An emergency resource intelligent scheduling optimization method and system based on large model analysis

Abstract

This invention provides an intelligent scheduling and optimization method and system for emergency resources based on large-scale model analysis, belonging to the field of emergency resource scheduling technology. The method includes collecting emergency event data and environmental data; extracting emergency event features and environmental features, constructing a prediction model based on a long short-term memory network, and obtaining a list of required emergency resources; constructing an emergency resource evaluation model based on the analytic hierarchy process (AHP) to obtain priority scores for the required emergency resources; constructing a multi-objective optimization model and solving the multi-objective optimization model using a particle swarm optimization algorithm to generate a scheduling scheme; and constructing a decision function and solving it using a genetic algorithm to obtain the scheduling decision. This invention, through accurate prediction, scientific evaluation, multi-objective optimization, and priority-based decision-making, can significantly improve the accuracy, efficiency, and resource utilization efficiency of emergency resource scheduling, providing strong support for emergency management work and possessing significant practical significance and application value.

Description

Technical Field

[0001] This invention relates to the field of emergency resource scheduling technology, and in particular to an intelligent emergency resource scheduling optimization method and an intelligent emergency resource scheduling optimization system based on large model analysis. Background Technology

[0002] Emergency resource dispatch refers to the rational allocation and deployment of various emergency resources (such as personnel, materials, equipment, and transportation) in the event of a sudden incident or emergency, to ensure a rapid and efficient response to and handling of accidents or disasters. Its core objective is to optimize resource allocation, shorten response time, and minimize losses and risks.

[0003] Current scheduling methods primarily focus on the spatial layout, transportation routes, and time efficiency of resources, neglecting the different priorities arising from the critical roles resources play in emergencies. This means that scheduling systems cannot dynamically adjust priorities based on resource importance; they mechanically allocate resources according to quantity and location, lacking a scientific assessment and prioritization of "which resources are most critical" and "which can be postponed." This deficiency leads to a situation where, in emergencies, limited transportation and scheduling capabilities cannot be concentrated on ensuring the most critical resources, reducing overall emergency response efficiency and effectiveness. Summary of the Invention

[0004] This invention provides an intelligent scheduling and optimization method and system for emergency resources based on large model analysis, in order to overcome the deficiencies in the existing technology.

[0005] On the one hand, this invention provides an intelligent scheduling and optimization method for emergency resources based on large-scale model analysis, including:

[0006] Collect emergency event data and environmental data and preprocess them to obtain preprocessed data.

[0007] Emergency event features and environmental features are extracted from preprocessed data. A prediction model based on long short-term memory network is constructed. The emergency event features and environmental features are input, and the required emergency resource list is output.

[0008] Construct an emergency resource assessment model based on the analytic hierarchy process (AHP) to obtain priority scores for required emergency resources.

[0009] Collect emergency resource data for each resource in the required emergency resource list, construct a multi-objective optimization model based on the emergency resource data, emergency event data, and environmental data, and use the particle swarm optimization algorithm to solve the multi-objective optimization model to generate a scheduling scheme.

[0010] Based on the scheduling scheme, a decision function is constructed by combining priority scores, and a genetic algorithm is used to solve it to obtain the scheduling decision.

[0011] According to the present invention, an intelligent scheduling and optimization method for emergency resources based on large-scale model analysis is provided. Emergency event data includes event type, event time, and event scope. Emergency resource data includes resource type, resource quantity, and resource distribution. Environmental data includes geographic information and meteorological information.

[0012] According to the present invention, an intelligent scheduling and optimization method for emergency resources based on large model analysis includes the following process for extracting emergency event features and environmental features from preprocessed data:

[0013] The equal-width discretization method is used to transform continuous data in the preprocessed data into discrete categorical data. The transformation process includes:

[0014] The continuous data in the preprocessed data is divided into k intervals, and the interval width and the boundary of each interval are calculated to obtain discrete categorical data.

[0015] Using an information gain-based feature selection method, the information gain of each discrete category of data for emergency events and environmental classification is calculated.

[0016] Category data with information gain greater than a preset threshold are selected as the final emergency event features and environmental features.

[0017] According to the present invention, an intelligent scheduling and optimization method for emergency resources based on large model analysis includes the following process for constructing a prediction model based on a long short-term memory network:

[0018] Collect historical emergency event data and historical environmental data.

[0019] Historical emergency event data and historical environmental data are preprocessed to extract historical emergency event characteristics and historical environmental characteristics, and a list of historical emergency resources is annotated.

[0020] A Long Short-Term Memory (LSTM) network structure is established to obtain the basic model, which includes an input layer, hidden layers, and an output layer. The input layer receives historical emergency event features and historical environmental features. The hidden layer contains multiple LTM network units to capture long-term dependencies in the historical emergency event features and historical environmental features. The output layer outputs a predicted list of emergency resources.

[0021] The basic model is trained by taking the characteristics of historical emergency events and historical environment as inputs and the corresponding list of historical emergency resources as outputs.

[0022] The model parameters are updated using the stochastic gradient descent algorithm, and the mean squared error is used as the loss function. The model parameters that meet the preset accuracy are retained to obtain the prediction model.

[0023] According to the present invention, an intelligent scheduling and optimization method for emergency resources based on large model analysis is provided. The process of obtaining the priority score of the required emergency resources includes:

[0024] The emergency resource assessment problem is decomposed into three layers: the objective layer, the criteria layer, and the alternative layer. The objective layer determines the priority of the required emergency resources. The criteria layer includes three criteria, corresponding to the importance, availability, and timeliness of the required emergency resources. The alternative layer represents each emergency resource in the list of required emergency resources.

[0025] For each criterion in the criterion layer, the relative importance of various resources in the scheme layer is compared to construct a judgment matrix.

