Intelligent quantitative evaluation method and system for community fire-fighting toughness

Abstract

The invention discloses an intelligent quantitative evaluation method and system for community fire-fighting toughness, and the method comprises the steps: collecting community fire accident cases, building a community fire accident text database in combination with a community fire-fighting toughness key stage, and recognizing community fire-fighting toughness related entities and mutual relationships; determining a community fire accident knowledge graph mode layer, establishing a knowledge graph, and concluding and extracting community fire-fighting toughness influence factors; a TOPSIS method is improved based on community fire-fighting toughness characteristics, key influence factors are screened, a community fire-fighting toughness evaluation index system is established, and a toughness network is constructed; and in combination with parameter learning and a Delphi method, establishing a Bayesian network evaluation model, carrying out causal reasoning and sensitivity analysis, quantitatively evaluating the community fire-fighting toughness level, identifying weak links and key influence indexes of the community fire-fighting toughness, and pointedly providing a promotion strategy. According to the method and the system, a practical tool is provided for a decision department to strengthen fire safety management and promote flexible sustainable community construction.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of emergency management, in particular to a community fire resilience intelligent quantitative evaluation method and system. BACKGROUND

[0002] With the acceleration of urbanization process, the number of urban communities in China has increased dramatically. Compared with traditional rural settlements in China, urban communities have high concentration of people, financial resources and material resources, more fire hazards, frequent accidents and increasingly prominent fire safety problems. As an important part of the community resilience system, community fire resilience reflects the prevention, response, recovery and adaptation capacity of the community under the impact of fire, and is a key indicator for evaluating the level of community safety governance. Improving community fire resilience has become a key issue for sustainable development and disaster management of communities. Community fire resilience evaluation, as an important tool for understanding and improving resilience level, helps community managers identify weak links and key influencing factors in fire management, and provides data support and decision basis for developing targeted resilience improvement measures.

[0003] For resilience evaluation, some scholars have proposed relevant methods. Su Xin et al. proposed a flood resilience efficient evaluation method and system based on natural language processing and data-driven model (authorized publication number CN 118643997 B), which collects multi-source data, constructs a flood resilience knowledge graph, screens indicators based on correlation and causality analysis, establishes a resilience index system, and uses machine learning algorithm to establish a resilience evaluation model. Wang Jian et al. proposed a community fire resilience evaluation method and system based on association rules and complex networks (application publication number CN 119515157 A), which analyzes the disaster-causing reasons of historical accident cases, establishes a community fire resilience evaluation index system, and uses association rule algorithm and complex network to identify key indicators. Huang Ying et al. proposed a community disaster prevention resilience evaluation method based on improved entropy weight-CRTIC method (authorized publication number CN 114565249 B), which established a community disaster prevention resilience evaluation index system, combined with field data and improved entropy weight-CRITIC method to calculate the index weight, and used TOPSIS method to calculate the resilience level of the evaluation object. From the above patents, it can be seen that the index system in the existing evaluation method mainly depends on existing data and expert experience, lacks in-depth analysis of historical accident cases, and the index screening process mainly analyzes the correlation and causality between indicators, without considering the influence degree of indicators on disasters. In addition, most evaluation methods focus on comprehensive score or grading, and fail to effectively link the evaluation results with specific improvement measures, lacking systematic identification of key weak links, which limits its value in actual management and decision-making.

[0004] The analysis and screening of influencing factors are the core links of constructing the evaluation index system. In the aspect of influencing factor analysis, Yang Xin et al. proposed a prediction method of rail transit accidents based on knowledge graph (application publication number: CN 116484056A), based on rail transit accident data, relevant knowledge entities and their mutual relationships were identified, a rail transit hazard source correlation analysis model based on knowledge graph theory was established, and its analysis indicators were clarified. Zhang Yibin et al. proposed a safety risk management and control method and system based on knowledge graph (application publication number: CN 117787725 A), by analyzing the power grid infrastructure construction safety related data in infrastructure accident cases, and carrying out related entity and relationship extraction, a power grid infrastructure construction safety management and control knowledge graph with multi-modal characteristics was constructed, and a risk identification index system was established. In the aspect of key influencing factor screening, Shi Zheqi et al. carried out a research on the environmental risk screening of petrochemical enterprises based on TOPSIS-AHP method, the weighted distance square sum between positive and negative ideal points was calculated by TOPSIS method, and the risk indicators were ranked in terms of advantages and disadvantages, and the key risk indicators were screened. Tong Junhao et al. proposed a highway tunnel fire resistance evaluation method (application publication number: CN 118195133 A), based on the highway tunnel fire resistance evaluation index system, the importance, discriminability and influence degree of each indicator were calculated by using the improved TOPSIS method, and then the key indicators were screened. From the above patents, it can be seen that the current applied or granted patents mainly focus on the extraction of accident causes and the causal relationship modeling of disaster-causing path, and pay less attention to the system resilience characteristics reflected by the accident, lack of in-depth mining of resilience characteristics in historical accidents, and it is difficult to support the resilience-oriented index system construction identification. In addition, TOPSIS as a multi-objective decision-making method, lacks attention to the characteristics of specific research objects in existing research, especially lacks analysis of the dynamic time sequence characteristics of resilience, and has not formed a targeted and systematic index screening method, which limits the applicability and effectiveness of the method in the field of resilience evaluation.

