Improved Bayesian network old travel building facility state evaluation system and method based on data imbalance processing, and storage medium

Abstract

The invention provides a Bayesian network state evaluation system and method based on data imbalance processing improvement for old travel building facilities, and a storage medium. The evaluation method comprises the following steps: extracting multi-dimensional influence factors and index data of to-be-detected old travel building facilities; carrying out data cleaning, coding and unbalanced data processing to obtain a trainable data set; key characteristic parameters are determined through correlation analysis and stepwise regression analysis; and constructing a Bayesian network model, and importing the key feature parameters into the trained model to obtain an old travel building facility state evaluation result. According to the method, key feature parameters are screened in advance, correlation and stepwise regression analysis is carried out, then an improved sample resampling technology is applied, the model can better reflect the real condition of the state of the old travel building facility, an improved data balance method is applied to weaken poor evaluation model performance caused by data imbalance, and the evaluation accuracy is improved. And comprehensive assessment of the state of the old travel building facility is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of building facility status evaluation, and in particular relates to a Bayesian network old cultural tourism building facility status evaluation system, method, and storage medium based on improved data imbalance processing. Background Art

[0002] With the continuous development of social economy and the booming cultural tourism industry, the status of facilities in old cultural and tourism buildings has attracted more and more attention. Over time, the problem of facility aging has become increasingly prominent. Environmental factors such as tourist use, natural disasters, climate change, and management problems such as irregular management and untimely maintenance are all important reasons for the aging and damage of facilities. The existing evaluation of facility status mostly relies on visual inspection or physical testing, which is easily affected by subjective factors and limitations, resulting in evaluation results that are not objective and scientific enough. In addition, the prior art does not fully consider the complex correlations between various potential factors when analyzing the status of building facilities, and lacks a comprehensive evaluation of the overall status, resulting in a lack of comprehensiveness and accuracy in the evaluation results, making it difficult to provide effective decision-making references for the maintenance and management of cultural and tourism building facilities. In order to solve the above technical problems, the present invention designs a Bayesian network old cultural and tourism building facility status evaluation system, method and storage medium based on data imbalance processing. Summary of the invention

[0003] In view of the shortcomings in the prior art, the present invention provides a Bayesian network old cultural and tourism building facility status evaluation system, method, and storage medium based on improved data imbalance processing, which pre-screens indicators to improve the accuracy of the status evaluation of old cultural and tourism building facilities and provide more effective and comprehensive decision-making support for their management. By comprehensively considering multi-factor data, correlation analysis and stepwise regression analysis are performed on the data to determine key characteristic parameters. In view of the data imbalance problem, an improved SMOTE technology is proposed, and then a facility status evaluation structure model based on the Bayesian network is constructed. Then, parameter learning is performed to obtain a facility status evaluation model, thereby achieving accurate evaluation of the status of old cultural and tourism building facilities.

[0004] The present invention achieves the above technical objectives through the following technical means.

[0005] A Bayesian network old cultural tourism building facility status evaluation method based on data imbalance processing improvement includes the following process:

[0006] Step 1: Extract multi-dimensional influencing factors and indicator data of the old cultural and tourism buildings and facilities to be tested;

[0007] Step 2: For the data extracted in step 1, obtain synthetic samples by performing missing value processing, outlier processing, encoding processing, and unbalanced data processing;

[0008] Step 3: Based on the original samples collected in step 1 and the synthetic samples generated in step 2, the key characteristic parameters of old cultural and tourism buildings and facilities are screened and determined through correlation analysis and stepwise regression analysis;

[0009] Step 4: Build a Bayesian network model;

[0010] Step 5: Input node status information into the trained model to evaluate the status of old cultural and tourism building facilities.