[0026] The eigenvalue method is used to calculate the largest eigenvalue of the judgment matrix and its corresponding eigenvector, and the eigenvector is normalized to obtain the weight vector.

[0027] The weight of each criterion is multiplied by the weight of the corresponding resource in the scheme layer, and the sum is obtained to obtain the comprehensive priority score of each resource.

[0028] According to the present invention, an intelligent scheduling and optimization method for emergency resources based on large model analysis is provided. The process of constructing a multi-objective optimization model includes:

[0029] Define an objective function that includes minimizing resource transportation costs, minimizing resource response time, and maximizing the matching degree between resource supply and demand.

[0030] By defining the constraints, a multi-objective optimization model is obtained. These constraints include supply constraints, demand constraints, and non-negativity constraints. Supply constraints state that the amount of each resource transported from each supply point cannot exceed its supply quantity. Demand constraints state that the amount of each resource obtained by each demand point must satisfy its basic demand. Non-negativity constraints state that the transport quantity cannot be negative.

[0031] According to the present invention, an intelligent scheduling optimization method for emergency resources based on large model analysis is provided. The process of solving the multi-objective optimization model using the particle swarm optimization algorithm includes:

[0032] A group of particles is randomly initialized, and each particle represents a possible scheduling scheme.

[0033] For each particle, its fitness value is calculated based on the objective function of the multi-objective optimization model.

[0034] The algorithm updates the particle position and velocity, terminates when the preset maximum number of iterations is reached, and outputs the scheduling scheme corresponding to the globally optimal position as the approximate optimal solution of the multi-objective optimization model.

[0035] The emergency resource intelligent scheduling optimization method based on large model analysis provided by the present invention includes the following process for constructing a decision function by combining priority scoring:

[0036] Define priority weight coefficients, and based on the multi-objective optimization model, introduce priority weights to adjust the objective function, and construct decision functions: resource supply cost decision function, resource scheduling time decision function, and resource demand satisfaction degree decision function.

[0037] The decision function is solved using a genetic algorithm to obtain the optimized scheduling decision.

[0038] According to the present invention, an intelligent scheduling optimization method for emergency resources based on large model analysis includes the following process for solving the decision function using a genetic algorithm:

[0039] Using real-number encoding, the scheduling quantity of each resource in the scheduling scheme is used as the gene of the chromosome. A certain number of individuals are randomly generated in the feasible solution space to form an initial population, and each individual represents a possible solution of a decision function.

[0040] A weighted summation method is used to transform multiple decision functions into a single-objective fitness function.

[0041] Randomly select k individuals from the current population to form a tournament group, compare the fitness values ​​of the k individuals, and select the individual whose fitness reaches a preset threshold as the parent.

[0042] Randomly pair up the selected parent individuals. For each pair of parent individuals, randomly select a crossover point and exchange the genes of the two parent individuals after the crossover point to generate two offspring individuals.

[0043] Set a crossover probability, and perform a crossover operation for each pair of parent individuals according to the crossover probability.

[0044] For each individual in the offspring population, a predetermined mutation probability is used to determine whether to mutate. If the individual needs to mutate, one or more genes on the chromosome are randomly selected for mutation. For the selected gene, a random mutation value is generated according to a Gaussian distribution to update the gene value.

[0045] For each individual in the mutated offspring population, the fitness value is recalculated. The offspring population and the parent population are merged, and individuals are selected in descending order of fitness value to form the next generation population, keeping the population size unchanged.

[0046] When the maximum number of iterations is reached, the iteration stops, and the individual with the highest fitness in the population is output, which is the optimal scheduling scheme.

[0047] On the other hand, the present invention also provides an intelligent emergency resource scheduling and optimization system based on large model analysis, comprising:

[0048] The data acquisition module is used to collect emergency event data and environmental data and preprocess them to obtain preprocessed data.

[0049] The emergency resource prediction module is used to extract emergency event features and environmental features from preprocessed data, build a prediction model based on long short-term memory networks, input emergency event features and environmental features, and output a list of required emergency resources.

[0050] The resource priority assessment module is used to build an emergency resource assessment model based on the analytic hierarchy process (AHP) and obtain priority scores for the required emergency resources.

[0051] The scheduling scheme generation module is used to collect emergency resource data for each resource in the required emergency resource list, construct a multi-objective optimization model based on the emergency resource data, emergency event data, and environmental data, and solve the multi-objective optimization model using the particle swarm optimization algorithm to generate a scheduling scheme.

[0052] The scheduling decision generation module is used to construct a decision function based on the scheduling scheme and priority score, and then solve it using a genetic algorithm to obtain the scheduling decision.

[0053] This invention provides an intelligent emergency resource scheduling optimization method and system based on large-scale model analysis. Utilizing the sequence data processing capabilities of Long Short-Term Memory (LSTM) networks, it can capture complex patterns of emergency events and environmental factors changing over time, accurately predicting the required emergency resource list. This avoids over- or under-allocation of resources, providing a reliable foundation for subsequent scheduling work. An emergency resource evaluation model based on the Analytic Hierarchy Process (AHP) is constructed to comprehensively and scientifically determine the priority scores of required emergency resources, ensuring that, under limited resource conditions, the most important resources are prioritized, improving the effectiveness of emergency response. By constructing a multi-objective optimization model and using the particle swarm optimization algorithm, the optimal solution can be quickly found among numerous possible scheduling schemes. This not only reduces the cost of emergency resource scheduling but also shortens scheduling time and improves resource utilization efficiency. After generating scheduling schemes, a decision function is constructed based on priority scores and solved using a genetic algorithm, making the final scheduling decision more in line with actual needs and ensuring that important resources are prioritized. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0055] Figure 1This is a flowchart illustrating an intelligent emergency resource scheduling optimization method based on large model analysis provided in an embodiment of the present invention.

