Disaster emergency material distribution method based on multi-agent collaborative optimization

Abstract

The present invention discloses a method for allocating disaster emergency supplies based on multi-agent collaborative optimization, comprising the following steps: S1, acquiring real-time data from the disaster area to generate a multidimensional data set; S2, constructing a collaborative optimization model based on game theory to generate a multi-level risk assessment model; S3, using a spatiotemporal convolutional neural network to perform demand forecasting; S4, optimizing material distribution paths and resource scheduling; S5, automatically adjusting the scheduling strategy using a policy gradient algorithm; S6, performing global optimization using particle swarm optimization to adjust the transportation path; and S7, dynamically adjusting the material distribution plan and path planning. The present invention uses a multi-level risk assessment model and optimization algorithm to adjust the material distribution path and scheduling strategy in real time, thereby improving the efficiency, accuracy, and emergency response capabilities of disaster emergency supply distribution.

Description

Technical Field

[0001] The present invention relates to the technical field of emergency management and material dispatching, and in particular to a disaster emergency material distribution method based on multi-agent collaborative optimization. Background Art

[0002] With the increasing frequency of natural disasters and the advancement of urbanization, the dispatch and distribution of emergency supplies are becoming increasingly important. After a disaster, timely and accurate delivery of supplies to the disaster area and their rational distribution are key to reducing the impact of the disaster, protecting the lives of victims, and restoring post-disaster order. Traditional methods of dispatching supplies often rely on experience and manual judgment, often leading to problems such as uneven distribution of supplies, transportation delays, and wasted resources. Especially in the face of large-scale, multi-type, and cross-regional disasters, existing emergency supply dispatch systems are unable to effectively cope with the changing nature of disaster situations, resulting in inefficient emergency response and potentially exacerbating the negative impact of the disaster. Therefore, optimizing the distribution and dispatch of emergency supplies and improving dispatch efficiency have become important issues that need to be urgently addressed in the current field of emergency management.

[0003] Currently, many traditional emergency material dispatch systems rely on static scheduling strategies and simple material allocation models, failing to fully account for dynamic changes and real-time feedback in disaster areas. For example, many systems allocate materials based solely on preset material demand quantities, ignoring changes in actual needs in disaster areas, such as changes in traffic conditions, weather, personnel safety, and material inventory levels, all of which can affect the timely supply and rational allocation of materials. Furthermore, many systems employ simple optimization methods, such as linear programming or rule-based allocation strategies. While these methods can provide a certain degree of scheduling support, they struggle to achieve optimal allocation when dealing with complex multi-objective, multi-constrained emergency material dispatch problems.

[0004] To address the shortcomings of traditional methods, a growing number of studies in recent years have begun to explore the use of intelligent technologies, particularly resource dispatch methods based on artificial intelligence (AI) and big data. These methods primarily analyze real-time data from disaster areas, dynamically adjust resource allocation plans, and continuously optimize dispatch strategies through intelligent algorithms, striving to achieve efficient distribution of emergency supplies. For example, some studies use machine learning algorithms to predict resource needs in disaster areas and allocate supplies based on the predicted results. However, while these methods can account for real-time data and changing demand in disaster areas, some challenges remain. For example, existing methods often rely on a single model and struggle to comprehensively consider the complex multi-dimensional situation in disaster areas, resulting in low efficiency and insufficient flexibility in practical applications.

[0005] Existing technologies are particularly deficient in multi-level risk assessment and multi-objective optimization. Risk factors in disaster areas are multi-layered and multi-dimensional, encompassing traffic conditions, weather changes, resource inventory, personnel safety, social stability, and other aspects. Existing technologies often overlook the interrelationships and dynamic changes between these factors, making it impossible to make efficient and flexible decisions in complex emergency environments. For example, some existing dispatch systems may rely on a single risk assessment model, ignoring the comprehensive consideration of the multi-level risks in the disaster area, resulting in material dispatch plans that are unable to fully respond to actual conditions.

[0006] Furthermore, existing optimization methods also face challenges when handling large-scale, multi-objective scheduling problems. Traditional optimization methods often struggle to simultaneously consider multiple optimization objectives, such as transportation costs, delivery times, resource utilization, and risk control. Furthermore, many existing systems lack adaptive scheduling capabilities and are unable to dynamically adjust and optimize based on the real-time conditions in the disaster area. For example, when emergencies occur in a disaster area, such as traffic congestion, weather changes, or insufficient supplies in a warehouse, existing systems are often unable to make timely adjustments, resulting in reduced resource allocation efficiency and even shortages in certain areas of the disaster area.

[0007] Therefore, how to provide a disaster emergency material distribution method based on multi-agent collaborative optimization is an urgent problem that technicians in this field need to solve. Summary of the Invention

[0008] One objective of this invention is to propose a method for allocating disaster emergency supplies based on multi-agent collaborative optimization. This method leverages intelligent optimization technologies such as multi-level risk assessment models, particle swarm optimization algorithms, genetic algorithms, and deep reinforcement learning. It describes in detail how to dynamically optimize the distribution of supplies by combining real-time data from the disaster area, along with multi-dimensional risk factors such as transportation, weather, safety, and inventory. This method enables real-time adjustments to the distribution path when a disaster strikes, ensuring efficient, timely, and accurate distribution of supplies to the disaster area.

[0009] The present invention has a number of significant advantages. First, through a multi-level risk assessment model, it can comprehensively consider multiple factors such as transportation, weather, safety and inventory in the disaster area, avoiding the shortcomings of traditional methods that ignore important risk factors. Secondly, combined with the particle swarm optimization algorithm and the genetic algorithm, the present invention can optimize the material distribution path in a complex scheduling environment, thereby reducing transportation costs and resource waste, and improving distribution efficiency. In addition, the application of deep reinforcement learning algorithms enables the system to make dynamic adjustments based on real-time changes in the disaster area, ensuring the flexibility of material distribution and the efficiency of emergency response. Through this comprehensive approach, the present invention ensures that the material distribution plan not only has a high degree of accuracy, but can also respond in the shortest time, thereby improving the response speed and effect of material scheduling.