[0011] Furthermore, the data extracted in step 1 include building condition indicators, environmental condition indicators, operating condition indicators, facility management situation indicators, and evaluation result indicators; the building condition indicators include seven secondary indicators: main building materials, structural type, building age, building area, building height, building crack conditions, and building tilt conditions; the environmental condition indicators include five secondary indicators: geological stability, humidity impact, temperature impact, corrosion risk, and traffic vibration impact; the operating condition indicators include three secondary indicators: maximum number of people, daily passenger density, and peak passenger density; the facility management situation indicators include three secondary indicators: facility inspection frequency, facility maintenance frequency, and completeness of emergency plans; the evaluation result indicators include three indicators: safety, usability, and comprehensive status; main building materials, structural type, building crack conditions, building tilt conditions, geological stability, humidity impact, temperature impact, corrosion risk, traffic vibration impact, completeness of emergency plans, safety, usability, and comprehensive status are categorical variables, and the rest are continuous variables.

[0012] Furthermore, in step 2, the specific process of unbalanced data processing is as follows:

[0013] 80% of the samples are taken as the training set, and the remaining 20% ​​are used as the test set. In the face of the imbalanced data categories in the training set, the oversampling technology based on k-means clustering is used to balance the category distribution. First, the k-means clustering algorithm is applied to the minority class and majority class samples respectively, and small clusters are filtered. Samples are generated for large clusters to avoid expanding noise, so that the number of majority class samples is increased to 1.2 times the original number, so that the number of minority class samples is balanced with the number of majority class samples;

[0014] Set the number of majority class samples to D and the number of large clusters after clustering to d k , each large cluster contains X samples d1 ,X d2 ,…,X dk , the number of samples in the small cluster is X d′ ; The number of minority class samples is E, and the number of large clusters after clustering is e k , each large cluster contains X samples e1 ,X e2 ,…,Xek , the number of samples in the small cluster is X e′ Then, the synthetic minority class oversampling algorithm is used to generate synthetic samples. The specific algorithm flow is as follows:

[0015] x new =x+rand(0,1)×(x′-x)

[0016] In the formula, x new is a synthetic sample, x is an original sample, and x′ is one of the q nearest neighbors to sample x;

[0017] For the majority class samples, the number of new samples synthesized for each large cluster is

[0018] For each large cluster of minority class samples, the number of new samples synthesized is

[0019] Furthermore, in step 3, the specific process of correlation analysis is as follows:

[0020] Categorical variables are divided into unordered categorical variables and ordered categorical variables. Unordered categorical variables include the main building materials and structural types, and the rest are ordered categorical variables. The Spearman correlation coefficient of continuous variables and ordered categorical variables is calculated, and the indicators with weak correlation of old cultural and tourism building facilities are eliminated according to the calculation results.

[0021] Then, the Cramer correlation coefficient was used to determine the degree of correlation between the unordered categorical variables and the evaluation result indicators in the sample of old cultural and tourism building facilities: first, a chi-square test was performed to determine whether there was a correlation between the unordered categorical variables and the evaluation result indicators in the status indicators of old cultural and tourism building facilities, and then the chi-square distribution table was queried to obtain the P value to determine whether the difference between the actual data and the expected data was statistically significant; based on the above results, the Cramer correlation coefficient was calculated to obtain the relationship degree information. The Cramer correlation coefficient V ranged from [0,1]. When V=0, it means that the unordered categorical variables have no correlation with the evaluation result indicators; when V=1, it means that the unordered categorical variables are completely correlated with the evaluation result indicators; finally, according to the rules, variables with less correlation with the facility status were eliminated.

[0022] Furthermore, in step 3, the specific process of stepwise regression analysis is as follows:

[0023] Set the dependent variable Y and p independent variables X1, X2, …, X j ,…,X p , consider the two processes of introducing new variables into the regression model and removing insignificant variables from it;

[0024] Consider the steps to introduce new variables: First calculate the independent variable X that is not in the regression equation. j The partial regression sum of squares Pj·(1,2,…,p) ;P j·(1,2,…,p) The larger the independent variable is, the greater the contribution of the corresponding independent variable to the dependent variable is. After finding the independent variable that has the greatest impact on the dependent variable Y, perform a significance test on the regression coefficient. If the result is significant, introduce the variable and proceed to the step of considering whether the variable can be eliminated. If it is not significant, the process of gradually screening variables ends.