[0005] Therefore, based on community fire accident cases, combined with the key characteristics of community fire resilience, the application develops a community fire resilience intelligent quantitative evaluation method and system by using knowledge graph and improved TOPSIS method, to solve the shortcomings of existing evaluation methods in the aspects of utilization of historical experience, phased feature identification and index disaster correlation analysis, which has great practical significance and social value for improving the comprehensive response ability of community fire and promoting fine community fire safety management. SUMMARY

[0006] The present application aims to provide a community fire resilience intelligent quantitative evaluation method and system, based on the key stages of community fire resilience, deeply analyzes the accident investigation report, establishes a community fire accident knowledge graph, and extracts the influencing factors of community fire resilience. Using the improved TOPSIS method, the influence of the influencing factors on community fire is analyzed, the key factors are screened, the community fire resilience evaluation index system and resilience network are established, and based on this, the community fire resilience Bayesian evaluation model is constructed, the causal reasoning and sensitivity analysis are carried out, the quantitative evaluation of community fire resilience is realized, and the weak links and key indicators of community fire safety management are identified.

[0007] In one aspect, the present application provides a community fire resilience intelligent quantitative evaluation method, which comprises the following steps:

[0008] Step S1, collect community fire accident cases, establish a community fire accident text database combined with the key stages of community fire resilience, and identify community fire resilience related entities and their mutual relationships.

[0009] Further, the key stages of community fire resilience are divided into fire prevention stage, emergency response stage, post-disaster recovery stage and improvement stage based on resilience theory and combined with the actual community fire management work.

[0010] Further, the community fire accident cases are stored in the community fire accident text database after format preprocessing, and the community fire accident text database at least contains accident basic information column, accident cause column, emergency disposal measure column, after-disposal measure column and prevention measure column.

[0011] Optionally, the community fire resilience related entities and their relationships can be selected using natural language processing tools, which are not limited here.

[0012] Step S2, based on the community fire resilience entities and their relationships, the mode layer structure of the community fire accident knowledge graph is determined, the community fire accident knowledge graph is established, and the community fire resilience influencing factors are extracted.

[0013] Further, the mode layer of the community fire accident knowledge graph includes basic information layer, resilience stage layer, accident cause layer and influencing factor layer, which is used to describe the constituent elements and logical relationships of the community fire accident cases.

[0014] Further, the community fire resilience influencing factors include fire prevention stage influencing factors, emergency response stage influencing factors, recovery stage influencing factors and improvement stage influencing factors, each type of influencing factor can be further divided into personnel factors, building factors, environmental factors and management factors, which specifically include:

[0015] The fire prevention stage influencing factors mainly include personnel factors: insufficient fire risk prevention ability of residents; building factors: old buildings, high buildings, low fire resistance rating, missing or failed fire separation, abnormal internal equipment operation state, and use of flammable decoration materials; environmental factors: accumulation of combustible materials, illegal addition of facilities; management factors: insufficient supervision of six types of people, insufficient hidden danger investigation and rectification, insufficient fire safety work funds, imperfect fire safety management system, inadequate fire safety propaganda education and training, and inadequate fire safety facility maintenance;

[0016] The emergency response stage influencing factors mainly include personnel factors: insufficient emergency disposal and escape ability; building factors: insufficient evacuation passage smoothness, incomplete or failed fire safety facilities; environmental factors: missing or failed outdoor fire safety facilities, occupied fire truck access or climbing site; management factors: inadequate basic rescue force construction, no establishment of emergency plan or imperfect plan;

[0017] The recovery stage influencing factors mainly include management factors: insurance claims, public opinion guidance, compensation for the distressed, medical rescue support, post-disaster resident resettlement, and post-disaster equipment repair and replenishment;

[0018] The improvement stage influencing factors mainly include management factors: perfecting fire management system and local supervision mechanism, deepening resident fire safety education and cognitive construction, strengthening multi-dimensional fire force construction and emergency mobilization capacity, application of new fire technology, and historical accident case learning system.