[0010] The disaster emergency material distribution method based on multi-agent collaborative optimization according to an embodiment of the present invention includes the following steps:

[0011] S1. Obtain real-time data from the disaster area and generate a multi-dimensional data set through data cleaning and standardization;

[0012] S2. Construct a multi-agent collaborative optimization model based on game theory, define the multi-agent collaborative optimization goals through information sharing and dynamic adjustment mechanisms, and generate a multi-level risk assessment model;

[0013] S3. Based on multi-dimensional data sets and multi-agent collaborative optimization objectives, a spatiotemporal convolutional neural network is used to analyze the multi-scale spatiotemporal characteristics of the disaster area and generate a dynamic demand forecasting model for the disaster area.

[0014] S4. Combine the dynamic demand forecast model and spatiotemporal characteristics of the disaster area to optimize material distribution paths and resource scheduling, generate material distribution plans, and adjust resource configuration and allocation strategies in real time;

[0015] S5. Based on the preliminary material allocation plan, the policy gradient algorithm is used to automatically adjust the scheduling strategy according to changes in the disaster situation and resource demand;

[0016] S6. Combine the multi-level risk assessment model and scheduling strategy to globally optimize the material distribution plan through the particle swarm optimization algorithm, predict potential risks and bottlenecks, and adjust transportation routes and resource allocation;

[0017] S7. Dynamically adjust the material distribution plan based on real-time disaster situation changes and feedback information, and use the self-repair mechanism to plan routes and reallocate resources for unforeseen emergencies.

[0018] Optionally, the S2 specifically includes:

[0019] S21. Define multiple subject roles, including logistics companies and rescue teams, and determine the resources, goals, and constraints of each subject;

[0020] S22. Establish a multi-agent collaborative optimization model and set the profit function of each agent:

[0021] R i =f i (x1,x2,...,x n ,t)+λ i ·g i (x1,x2,...,x n );

[0022] Among them, R i represents the income of the i-th entity, f i () is the objective function of the i-th subject, gi () is the constraint function of the ith subject, x1,x2,...,x n is the resource allocation and scheduling strategy variable, t represents the time factor, λ i is the Lagrange multiplier;

[0023] S23. Adopting an adaptive game algorithm based on multi-agent reinforcement learning, the deep Q network is used to dynamically adjust the strategies of each agent in the game model, optimize the cooperation and competition relationship between multiple agents, and obtain the Q-value function of each agent:

[0024] Q i (x1,x2,...,x n ,t)=r i +γ·max a' Q i (x1',x2',...,x n ',t');

[0025] Among them, Q i () is the Q value function of the i-th subject, r i is the immediate reward of the current strategy, γ is the discount factor, x1',x2',...,x n ' represents the future state, t' is the time of the future moment, a' is the future action selection, and max is the maximum value operation;

[0026] S24. Through the learning mechanism of the deep Q network, the Q-value function is updated according to real-time data feedback and environmental changes, and the ε-greedy strategy is used for action selection:

[0027] Q i (x t ,a t )←Q i (x t ,a t )+α(r t +γ·max a' Q i (x t+1 ,a′)-Q i (x t ,a t ));

[0028] Among them, Q i (x t ,a t ) is the i-th subject in state x t Next select action a t Q value, r t is the immediate reward, α is the learning rate, max a' Q i (xt+1 ,a') is the maximum Q value in the future state, x t+1 For status;

[0029] S25. Based on the dynamic adjustment mechanism, a multi-level risk assessment model is generated. Through a comprehensive assessment of the material demand, transportation capacity, and time constraints in the disaster area, potential resource bottlenecks and risk points are predicted. The feedback adjustment strategy of the game model is combined to form an optimization goal.

[0030] Optionally, the S3 specifically includes:

[0031] S31. Obtain real-time data on the disaster area, including historical material needs, traffic conditions, weather information, and geographic information;

[0032] S32. Use the spatiotemporal convolutional neural network to analyze the spatiotemporal characteristics of the disaster area. The spatial convolution layer extracts spatial features, and the temporal convolution layer captures temporal features to generate a spatiotemporal feature expression of the disaster area's needs:

[0033] X spatial =Conv(X input ,W1);

[0034] Among them, X input is the input data, W1 is the spatial convolution kernel, X spatial It is the spatial feature extracted by convolution operation, and Conv is the convolution operation;

[0035] S33. Combine the graph attention network to model the geographical information and transportation network of the disaster area, extract the heterogeneous features of the graph data, and use the self-attention mechanism to perform weighted learning on different adjacent nodes of the graph data to generate a graph representation of the disaster area's transportation and geographical features:

[0036] X geo =GAT(X spatial ,A,W2);

[0037] Among them, A is the adjacency matrix of the graph, X geo represents the geographical and traffic features extracted by graph convolution, W2 is the weight matrix in the graph attention network, and GAT is the graph attention network;

[0038] S34. By combining the spatiotemporal convolutional neural network with the graph attention network, the spatiotemporal and graph features are integrated to generate the predicted value of the material demand in the disaster area.

[0039] S35. Combined with the adaptive weighting mechanism, the temporal and spatial characteristics and geographical and traffic characteristics of the disaster area's material demand are weighted, the impact of different characteristics on the prediction results is adjusted, and a dynamic demand prediction model for the disaster area is generated:

[0040]

[0041] Among them, ω1 and ω2 are adaptive weighting factors, W3 and W4 are weight matrices for spatiotemporal and geographical features, and X temporal is the temporal feature extracted by spatiotemporal convolution, The material demand forecast value.