[0025] Consider whether to eliminate variables: Calculate the partial regression sum of squares of the independent variables in the regression equation, find the independent variable with the smallest impact on the dependent variable Y, and then perform a significance test on the regression coefficient. If it is not significant, eliminate the variable, re-establish the regression equation of Y and the remaining variables, and then test whether the independent variable with the smallest impact on the dependent variable Y in the equation can be eliminated, until there are no variables that can be eliminated in the equation, and then proceed to consider whether to introduce new variables; if it is significant, the variable cannot be eliminated, and proceed to consider whether to introduce new variables;

[0026] Repeat the above two steps until no more variables are introduced or eliminated.

[0027] Furthermore, in step 4, after constructing the Bayesian network model, the key feature parameters obtained in step 3 are used as parent nodes, safety and usability are used as child nodes, and the comprehensive status is used as the child nodes of safety and usability. Then, the network is drawn according to the correlation analysis results; the continuous variables in the key feature parameters obtained in step 3 are screened out, including building age, building height, maximum number of people, and daily crowd density, and the intervals are divided according to their value ranges and label encoding is performed; then, the sample data set containing all the key feature parameters is sorted out, divided into a training set and a test set, and the training set samples are imported into the model for parameter learning to obtain a probability distribution table for each node, which lists the probability of each node occurring under all combinations of its parent node states as a basis for obtaining subsequent test results.

[0028] Furthermore, in step 5, the test set is imported, and the node status information with key feature parameters as nodes is entered, and finally the probability distribution result with each evaluation result indicator as the node is obtained. The result reflects the probability of usability, safety, and comprehensive status being rated as three levels of ABC respectively. The level with the largest probability value is taken as the final evaluation result to realize the evaluation of the status of old cultural and tourism building facilities.

[0029] A Bayesian network old cultural tourism building facility status evaluation system based on improved data imbalance processing, including:

[0030] Data collection module, used to collect multi-dimensional influencing factors and indicator data of old cultural and tourism buildings and facilities;

[0031] The preprocessing module is used to preprocess the data collected by the data collection module. The preprocessing includes missing value processing, outlier processing, text data label encoding processing, and unbalanced data processing in sequence, and finally obtains a synthetic sample;

[0032] The key characteristic parameter screening module, based on the original samples and synthetic samples, screens and determines the key characteristic parameters of old cultural and tourism buildings and facilities through correlation analysis and stepwise regression analysis;

[0033] The model building module is used to build a Bayesian network model, import key feature parameters into the trained model, and obtain the evaluation results of the status of old cultural and tourism building facilities.

[0034] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the improved Bayesian network method for evaluating the status of old cultural and tourism building facilities based on data imbalance processing as described in claim 1.

[0035] The present invention has the following beneficial effects:

[0036] The present invention evaluates the facility status of old cultural and tourism buildings based on Bayesian networks, making full use of the modeling ability of Bayesian networks for uncertainty and diversity factors in complex systems, and improving the comprehensiveness and accuracy of the evaluation. By pre-screening key feature parameters and performing correlation and stepwise regression analysis, and applying improved sample resampling technology after screening key feature parameters, the model can better reflect the true situation of the status of old cultural and tourism building facilities. Compared with the traditional method that relies on visual inspection and physical testing, the present invention can apply an improved data balancing method on the basis of comprehensively considering the complex correlation between various potential factors, weaken the poor performance of the evaluation model caused by data imbalance, and then realize a comprehensive evaluation of the status of old cultural and tourism building facilities, providing more effective decision support for protection and management. The present invention reduces the manual intervention and errors of traditional inspection methods through an automated Bayesian network model, and improves the convenience and accuracy of the evaluation. At the same time, since the evaluation of the status of old cultural and tourism building facilities is more accurate, potential problems can be predicted in advance and maintenance costs can be reduced.