[0056] Figure 2 This is a schematic diagram of the structure of an emergency resource intelligent scheduling and optimization system based on large model analysis provided in an embodiment of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0058] The following is combined with Figures 1-2 This invention describes an intelligent scheduling and optimization method and system for emergency resources based on large model analysis.

[0059] Figure 1 This is a flowchart illustrating an intelligent scheduling and optimization method for emergency resources based on large model analysis, provided in an embodiment of the present invention.

[0060] like Figure 1 As shown in the embodiment of the present invention, an intelligent scheduling and optimization method and system for emergency resources based on large model analysis is provided. The executing entity can be an intelligent scheduling and optimization method for emergency resources based on large model analysis, and the method includes:

[0061] Collect emergency event data and environmental data and preprocess them to obtain preprocessed data.

[0062] Emergency event data includes event type, event time, and event scope. Emergency resource data includes resource type, resource quantity, and resource distribution.

[0063] Emergency event data and environmental data are collected extensively from various channels. Emergency event data may come from reports by emergency management departments, feedback from on-site rescue personnel, and records from monitoring equipment; environmental data can be obtained from meteorological departments, geographic information systems, transportation departments, etc.

[0064] The collected data from multiple sources is integrated and stored uniformly in a data warehouse or database. Since data formats and standards may differ across data sources, data format conversion and standardization are necessary to ensure data consistency and compatibility.

[0065] Check the data for missing values. Different methods can be used to handle missing values. If there are few missing values, records containing missing values ​​can be deleted directly. If there are many missing values, statistical measures such as the mean, median, and mode can be used to fill in the missing values, or interpolation methods can be used for estimation.

[0066] Identify outliers in the data. Outliers may be caused by data entry errors, sensor malfunctions, or other reasons. For outliers, you can choose to delete, correct, or smooth them out.

[0067] Check the data for duplicate records. If there are duplicate records, keep only one of them to avoid data redundancy interfering with subsequent analysis.

[0068] Transform the data into a standard normal distribution with a mean of 0 and a standard deviation of 1. Standardization eliminates dimensional differences between different features, making them comparable.

[0069] Emergency event features and environmental features are extracted from preprocessed data. A prediction model based on long short-term memory network is constructed. The emergency event features and environmental features are input, and the required emergency resource list is output.

[0070] Extract features relevant to emergency resource allocation from the preprocessed data. For example, extract features such as road congestion level and average vehicle speed from traffic data.

[0071] The process of extracting emergency event features and environmental features from preprocessed data includes:

[0072] The equal-width discretization method is used to transform continuous data in the preprocessed data into discrete categorical data. Continuous data includes specific timestamps in event time, and temperature and wind speed in meteorological information. The transformation process includes:

[0073] Let the continuous data in the preprocessed data be... Divide it into k intervals and calculate the interval width, as shown in the formula:

[0074]

[0075] Calculate the boundaries of each interval to obtain discrete categorical data, expressed by the formula:

[0076]

[0077] In the formula, This represents the maximum value in continuous data. This represents the minimum value in continuous data, i = 0, 1, ..., k.

[0078] Using an information gain-based feature selection method, the information gain of each discrete category of data for emergency events and environmental classification is calculated, as expressed by the formula:

[0079]

[0080] Where H(S) is the information entropy of dataset S, expressed by the formula:

[0081]

[0082] In the formula, Let m represent the probability of category i appearing in dataset S, and m represent the number of categories. Representing discrete categorical data A subset of values ​​v Representing a subset The number of samples, This represents the number of samples in dataset S.

[0083] By calculating the information gain of each category of data, the category data with information gain greater than a preset threshold are selected as the final emergency event features and environmental features.

[0084] For discrete categorical features, such as event type, resource type, and region category in geographic information, one-hot encoding is used to convert them into numerical vectors. Suppose a categorical feature has n distinct values; each value is represented as an n-dimensional vector, where only the position corresponding to that value is 1, and the rest are 0.

[0085] The process of constructing a prediction model based on a long short-term memory network includes:

[0086] Collect historical emergency event data and historical environmental data.

[0087] Historical emergency event data and historical environmental data are preprocessed to extract historical emergency event characteristics and historical environmental characteristics, and a list of historical emergency resources is annotated.

[0088] A Long Short-Term Memory (LSTM) network structure is established to obtain the basic model, which includes an input layer, hidden layers, and an output layer. The input layer receives historical emergency event features and historical environmental features. The hidden layer contains multiple LTM network units to capture long-term dependencies in the historical emergency event features and historical environmental features. The output layer outputs a predicted list of emergency resources.

[0089] The basic model is trained by taking the characteristics of historical emergency events and historical environment as inputs and the corresponding list of historical emergency resources as outputs.

[0090] The stochastic gradient descent algorithm is used to update the model parameters, and the mean squared error is used as the loss function. Model parameters that meet the preset accuracy are retained to obtain the prediction model. The mean squared error formula is expressed as:

[0091]

[0092] In the formula, This indicates a predicted list of emergency resources. This represents a list of historical emergency resources, where n represents the number of samples.

[0093] Taking flood events as an example, obtain records of similar flood events that have occurred in the past 10 years, including event type (flood), scale (affected area, water level), time of occurrence, duration, and affected population.

[0094] Collect historical environmental data: Collect relevant environmental factors, including rainfall, temperature, humidity, topographic data (such as river slope and city elevation), and infrastructure information (such as drainage system status).

[0095] For each historical event, a list of resources actually used is labeled, which serves as the output label for supervised learning.

[0096] Before training the model, the emergency management department compiled data on 20 historical flood events. Each event included 10 dimensions of emergency event characteristics and 8 dimensions of environmental characteristics, as well as a list of 4 types of resources (tents, food, medical teams, and rescue boats) to form a training dataset.

