Emergency field multi-dimensional inference knowledge graph construction and real-time response method

Abstract

The invention relates to an emergency field multi-dimensional inference knowledge graph construction and real-time response method, which comprises the following steps: defining entity attributes and semantic relationships of a disaster-bearing body layer, a disaster-inducing mechanism layer, an emergency resource layer and a disposal process layer, and constructing a multi-dimensional inference knowledge graph; performing adjacent matrix optimization on the multi-dimensional inference mapping knowledge domain by adopting a diffusion model; a disaster-bearing body entity and a disaster-inducing factor entity in the implementation intelligence are identified, spatial semantics are analyzed to GIS coordinates, and an event association strength index is generated in combination with historical cases; performing three-level propagation on the basis of the optimized knowledge graph by taking an entity of which the association strength index meets a threshold value as a seed node; and sorting and outputting according to the comprehensive risk values of the propagation paths, and generating a resource scheduling instruction. Structural defects of an existing system can be thoroughly overcome, and it is ensured that an emergency scheme is always matched with disaster evolution.

Description

Technical Field

[0001] This invention relates to the field of emergency response technology, and more specifically to a method for constructing and responding to a multidimensional reasoning knowledge graph in the emergency field. Background Technology

[0002] With the development of artificial intelligence technology, knowledge graphs for emergency scenarios play a crucial role in emergency management. Through the construction and application of knowledge graphs, functions such as rapid information retrieval, entity relationship analysis, risk prediction and assessment, and scenario simulation and deduction can be achieved, thus providing strong support for decision-makers. However, current knowledge graph systems in the field of emergency management have significant limitations when dealing with complex disaster scenarios:

[0003] Traditional disaster modeling methods only associate disaster events with geographic information through static knowledge graphs. This one-dimensional modeling leads to a serious lack of secondary disaster prediction. For example, when a chemical plant leak occurs, the existing system cannot automatically associate the risk of downstream water pollution along the underground pipeline topology.

[0004] Meanwhile, the construction of disaster propagation paths also has fundamental flaws. Existing simulation engines overemphasize geographical proximity propagation (such as flood spread) while neglecting the crucial impact of engineering dependency chains and social relationship networks. When a typhoon causes a power grid failure, the system cannot predict the derivative life risks caused by power outages in ICU wards because the dependence of hospitals on the power system has not been quantified.

[0005] Furthermore, in the response decision-making stage, resource scheduling relies heavily on human experience and lacks quantitative models.

[0006] Therefore, overcoming the above problems is an urgent issue that needs to be addressed by those skilled in the art. Summary of the Invention

[0007] In view of this, in order to at least partially solve the above-mentioned technical problems, improve the accuracy of decision-making, and enhance the timeliness of deduction, this invention discloses a method for constructing and responding to a multidimensional reasoning knowledge graph in the emergency domain.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] A method for constructing and responding in real time to multidimensional reasoning knowledge graphs in the emergency response domain, comprising the following steps:

[0010] Define the entity attributes and semantic relationships of the disaster-bearing body layer, disaster-causing mechanism layer, emergency resource layer, and response process layer, and construct a multi-dimensional reasoning knowledge graph;

[0011] The adjacency matrix of the multidimensional reasoning knowledge graph is optimized using a diffusion model.

[0012] Identify disaster-bearing entities and disaster-causing factor entities in the implementation intelligence, parse spatial semantics to GIS coordinates, and generate an event correlation strength index by combining historical cases;

[0013] Entities whose association strength index meets the threshold are used as seed nodes for three-level propagation based on the optimized knowledge graph;

[0014] The output is sorted according to the comprehensive risk value of the propagation path, and resource scheduling instructions are generated.

[0015] In one optional embodiment, an adjacency matrix optimization is performed on the multidimensional reasoning knowledge graph using a diffusion model, including:

[0016] Convert entity relationships into a binary adjacency matrix;

[0017]

[0018] Noise is injected using the following closed-loop diffusion formula;

[0019]

[0020] β represents the cumulative noise attenuation coefficient, ∈ represents standard Gaussian noise; s It is a coefficient dependent on s, used to control noise intensity.

[0021] Inverse denoising involves performing dual filtering on the denoised output: retaining the k edges with the highest probability of existence, and removing edges whose spatial distance is greater than the first threshold and whose connectivity probability is less than the second threshold.