[0042] Optionally, the S4 specifically includes:

[0043] S41. Based on the generated dynamic demand forecast model for the disaster area, obtain the material demand in each area of the disaster area and construct a preliminary material allocation objective function F:

[0044]

[0045] Among them, c ij represents the transportation cost from warehouse i to destination j, x ij is the number of materials allocated, m is the number of warehouses, n is the number of destinations, λ i is the Lagrange multiplier, s i is the storage capacity of warehouse i, min is the minimum value operation;

[0046] S42. Based on the traffic conditions, infrastructure damage, and weather conditions in the disaster area, combined with spatiotemporal feature data, establish constraints on the material transportation routes.

[0047] S43. Use multi-objective reinforcement learning algorithm to globally optimize material distribution paths and resource scheduling:

[0048] Q i (x t ,a t )=r t +γ·max a' Q i (x t+1 ,a′);

[0049] Among them, Q i (x t ,a t ) is the i-th subject in state x t Next select action a t The Q value, γ is the discount factor, a' is the future action selection, r t is the immediate reward, max is the maximum value operation, x t+1 For status;

[0050] S44. Add Kalman filtering to the particle swarm optimization algorithm for dynamic path adjustment, reduce prediction errors through state estimation, and update the material distribution path in real time:

[0051]

[0052] P t|t =(IK t H t )P t|t-1 ;

[0053] in, is the estimated state, K t is the Kalman gain, z t is the measured value, P t|t-1 is the previous estimated covariance, H t is the observation matrix, I is the identity matrix, is the state estimate before time t, P t|t is the updated covariance matrix;

[0054] S45. Based on the material allocation plan, adjust resource configuration and allocation strategies in real time, and replan the path through the scheduling optimization algorithm.

[0055] Optionally, the S5 specifically includes:

[0056] S51. Based on the generated preliminary material allocation plan, obtain real-time data feedback from the disaster area, including traffic information, weather changes, and resource inventory, establish a real-time adjustment model, and combine different data sources using a weighted summation method to generate adjustment factors for resource demand and material dispatch in the disaster area:

[0057] D adjusted =τ1·D traffic +τ2·D weather +τ3·D inventory ;

[0058] Among them, D adjusted is the adjusted resource demand vector of the disaster area, D traffic 、D weather and D inventory They represent the traffic information, weather changes and resource inventory data of the disaster area respectively, and τ1, τ2 and τ3 are weighting coefficients;

[0059] S52. Based on the real-time adjustment model and the feedback data from the disaster area, a dynamic scheduling objective function D is established. f :

[0060]

[0061] Among them, c ij is the transportation cost from warehouse i to destination j, t ij Due to delivery delay, u ij is the resource utilization rate, α1, α2 and α3 are weighted coefficients, x ijis the quantity of materials allocated, m is the number of warehouses, and n is the number of destinations;

[0062] S53. Use the policy gradient algorithm in deep reinforcement learning to optimize the material dispatch plan by updating the policy function, learn the optimal dispatch strategy, and dynamically adjust the dispatch strategy based on real-time feedback from the disaster area to optimize material distribution efficiency and resource utilization:

[0063]

[0064] in, is the policy gradient, E is the expected operation, π θ (a t |x t ) is the current strategy, a t is the action selected at time t, x t is the state of the system at time t, V(x t+1 ) is the value of the next state, V(x t is the value of the current state, r t is the immediate reward, γ is the discount factor, ln is the logarithmic function, θ is the parameter of the strategy, x t+1 For the next state;

[0065] S54. Based on the optimization results of the policy gradient algorithm, adjust the material scheduling plan, and automatically adjust the distribution path based on the generated scheduling strategy with real-time feedback;

[0066] S55. Re-plan the material distribution and dispatch plan in the disaster area based on the material dispatch strategy.

[0067] Optionally, the S6 specifically includes:

[0068] S61. Based on the generated material dispatch strategy and real-time disaster area demand forecast data, obtain the risk assessment data of each area in the disaster area. Through the multi-level risk assessment model, integrate the risk factors at different levels, and use the fuzzy logic reasoning method to obtain the fuzziness of each risk factor, and finally obtain the comprehensive risk assessment value R total As input for material dispatch:

[0069]

[0070] Among them, μ1, μ2, μ3 and μ4 are weighting coefficients, and are the transportation, weather, safety and inventory risks of the i-th region in the disaster area, respectively, and n is the number of regions in the disaster area;

[0071] S62. Based on comprehensive risk assessment, a multi-objective genetic algorithm is used to optimize material distribution routes. A fitness function is constructed to evaluate the pros and cons of various material distribution schemes. The fitness function comprehensively considers transportation costs, risk assessment, and resource utilization efficiency:

[0072]

[0073] Among them, F i represents the fitness of the i-th material allocation plan, c ik is the transportation cost from warehouse i to destination k, x ik is the quantity of materials allocated from warehouse i to destination k, u ik is the resource utilization rate, ρ1, ρ2 and ρ3 are weighting coefficients;

[0074] S63. Use particle swarm optimization algorithm to perform global optimization and adjust the delivery path:

[0075]

[0076] in, represents the velocity of particle i in the kth generation, represents the position of particle i in the kth generation, is the individual best historical position of particle i, gbest k is the historical optimal position of all particles, ω is the inertia weight, c1 and c2 are acceleration constants, r1 and r2 are random factors, is the updated particle velocity, is the updated particle position;

[0077] S64. By optimizing material distribution routes, adjusting scheduling strategies in real time, and replanning distribution routes, comprehensively considering the impact of traffic and weather changes in the disaster area, we can maximize resource utilization and efficiency.

[0078] S65. Based on the optimization results, the material distribution plan is adjusted through the adaptive path correction mechanism, and the material distribution path is replanned in combination with real-time disaster changes.