[0097] The collected raw data is cleaned (e.g., missing values ​​are handled by filling missing rainfall data with the mean) and normalized (e.g., the affected area, population, etc. are scaled to the [0, 1] interval to ensure data comparability). For text-based data (e.g., event types), it is converted into numerical encoding.

[0098] Key features are extracted from the preprocessed data, including event scale (affected area, affected population), severity (peak water level, event duration) and dynamic changes (time-series data of the event, such as hourly water level changes).

[0099] Extract environmental factors that influence resource demand, such as meteorological conditions (average rainfall, temperature variation), geographical conditions (topographic complexity, altitude), and external factors, such as seasonal information.

[0100] The extracted historical features include: emergency event features (affected area, affected population, peak water level, duration) and environmental features (average rainfall, temperature, terrain complexity index). These features are combined into an input vector, which corresponds to an labeled output resource list.

[0101] Define a basic LSTM model, including:

[0102] Input layer: Receives historical emergency event characteristics and historical environmental characteristics (18 input dimensions, including event scale, environmental factors, etc.).

[0103] Hidden layer: Contains 128 LSTM units (stacked in two layers) to capture long-term dependencies between features (such as the lagged impact of the duration of a flood event on resource demand).

[0104] Output layer: Outputs a predicted list of emergency resources (a 4-dimensional vector, corresponding to the number of tents, food, medical teams, and rescue boats, respectively).

[0105] Input feature vectors for 20 historical events (each event corresponds to a time step sequence, such as from 24 hours before the event to the end of the event).

[0106] Output a list of historical emergency resources with annotations.

[0107] The stochastic gradient descent optimizer was used with a learning rate of 0.001; the mean squared error loss function was used to calculate the difference between the predicted resource list and the actual resource list.

[0108] The model is trained iteratively for 1000 rounds, retaining model parameters with an MSE lower than 0.05 (preset precision), and finally obtaining the trained prediction model.

[0109] After training the model based on historical data, the model learned a pattern: when rainfall exceeds a certain threshold and affects a larger population, the demand for medical teams increases significantly (because floods easily trigger diseases). The model, through the time-series processing capabilities of LSTM, captured the dynamic changes in resource demand as events unfold (e.g., a sharp increase in food demand 24 hours after the water level peak).

[0110] When a flood occurs in city B, collect current emergency event data (affected area, affected population, peak water level) and environmental data (average rainfall, temperature, topographic complexity).

[0111] Perform the same preprocessing (cleaning and normalization) on the new data to extract feature vectors (e.g., the input sequence includes hourly data from the start of the event to the current time).

[0112] The new feature vectors are input into the trained LSTM prediction model. The model uses LSTM units in the hidden layers to analyze long-term dependencies between features (such as the relationship between heavy rainfall and subsequent medical needs) and outputs a predicted list of emergency resources.

[0113] For a flood event in City B, the model outputs the following list of required emergency resources: P4 tents, P3 food packages, P1 medical team, and P2 rescue boats. Higher water levels and population impact increase the demand for tents and medical resources, while continuous rainfall increases the demand for rescue boats (due to complex terrain).

[0114] An emergency resource assessment model based on the analytic hierarchy process (AHP) is constructed to obtain priority scores for required emergency resources. The process includes:

[0115] The emergency resource assessment problem is decomposed into three layers: the objective layer, the criteria layer, and the alternative layer. The objective layer determines the priority of the required emergency resources. The criteria layer includes three criteria, corresponding to the importance, availability, and timeliness of the required emergency resources. The alternative layer represents each emergency resource in the list of required emergency resources.

[0116] For each criterion in the criterion layer, the relative importance of various resources in the scheme layer is compared to construct a judgment matrix. The formula is expressed as:

[0117]

[0118] In the formula, The value of resource i is relative to that of resource j, and z represents the type of resource, with values ​​ranging from 1 to 9 and their reciprocals.

[0119] The eigenvalue method is used to calculate the largest eigenvalue and its corresponding eigenvector of the judgment matrix A. The eigenvector is then normalized to obtain the weight vector, expressed by the following formula:

[0120]

[0121] In the formula, and Let i and j represent the i-th and j-th eigenvectors, respectively.

[0122] Multiply the weight of each criterion by the weight of the corresponding resource in the scheme layer, and sum them to obtain the comprehensive priority score for each resource. Let the weight vector of the criterion layer be... The weight matrix for each criterion in the scheme layer is as follows: The comprehensive priority scoring formula for each resource is expressed as:

[0123]

[0124] In the formula, This represents the priority score of the i-th resource. This represents the weight vector of the k-th criterion layer. Let represent the weight matrix for the i-th criterion.

[0125] After the predictive model outputs the list of required emergency resources, the emergency response center needs to prioritize the four types of resources. The goal is to prioritize the allocation of critical resources when resources are limited (such as insufficient transport capacity).

[0126] The key challenge is that transporting all resources simultaneously would exceed the current capacity of available transport vehicles (e.g., only 70% of the resources can be transported), so priorities must be prioritized.

[0127] Priority scoring process: based on the analytic hierarchy process.

[0128] The process is completed in four steps, including establishing a hierarchical structure, constructing a judgment matrix, calculating weights, and synthesizing priority scores.

[0129] Establishing a hierarchical structure includes:

[0130] At the target level, the priority of resource scheduling (highest target) is determined.

[0131] At the criteria level, there are three assessment criteria: Importance (C1): The direct impact of resources on the survival / safety of disaster victims (e.g., medical teams saving lives); Availability (C2): Current inventory and difficulty of acquisition (e.g., whether there are reserves in local warehouses); Timeliness (C3): Whether resources need to be delivered within the golden time frame (e.g., rescue boats need to arrive within 6 hours).