[0019] Optionally, the community fire accident knowledge graph can be established using a Neo4j graph database, which is not limited here.

[0020] Step S3, based on the characteristics of community fire resilience, the TOPSIS method is improved, the key influencing factors are screened, the community fire resilience evaluation index system is established, and the community fire resilience network is constructed.

[0021] Further, based on the characteristics of community fire resilience, the TOPSIS method is improved, and the specific steps are as follows:

[0022] The characteristics of community fire resilience are analyzed, and the attributes of the community fire resilience influencing factors are determined, wherein the influencing factor attributes of the fire prevention stage (t=1) are the likelihood of occurrence, the severity of consequences, and controllability; the influencing factor attributes of the emergency response stage (t=2) are the severity of consequences, controllability, and timeliness; the influencing factor attributes of the post-disaster recovery stage (t=3) are recovery speed, resource input, and recovery degree; the influencing factor attributes of the improvement stage (t=4) are experience conversion rate, ability improvement degree, and system redundancy;

[0023] K experts are invited to fill in the questionnaire to collect their knowledge, and the attribute values of the community fire resilience influencing factors in stage t are quantitatively scored to establish an initial decision matrix X:

[0024]

[0025] In the formula, X (t) is the initial decision matrix of stage t, is the jth attribute value of the Kth expert for the community fire resilience influencing factor i in stage t, wherein t = 1, 2, 3, 4; i = 1, 2, 3…m; j = 1, 2, 3;

[0026] The initial decision matrix is normalized to establish a normalized matrix R (t) :

[0027]

[0028] In the formula, is the normalized value of the jth attribute value of the community fire resilience influencing factor i in stage t;

[0029] K experts are invited to fill in the questionnaire to collect their knowledge, and the influence degree evaluation (score range [-1, 1]) of attribute q on attribute j in stage t is established to establish an attribute correlation matrix Γ (t) :

[0030]

[0031] In the formula, represents the influence degree evaluation of the Kth expert for the attribute q on the attribute j in stage t, q = 1, 2, 3; represents the average value of the influence degree evaluation, represents the enhancement effect, represents the inhibition effect;

[0032] The influence of adjacent attributes on the attributes of the community fire resilience influencing factors in stage t is introduced to enhance the expression effect, and an enhanced attribute value matrix Ψ (t) is established:

[0033]

[0034] Ψ (t) = R (t) ×(Γ (t) ) T

[0035] In the formula, represents the attribute value considering the mutual influence between attributes;

[0036] The enhanced attribute value matrix Ψ (t) is normalized to establish a normalized enhanced attribute value matrix

[0037]

[0038] wherein, is the normalized result of the attribute value after considering the mutual influence between attributes;

[0039] The attribute weight in the stage t is :

[0040]

[0041] wherein, and the sum of the attribute weights in the stage t is 1; is the attribute information quantity, indicating the reciprocal of the sum of the squared distances from all the influence factor values of the attribute j in the stage t to the boundary points (0 and 1);

[0042] is the attribute optimal weight vector ω of the community fire resilience influence factor in the stage t (t) When distance is used as a limiting condition, the sum of the weighted squared distances of all the influence factor values in the stage t should satisfy the minimum;

[0043] The objective function is:

[0044]

[0045] For the community fire resilience influence factor i in the stage t, the influence index is calculated

[0046]

[0047] For the community fire resilience influence factor i in the stage t, the recognition degree is calculated

[0048]

[0049] wherein, is the entropy weight coefficient of the community fire resilience influence factor i in the stage t, is the entropy value of the community fire resilience influence factor i in the stage t, and the specific calculation formula is as follows:

[0050] wherein when ,

[0051]

[0052] In the formula, is the attribute influence proportion of the community fire resilience influence factor i in the stage t, and the specific calculation formula is as follows:

[0053] In the formula,

[0054] In the formula, is the global influence proportion coefficient of the community fire resilience influence factor i in the stage t;

[0055] The influence factor contribution degree is calculated

[0056]

[0057] For the community fire resilience influence factor attribute of the stage t, a threshold is set, and the corresponding and When , the community fire resilience influence factor is determined to have low influence, that is, it is not important; when , the community fire resilience influence factor is determined to have no significant influence on decision-making; when , the community fire resilience influence factor is determined to have low contribution degree.