[0079] The beneficial effects of the present invention are:

[0080] The global optimization method for material distribution plans proposed in this paper, based on a multi-level risk assessment model and optimized scheduling strategies, addresses the low efficiency, slow response, and resource waste inherent in traditional emergency material dispatch. This method utilizes a multi-level risk assessment model to comprehensively consider multi-dimensional risk factors such as transportation, weather, safety, and inventory in the disaster area, enabling material distribution plans to more accurately reflect the actual needs and risk profile of the disaster area, thereby optimizing the scheduling and distribution of materials. This innovation effectively addresses the shortcomings of existing technologies in emergency material dispatch, which fail to fully consider dynamic risk changes and multi-objective optimization.

[0081] In addition, the present invention uses a particle swarm optimization algorithm and a genetic algorithm to globally optimize the material distribution path, ensuring that the optimal solution can be found in large-scale, multi-objective scheduling problems. This optimization not only reduces transportation costs and resource waste, but also improves the efficiency of material distribution, ensuring that materials can be delivered to the disaster area in a more timely manner. Through the introduction of a deep reinforcement learning algorithm, the present invention further improves the adaptive ability of the scheduling strategy, enabling the system to adjust the material distribution plan in real time according to changes in the disaster situation and emergencies. This dynamic adjustment capability can maximize the timely supply and effective distribution of materials when emergencies or environmental changes occur in the disaster area, avoiding material shortages or uneven distribution that may be caused by scheduling delays.

[0082] In general, the present invention significantly improves the flexibility, accuracy and efficiency of the material dispatch system by comprehensively utilizing multi-level risk assessment models and advanced optimization algorithms. It can optimize material distribution paths and resource scheduling strategies in real time in complex and dynamic disaster environments, thereby effectively improving the response speed and coping capabilities of emergency material dispatch in disaster areas, ensuring the safety of disaster victims and the restoration of social order. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0084] Figure 1 This is a flow chart of the disaster emergency material distribution method based on multi-agent collaborative optimization proposed by the present invention;

[0085] Figure 2 This is a schematic diagram of the material dispatch path of the disaster emergency material distribution method based on multi-agent collaborative optimization proposed in this invention. DETAILED DESCRIPTION

[0086] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0087] refer to Figure 1 and Figure 2 ,The disaster emergency material distribution method based on multi-agent collaborative optimization, includes the following steps:

[0088] S1. Obtain real-time data from the disaster area and generate a multi-dimensional data set through data cleaning and standardization;

[0089] S2. Construct a multi-agent collaborative optimization model based on game theory, define the multi-agent collaborative optimization goals through information sharing and dynamic adjustment mechanisms, and generate a multi-level risk assessment model;

[0090] S3. Based on multi-dimensional data sets and multi-agent collaborative optimization objectives, a spatiotemporal convolutional neural network is used to analyze the multi-scale spatiotemporal characteristics of the disaster area and generate a dynamic demand forecasting model for the disaster area.

[0091] S4. Combine the dynamic demand forecast model and spatiotemporal characteristics of the disaster area to optimize material distribution paths and resource scheduling, generate material distribution plans, and adjust resource configuration and allocation strategies in real time;

[0092] S5. Based on the preliminary material allocation plan, the policy gradient algorithm is used to automatically adjust the scheduling strategy according to changes in the disaster situation and resource demand;

[0093] S6. Combine the multi-level risk assessment model and scheduling strategy to globally optimize the material distribution plan through the particle swarm optimization algorithm, predict potential risks and bottlenecks, and adjust transportation routes and resource allocation;

[0094] S7. Dynamically adjust the material distribution plan based on real-time disaster situation changes and feedback information, and use the self-repair mechanism to plan routes and reallocate resources for unforeseen emergencies.

[0095] In this embodiment, S2 specifically includes:

[0096] S21. Define multiple subject roles, including logistics companies and rescue teams, and determine the resources, goals, and constraints of each subject;

[0097] S22. Establish a multi-agent collaborative optimization model and set the profit function of each agent:

[0098] R i =f i (x1,x2,...,x n ,t)+λ i ·g i (x1,x2,...,x n );

[0099] Among them, R irepresents the income of the i-th entity, f i () is the objective function of the i-th subject, g i () is the constraint function of the ith subject, x1,x2,...,x n is the resource allocation and scheduling strategy variable, t represents the time factor, λ i is the Lagrange multiplier;

[0100] S23. Adopting an adaptive game algorithm based on multi-agent reinforcement learning, the deep Q network is used to dynamically adjust the strategies of each agent in the game model, optimize the cooperation and competition relationship between multiple agents, and obtain the Q-value function of each agent:

[0101] Q i (x1,x2,...,x n ,t)=r i +γ·max a' Q i (x1',x2',...,x n ',t');

[0102] Among them, Q i () is the Q value function of the i-th subject, r i is the immediate reward of the current strategy, γ is the discount factor, x1',x2',...,x n ' represents the future state, t' is the time of the future moment, a' is the future action selection, and max is the maximum value operation;

[0103] S24. Through the learning mechanism of the deep Q network, the Q-value function is updated according to real-time data feedback and environmental changes, and the ε-greedy strategy is used for action selection:

[0104] Q i (x t ,a t )←Q i (x t ,a t )+α(r t +γ·max a' Q i (x t+1 ,a′)-Q i (x t ,a t ));

[0105] Among them, Q i (x t ,a t ) is the i-th subject in state x t Next select action a t Q value, r tis the immediate reward, α is the learning rate, max a' Q i (x t+1 ,a') is the maximum Q value in the future state, x t+1 For status;

[0106] S25. Based on the dynamic adjustment mechanism, a multi-level risk assessment model is generated. Through a comprehensive assessment of the material demand, transportation capacity, and time constraints in the disaster area, potential resource bottlenecks and risk points are predicted. The feedback adjustment strategy of the game model is combined to form an optimization goal.