[0132] At the solution level, there are four types of resources to be evaluated: R1: medical team (P1 team), R2: rescue boat (P2 boats), R3: food packs (P3 sets), and R4: tents (P4 tents).

[0133] A judgment matrix was constructed, and the emergency expert team compared the criteria and resources pairwise, assigning values ​​according to a scale of 1-9 (Table 1).

[0134] Table 1: Meaning of scale 1-9 in chromatography:

[0135]

[0136] Criterion-based comparison (objective: resource priority):

[0137] Importance (C1) has the most direct impact on survival (e.g., lack of medical teams can lead to death), and is significantly more important than timeliness (C3) (scale 5).

[0138] Timeliness (C3) is slightly more important than availability (C2) (scale 3);

[0139] Generate the criterion-level judgment matrix A:

[0140]

[0141] Among them, the main diagonal = 1 (self-comparison), C1 vs C2 = 5 (importance is significantly more important than availability), and C3 vs C2 = 3 (timeliness is slightly more important than availability).

[0142] Resource layer comparison (for each criterion), a. Criterion: Importance (C1):

[0143] The medical team (R1) is the most critical to saving lives, significantly more important than the rescue boat (R2) (scale 5); the medical team (R1) is strongly more important than food (R3) (scale 7); the rescue boat (R2) is slightly more important than the tent (R4) (scale 3).

[0144] Generating matrix B1:

[0145]

[0146] b. Criteria: Availability (C2), Data Support:

[0147] The local warehouse has ample food packages (high availability, low cost), while rescue boats need to be transported from elsewhere (low availability, high cost).

[0148] Based on the data:

[0149] Food (R3) has the highest availability and is significantly more important than medical teams (R1) (scale 5). Tents (R4) are slightly more important than rescue boats (R2) (scale 3). Generating matrix B2:

[0150]

[0151] c. Criteria: Timeliness (C3), Current flood situation: Water level is still rising, and trapped people need to be evacuated within 6 hours. Rescue boats (R2) need to be in place immediately, slightly more important than medical teams (R1) (scale 3). Food (R3) can be delivered 12 hours later, equally important than tents (R4) (scale 1).

[0152] Generating matrix B3:

[0153]

[0154] Calculate the weights of each layer and use the eigenvalue method to solve for the weight vector.

[0155] Criterion layer weight W c Calculate the largest eigenvalue λ for matrix A. max ≈3.085, corresponding to the normalized eigenvector:

[0156]

[0157] Importance (63.7%) > Timeliness (25.8%) > Availability (10.5%).

[0158] Resource layer weights (for each criterion), importance criterion (C1) weight W1:

[0159]

[0160] That is, medical teams (60%) are the highest priority, followed by rescue boats (21.7%).

[0161] Availability Criterion (C2) Weight W2:

[0162]

[0163] That is, food (48.9%) is the easiest to obtain, while rescue boats (13.2%) are the most difficult to transport.

[0164] Timeliness criterion (C3) weight W3:

[0165]

[0166] That is, rescue boats (51.1%) need to be delivered as quickly as possible, while tents (12.8%) can be delayed.

[0167] Calculate the overall priority score for each resource based on the priority score synthesis formula.

[0168] Table 2: Comprehensive Priority Score for Emergency Resources

[0169]

[0170] Final score ranking: Medical team (0.468) > Rescue boat (0.284) > Food pack (0.145) > Tent (0.102).

[0171] Collect emergency resource data for each resource in the required emergency resource list, construct a multi-objective optimization model based on the emergency resource data, emergency event data, and environmental data, and use the particle swarm optimization algorithm to solve the multi-objective optimization model to generate a scheduling scheme.

[0172] Environmental data includes geographic information and meteorological information.

[0173] The process of constructing a multi-objective optimization model includes:

[0174] Define an objective function that includes minimizing resource transportation costs, minimizing resource response time, and maximizing resource supply-demand matching. Minimizing resource transportation costs includes:

[0175] Let the transportation cost of emergency resource i from supply point j to demand point k be... The quantity transported is The objective function for resource transportation costs is then expressed as:

[0176]

[0177] Minimizing resource response time includes:

[0178] Let the transportation time of emergency resource i from supply point j to demand point k be . The objective function for resource response time is then expressed as:

[0179]

[0180] Maximizing the matching degree of resource supply and demand includes:

[0181] Let the quantity of emergency resource i required by demand point k be... The supply quantity of emergency resource i from supply point j is The objective function for the resource supply and demand matching degree is expressed as:

[0182]

[0183] Define the constraints, including supply constraints, demand constraints, and non-negativity constraints. The supply constraint is expressed as:

[0184]

[0185] This indicates that the amount of each resource transported from each supply point does not exceed its supply.

[0186] The demand constraint is expressed as:

[0187]

[0188] This means that the amount of each type of resource obtained at each demand point must meet its basic requirements.

[0189] Nonnegativity constraints are expressed as:

[0190]

[0191] This indicates that the quantity of goods transported cannot be negative.

[0192] In this embodiment, city B is divided into severely affected, moderately affected, and lightly affected areas based on the severity of the flood disaster. Resource needs are matched according to the specific disaster situation, for example:

[0193] D1: Severely Affected Area (requires 60% resources), including 15 medical teams, 36 rescue boats, 9,000 food packages, and 1,800 tents. D2: Moderately Affected Area (requires 30% resources), including 7 medical teams, 18 rescue boats, 4,500 food packages, and 900 tents. D3: Lightly Affected Area (requires 10% resources), including 3 medical teams, 6 rescue boats, 1,500 food packages, and 300 tents.

[0194] Obtain information on supply points that can provide the necessary emergency resources, including:

[0195] S1: Local warehouse (limited supply), including 10 medical teams, 20 rescue boats, 5,000 food packages, and 1,000 tents. S2: Neighboring city warehouse (replenishable), including 20 medical teams, 40 rescue boats, 10,000 food packages, and 2,000 tents.