[0022] In one optional embodiment, the loss function during the reverse denoising learning process is as follows:

[0023]

[0024] In the formula, This represents the loss from basic reconstruction, and φ represents the dominant factor based on common disaster knowledge. This indicates the constraints of common sense regarding disasters. Represents the disaster attenuation factor, ||λ-P E ||2 represents the disaster dynamic adaptation term, λ represents the learnable parameter, and P E A quantitative indicator representing the real-time extent of disaster damage.

[0025] In one optional embodiment, parsing spatial semantics to GIS coordinates includes: converting natural language location descriptions into WGS84 coordinates, the method being:

[0026] If the text contains "X kilometers from [reference point]", then a random point within a buffer with a radius of X kilometers will be generated based on the coordinates of the reference point.

[0027] If the text contains "[×× area]", then the centroid coordinates of the GIS boundary of that area will be retrieved.

[0028] In one optional embodiment, generating an event association strength index includes the following steps:

[0029] Based on the disaster-bearing entities and their corresponding GIS coordinates, determine the number of people affected, economic losses, scope of impact, number of affected units, and average population density.

[0030] And based on the causative factor entities, match similar events from historical cases to generate initial weights α, β, γ and weight coefficient w. d w l and w a ;in,

[0031] The formula for calculating the event correlation strength index is:

[0032]

[0033] In the formula, t represents hourly decay, S represents the severity of the event, D represents the risk of spread, and T represents the urgency of response.

[0034] S = w d ×Death+w l ×Loss+w a ×A

[0035] w d w represents the weighting coefficient for the number of people affected by the disaster. l The weighting coefficient representing economic losses, w a Weighting coefficients representing the scope of influence.

[0036] D = log 10 (N×ρ)

[0037] N represents the number of affected units, and ρ represents the average population density.

[0038] In one alternative embodiment, α, β, and γ are adaptively adjusted as follows:

[0039]

[0040] γ=n2-β

[0041] In the formula, n1 represents the baseline value of weight β, and n2 represents the baseline value of weight γ.

[0042] In one alternative embodiment, the three-level propagation includes:

[0043] Diffusion occurs along the geographic topology network, with a diffusion probability of:

[0044]

[0045] m1 represents the basic probability scaling factor for the diffusion of the geographic topology network, and d represents the distance between adjacent entities in the geographic topology network.

[0046] Diffusion occurs along the infrastructure dependency network, with a diffusion probability of:

[0047] O2=min(m 21 ,m 22 ×Adjacent Facility Dependency Level)

[0048] m 21 m represents the upper limit of the probability of infrastructure dependency network diffusion. 22 The dependency conversion coefficient represents the degree of dependence on the diffusion of infrastructure reliance networks;

[0049] The propagation network spreads, and the propagation probability is:

[0050] P3 = m3 × User Forwarding Index

[0051] In the formula, m3 represents the influence conversion coefficient of the propagation network.

[0052] In one optional embodiment, a comprehensive risk value is determined based on the product of the diffusion probabilities in each propagation path, and a disaster chain path map with probability weights is output based on the comprehensive risk value. (Node color depth represents risk level) (e.g., typhoon → rainstorm (0.95) → flooding (0.88) → hospital power outage (0.82)), and a predefined action library is associated with each risk point (e.g., hospital power outage → start generator + request power repair).

[0053] In one optional embodiment, generating resource scheduling instructions includes the following steps:

[0054] Quantify the gaps in the disaster chain path diagram and perform multi-level resource matching:

[0055] ① Draw concentric circles around the disaster site for the search;

[0056] ② Prioritize matching warehouses within the same region, followed by adjacent warehouses; when local resources are insufficient, call upon reserves from higher levels according to zoning priority.

[0057] As can be seen from the above technical solutions, the present invention discloses a method for constructing and responding to multi-dimensional reasoning knowledge graphs in the field of emergency response, which aims to completely solve the structural defects of the existing system and ensure that emergency response plans always match the evolution of the disaster situation.

[0058] Specifically, a four-dimensional knowledge graph was used to clarify the causal chain from disaster-bearing body to disaster-causing mechanism to emergency resources to response process; it was optimized to adapt to real-time disaster changes; at the same time, the intensity of disaster evolution was quantified by a dynamically decaying correlation strength index; and finally, a three-level propagation probability model of natural path, engineering dependence, and social network was innovatively integrated.