[0107] In this embodiment, S3 specifically includes:

[0108] S31. Obtain real-time data on the disaster area, including historical material needs, traffic conditions, weather information, and geographic information;

[0109] S32. Use the spatiotemporal convolutional neural network to analyze the spatiotemporal characteristics of the disaster area. The spatial convolution layer extracts spatial features, and the temporal convolution layer captures temporal features to generate a spatiotemporal feature expression of the disaster area's needs:

[0110] X spatial =Conv(X input ,W1);

[0111] Among them, X input is the input data, W1 is the spatial convolution kernel, X spatial It is the spatial feature extracted by convolution operation, and Conv is the convolution operation;

[0112] S33. Combine the graph attention network to model the geographical information and transportation network of the disaster area, extract the heterogeneous features of the graph data, and use the self-attention mechanism to perform weighted learning on different adjacent nodes of the graph data to generate a graph representation of the disaster area's transportation and geographical features:

[0113] X geo =GAT(X spatial ,A,W2);

[0114] Among them, A is the adjacency matrix of the graph, X geo represents the geographical and traffic features extracted by graph convolution, W2 is the weight matrix in the graph attention network, and GAT is the graph attention network;

[0115] S34. By combining spatiotemporal convolutional neural networks with graph attention networks, spatiotemporal and graph features are integrated to generate a forecast of material demand in the disaster area.

[0116] S35. Combined with the adaptive weighting mechanism, the temporal and spatial characteristics and geographical and traffic characteristics of the disaster area's material demand are weighted, the impact of different characteristics on the prediction results is adjusted, and a dynamic demand prediction model for the disaster area is generated:

[0117]

[0118] Among them, ω1 and ω2 are adaptive weighting factors, W3 and W4 are weight matrices for spatiotemporal and geographical features, and X temporal is the temporal feature extracted by spatiotemporal convolution, The material demand forecast value.

[0119] In this embodiment, the S4 specifically includes:

[0120] S41. Based on the generated dynamic demand forecast model for the disaster area, obtain the material demand in each area of the disaster area and construct a preliminary material allocation objective function F:

[0121]

[0122] Among them, c ij represents the transportation cost from warehouse i to destination j, x ij is the number of materials allocated, m is the number of warehouses, n is the number of destinations, λ i is the Lagrange multiplier, s i is the storage capacity of warehouse i, min is the minimum value operation;

[0123] S42. Based on the traffic conditions, infrastructure damage, and weather conditions in the disaster area, combined with spatiotemporal feature data, establish constraints on the material transportation routes.

[0124] S43. Use multi-objective reinforcement learning algorithm to globally optimize material distribution paths and resource scheduling:

[0125] Q i (x t ,a t )=r t +γ·max a' Q i (x t+1 ,a′);

[0126] Among them, Q i (x t ,a t ) is the i-th subject in state x t Next select action a t The Q value, γ is the discount factor, a' is the future action selection, r t is the immediate reward, max is the maximum value operation, x t+1 For status;

[0127] S44. Add Kalman filtering to the particle swarm optimization algorithm for dynamic path adjustment, reduce prediction errors through state estimation, and update the material distribution path in real time:

[0128]

[0129] P t|t =(IK t H t )P t|t-1 ;

[0130] in, is the estimated state, K t is the Kalman gain, z t is the measured value, P t|t-1 is the previous estimated covariance, H t is the observation matrix, I is the identity matrix, is the state estimate before time t, P t|t is the updated covariance matrix;

[0131] S45. Based on the material allocation plan, adjust the resource configuration and allocation strategy in real time, and replan the path through the scheduling optimization algorithm.

[0132] In this embodiment, the S5 specifically includes:

[0133] S51. Based on the generated preliminary material allocation plan, obtain real-time data feedback from the disaster area, including traffic information, weather changes, and resource inventory, establish a real-time adjustment model, and combine different data sources using a weighted summation method to generate adjustment factors for resource demand and material dispatch in the disaster area:

[0134] D adjusted =τ1·D traffic +τ2·D weather +τ3·D inventory ;

[0135] Among them, D adjusted is the adjusted resource demand vector of the disaster area, D traffic 、D weather and D inventory They represent the traffic information, weather changes and resource inventory data of the disaster area respectively, and τ1, τ2 and τ3 are weighting coefficients;

[0136] S52. Based on the real-time adjustment model and the feedback data from the disaster area, a dynamic scheduling objective function D is established. f :

[0137]

[0138] Among them, c ij is the transportation cost from warehouse i to destination j, t ij Due to delivery delay, u ij is the resource utilization rate, α1, α2 and α3 are weighted coefficients, x ij is the quantity of materials allocated, m is the number of warehouses, and n is the number of destinations;

[0139] S53. Use the policy gradient algorithm in deep reinforcement learning to optimize the material dispatch plan by updating the policy function, learn the optimal dispatch strategy, and dynamically adjust the dispatch strategy based on real-time feedback from the disaster area to optimize material distribution efficiency and resource utilization:

[0140]

[0141] in, is the policy gradient, E is the expected operation, π θ (a t |x t ) is the current strategy, a t is the action selected at time t, x t is the state of the system at time t, V(x t+1 ) is the value of the next state, V(x t is the value of the current state, r t is the immediate reward, γ is the discount factor, ln is the logarithmic function, θ is the parameter of the strategy, x t+1 For the next state;

[0142] S54. Based on the optimization results of the policy gradient algorithm, adjust the material scheduling plan, and automatically adjust the distribution path based on the generated scheduling strategy with real-time feedback;

[0143] S55. Re-plan the material distribution and dispatch plan in the disaster area based on the material dispatch strategy.