[0196] The process of solving a multi-objective optimization model using the particle swarm optimization algorithm includes:

[0197] A randomized group of particles is initialized, with each particle representing a possible scheduling scheme. Let the position vector of the particle be... Let represent the position of the i-th particle in the d-dimensional search space, where This represents the value of the i-th particle in the j-th dimension, corresponding to the decision variable in the multi-objective optimization model. Initialize the particle velocity vector .

[0198] For each particle, its fitness value is calculated based on the objective function of the multi-objective optimization model.

[0199] Update particle position and velocity; the velocity update formula is expressed as:

[0200]

[0201] In the formula, Indicates the current speed. R represents the updated velocity, and R represents the inertia weight. and Represents the acceleration constant. and It is a random number between [0,1]. Let represent the optimal position of the i-th particle. This indicates the globally optimal position.

[0202] The position update formula is expressed as:

[0203]

[0204] In the formula, This represents the current position of particle i. This represents the updated position of particle i.

[0205] The algorithm continuously updates the particle position and velocity, terminates when the preset maximum number of iterations is reached, and outputs the scheduling scheme corresponding to the globally optimal position as the approximate optimal solution of the multi-objective optimization model.

[0206] Based on the emergency resource needs and supply points available for these resources during the flood event in City B, a multi-objective optimization model was constructed and solved using the particle swarm optimization algorithm. The resulting scheduling plan is as follows: Medical Teams: S1 to D1: 8 teams, transportation cost 1600 yuan, time 3 hours. S2 to D1: 7 teams, transportation cost 2800 yuan, time 6 hours. S1 to D2: 2 teams, transportation cost 300 yuan, time 2 hours. S2 to D2: 5 teams, transportation cost 1500 yuan, time 4 hours. S2 to D3: 3 teams, transportation cost 600 yuan, time 3 hours.

[0207] Rescue boats: S1 to D1, 15 boats, transportation cost 7500 yuan, time required 3 hours. S2 to D1, 21 boats, transportation cost 16800 yuan, time required 6 hours. S2 to D2, 18 boats, transportation cost 10800 yuan, time required 4 hours. S2 to D3, 6 boats, transportation cost 2400 yuan, time required 3 hours.

[0208] Food packages: S1 to D2, 4000 packages, transportation cost 32000 yuan, time required 2 hours. S2 to D1, 9000 packages, transportation cost 135000 yuan, time required 6 hours. S2 to D3, 1500 packages, transportation cost 12000 yuan, time required 3 hours.

[0209] Tents: From S1 to D1, 1000 tents to be transported, costing 100,000 yuan, taking 3 hours. From S2 to D3, 300 tents to be transported, costing 24,000 yuan, taking 3 hours.

[0210] Based on the scheduling scheme, a decision function is constructed by combining priority scores, and a genetic algorithm is used to solve it to obtain the scheduling decision.

[0211] The process of constructing a decision function by combining priority scores includes:

[0212] The priority weight coefficient is defined by the following formula:

[0213]

[0214] In the formula, z represents the type of resource. This represents the priority score of the j-th resource in the scheduling scheme.

[0215] Based on the multi-objective optimization model, priority weights are introduced to adjust the objective function, and a decision function is constructed, including:

[0216] The resource supply cost decision function is expressed as follows:

[0217]

[0218] The weighting factor for tent transportation costs is only 0.102, reducing its cost focus. The weighting factor for medical team costs is 0.468, strengthening its cost control.

[0219] The resource scheduling time decision function is expressed as follows:

[0220]

[0221] A 1-hour delay for the medical team incurs a penalty of 0.468 × penalty value. A 1-hour delay for the tent incurs a penalty of 0.102 × penalty value, with a higher tolerance.

[0222] The decision function for the degree to which resources meet demand is expressed by the formula:

[0223]

[0224] min(⋅) ensures that over-delivery does not increase the score, focusing on the basic needs satisfaction rate. The medical team's satisfaction rate is 95%, contributing 0.468×0.95; the tent satisfaction rate is 30%, contributing only 0.102×0.30.

[0225] In the formula, This represents the transportation cost of emergency resource i from supply point j to demand point k. This represents the number of times the i-th type of resource is scheduled in the scheduling scheme. This represents the optimized resource supply cost decision function. This represents the optimized resource scheduling time decision function. This represents the transportation time of emergency resource i from supply point j to demand point k. This represents the quantity of emergency resource i required by demand point k. This represents the decision function for the degree to which the optimized resources meet the demand, where... Indicates taking and The minimum value in.

[0226] The decision function is solved using a genetic algorithm to obtain the optimized scheduling decision.

[0227] The process of solving the decision function using a genetic algorithm includes:

[0228] Using real-number encoding, the scheduling quantity of each resource in the scheduling scheme is used as the gene of the chromosome. A certain number of individuals are randomly generated in the feasible solution space to form an initial population, and each individual represents a possible solution of a decision function.

[0229] The weighted summation method is used to transform multiple decision functions into a single-objective fitness function, expressed by the following formula:

[0230]

[0231] In the formula, , , It is the weighting coefficient.

[0232] Randomly select k individuals from the current population to form a tournament group, compare the fitness values ​​of the k individuals, and select the individual whose fitness reaches a preset threshold as the parent.

[0233] Randomly pair up the selected parent individuals. For each pair of parent individuals, randomly select a crossover point and exchange the genes of the two parent individuals after the crossover point to generate two offspring individuals.

[0234] Set a crossover probability, and perform a crossover operation for each pair of parent individuals according to the crossover probability.

[0235] For each individual in the offspring population, a predetermined mutation probability is used to determine whether to mutate. If the individual needs to mutate, one or more genes on the chromosome are randomly selected for mutation. For the selected gene, a random mutation value is generated according to a Gaussian distribution to update the gene value.