[0058] Step S4, based on the community fire resilience network, combining parameter learning and Delphi method to calculate node probability, and establishing a community fire resilience Bayesian network evaluation model.

[0059] Further, the parameter learning method is based on the community fire accident case to construct a parameter learning training set, and the maximum expectation algorithm is used to calculate the node prior probability and conditional probability; the Delphi method uses questionnaire survey to collect expert knowledge, and uses the defuzzification and Noisy-MAX model to supplement the node probability which is difficult to quantify in parameter learning;

[0060] Step S5, using the community fire resilience Bayesian network evaluation model for causal reasoning and sensitivity analysis, quantitatively evaluating the community fire resilience level, and identifying the weak links and key indicators affecting the community fire resilience.

[0061] Further, the quantitative evaluation result of the community fire resilience level is the marginal probability of the top node in the community fire resilience Bayesian network evaluation model.

[0062] Optionally, the sensitivity analysis can select to use the mutual information method, which is not limited here.

[0063] Step S6, the community fire resilience weak link and key indicator is proposed for the targeted community fire resilience improvement strategy.

[0064] In another aspect, the present application provides a kind of community fire resilience intelligent quantitative evaluation system, the system includes: historical fire accident analysis module, index system construction module, community fire resilience evaluation module and community fire resilience promotion module, each module is sequentially connected.

[0065] Further, the historical fire accident analysis module is used to process the collected community fire accident case, extracts community fire resilience related entity and relationship, and determines the community fire resilience influencing factor;The index system construction module is used to filter the key influencing factors with high influence on community fire resilience, strong distinguishing characteristics and high contribution degree in influencing factors, and establish community fire resilience evaluation index system;The community fire resilience evaluation module is used to carry out community fire resilience quantitative evaluation, and filter community fire resilience weak link and key indicator;The community fire resilience promotion module is based on weak link and key indicator, and proposes targeted improvement strategy.

[0066] Compared with the prior art, the present application has the following advantages:

[0067] The present application analyzes community historical fire accidents from the perspective of community fire resilience, establishes community fire accident knowledge graph, deeply mines the resilience elements and their internal relations reflected in the accident, realizes the structured integration and semantic expression of unstructured accident information;Based on the key stage of resilience, the improved TOPSIS method is used, the influence factor attributes strongly related to the accident are introduced, the key influencing factors are screened, the community fire resilience evaluation index system is constructed, the pertinence and scientificity of index screening are improved, and the problems of fuzzy stage division and incomplete index coverage in traditional evaluation system are overcome;The resilience evaluation model based on Bayesian network supports causal reasoning and sensitivity analysis, realizes the dynamic quantitative evaluation of fire resilience level in complex community system, effectively identifies the weak link of fire management and key influencing factors;Finally, the optimization of community fire resource allocation and governance strategy is realized, and the fine level of community fire comprehensive response ability and fire safety governance is improved. BRIEF DESCRIPTION OF DRAWINGS

[0068] Figure 1 A community fire resilience intelligent quantitative evaluation method flow chart of the embodiment of the present application;

[0069] Figure 2 The entity and relationship extraction process of the embodiment of the present application;

[0070] Figure 3 The knowledge graph mode layer schematic diagram of the embodiment of the present application;

[0071] Figure 4 This is a schematic diagram of a knowledge graph according to an embodiment of the present invention;

[0072] Figure 5 These are the factors influencing community fire resilience in embodiments of the present invention;

[0073] Figure 6 The results of the improved TOPSIS method in this embodiment of the invention are shown.

[0074] Figure 7 This is a Bayesian assessment model for community fire resilience in an embodiment of the present invention;

[0075] Figure 8 This is a schematic diagram of the structural composition of a community fire resilience intelligent quantitative assessment system according to an embodiment of the present invention. Detailed Implementation

[0076] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0077] Example 1

[0078] like Figure 1 As shown, the intelligent quantitative assessment method for community fire resilience in this embodiment includes the following steps:

[0079] Step S1: Collect community fire accident cases, combine them with the key stages of community fire resilience, establish a community fire accident text database, and identify relevant entities and their interrelationships related to community fire resilience.

[0080] The key stages of community fire resilience, based on resilience theory and combined with the actual situation of community fire management, are divided into the fire prevention stage, the emergency response stage, the post-disaster recovery stage, and the improvement and enhancement stage.

[0081] In this embodiment, the community fire accident cases are sourced from unstructured text data such as government public databases, news reports, accident investigation reports, and industry annual reports. A total of 175 community fire accident cases were collected. After format preprocessing, the accident cases are stored in the community fire accident text database. The community fire accident text database includes columns for accident time, accident location, accident consequences, accident causes, emergency response measures, post-accident handling measures, and prevention measures.

