasks and repair points, ensuring that high-risk nodes are prioritized in resource-constrained situations. The second layer describes the material-carrying paths and task sequences of each dispatched vehicle, reflecting the transportation sequence and transfer connections. The third layer encodes the "start-middle-end" node mapping information of material flow in the form of path node pairs, guiding path search and load control. During decoding, the state body is progressively restored to a complete scheduling scheme, ensuring logical closure and execution feasibility of the scheduling task. This three-level strategy balances structural expressiveness and solution feasibility, effectively supporting the state optimization algorithm's construction of encoding constraints and global search efficiency in complex emergency dispatch problems.

[0129] This three-level encoding and decoding strategy effectively compresses the information structure in emergency dispatch tasks, reduces redundancy in information transmission, and improves the efficiency of dispatch data processing and algorithm execution speed. In the encoding phase, "emergency order number" is used as the problem dimension, and "number of available transport vehicles" is used as the variable. By constructing a mapping relationship between orders and vehicles, the system achieves reasonable allocation and load balancing of tasks among various transportation resources. Each code not only includes the corresponding material type, starting supply node, and target repair point information, but also embeds the key node pair structure in the path network, facilitating subsequent path planning algorithm calls. During the decoding process, the system relies on the updated "order-vehicle-path" triplet mapping to gradually restore a complete emergency dispatch plan. Figure 4 The diagram illustrates this mapping structure. Taking an emergency order with task number 3 as an example, its origin is a primary storage node A, and its destination is a disaster relief and repair site B. The scheduling system uses an improved SBO algorithm to solve for the shortest path and generate the optimal transportation route for this task. This strategy balances information compression and route planning flexibility, providing structured support for building efficient coding models for complex emergency logistics systems.

[0130] like Figure 5 As shown, the improved ISBO algorithm steps are as follows:

[0131] (1) Input N teams, D-dimensional decision variables, the maximum number of evaluations MaxFEs, and the upper and lower bounds of the decision variables lb, ub and the objective function;

[0132] (2) An "elite-resource" hybrid initialization strategy is adopted to randomly generate and evaluate the initial population X, and at the same time establish elite profiles X.e and the global optimal individual b;

[0133] (3) Calculate the transportation cost and time cost of each task based on the initial population, and form the initial fitness Fit and elite fitness Fit e ;

[0134] (4) Start the IZOA main loop: If the current evaluation number FEs < MaxFEs, enter the elite participation stage; otherwise, jump to step (9);

[0135] (5) Elite participation stage: Select the elite individual e through roulette wheel selection, dynamically update the weights w1, w2, w3, w4, generate a new position according to formula (26), and perform boundary control;

[0136] (6) Resource acquisition stage: Copy the current solution as X e , initialize the resource vector m and update it according to formula (31);

[0137] (7) Resource evaluation stage: Update the population position according to the encoding and decoding strategy results, update the success flag according to formula (30), and superimpose the delay penalty in the fitness calculation;

[0138] (8) Integration stage: Update the elite file X e and the global optimal b, accumulate the evaluation number FEs ← FEs + 2N, and return to step (4) to continue iteration;

[0139] (9) Repeat the above steps. After reaching the maximum evaluation number, output the optimal order allocation scheme corresponding to the global optimal individual b and the visualized scheduling path.

[0140] Table 2 Symbol Explanation of the Model of the Present Invention

[0141]

[0142]

[0143] Table 3 0-1 Variables of the Present Invention and Their Explanations

[0144]

[0145] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0146] It should be understood that the above detailed description of the technical solutions of the present invention with reference to preferred embodiments is illustrative and not restrictive. Those skilled in the art can modify the technical solutions described in the embodiments or make equivalent substitutions for some of the technical features based on reading this specification; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for efficient dispatching and logistics optimization of power resources in emergency scenarios, characterized in that: The method includes: A mathematical description is provided for the material dispatching system problem during the emergency response of the power system. Establish an efficient scheduling and logistics optimization model for power materials, including assumptions, constraints, and objective function; An algorithm for solving the efficient scheduling and logistics optimization model of power materials is constructed, and an improved state optimization algorithm is used to solve the model.