[0236] For each individual in the mutated offspring population, the fitness value is recalculated. The offspring population and the parent population are merged, and individuals are selected in descending order of fitness value to form the next generation population, keeping the population size unchanged.

[0237] When the maximum number of iterations is reached, the iteration stops, and the individual with the highest fitness in the population is output, which is the optimal scheduling scheme.

[0238] In this embodiment, a decision function is constructed by combining priority scoring and solved using a genetic algorithm. The final scheduling decision is as follows: Medical Teams: The original 8 teams transported from S1 to D1 are increased to 10 teams, seizing local resources in S1 to reduce costs and improve efficiency. The original 7 teams transported from S2 to D1 are reduced to 5 teams, freeing up capacity for rescue boats. Rescue Boats: 8 new rescue boats are added from S1 to D2 to ensure timely delivery to disaster areas. Food Packages: The transport volume from S2 to D3 is reduced from 1500 to 1000 packages. Tents: The transport volume from S1 to D1 is changed from 1000 to 0 tents. The rest remain unchanged.

[0239] The optimized scheduling decision reduced the average delivery time of medical teams from 4.2 hours to 3.1 hours, a 27% reduction. Total transportation costs decreased from RMB 225,400 to RMB 201,500, an 11% reduction.

[0240] In summary, this embodiment provides an intelligent emergency resource scheduling optimization method based on large-scale model analysis. It overcomes the limitations of static experience by using a prediction model based on an LSTM network. By analyzing disaster evolution patterns (such as the nonlinear correlation between flood level rise rate and medical resource demand), environmental coupling factors (the potential impact of geological conditions and the wear coefficient of rescue equipment), and historical response data, it constructs a multi-dimensional spatiotemporal mapping of resource demand. Compared to traditional manual estimation, this significantly reduces resource waste caused by misjudgments (such as excessive allocation of tents crowding out life-saving equipment capacity).

[0241] By transforming the importance of resources into a computable priority scoring system through the Analytic Hierarchy Process (AHP), a hierarchical decision-making benchmark is constructed under complex constraints, forcing the scheduling model to prioritize lifeline resources and avoiding strategic errors caused by human emotional bias.

[0242] Particle swarm optimization (PSO) discovers Pareto optimal solutions in a multi-dimensional objective (cost / time / matching degree) search space that traditional scheduling methods cannot reach. When a genetic algorithm is incorporated with resource weight coefficients, in crisis scenarios with capacity shortages, the system automatically reduces low-priority resources to ensure 100% delivery of medical equipment; the priority weight coefficients inject the results of hierarchical analysis into the cost function, enabling high-value resources to receive excess capacity guarantees.

[0243] The priority scoring system and objective function transformation process in the plan construct a traceable, verifiable, and explainable decision-making chain. Auditing departments can verify the scientific validity of the scheduling plan by replaying the genetic algorithm iteration process, significantly enhancing the credibility of emergency management.

[0244] Based on the same general inventive concept, this invention also protects an intelligent emergency resource scheduling and optimization system based on large model analysis. The following describes an intelligent emergency resource scheduling and optimization system based on large model analysis provided by this invention. The intelligent emergency resource scheduling and optimization system based on large model analysis described below and the intelligent emergency resource scheduling and optimization method based on large model analysis described above can be referred to and correspond to each other.

[0245] Figure 2 This is a schematic diagram of the structure of an emergency resource intelligent scheduling and optimization system based on large model analysis provided in an embodiment of the present invention.

[0246] like Figure 2As shown, an emergency resource intelligent scheduling and optimization system based on large model analysis includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. The processor includes a data acquisition module, an emergency resource prediction module, a resource priority evaluation module, a scheduling scheme generation module, and a scheduling decision generation module.

[0247] The data acquisition module is used to collect emergency event data and environmental data and preprocess them to obtain preprocessed data.

[0248] The emergency resource prediction module is used to extract emergency event features and environmental features from preprocessed data, build a prediction model based on long short-term memory networks, input emergency event features and environmental features, and output a list of required emergency resources.

[0249] The resource priority assessment module is used to build an emergency resource assessment model based on the analytic hierarchy process (AHP) and obtain priority scores for the required emergency resources.

[0250] The scheduling scheme generation module collects emergency resource data for each resource in the required emergency resource list, constructs a multi-objective optimization model based on the emergency resource data, emergency event data, and environmental data, and uses the particle swarm optimization algorithm to solve the multi-objective optimization model to generate a scheduling scheme.

[0251] The scheduling decision generation module is used to construct a decision function based on the scheduling scheme and priority score, and then solve it using a genetic algorithm to obtain the scheduling decision.

[0252] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0253] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent scheduling and optimization method for emergency resources based on large-scale model analysis, characterized in that, include: Collect emergency event data and environmental data and preprocess them to obtain preprocessed data; Emergency event features and environmental features are extracted from the preprocessed data, and a prediction model based on long short-term memory network is constructed. The emergency event features and environmental features are input, and a list of required emergency resources is output. Construct an emergency resource assessment model based on the analytic hierarchy process (AHP) to obtain priority scores for the required emergency resources; Collect emergency resource data for each resource in the required emergency resource list, construct a multi-objective optimization model based on the emergency resource data, the emergency event data, and the environmental data, and solve the multi-objective optimization model using the particle swarm optimization algorithm to generate a scheduling scheme; Based on the scheduling scheme, a decision function is constructed by combining the priority score, and a genetic algorithm is used to solve it to obtain the scheduling decision.

2. The emergency resource intelligent scheduling optimization method based on large model analysis according to claim 1, characterized in that, The emergency event data includes event type, event time, and event scope; the emergency resource data includes resource type, resource quantity, and resource distribution; and the environmental data includes geographic information and meteorological information.