[0082] like Figure 2 As shown, this embodiment utilizes the Natural Language Processing Platform (LTP) and employs modules such as word segmentation, part-of-speech tagging, dependency parsing, and semantic role labeling to collaboratively process and extract the community fire resilience-related entities and their relationships.

[0083] Step S2, based on the community fire resilience entity and its relationship, the mode layer structure of the community fire accident knowledge graph is defined, the community fire accident knowledge graph is established, and the community fire resilience influencing factors are summarized and extracted.

[0084] The mode layer of the community fire accident knowledge graph is used to define the semantic structure of the graph, including a basic information layer, a resilience key stage layer, an accident cause layer and an influencing factor layer. The mode layer of the community fire accident knowledge graph of the embodiment is as shown in Figure 3 .

[0085] As shown in Figure 4 , the embodiment uses a Neo4j graph database to realize the storage and visual display of the community fire accident.

[0086] As shown in Figure 5 , the embodiment defines the community fire resilience influencing factors based on the community fire resilience related entities and the fire supervision and inspection records.

[0087] Step S3, based on the characteristics of community fire resilience, the TOPSIS method is improved, the community fire resilience influencing factors are quantitatively analyzed, the key influencing factors are screened, the community fire resilience evaluation index system is established, and the community fire resilience network is constructed.

[0088] The improved TOPSIS method has the following specific steps:

[0089] The characteristics of community fire resilience stages are analyzed, and the attributes of the community fire resilience influencing factors are determined, wherein the influencing factor attributes of the fire prevention stage (t=1) are the possibility of occurrence, the consequence severity and the controllability; the influencing factor attributes of the emergency response stage (t=2) are the consequence severity, the controllability and the timeliness; the influencing factor attributes of the post-disaster recovery stage (t=3) are the recovery speed, the resource input and the recovery degree; the influencing factor attributes of the improvement promotion stage (t=4) are the experience conversion rate, the ability improvement degree and the system redundancy;

[0090] The knowledge of K experts is collected by a questionnaire, the attributes of the community fire resilience influencing factors in stage t are quantitatively scored, and an initial decision matrix X is established:

[0091]

[0092] In the formula, X (t) is the initial decision matrix of stage t, is the jth attribute value of the community fire resilience influencing factor i in stage t for the Kth expert, wherein t=1, 2, 3, 4; i=1, 2, 3…m; j=1, 2, 3.

[0093] normalizing the initial decision matrix to establish a normalized matrix R (t) :

[0094]

[0095] wherein, is the normalized value of the jth attribute value of the community fire resilience influencing factor i in the stage t;

[0096] K experts' knowledge is collected by a questionnaire to evaluate the influence degree of the attribute q on the attribute j in the stage t (score range [-1, 1]), and an attribute correlation matrix Γ is established (t) :

[0097]

[0098] wherein, represents the Kth expert's evaluation of the influence degree of the attribute q on the attribute j in the stage t, q = 1, 2, 3; represents the average value of the influence degree evaluation, represents the enhancement effect, represents the inhibition effect;

[0099] The influence of adjacent attributes on the attributes of the community fire resilience influencing factor in the stage t is introduced to enhance the expression effect, and an enhanced attribute value matrix Ψ is established (t) :

[0100]

[0101] Ψ (t) = R (t) ×(Γ (t) ) T

[0102] wherein, represents the attribute value considering the mutual influence between attributes;

[0103] The enhanced attribute value matrix Ψ (t) is normalized to establish a normalized enhanced attribute value matrix

[0104]

[0105] wherein, is the normalized result of the attribute value considering the mutual influence between attributes;

[0106] The attribute weight in the stage t is:

[0107] wherein,

[0108] wherein, and the sum of attribute weights in the stage t is 1; is the attribute information amount, indicating the reciprocal of the sum of the squared distances of all the influence factor values of the attribute j in the stage t to the boundary points (0 and 1);

[0109] is the attribute optimal weight vector ω of the community fire resilience influence factor in the stage t, and is calculated as follows: (t) When distance is taken as the limiting condition, the sum of the weighted squared distances of all the influence factor values in the stage t should satisfy the following condition: is the minimum;

[0110] The objective function is:

[0111]

[0112] For the community fire resilience influence factor i in the stage t, the influence index is calculated as follows:

[0113]

[0114] For the community fire resilience influence factor i in the stage t, the recognition degree is calculated as follows:

[0115]