2. The method for efficient dispatching and logistics optimization of power resources in emergency scenarios according to claim 1, characterized in that: The mathematical description of the aforementioned material scheduling system problem is as follows: (1) Minor disaster scenario: rapid response from a single warehouse; when the disaster scenario is relatively localized, the scale of emergency repair needs is small, and the response time requirement is medium, priority should be given to using the primary transfer warehouse near the disaster area for task response; (2) Moderate disaster scenario: multi-warehouse collaborative replenishment; when a single transfer warehouse cannot meet the needs of the disaster area for the types or quantities of emergency repair materials, the system will automatically activate the nearby multi-warehouse replenishment mechanism; (3) Major disaster scenarios: centralized support across the entire network; when the disaster affects a wide area, the emergency repair tasks are intensive, and the demand for materials is large or the types are complex, the system will trigger a network-wide collaborative + central warehouse support mechanism; The entire optimization decision includes three aspects: 1) identifying the transshipment nodes that meet the demand nodes, 2) establishing transportation paths between the three types of nodes, and 3) determining the quantity of supply received by the supply nodes and the quantity allocated to the demand nodes.

3. The method for efficient dispatching and logistics optimization of power resources in emergency scenarios according to claim 1, characterized in that: The assumptions mentioned include: (1) The locations of supply, transshipment and demand nodes are known, and the supply quantity of the supply nodes is known; (2) Each demand node can only be accessed once by a vehicle from a transit node; (3) There is no supply and demand relationship between each demand level node; (4) All vehicles provided by nodes of the same type have the same size and constant speed; (5) Loading and distribution can only begin after all supplies have arrived at the transfer nodes, and the loading order depends on the urgency of the demand nodes.

4. The method for efficient dispatching and logistics optimization of power resources in emergency scenarios according to claim 1, characterized in that: The improved state optimization algorithm described above has been improved in three aspects: First, a Sine mapping population initialization strategy is adopted, and an elite retention strategy is used to improve the distribution balance and coverage of the initial solution, thereby enhancing the quality of the algorithm's global search starting point. Secondly, a differential mutation perturbation mechanism is introduced. By constructing the difference vector between individuals and superimposing the perturbation, the search is effectively guided to escape the local optimum trap. Finally, by combining a dynamic reverse learning strategy, adaptive reverse updates are implemented based on the individual evolutionary state, guiding the algorithm out of stagnation areas and improving population diversity and search robustness.

5. The method for efficient dispatching and logistics optimization of power resources in emergency scenarios according to claim 1, characterized in that: The objective function includes: To minimize transportation costs, including time and vehicle costs, and to allocate emergency supplies appropriately; Minimize total transit time; Minimize the degree of shortage.

6. The method for efficient dispatching and logistics optimization of power resources in emergency scenarios according to claim 1, characterized in that: The constraints include: Vehicle transport capacity constraints that represent two transport processes at any given time; Limits on the number of vehicles a node can dispatch at any given time; Vehicles carrying a certain type of supply may visit a transit node at most once at any given time. The supply that arrives at transit and demand nodes at any given time; The demand at any given time will not exceed the actual demand. The total amount of supply sent from the transport node to the demand node in any cycle is equal to the total amount at the transport node; At any given time, vehicles dispatched by transit nodes only visited each demand node once. When a vehicle reaches a node at any time, the amount of goods carried by the vehicle will not exceed the vehicle's transport capacity. At any given time, all types of supplies required by a demand node are transported by the same vehicle.

7. The method for efficient dispatching and logistics optimization of power resources in emergency scenarios according to claim 4, characterized in that: The Sine mapping population initialization strategy is as follows: Based on the sine function, by selecting initial values ​​and parameters, the Sine mapping effectively avoids the problem of the initial population being too concentrated in a local area, thereby improving the optimization algorithm's coverage and diversity of the solution space.

8. A method for efficient dispatching and logistics optimization of power resources for emergency scenarios according to claim 4, characterized in that: The introduction of the differential mutation perturbation mechanism specifically involves: introducing an elite retention strategy to improve the algorithm; in each iteration, the individual with the best fitness is directly retained to the next generation to avoid being replaced in crossover or mutation operations, thereby effectively avoiding the loss of high-quality solutions.

9. A method for efficient dispatching and logistics optimization of power resources for emergency scenarios according to claim 4, characterized in that: The dynamic reverse learning strategy is as follows: by constructing the reverse solution of the current individual on the boundary of the search space, and combining the current iteration number and the maximum iteration number, the frequency and magnitude of the reverse solution are adaptively adjusted. In the early stages of the search, the reversed solution participates in the population update with a high probability, guiding the search to expand towards potentially high-quality regions; in the later stages of the search, the role of reverse learning gradually weakens to ensure the accuracy and stability of the local search.

Images