Research on fire risk assessment model in commercial and residential communities

: The causes of fire in commercial and residential communities are complex, rescue is difficult, once a fire occurs, the loss is difficult to estimate, and a reasonable fire risk assessment needs to be made. This paper establishes a complete index system for fire risk assessment in commercial and residential communities, establishes a cloud model by using the ISM-Shapley value method, evaluates the safety level of a commercial and residential community in Jinan


GENERAL INSTRUCTIONS
With the acceleration of China's urbanization process, ordinary residential communities can no longer meet the new lives of current residents, and residential communities suitable for living and business have become the current development trend in China.However, according to statistics, in the past 10 years of national fire situation, residential places are more prominent.From 2012 to 2021, a total of 1.324 million residential fires occurred nationwide, resulting in 11,634 deaths, 6,738 injuries and direct property losses of 7.77 billion yuan, including 429 large fires, resulting in 1,579 deaths, and 2 major fires, resulting in 26 deaths.As a high-incidence fire accident area, commercial and residential small is bound to cause casualties and property damage, so assessing the risk and safety level of fire occurrence has become an important means to judge whether the building is safe.
Scholars at home and abroad have made the following achievements in building fire risk assessment: A.Bell [1] proposed a multi-mode hybrid neural network for building fire risk prediction, using Shapley value and ablation research to evaluate traditional and novel image features.O. Rachid [2] using artificial neural network models to predict potential fire impacts helps decisionmakers strengthen and invest in specific fire safety strategies accordingly.M.D.B [3] developed the BN model to study fire development in ordinary homes down to advanced fire situations.Chen [4] A multi-layered regional model for predicting the development of a single indoor fire was established.The control equations for each layer are established through conservation of energy.Mi [5] conducted a comprehensive assessment of the two-dimensional quantity of building fire risk, and constructed the Fuzzy-DS model, which verified the effectiveness of the model.Zhou [6] used Bayes method and system fault tree analysis to obtain the probability of failure of large and medium-sized commercial buildings in China to control fire spread in the event of fire, which provided a basis for preventing and controlling fire spread.Zhang [7] constructed an assessment system for fire risk correlation factors in urban large-scale public buildings, and explored the mutual influence relationship between key risk factors.Yu [8] constructed a dynamic intelligent fire risk assessment model for old urban communities based on AHP-Bayes, realized the dynamic risk assessment of urban old communities.
In view of the current research status at home and abroad, most scholars have conducted fire risk research on single-function buildings, but there are few studies on fire risk assessment in commercial and residential communities, and there is a lack of fire risk model based on the logical relationship of community factors.In view of this, this paper establishes a complete index system for fire risk assessment in commercial and residential communities, and a cloud model based on the ISMshapley value method is a fire risk assessment model for commercial and residential communities.It provides a quantifiable means for its fire risk analysis.

Characteristics of fire in commercial and residential communities
The commercial and residential community in this article refers to a number of commercial and residential buildings with a commercial ground floor or several floors and residential buildings on it, forming a relatively independent area through commercial podiums and other buildings along the street, meeting the needs of residents' residence, entertainment and commerce, and its fire has the following characteristics:(1) The structure is complex, and there are many ways to spread fire.(2) The area is narrow, and the fire protection occupies many roads.(3) The fire load is large.(4) The personnel are complex, the evacuation difficulty is large.(5) Fire safety

Indicator system construction
Through the analysis of fire risk factors in commercial and residential communities, reference to research literature and regulations, combined with expert opinions, the fire risk assessment index system of commercial and residential communities was determined.There are 8 level 1 indicators and 34 level 2 indicators.See Table 1

Introduction to ISM model
The ISM GAO [9] model decomposes complex problems into several subsystem elements, analyzes the mutual influence relationship of constituent elements, clarifies the level and structure of problems, and forms a multilevel ladder model for the system.The steps are as follows: (1) Determine the adjacency matrix Use 0 and 1 to represent the relationship between the two elements and establish the adjacency matrix.If  has a direct effect on  ,  is 1; If  has no direct effect on  ,  is 0. When  ,  1 each element is autocorrelated.The result is expressed as an adjacency matrix A=  .
(2) Calculate the reachability matrix The reachability matrix represents the indirect effects between factors.A matrix can be obtained by adding A to the identity matrix I.The formula is as follow:

) (3) Hierarchical division
The intersection of the reachability set   of a constraint and the antecedent set   is compared with the reachability set, and if   ⋃    , the influencing factor is deleted at the top level, and the above steps are repeated to obtain an explanatory structural model.

Introduction to Shapley value
Some index factors are interrelated due to some interaction between fire risk indicators in commercial and residential communities.The Shapley method, based on the ISM explanatory structural model, was chosen to express the degree to which multiple factors influence the overall goal [10].
(1) Normalize the mean of the original data and calculate the risk value of the index  ∑ (2) Calculate the impact value of each indicator combination (3) Calculate the weights of each indicator ℎ ,  ∑

Introduction to cloud models
The basic idea of the cloud model is: assuming that  is a quantitative domain and  is a qualitative expression on , if  ∈ , and  is still a random implementation on , and satisfies the membership degree of x to    ∈ 0,1 is a random number with a stable tendency.Then the distribution of   satisfies :  → 0,1 , ∀ ∈ ,  →   is called a single x droplet, and the distribution state of all  on  is called a cloud [11]: (1) Determine the collection of comments The evaluation index set  and comment set  are established, and the value range corresponding to the comment set  is [0,100].
(2) Build a cloud map of evaluation ratings The comment set is reduced to the standard cloud parameter  ,  ,  , and the transformation process is as follows.As shown in Table 2. (5)

Project overview
The project is located in Longshan Street, Jinan.The community has a planned land area of 11,214 square meters, a total construction area of 46,800 square meters, a total of 3 16-storey residential buildings, a total of 226 sets, the community is equipped with small four-storey commercial buildings along the street, and various supporting public facilities.

Build the ISM model diagram
(1) Using a matrix of 34×34, the adjacency matrix  is obtained as shown in Figure 1: (3) (2) Entering the adjacency matrix into the MATLAB program, and the resulting matrix  is: (4) The reachability matrix is divided into five levels through MATLAB software as shown in Table 3.The hierarchical diagram of the ISM model is shown in Figure 2.

Shapley value method weight determination
Based on the index portfolio association impact degree of the ISM explanatory structure model, the combined risk value is obtained by formula, as shown in the Table 4.

( 3 ) 4 )( 5 )
Calculate the cloud assembly of a single indicator The reverse cloud generator is used to convert the score of each indicator  into the characteristics of the cloud model  ,  ,  .Scoring data is based on the design of questionnaires that invite experts to assign scores to each risk indicator.Expert evaluation matrix for each indicator: cloud vector is calculated using the following equation   ,  ,  : Integrate evaluation cloud parameters Combined with the weight result  , the fire cloud parameters of commercial and residential communities are calculated as follows: Determine the level of risk MATLAB is used to generate risk cloud maps, and compare them with standard cloud maps to determine the fire risk level of commercial and residential communities.

Table 1 .
: Fire risk assessment index system table for commercial and residential communities.

Table 2 .
Evaluation parameters and standard cloud map.