[0116] wherein, is the entropy weight coefficient of the community fire resilience influence factor i in the stage t, is the entropy value of the community fire resilience influence factor i in the stage t, and the specific calculation formula is as follows:

[0117] wherein when ,

[0118]

[0119] wherein, is the attribute influence proportion of the community fire resilience influence factor i in the stage t, and the specific calculation formula is as follows:

[0120] wherein,

[0121] wherein, This is the global influence ratio coefficient of the community fire resilience influencing factor i in stage t;

[0122] Calculate the contribution of the influencing factors

[0123]

[0124] For the community fire resilience influencing factors at stage t, a threshold is set, and the corresponding values ​​are calculated. and when At that time, the factors affecting community fire resilience were determined to have low influence, i.e., unimportant; when When the factors affecting community fire resilience are determined to have no significant impact on decision-making; when At that time, the factors affecting the community's fire resilience were determined to have a low contribution.

[0125] This embodiment invited 33 experts in the field of fire resilience to use a five-point Likert scale to quantify and score the attributes of the influencing factors and the degree of influence between their respective attributes, and to use... For the threshold, such as Figure 6 The figure shows the screening results of the community fire resilience influencing factors using the improved TOPSIS method in this embodiment. A total of 23 key influencing factors were extracted to establish the community fire resilience assessment index and the community fire resilience network.

[0126] Step S4: Based on the community fire resilience network, and combining parameter learning and the Delphi method to calculate node probabilities, establish the community fire resilience Bayesian network evaluation model.

[0127] The parameter learning method described above constructs a parameter learning training set based on the community fire accident case, and uses the expectation-maximization algorithm to calculate the prior probability and conditional probability of nodes; the Delphi method uses a questionnaire survey to collect expert knowledge, and uses defuzzification and Noisy-MAX models to supplement the node probabilities that are difficult to quantify in parameter learning.

[0128] like Figure 7 As shown, this embodiment uses the "GeNIe 4.0 Academic" software to establish the Bayesian network evaluation model based on the community fire resilience network.

[0129] Step S5: Use the Bayesian network evaluation model to perform causal reasoning and sensitivity analysis to quantitatively assess the community's fire resilience level and identify weak links and key indicators affecting the community's fire resilience.

[0130] The quantitative assessment result of the community's fire resilience level is the edge probability of the top node in the BN model.

[0131] The community fire resilience Bayesian network evaluation model of this embodiment carries out the causal reasoning discovery: the probability of Chinese community fire resilience in the high resilience state is 62%, the sound fire governance system and the local supervision mechanism and the failure probability of insufficient emergency response and escape ability are the highest, both are 6%, which are the key weak links for improving community fire resilience at the present stage. In addition, electrical equipment failure, lack of hidden danger investigation and governance, poor residents' emergency response and escape ability, and occupation of fire rescue passage are the main reasons for reducing community fire resilience. Using the mutual information method, sensitivity analysis is carried out on the fire prevention stage, the emergency response stage, the post-disaster recovery stage and the improvement and promotion stage respectively. It is found that the accumulation of flammable and combustible materials is most sensitive to fire prevention ability, with an entropy reduction value of 5.68%, followed by abnormal internal equipment operation state, with an entropy reduction value of 4.47%. Insufficient emergency response and escape ability is the key factor to improve community fire emergency response ability, with an entropy reduction value of 5.65%. The sensitivity of medical rescue support in recovery ability is the highest, with an entropy reduction of 4.38%. In the improvement and promotion ability, the sensitivity of sound fire governance system and local supervision mechanism and historical accident case learning system is higher, with an entropy reduction value of 3.51% and 3.81% respectively.

[0132] Step S6, the community fire resilience improvement strategy is proposed for the weak links and key indicators affecting community fire resilience.

[0133] Based on the key weak links and key indicators of this embodiment, taking the fire prevention stage as an example, the targeted community fire resilience improvement strategy is proposed: for fire prevention ability, the community should broaden the channels of propaganda and education, and combined with the actual situation, establish a hierarchical and classified fire safety propaganda and education mechanism for key groups such as the elderly, carry out differentiated education, and improve the residents' fire risk awareness and prevention consciousness. At the same time, develop and implement a hidden danger investigation and rectification system based on the grid management system, 100% cover key inspection areas such as corridors and fire truck passages, and form a hidden danger closed-loop management system.

[0134] Embodiment 2

[0135] This embodiment provides a community fire resilience intelligent quantitative evaluation system based on embodiment 1, as shown in Figure 8 The composition schematic diagram of the system disclosed by this embodiment, the system comprises: a historical fire accident analysis module, an index system construction module, a community fire resilience evaluation module and a community fire resilience improvement module, and the modules are connected in sequence.