[0037] By applying the method of the present invention to the cultural tourism industry, it is not only possible to effectively extend the life of old cultural tourism buildings and facilities, but also to ensure their sustainable development. Through scientific and comprehensive evaluation results, relevant departments can take effective measures to maintain them in a timely manner, thereby promoting the sustainable development of old cultural tourism buildings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flow chart of a method for evaluating the condition of old cultural and tourism building facilities based on a Bayesian network in Example 1;

[0039] Figure 2(a) shows the basic parameters in the Spearman correlation coefficient of the sequential variable parameters of the status characteristics of old cultural and tourism building facilities;

[0040] Figure 2(b) shows the evaluative parameters in the Spearman correlation coefficient of the ordinal variable parameters of the status characteristics of old cultural and tourism building facilities;

[0041] Figure 3 The chi-square test results of the parameters of the unordered categorical variables of the status characteristics of old cultural and tourism building facilities;

[0042] Figure 4 A Bayesian network model for the status of old cultural and tourism building facilities;

[0043] Figure 5 This is the evaluation result of the status of old cultural and tourism building facilities. DETAILED DESCRIPTION

[0044] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments, but the protection scope of the present invention is not limited thereto.

[0045] Reference Figure 1 The improved Bayesian network old cultural tourism building facility status evaluation method based on data imbalance processing of the present invention includes the following process:

[0046] Step 1: Extract the multi-dimensional influencing factors and indicators of the old cultural and tourism buildings and facilities to be tested and collect data;

[0047] The multi-dimensional influencing factors and indicators of the old cultural and tourism buildings and facilities are comprehensively determined based on relevant policy documents, journal articles, expert interviews, questionnaires and other diversified methods, mainly considering that the status of cultural and tourism facilities is affected by the first-level indicators of the building itself, environment, operation and management; among them, the building condition indicators include seven secondary indicators: main building materials, structural type, building age, building area, building height, building cracks, and building tilt; environmental condition indicators include five secondary indicators: geological stability, humidity impact, temperature impact, corrosion risk, and traffic vibration impact; operational condition indicators include three secondary indicators: maximum number of people, daily passenger density, and peak passenger density; facility management indicators include three secondary indicators: facility inspection frequency, facility maintenance frequency, and completeness of emergency plans; evaluation result indicators include three indicators: safety, usability, and comprehensive status. Among them, main building materials, structural type, building cracks, building tilt, geological stability, humidity impact, temperature impact, corrosion risk, traffic vibration impact, completeness of emergency plans, safety, usability, and comprehensive status are categorical variables, and the rest are continuous variables.

[0048] The multi-dimensional indicator data of old cultural and tourism buildings and facilities collected comes from the cultural and tourism management departments and is mainly obtained through surveys.

[0049] Step 2: For the data collected in step 1, obtain an effective trainable data set by performing data cleaning and encoding preprocessing, which specifically includes the following steps:

[0050] Step 2.1: Missing value processing;

[0051] Check the missing values ​​of the data. For data with less missing values, use the mean (numeric data) or mode (categorical data) to fill in the missing values, and delete the data with serious missing values.

[0052] Step 2.2: Outlier processing;

[0053] Identify outliers in the indicator data of old cultural and tourism buildings and facilities through box plots and standard deviation methods, and delete or replace the detected outliers to ensure that the accuracy of the data is not affected by outliers;

[0054] Step 2.3: Text data label encoding;

[0055] The categorical variables described in step 1 are divided into unordered categorical variables and ordered categorical variables. The unordered categorical variables include the main building materials and structural types, and the rest are ordered categorical variables. The categorical variables in the above text form are converted into numerical data through label encoding for subsequent model processing;

[0056] Step 2.4: Imbalanced data processing;

[0057] 80% of the samples are taken as the training set and the remaining 20% ​​as the test set. In the face of the imbalanced data categories in the training set, the oversampling technology based on k-means clustering is used to balance the category distribution. First, the k-means clustering algorithm is applied to the minority class and majority class samples respectively, and small clusters are filtered out. Samples are generated for large clusters to avoid expanding noise, and the number of majority class samples is increased to 1.2 times the original number, so that the number of minority class samples is balanced with the number of majority class samples.

[0058] Set the number of majority class samples to D and the number of large clusters after clustering to d k , each large cluster contains X samples d1 ,X d2 ,…,X dk , the number of samples in the small cluster is X d′ ; The number of minority class samples is E, and the number of large clusters after clustering is e k , each large cluster contains X samples e1 ,X e2 ,…,X ek , the number of samples in the small cluster is X e′Then, the synthetic minority oversampling technique (SMOTE) algorithm is used to generate synthetic samples. The specific algorithm flow is as follows:

[0059] x new =x+rand(0,1)×(x′-x)

[0060] In the formula, x new is a synthetic sample, x is an original sample, and x′ is one of the q nearest neighbors to sample x.