[0059] Compared with existing technologies, the beneficial effects include:

[0060] 1) By incorporating human error, environmental factors, resource dynamics, and response processes into a unified ontology framework, the limitations of traditional single-disaster modeling are overcome, and a systematic expression of cross-domain disaster-causing links is achieved;

[0061] 2) Based on external disaster signals, adaptive reconstruction of knowledge graphs effectively improves the completeness of causal chains, thereby reducing resource scheduling errors;

[0062] 3) By correcting for time decay in the correlation strength index, the efficiency of resource scheduling within the golden time window can be improved, thereby enhancing response timeliness;

[0063] 4) By expanding the single-path disaster chain into a three-dimensional propagation network of "natural geography × engineering facilities × social relations", the coverage of secondary disaster prediction can be effectively improved. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of the present 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 only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0065] Figure 1 This is a flowchart of the method for constructing and responding to multi-dimensional reasoning knowledge graphs in the emergency field according to the present invention;

[0066] Figure 2 This is the disaster chain path diagram with probability weights according to the present invention. Detailed Implementation

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0068] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0069] like Figure 1 This invention discloses a method for constructing and responding in real time to a multidimensional reasoning knowledge graph in the emergency response field, comprising the following steps:

[0070] Step 1: Define the entity attributes and semantic relationships of the disaster-bearing body layer, disaster-causing mechanism layer, emergency resource layer, and response process layer, and construct a multi-dimensional reasoning knowledge graph;

[0071] Step 2: Optimize the adjacency matrix of the multidimensional reasoning knowledge graph using a diffusion model;

[0072] Step 3: Identify the disaster-bearing entities and disaster-causing factor entities in the implementation intelligence, and parse the spatial semantics to GIS coordinates; generate an event correlation strength index by combining historical cases;

[0073] Step 4: Using entities whose association strength index meets the threshold as seed nodes, perform three-level propagation based on the optimized knowledge graph;

[0074] Step 5: Sort and output the results according to the comprehensive risk value of the propagation path, and generate resource scheduling instructions.

[0075] In one embodiment, this application first constructs a four-dimensional emergency event ontology model including a disaster-bearing body layer, a disaster-causing mechanism layer, an emergency resource layer, and a response process layer, wherein,

[0076] The disaster-bearing layer includes the physical and spatial attributes of people (density), infrastructure (seismic resistance level), and ecological environment (sensitive area);

[0077] Disaster-causing mechanism layer: Integrating four types of disaster-causing factors: natural disasters, accidents, public health, and social security, and linking human error (such as command error) and environmental factors (such as extreme weather) to establish a causal chain of "human error - environmental factors - primary disaster";

[0078] Emergency Resource Layer: This layer provides a structured description of the distribution and status of rescue teams, supply warehouses, and medical resources; and includes annotations on rescue team response times, supply warehouse capacity fluctuation thresholds, and medical resource shortage coefficients.

[0079] The handling process layer defines the handling action entities and their logical flow in three stages: monitoring and early warning, emergency response, and recovery and reconstruction, according to the timeline. (The action entities of the three stages of monitoring and early warning (T-72h), emergency response (T0~T+48h), and recovery and reconstruction (T+49h~) are serialized according to the timestamp).

[0080] This application can break through the limitations of traditional single disaster modeling by incorporating human error, environmental factors, resource dynamics, and response processes into a unified ontology framework, thereby achieving a systematic expression of cross-domain disaster-causing chains.

[0081] In a preferred embodiment, since static knowledge graphs cannot adapt to real-time disaster evolution, the present invention employs a diffusion model to optimize the adjacency matrix of the multidimensional reasoning knowledge graph; the optimization steps include:

[0082] S1. Convert entity relationships in the multidimensional reasoning knowledge graph into a binary adjacency matrix to transform structured knowledge into a computable graph topology while preserving key associations in the disaster chain;

[0083]

[0084] S2. To avoid the high time consumption of step-by-step iterative calculation, the forward diffusion process in this embodiment uses a closed-form solution to directly calculate the noisy adjacency matrix at any time t step, so as to simulate the gradual damage of disasters to infrastructure nodes; the formula is expressed as follows:

[0085]

[0086] β represents the cumulative noise attenuation coefficient, ∈ represents standard Gaussian noise; s It is a coefficient dependent on s, used to control noise intensity.