[0136] The historical fire accident analysis module is used for processing the collected community fire accident cases, extracting community fire resilience related entities and relations, and determining the community fire resilience influencing factors. The index system construction module is used for screening the key influencing factors with high influence on the community fire resilience, strong distinguishing characteristics and high contribution degree, and establishing the community fire resilience evaluation index system. The community fire resilience evaluation module is used for carrying out the community fire resilience quantitative evaluation, and screening the weak links and key indicators of the community fire resilience. The community fire resilience improvement module proposes the targeted improvement strategies based on the weak links and key indicators.

[0137] The above specific embodiments further explain the purposes, technical solutions and advantages of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A community fire resilience intelligent quantitative evaluation method, characterized in that, The method comprises the following steps: Step S1, collecting community fire accident cases, combining community fire resilience key stages, establishing a community fire accident text database, identifying community fire resilience related entities and their mutual relationships; Step S2, based on the community fire resilience entities and their relationships, defining the mode layer structure of the community fire accident knowledge graph, establishing the community fire accident knowledge graph, and summarizing and extracting the community fire resilience influencing factors; Step S3, based on the characteristics of community fire resilience, improving the TOPSIS method, quantitatively analyzing the community fire resilience influencing factors, screening key influencing factors, establishing a community fire resilience evaluation index system, and constructing a community fire resilience network; Step S4, based on the community fire resilience network, combining parameter learning and Delphi method to calculate node probability, and establishing a community fire resilience Bayesian network evaluation model; Step S5, using the community fire resilience Bayesian network evaluation model for causal reasoning and sensitivity analysis, quantitatively evaluating the community fire resilience level, and identifying the weak links and key indicators affecting community fire resilience; Step S6, proposing targeted community fire resilience improvement strategies for the weak links and key indicators affecting community fire resilience.

2. The method of claim 1, wherein, The community fire resilience key stages in step S1 are divided into fire prevention stage, emergency response stage, post-disaster recovery stage and improvement stage based on resilience theory and combined with the actual community fire management work.

3. The method of claim 1, wherein, The community fire accident cases in step S1 are stored in the community fire accident text database after format preprocessing, and the community fire accident text database at least includes an accident basic information column, an accident cause column, an emergency disposal measure column, a post-disposal measure column and a prevention measure column.

4. The method of claim 1, wherein, The mode layer of the community fire accident knowledge graph in step S2 includes a basic information layer, a resilience stage layer, an accident cause layer and an influencing factor layer, which are used to describe the constituent elements and logical relationships of the community fire accident cases.

5. The method of claim 1, wherein, The community fire resilience influencing factors in step S2 include fire prevention stage influencing factors, emergency response stage influencing factors, recovery stage influencing factors and improvement stage influencing factors, and each type of influencing factor can be further divided into personnel factors, building factors, environmental factors and management factors, which specifically include: The fire prevention stage influencing factors mainly include personnel factors: insufficient fire risk prevention ability of residents; building factors: old buildings, high buildings, low fire resistance rating, missing or failed fire separation, abnormal internal equipment operation state and use of flammable decorative materials; environmental factors: accumulation of combustible materials, illegal addition of facilities; management factors: inadequate supervision of six types of people, insufficient hidden danger rectification efforts, insufficient fire safety work funds, imperfect fire safety management system, inadequate fire safety propaganda education and training work and inadequate fire safety facility maintenance; The emergency response stage influencing factors mainly include personnel factors: insufficient emergency treatment and escape ability; building factors: insufficient evacuation passage smoothness, incomplete or failed fire-fighting facilities; environmental factors: lack or failure of outdoor fire-fighting facilities, occupied fire truck passage or climbing site; management factors: insufficient basic rescue force construction, no or imperfect emergency plan; The recovery stage influencing factors mainly include management factors: insurance claim, public opinion guidance, compensation for the victims, medical aid guarantee, post-disaster resident resettlement and post-disaster equipment repair and supplement; The improvement stage influencing factors mainly include management factors: perfect fire control system and local supervision mechanism, deepening of resident fire safety education and cognitive construction, strengthening of multi-element fire force construction and emergency mobilization ability, application and use of new fire-fighting technology and historical accident case learning system.