3. The emergency resource intelligent scheduling optimization method based on large model analysis according to claim 1, characterized in that, The process of extracting emergency event features and environmental features from the preprocessed data includes: The continuous data in the preprocessed data is transformed into discrete categorical data using an equal-width discretization method. The transformation process includes: The continuous data in the preprocessed data is divided into k intervals, and the interval width and the boundary of each interval are calculated to obtain discrete category data; Using an information gain-based feature selection method, the information gain of each discrete category of data for emergency events and environmental classification is calculated. Category data with information gain greater than a preset threshold are selected as the final emergency event features and environmental features.

4. The emergency resource intelligent scheduling optimization method based on large model analysis according to claim 1, characterized in that, The process of constructing a prediction model based on a long short-term memory network includes: Collect historical emergency event data and historical environmental data; The historical emergency event data and the historical environmental data are preprocessed to extract historical emergency event features and historical environmental features, and a historical emergency resource list is annotated. A long short-term memory network structure is established to obtain a basic model, which includes an input layer, a hidden layer, and an output layer. The input layer is used to receive historical emergency event features and historical environment features. The hidden layer contains multiple long short-term memory network units to capture long-term dependencies in the historical emergency event features and historical environment features. The output layer is used to output a predicted list of emergency resources. The basic model is trained by taking the characteristics of historical emergency events and historical environment as inputs and the corresponding list of historical emergency resources as outputs. The model parameters are updated using the stochastic gradient descent algorithm, and the mean squared error is used as the loss function. The model parameters that meet the preset accuracy are retained to obtain the prediction model.

5. The emergency resource intelligent scheduling optimization method based on large model analysis according to claim 1, characterized in that, The process of obtaining the priority score of the required emergency resources includes: The emergency resource assessment problem is decomposed into an objective layer, a criterion layer, and a solution layer. The objective layer is used to determine the priority of the required emergency resources. The criterion layer includes three criteria, which correspond to the importance, availability, and timeliness factors of the required emergency resources, respectively. The solution layer is used to represent each emergency resource in the list of required emergency resources. For each criterion in the criterion layer, the relative importance of various resources in the scheme layer is compared, and a judgment matrix is ​​constructed. The eigenvalue method is used to calculate the largest eigenvalue of the judgment matrix and its corresponding eigenvector, and the eigenvector is normalized to obtain the weight vector. The weight of each criterion is multiplied by the weight of the corresponding resource in the scheme layer, and the sum is obtained to obtain the comprehensive priority score of each resource.

6. The emergency resource intelligent scheduling optimization method based on large model analysis according to claim 1, characterized in that, The process of constructing a multi-objective optimization model includes: Define an objective function, which includes minimizing resource transportation costs, minimizing resource response time, and maximizing the matching degree of resource supply and demand. By defining the constraints, a multi-objective optimization model is obtained. The constraints include supply constraints, demand constraints, and non-negativity constraints. The supply constraint states that the amount of each resource transported from each supply point does not exceed its supply amount. The demand constraint states that the amount of each resource obtained by each demand point must meet its basic demand. The non-negativity constraint states that the transport quantity cannot be negative.

7. The emergency resource intelligent scheduling optimization method based on large model analysis according to claim 1, characterized in that, The process of solving the multi-objective optimization model using the particle swarm optimization algorithm includes: A group of particles is randomly initialized, and each particle represents a possible scheduling scheme. For each particle, its fitness value is calculated based on the objective function of the multi-objective optimization model; The algorithm updates the particle position and velocity, terminates when the preset maximum number of iterations is reached, and outputs the scheduling scheme corresponding to the globally optimal position as the approximate optimal solution of the multi-objective optimization model.

8. The emergency resource intelligent scheduling optimization method based on large model analysis according to claim 1, characterized in that, The process of constructing a decision function based on the priority scores includes: Define priority weight coefficients, and based on the multi-objective optimization model, introduce priority weights to adjust the objective function, and construct decision functions: resource supply cost decision function, resource scheduling time decision function, and resource demand satisfaction degree decision function; The decision function is solved using a genetic algorithm to obtain the optimized scheduling decision.

9. The emergency resource intelligent scheduling optimization method based on large model analysis according to claim 8, characterized in that, The process of solving the decision function using a genetic algorithm includes: Using real-number encoding, the scheduling quantity of each resource in the scheduling scheme is used as the gene of the chromosome. A certain number of individuals are randomly generated in the feasible solution space to form an initial population. Each individual represents a possible solution of a decision function. A weighted summation method is used to transform multiple decision functions into a single-objective fitness function; Randomly select k individuals from the current population to form a tournament group, compare the fitness values ​​of the k individuals, and select the individual whose fitness reaches a preset threshold as the parent. Randomly pair up the selected parent individuals. For each pair of parent individuals, randomly select a crossover point and exchange the genes of the two parent individuals after the crossover point to generate two offspring individuals. Set the crossover probability, and perform crossover operation according to the crossover probability for each pair of parent individuals; For each individual in the offspring population, whether to mutate is determined by a preset mutation probability. If the individual needs to mutate, one or more genes on the chromosome are randomly selected for mutation. For the selected gene, a random mutation amount is generated according to a Gaussian distribution, and the gene value is updated. For each individual in the mutated offspring population, the fitness value is recalculated, the offspring population and the parent population are merged, and individuals are selected in descending order of fitness value to form the next generation population, while keeping the population size unchanged. When the maximum number of iterations is reached, the iteration stops, and the individual with the highest fitness in the population is output, which is the optimal scheduling scheme.

10. An intelligent emergency resource scheduling and optimization system based on large model analysis, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements an emergency resource intelligent scheduling optimization method based on large model analysis as described in any one of claims 1 to 9.

Images