[0144] In this embodiment, S6 specifically includes:

[0145] S61. Based on the generated material dispatch strategy and real-time disaster area demand forecast data, obtain the risk assessment data of each area in the disaster area. Through the multi-level risk assessment model, integrate the risk factors at different levels, and use the fuzzy logic reasoning method to obtain the fuzziness of each risk factor, and finally obtain the comprehensive risk assessment value R total As input for material dispatch:

[0146]

[0147] Among them, μ1, μ2, μ3 and μ4 are weighting coefficients, and are the transportation, weather, safety and inventory risks of the i-th region in the disaster area, respectively, and n is the number of regions in the disaster area;

[0148] S62. Based on comprehensive risk assessment, a multi-objective genetic algorithm is used to optimize material distribution routes. A fitness function is constructed to evaluate the pros and cons of various material distribution schemes. The fitness function comprehensively considers transportation costs, risk assessment, and resource utilization efficiency:

[0149]

[0150] Among them, F i represents the fitness of the i-th material allocation plan, c ik is the transportation cost from warehouse i to destination k, x ik is the quantity of materials allocated from warehouse i to destination k, u ik is the resource utilization rate, ρ1, ρ2 and ρ3 are weighting coefficients;

[0151] S63. Use particle swarm optimization algorithm to perform global optimization and adjust the delivery path:

[0152]

[0153] in, represents the velocity of particle i in the kth generation, represents the position of particle i in the kth generation, pbest i k is the individual best historical position of particle i, gbest k is the historical optimal position of all particles, ω is the inertia weight, c1 and c2 are acceleration constants, r1 and r2 are random factors, is the updated particle velocity, is the updated particle position;

[0154] S64. By optimizing material distribution routes, adjusting scheduling strategies in real time, and replanning distribution routes, comprehensively considering the impact of traffic and weather changes in the disaster area, we can maximize resource utilization and efficiency.

[0155] S65. Based on the optimization results, the material distribution plan is adjusted through the adaptive path correction mechanism, and the material distribution path is replanned in combination with real-time disaster changes.

[0156] Example 1:

[0157] To verify the feasibility of the present invention in practice, the present invention was applied to the post-disaster emergency material dispatch in a certain region. A severe typhoon struck the region, disrupting transportation, destroying buildings, causing power outages, and paralyzing communications. This led to a sharp increase in the demand for materials in multiple cities and towns. After the typhoon, the affected areas faced demands for water, food, tents, and medical equipment. Due to post-disaster transportation damage and interregional logistics bottlenecks, the supply of materials to some affected areas was delayed. Optimizing material distribution plans and efficiently completing dispatch within a short period of time became a key issue in emergency management in the region.

[0158] To address this issue, the proposed material dispatch optimization method was applied in a real-world scenario. First, the post-disaster emergency material dispatch center collected real-time data from the disaster area, including road traffic conditions, weather forecasts, material inventory levels, and specific needs within the affected area. Needs varied across the disaster area, with some areas experiencing severe traffic congestion, while others were in urgent need of relief supplies such as water and food due to insufficient inventory. The system used real-time data collection and risk assessment models to comprehensively analyze the material needs of each region.

[0159] In practice, the system first evaluates each affected area using a multi-level risk assessment model, calculating a risk value for each area. This risk assessment model takes into account factors such as traffic conditions, weather conditions, resource inventory, and supply needs. Through this model, the risk value of a disaster area reflects the severity of the disaster and the urgency of supply dispatch. For example, some areas may be high-risk due to impassable roads caused by heavy rain and storms, while other areas may be low-risk due to sufficient supply inventory.

[0160] The system then applied particle swarm optimization and genetic algorithms to globally optimize the material distribution paths. Based on the risk assessment results, the material distribution paths and resource scheduling plans were replanned to reduce transportation costs, optimize transportation routes, and avoid congested areas. This optimization algorithm significantly improved the timeliness of material transportation, ensuring that supplies reached the places where they were most needed, on time and accurately.

[0161] For example, the initial transportation route planning was estimated to take 48 hours, but after optimization, the material distribution time was shortened to 36 hours. The optimized plan avoided traffic bottlenecks and took into account the impact of weather changes in the disaster area, enabling the delivery of essential supplies to high-risk areas before heavy rains arrived. Through deep reinforcement learning algorithms, the system can dynamically adjust the material distribution plan based on real-time feedback. When an emergency occurs in a certain area (such as a road closure or weather changes), the system will immediately adjust the route and re-plan the resource scheduling strategy.

[0162] In this embodiment, the optimized material dispatch system significantly improves the efficiency and accuracy of emergency supply delivery. For example, within the first day after the disaster, the system optimized the material delivery routes, reducing the actual transportation time from the initially planned 72 hours to 48 hours, a 33% reduction in transportation time. For areas significantly affected by traffic and weather, the material delivery time was reduced from the estimated 96 hours to 72 hours, saving 24 hours of transportation time. This method ensures that materials can reach the disaster area in the shortest possible time, significantly improving the efficiency of post-disaster emergency response.

[0163] Table 1 Material demand and distribution plan in disaster areas

[0164]

[0165] The data in the table above can be used to analyze material demand and dispatch efficiency improvements. The table lists material demand, initial dispatch time, optimized dispatch time, and dispatch efficiency improvements for the five affected areas.

[0166] First, in terms of material demand, Region 5 has the greatest need, including 300 tons of water, 200 tons of food, 800 tents, and 500 pieces of medical equipment. Region 1, on the other hand, has less demand, requiring 150 tons of water, 50 tons of food, 300 tents, and 100 pieces of medical equipment. This demonstrates the significant disparity in material demand across regions, necessitating the optimization of dispatch routes.

[0167] In terms of initial dispatch time, Region 4 had the longest material delivery time, at 120 hours, indicating that factors such as traffic and weather had a significant impact on dispatch. Region 2 had a relatively short initial dispatch time of 72 hours, indicating that material dispatch in this region was in good condition, but there was still room for improvement.