6. The method of claim 2, wherein, The step S3 improves the TOPSIS method based on the community fire resilience characteristics, and the specific steps are as follows: The community fire resilience stage characteristics are analyzed, and the attributes of the community fire resilience influencing factors are determined, wherein the influencing factor attributes of the fire prevention stage (t=1) are the occurrence possibility, the consequence severity and the controllability; the influencing factor attributes of the emergency response stage (t=2) are the consequence severity, the controllability and the timeliness; the influencing factor attributes of the post-disaster recovery stage (t=3) are the recovery speed, the resource input and the recovery degree; and the influencing factor attributes of the improvement stage (t=4) are the experience conversion rate, the ability improvement degree and the system redundancy. The knowledge of K experts is collected by using a questionnaire, the attributes of the community fire resilience influencing factors in the stage t are quantitatively scored, and an initial decision matrix X is established: wherein X (t) is the initial decision matrix of the stage t, is the jth attribute value of the ith community fire resilience influencing factor in the stage t by the Kth expert, wherein t = 1, 2, 3, 4; i = 1, 2, 3…m; j = 1, 2, 3. normalizing the initial decision matrix to establish a normalized matrix R (t) : In the formula, is the normalized value of the jth attribute value of the community fire resilience influencing factor i in the stage t. A questionnaire is used to collect K expert knowledge, and the influence degree of the attribute q on the attribute j at stage t is evaluated (score range [-1, 1]) to establish an attribute correlation matrix Γ (t) : wherein represents the degree of influence evaluation of the Kth expert on the attribute j of the attribute q of the stage t, q = 1, 2, 3; represents the average value of the degree of influence evaluation, represents an enhancing action, represents an inhibiting action; Introducing the influence of adjacent attributes to the attributes of the community fire resilience influencing factors in the stage t, enhancing the expression effect, establishing the enhanced attribute value matrix Ψ (t) : Ψ (t) = R (t) x (Γ (t) ) T In the formula, represents the attribute value after considering the mutual influence among attributes; normalizing the enhanced attribute value matrix Ψ (t) ​ In the formula, is the normalized result of the attribute value after considering the mutual influence between attributes; The attribute weight in the stage t is: wherein, wherein and the sum of the attribute weights in the stage t is 1; is the attribute information quantity, indicating the reciprocal of the sum of the squared distances of all the influence factor values of the attribute j in the stage t to the boundary points (0 and 1); To calculate the attribute optimal weight vector ω of the community fire resilience influencing factor in the stage t (t) When distance is used as a limiting condition, the sum f of the weighted square distances of all the influencing factor values in the stage t should be satisfied i (t) (ω)min; The objective function is: calculating an influence index for said community fire resilience influencing factor i in said phase t calculating a recognition for the community fire resilience influence factor i in the phase t In the formula, is the entropy weight coefficient of the community fire resilience influencing factor i in the stage t, is the entropy value of the community fire resilience influencing factor i in the stage t, and the specific calculation formula is as follows: wherein when time, In the formula, is the attribute influence proportion of the community fire resilience influence factor i in the stage t, and the specific calculation formula is as follows: wherein, In the formula, is the global impact proportionality coefficient of the community fire resilience impact factor i in the phase t. calculating the contribution degree of the influence factor For the community fire resilience influence factor attribute of the stage t, a threshold is set, and the corresponding and When , the community fire resilience influence factor is determined as not important; when , the community fire resilience influence factor is determined as having no significant influence on decision-making; and when , the community fire resilience influence factor is determined as having low contribution.

7. The method of claim 1, wherein, The parameter learning method of the step S4 is based on the community fire accident cases to construct a parameter learning training set, and the maximum likelihood algorithm is used to calculate the node prior probability and the conditional probability; the Delphi method collects expert knowledge by using a questionnaire, and uses a de-fuzzification and Noisy-MAX model to supplement the node probability which is difficult to quantify in parameter learning.

8. The method of claim 1, wherein, The quantitative evaluation result of the community fire resilience level in the step S5 is the marginal probability of the top node in the community fire resilience Bayesian network evaluation model.

9. A community fire resilience smart quantification assessment system for performing the method of any one of claims 1-8, characterized in that, The system comprises a historical fire accident analysis module, an index system construction module, a community fire resilience evaluation module and a community fire resilience improvement module, and the modules are sequentially connected. The historical fire accident analysis module is used for processing the collected community fire accident cases, extracting community fire resilience related entities and relationships, and determining community fire resilience influencing factors; The index system construction module is used for screening key influencing factors with high influence on community fire resilience, strong distinguishing characteristics and high contribution degree, and establishing a community fire resilience evaluation index system; The community fire resilience evaluation module is used for carrying out community fire resilience quantitative evaluation, and screening community fire resilience weak links and key indicators; The community fire resilience improvement module proposes targeted improvement strategies based on the weak links and key indicators.

Images