[0087] S3. Reverse denoising: Perform double filtering on the denoised output: retain the k edges with the highest probability of existence, and remove edges whose spatial distance is greater than a first threshold and whose connectivity probability is less than a second threshold. Preferably, the first threshold is 50km and the second threshold is 0.3.

[0088] In one optional embodiment, the loss function during the reverse denoising learning process is as follows:

[0089]

[0090] In the formula, This represents the loss from basic reconstruction, and φ represents the dominant factor based on common disaster knowledge. This indicates the constraints of common sense regarding disasters. Represents the disaster attenuation factor, ||λ-P E ||2 represents the disaster dynamic adaptation term, used to dynamically balance knowledge preservation and disaster adaptation, λ represents the learnable parameter, and P E A quantitative indicator representing the real-time extent of disaster damage. For example:

[0091] In earthquake disasters:

[0092] In flood disasters:

[0093] In this application, P E It is a quantitative indicator of the real-time extent of disaster damage, expressed as ||λ-P E ||2 Injecting external disaster signals into the diffusion model forces the learnable parameter λ to approximate the actual disaster situation, enabling data-driven adaptive reconstruction and avoiding deviation from actual needs by pure mathematical optimization.

[0094] This invention realizes a dynamic degradation-adaptive reconstruction closed loop for knowledge graphs in disaster scenarios. Specifically, through the diffusion process coupled with disaster and the knowledge-guided reconstruction mechanism, it effectively improves the integrity of causal chains and reduces resource scheduling errors, thereby achieving robust knowledge graph evolution modeling in emergency scenarios.

[0095] Furthermore, in one embodiment, the BERT-BiLSTM-CRF model is used to identify disaster-bearing entities (such as "Chemical Plant in City A") and disaster-causing entities (such as "Chlorine leakage rate 5kg / h") in the disaster report; and the geographical location is parsed based on spatial semantic enhancement rules.

[0096] In this practical embodiment, parsing spatial semantics to GIS coordinates includes: converting natural language location descriptions into WGS84 coordinates, the method being:

[0097] If the text contains "X kilometers from [reference point]", then a random point within a buffer with a radius of X kilometers will be generated based on the coordinates of the reference point.

[0098] If the text contains "[×× area]", then the centroid coordinates of the GIS boundary of that area will be retrieved.

[0099] In one specific embodiment, news / reports are scanned automatically:

[0100] Original report: "The Panyu District chemical plant is only 3 kilometers away from the Pearl River embankment."

[0101] System extraction:

[0102] Hazardous location: Chemical plant (chlorine storage)

[0103] Key location: Pearl River embankment (coordinates 113.36, 22.95)

[0104] Spatial positioning converts textual descriptions into precise coordinates:

[0105] "3 kilometers from the dam" → Draw a 3-kilometer radius circle with the dam as the center;

[0106] "Southern Panyu District" → Locate 5 kilometers south of the center of Panyu District.

[0107] In another embodiment, the step of generating the event association strength index includes:

[0108] Based on the disaster-bearing entities and their corresponding GIS coordinates, the number of affected people (Dth), economic losses (Loss), impact range (A), number of affected units (N), and average population density (ρ) are determined. The disaster-bearing entities are the direct source of "loss" data and determine the direct impact of the event.

[0109] And based on the causative factor entities, match similar events from historical cases to generate initial weights α, β, γ and weight coefficient w. d w l and w a ;

[0110] Specifically,

[0111] The formula for calculating the event correlation strength index is:

[0112]

[0113] In the formula, t represents hourly decay, S represents the severity of the event, which relies on the quantitative indicators of the disaster-affected entity to assess the severity of the event, D represents the spread risk, and T represents the urgency of response.

[0114] S = w d ×Dth+w l ×Loss+w a ×A

[0115] w d w represents the weighting coefficient for the number of people affected by the disaster. l The weighting coefficient representing economic losses, w a Weighting coefficients representing the scope of influence.

[0116] D = log 10 (N×ρ)

[0117] N represents the number of affected units, and ρ represents the average population density.

[0118] In one alternative embodiment, α, β, and γ are adaptively adjusted as follows:

[0119]

[0120] γ=n2-β

[0121] In the formula, n1 represents the baseline value of weight β and n2 represents the baseline value of weight γ. In this embodiment, n1 is 0.4 and n2 is 0.6.