[0168] After optimizing the dispatch time, delivery times in all regions were significantly shortened. For example, the optimized dispatch time in Region 1 was 36 hours, a reduction of 12 hours compared to the initial dispatch time, and a 25% increase in dispatch efficiency. The optimized dispatch time in Region 2 was 48 hours, a reduction of 24 hours, and a 33.33% increase in dispatch efficiency. These data demonstrate the significant effectiveness of this invention in optimizing dispatch routes and improving material distribution efficiency.

[0169] For Regions 3, 4, and 5, the optimized dispatch times were 72 hours, 90 hours, and 90 hours, respectively. Although the optimization effects in these regions were relatively small, they still achieved a 25% efficiency improvement. In particular, in regions with longer initial dispatch times, the optimized dispatch significantly improved the timeliness of material distribution.

[0170] In summary, the table data shows that the optimization method of the present invention significantly improves the efficiency of material dispatching, ensures that materials can be distributed to the disaster area in a timely manner, and thus improves the response speed and effect of post-disaster rescue.

[0171] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A disaster emergency material distribution method based on multi-agent collaborative optimization, characterized by: The steps include: S1. Obtain real-time data from the disaster area and generate a multi-dimensional data set through data cleaning and standardization; S2. Construct a multi-agent collaborative optimization model based on game theory, define the multi-agent collaborative optimization goals through information sharing and dynamic adjustment mechanisms, and generate a multi-level risk assessment model; S3. Based on multi-dimensional data sets and multi-agent collaborative optimization objectives, a spatiotemporal convolutional neural network is used to analyze the multi-scale spatiotemporal characteristics of the disaster area and generate a dynamic demand forecasting model for the disaster area. S4. Combine the dynamic demand forecast model and spatiotemporal characteristics of the disaster area to optimize material distribution paths and resource scheduling, generate material distribution plans, and adjust resource configuration and allocation strategies in real time; S5. Based on the preliminary material allocation plan, the policy gradient algorithm is used to automatically adjust the scheduling strategy according to changes in the disaster situation and resource demand; S6. Combine the multi-level risk assessment model and scheduling strategy to globally optimize the material distribution plan through the particle swarm optimization algorithm, predict potential risks and bottlenecks, and adjust transportation routes and resource allocation; S7. Dynamically adjust the material distribution plan based on real-time disaster situation changes and feedback information, and use the self-repair mechanism to plan routes and reallocate resources for unforeseen emergencies.

2. The disaster emergency material distribution method based on multi-agent collaborative optimization according to claim 1 is characterized in that: The S2 specifically includes: S21. Define multiple subject roles, including logistics companies and rescue teams, and determine the resources, goals, and constraints of each subject; S22. Establish a multi-agent collaborative optimization model and set the profit function of each agent: R i =f i (x1,x2,…,x n ,t)+λ i ·g i (x1,x2,…,x n ): Among them, R i represents the income of the i-th entity, f i () is the objective function of the i-th subject, g i () is the constraint function of the ith subject, x1,x2,...,x n is the resource allocation and scheduling strategy variable, t represents the time factor, λ i is the Lagrange multiplier; S23. Adopting an adaptive game algorithm based on multi-agent reinforcement learning, the deep Q network is used to dynamically adjust the strategies of each agent in the game model, optimize the cooperation and competition relationship between multiple agents, and obtain the Q-value function of each agent: Q i (x1,x2,...,x n ,t)=r i +γ·max a 'Q i (x1',x2',...,x n ',t'); Among them, Q i () is the Q value function of the i-th subject, r i is the immediate reward of the current strategy, γ is the discount factor, x1',x2',...,x n ' represents the future state, t' is the time of the future moment, a' is the future action selection, and max is the maximum value operation; S24. Through the learning mechanism of the deep Q network, the Q-value function is updated according to real-time data feedback and environmental changes, and the ε-greedy strategy is used for action selection: Q i (x t ,a t )←Q i (x t ,a t )+α(r t +γ·max a' Q i (x t+1 ,a)-Q i (x t ,a t )); Among them, Q i (x t ,a t ) is the i-th subject in state x t Next select action a t Q value, r t is the immediate reward, α is the learning rate, max a' Q i (x t+1 ,a') is the maximum Q value in the future state, x t+1 For status; S25. Based on the dynamic adjustment mechanism, a multi-level risk assessment model is generated. Through a comprehensive assessment of the material demand, transportation capacity, and time constraints in the disaster area, potential resource bottlenecks and risk points are predicted. The feedback adjustment strategy of the game model is combined to form an optimization goal.

3. The disaster emergency material distribution method based on multi-agent collaborative optimization according to claim 1 is characterized in that: The S3 specifically includes: S31. Obtain real-time data on the disaster area, including historical material needs, traffic conditions, weather information, and geographic information; S32. Use the spatiotemporal convolutional neural network to analyze the spatiotemporal characteristics of the disaster area. The spatial convolution layer extracts spatial features, and the temporal convolution layer captures temporal features to generate a spatiotemporal feature expression of the disaster area's needs: X spatial =Conv(X input ,W1); Among them, X input is the input data, W1 is the spatial convolution kernel, X spatial It is the spatial feature extracted by convolution operation, and Conv is the convolution operation; S33. Combine the graph attention network to model the geographical information and transportation network of the disaster area, extract the heterogeneous features of the graph data, and use the self-attention mechanism to perform weighted learning on different adjacent nodes of the graph data to generate a graph representation of the disaster area's transportation and geographical features: X geo =GAT(X spatial ,A,W2); Among them, A is the adjacency matrix of the graph, X geo represents the geographical and traffic features extracted by graph convolution, W2 is the weight matrix in the graph attention network, and GAT is the graph attention network; S34. By combining spatiotemporal convolutional neural networks with graph attention networks, spatiotemporal and graph features are integrated to generate a forecast of material demand in the disaster area. S35. Combined with the adaptive weighting mechanism, the temporal and spatial characteristics and geographical and traffic characteristics of the disaster area's material demand are weighted, the impact of different characteristics on the prediction results is adjusted, and a dynamic demand prediction model for the disaster area is generated: Among them, ω1 and ω2 are adaptive weighting factors, W3 and W4 are weight matrices for spatiotemporal and geographical features, and X temporal is the temporal feature extracted by spatiotemporal convolution, The material demand forecast value.