[0122] In one embodiment, entities whose association strength index meets a threshold are used as seed nodes for three-level propagation based on the optimized knowledge graph; in this embodiment, the threshold is preferably 0.8.

[0123] Furthermore, the three levels of transmission include:

[0124] The diffusion probability is as follows: (This refers to the diffusion along geographical topological networks, such as the propagation of floods through river systems.)

[0125]

[0126] m1 represents the basic probability scale factor for the diffusion of the geographic topology network, preferably 0.7; d represents the distance between adjacent entities in the geographic topology network, preferably d≤50km;

[0127] The spread along infrastructure-dependent networks, such as a power grid outage leading to hospital shutdown, has the following probability:

[0128] P2=min(m 21 ,m 22 ×Adjacent Facility Dependency Level)

[0129] m 21 The upper limit of the probability of infrastructure dependency network diffusion is preferably 0.9, m 22 The dependency transformation coefficient representing the degree of infrastructure dependence on network diffusion is preferably 0.3.

[0130] The spread along the transmission network, such as from a rumor of material shortages to panic buying, has a probability of spreading as follows:

[0131] P3 = m3 × User Forwarding Index

[0132] In the formula, m3 represents the influence conversion coefficient of the propagation network diffusion, which is preferably 0.5.

[0133] Furthermore, a comprehensive risk value is determined based on the product of the diffusion probabilities in each propagation path, and a disaster chain path map with probability weights is output based on the comprehensive risk value, referring to... Figure 2 .

[0134] In this embodiment, the risk level is represented by the node color depth; for example: typhoon → rainstorm (0.95) → urban flooding (0.88) → hospital power outage (0.82);

[0135] As a preferred approach, each risk point is associated with a predefined action library, such as: hospital power outage → start generator + request power repair.

[0136] Step four involves generating resource scheduling instructions, which includes the following steps:

[0137] Quantify the gaps in the disaster chain path diagram and perform multi-level resource matching:

[0138] ① Draw concentric circles around the disaster site for the search;

[0139] ② Prioritize matching warehouses within the same region, followed by adjacent warehouses; when local resources are insufficient, call upon reserves from higher levels according to zoning priority.

[0140] To further optimize the above technical solution, data on the implementation of the response plan were collected. The ECAII weights and diffusion model parameters were adjusted simultaneously using gradient descent, and newly discovered disaster links were added to the training dataset of the diffusion model.

[0141]

[0142] In the formula, -0.1 is a negative constant.

[0143] In one exemplary embodiment, an operational overview example for a typhoon emergency scenario is as follows:

[0144] Disaster Report: Typhoon Haikui Landfall Warning (Wind Speed ​​45 m / s)

[0145] Graph optimization: The diffusion model strengthens the "substation-hospital" dependency chain (weight 0.91→0.96).

[0146] Event correlation strength index calculation:

[0147] Severity: Typhoon caused 2 deaths → 0.25 points (out of 1)

[0148] Affected areas: Water and power outages in 3 districts → 4.56 points (logarithmic calculation)

[0149] Time: Only 2 hours until landing → Emergency factor × 0.67

[0150] Total score 1.07 (exceeding the red alert by 0.8 points)

[0151] Risk communication:

[0152] Level 1: Typhoon → Coastal breakwater (P1 = 0.85, distance 2km)

[0153] Level 2: Dam → Subway flooding (P2 = 0.73, underground connection)

[0154] Resource scheduling:

[0155] Generator requirement = 15 ICU beds × 0.38 = 6 units

[0156] Dispatch plan: 4 units in local warehouse + 2 units in Shenzhen support

[0157] Closed-loop optimization: Actual casualties were 12% lower than predicted → β coefficient was automatically adjusted down from 0.32 to 0.29.

[0158] Through the above specific implementation details, the system can complete the entire process from disaster information input to response plan generation within 10 minutes, which is 60% more efficient than traditional methods, and the key dependency identification accuracy rate can reach over 92%.

[0159] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0160] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing and responding in real time to multi-dimensional reasoning knowledge graphs in the emergency response domain, characterized in that, Define the entity attributes and semantic relationships of the disaster-bearing body layer, disaster-causing mechanism layer, emergency resource layer, and response process layer, and construct a multi-dimensional reasoning knowledge graph; The adjacency matrix of the multidimensional reasoning knowledge graph is optimized using a diffusion model. Identify disaster-bearing entities and disaster-causing factor entities in the implementation intelligence, parse spatial semantics to GIS coordinates, and generate an event correlation strength index by combining historical cases; Entities whose association strength index meets the threshold are used as seed nodes for three-level propagation based on the optimized knowledge graph; The output is sorted according to the comprehensive risk value of the propagation path, and resource scheduling instructions are generated.

2. The method according to claim 1, characterized in that, The adjacency matrix of the multidimensional reasoning knowledge graph is optimized using a diffusion model, including: Convert entity relationships into a binary adjacency matrix; Noise is injected using the following closed-loop diffusion formula; β represents the cumulative noise attenuation coefficient, ∈ represents standard Gaussian noise; s It is a coefficient dependent on s, used to control noise intensity. Inverse denoising involves performing dual filtering on the denoised output: retaining the k edges with the highest probability of existence, and removing edges whose spatial distance is greater than the first threshold and whose connectivity probability is less than the second threshold.

3. The method according to claim 2, characterized in that, The loss function in the reverse denoising learning process is as follows: In the formula, This represents the loss from basic reconstruction, and φ represents the dominant factor based on common disaster knowledge. This indicates the constraints of common sense regarding disasters. Represents the disaster attenuation factor, ||λ-P E ||2 represents the disaster dynamic adaptation term, λ represents the learnable parameter, and P E A quantitative indicator representing the real-time extent of disaster damage.

4. The method according to claim 1, characterized in that, Parsing spatial semantics to GIS coordinates includes: converting natural language location descriptions into WGS84 coordinates, using the following method: If the text contains "X kilometers from [reference point]", then a random point within a buffer with a radius of X kilometers will be generated based on the coordinates of the reference point. If the text contains "[×× district]", then the centroid coordinates of the GIS boundary of that district will be retrieved.

5. The method according to claim 4, characterized in that, The steps to generate an event correlation strength index include: Based on the disaster-bearing entities and their corresponding GIS coordinates, determine the number of people affected, economic losses, scope of impact, number of affected units, and average population density. And based on the causative factor entities, match similar events from historical cases to generate initial weights α, β, γ and weight coefficient w. d w l and w a ;in, The formula for calculating the event correlation strength index is: In the formula, t represents hourly decay, S represents the severity of the event, D represents the risk of spread, and T represents the urgency of response. S=w d ×Death+w l ×Loss+w a ×A w d w represents the weighting coefficient for the number of people affected by the disaster. l The weighting coefficient representing economic losses, w a Weighting coefficients representing the scope of influence. D=log 10 (N×ρ) N represents the number of affected units, and ρ represents the average population density.

6. The method according to claim 5, characterized in that, α, β, and γ are adaptively adjusted as follows: γ=n2-β In the formula, n1 represents the baseline value of weight β, and n2 represents the baseline value of weight γ.

7. The method according to claim 1, characterized in that, Third-level transmission includes: Diffusion occurs along the geographic topology network, with a diffusion probability of: m1 represents the basic probability scaling factor for the diffusion of the geographic topology network, and d represents the distance between adjacent entities in the geographic topology network. Diffusion occurs along the infrastructure dependency network, with a diffusion probability of: P2=min(m 21 ,m 22 ×Adjacent Facility Dependency Level) m 21 m represents the upper limit of the probability of infrastructure dependency network diffusion. 22 The dependency conversion coefficient represents the degree of dependence on the diffusion of infrastructure reliance networks; The propagation network spreads, and the propagation probability is: P3 = m3 × User Forwarding Index In the formula, m3 represents the influence conversion coefficient of the propagation network.

8. The method according to claim 1, characterized in that, The comprehensive risk value is determined by multiplying the diffusion probabilities in each propagation path, and the disaster chain path map with probability weights is output based on the comprehensive risk value.

9. The method according to claim 8, characterized in that, The steps for generating resource scheduling instructions include: Quantify the gaps in the disaster chain path diagram and perform multi-level resource matching: ① Draw concentric circles around the disaster site for the search; ② Prioritize matching warehouses within the same region, followed by adjacent warehouses; when local resources are insufficient, call upon reserves from higher levels according to zoning priority.

Images