4. The disaster emergency material distribution method based on multi-agent collaborative optimization according to claim 1 is characterized in that: The S4 specifically includes: S41. Based on the generated dynamic demand forecast model for the disaster area, obtain the material demand in each area of the disaster area and construct a preliminary material allocation objective function F: Among them, c ij represents the transportation cost from warehouse i to destination j, x ij is the number of materials allocated, m is the number of warehouses, n is the number of destinations, λ i is the Lagrange multiplier, s i is the storage capacity of warehouse i, min is the minimum value operation; S42. Based on the traffic conditions, infrastructure damage, and weather conditions in the disaster area, combined with spatiotemporal feature data, establish constraints on the material transportation routes. S43. Use multi-objective reinforcement learning algorithm to globally optimize material distribution paths and resource scheduling: Q i (x t ,a t )=r t +γ·max a' Q i (x t+1 ,a); Among them, Q i (x t ,a t ) is the i-th subject in state x t Next select action a t The Q value, γ is the discount factor, a' is the future action selection, r t is the immediate reward, max is the maximum value operation, x t+1 For status; S44. Add Kalman filtering to the particle swarm optimization algorithm for dynamic path adjustment, reduce prediction errors through state estimation, and update the material distribution path in real time: P t|t =(I-K t H t )P t|t-1 ; in, is the estimated state, K t is the Kalman gain, z t is the measured value, P t|t-1 is the previous estimated covariance, H t is the observation matrix, I is the identity matrix, is the state estimate before time t, P t|t is the updated covariance matrix; S45. Based on the material allocation plan, adjust the resource configuration and allocation strategy in real time, and replan the path through the scheduling optimization algorithm.

5. The disaster emergency material distribution method based on multi-agent collaborative optimization according to claim 1 is characterized in that: The S5 specifically includes: S51. Based on the generated preliminary material allocation plan, obtain real-time data feedback from the disaster area, including traffic information, weather changes, and resource inventory, establish a real-time adjustment model, and combine different data sources using a weighted summation method to generate adjustment factors for resource demand and material dispatch in the disaster area: D adjusted =τ1·D traffic +τ2·D weather +τ3·D inventory ; Among them, D adjusted is the adjusted resource demand vector of the disaster area, D traffic 、D weather and D inventory They represent the traffic information, weather changes and resource inventory data of the disaster area respectively, and τ1, τ2 and τ3 are weighting coefficients; S52. Based on the real-time adjustment model and the feedback data from the disaster area, a dynamic scheduling objective function D is established. f : Among them, c ij is the transportation cost from warehouse i to destination j, t ij Due to delivery delay, u ij is the resource utilization rate, α1, α2 and α3 are weighted coefficients, x ij is the number of materials allocated, m is the number of warehouses, and n is the number of destinations; S53. Use the policy gradient algorithm in deep reinforcement learning to optimize the material dispatch plan by updating the policy function, learn the optimal dispatch strategy, and dynamically adjust the dispatch strategy based on real-time feedback from the disaster area to optimize material distribution efficiency and resource utilization: in, is the policy gradient, E is the expected operation, π θ (a t |x t ) is the current strategy, a t is the action selected at time t, x t is the state of the system at time t, V(x t+1 ) is the value of the next state, V(x t is the value of the current state, r t is the immediate reward, γ is the discount factor, ln is the logarithmic function, θ is the parameter of the strategy, x t+1 For the next state; S54. Based on the optimization results of the policy gradient algorithm, adjust the material scheduling plan, and automatically adjust the distribution path based on the generated scheduling strategy with real-time feedback; S55. Re-plan the material distribution and dispatch plan in the disaster area based on the material dispatch strategy.

6. The disaster emergency material distribution method based on multi-agent collaborative optimization according to claim 1 is characterized in that: The S6 specifically includes: S61. Based on the generated material dispatch strategy and real-time disaster area demand forecast data, obtain the risk assessment data of each area in the disaster area. Through the multi-level risk assessment model, integrate the risk factors at different levels, and use the fuzzy logic reasoning method to obtain the fuzziness of each risk factor, and finally obtain the comprehensive risk assessment value R total As input for material dispatch: Among them, μ1, μ2, μ3 and μ4 are weighting coefficients, and are the transportation, weather, safety and inventory risks of the i-th region in the disaster area, respectively, and n is the number of regions in the disaster area; S62. Based on comprehensive risk assessment, a multi-objective genetic algorithm is used to optimize material distribution routes. A fitness function is constructed to evaluate the pros and cons of various material distribution schemes. The fitness function comprehensively considers transportation costs, risk assessment, and resource utilization efficiency: Among them, F i represents the fitness of the i-th material allocation plan, c ik is the transportation cost from warehouse i to destination k, x ik is the quantity of materials allocated from warehouse i to destination k, u ik is the resource utilization rate, ρ1, ρ2 and ρ3 are weighting coefficients; S63. Use particle swarm optimization algorithm to perform global optimization and adjust the delivery path: in, represents the velocity of particle i in the kth generation, represents the position of particle i in the kth generation, is the individual best historical position of particle i, gbest k is the historical optimal position of all particles, ω is the inertia weight, c1 and c2 are acceleration constants, r1 and r2 are random factors, is the updated particle velocity, is the updated particle position; S64. By optimizing material distribution routes, adjusting scheduling strategies in real time, and replanning distribution routes, comprehensively considering the impact of traffic and weather changes in the disaster area, we can maximize resource utilization and efficiency. S65. Based on the optimization results, the material distribution plan is adjusted through the adaptive path correction mechanism, and the material distribution path is replanned in combination with real-time disaster changes.

Images