[0061] For the majority class samples, the number of new samples synthesized for each large cluster is

[0062] For each large cluster of minority class samples, the number of new samples synthesized is

[0063] Step 3: Based on the original samples collected in step 1 and the synthetic samples generated in step 2.4, the key characteristic parameters of old cultural and tourism buildings and facilities are screened and determined through correlation analysis and stepwise regression analysis, and an indicator system is established; the specific steps include the following:

[0064] Step 3.1: Correlation analysis;

[0065] Referring to Figure 2(a) and Figure 2(b), the Spearman correlation coefficient of the ordinal variables (i.e., continuous variables and ordered categorical variables, such as building age, building cracks, etc.) is calculated. The Spearman correlation coefficient is a rank correlation coefficient and is applicable to ordered data. The coefficient range is [-1, 1]. Positive and negative represent positive and negative correlation. The closer the absolute value of the coefficient is to 1, the stronger the correlation between the variables. According to the calculation results, the indicators with weak correlation of old cultural and tourism building facilities are eliminated. The specific algorithm formula is as follows:

[0066]

[0067] Where ρ is the Spearman correlation coefficient, x ij is the jth indicator data of the i-th sample, x ik represents the kth indicator data of the i-th sample, is the mean value of the kth indicator, x j is the mean of the jth indicator; m and n represent the sample size and the number of indicators respectively;

[0068] Reference Figure 3 Then, Cramer's V was used to determine the correlation between the unordered categorical variables and the evaluation result indicators (such as the main building materials and comprehensive status, structural type and comprehensive status, etc.) in the samples of old cultural and tourism building facilities:

[0069] First, a chi-square test is performed to determine whether there is a correlation between the unordered categorical variables in the old cultural and tourism building facility status indicators and the evaluation result indicators. The calculation formula is as follows:

[0070]

[0071] In the formula, A is the actual frequency; E is the expected frequency; χ 2 represents the chi-square value; in addition, the degrees of freedom need to be calculated, the formula is as follows:

[0072] DF=(r-1)(c-1)

[0073] Where DF is the degree of freedom, r and c are the number of categories of the two variables.

[0074] Combined with the chi-square value calculated above, the chi-square distribution table is queried to obtain the P value to determine whether the difference between the actual data and the expected data is statistically significant. The calculation formula is as follows:

[0075] P = 1-F (χ 2 ,DF)

[0076] Where F is the cumulative distribution function of the chi-square distribution.

[0077] According to the above results, Cramer's V coefficient is calculated to obtain the relationship degree information. The calculation formula is as follows:

[0078]

[0079] In the formula, m is the sample size; r and c are the number of categories of the first and second variables respectively; Cramer's V coefficient (V) ranges from [0,1]. When V = 0, it means that the unordered categorical variable has no correlation with the evaluation result index; when V = 1, it means that the unordered categorical variable is completely correlated with the evaluation result index. Finally, according to the rules, the variables with less correlation with the facility status are eliminated.

[0080] Step 3.2: Stepwise regression analysis;

[0081] Set the dependent variable Y and p independent variables X1, X2, …, X j ,…,X p To meet the requirements of the multiple linear regression model, consider the two processes of introducing new variables into the regression model and removing insignificant variables from it;

[0082] Consider the steps to introduce new variables: First calculate the independent variable X that is not in the regression equation. j The partial regression sum of squares P j·(1,2,…,p) , the calculation formula is as follows:

[0083]

[0084] in, Indicates that X j The regression sum of squares when is the independent variable; represents the regression sum of squares when all independent variables are used as independent variables at the same time; m represents the sample size; u represents the number of independent variables; P j·(1,2,…,p) The larger it is, the greater the contribution of the corresponding independent variable to the dependent variable.

[0085] After finding the independent variable that has the greatest impact on the dependent variable Y, perform a t-test on the regression coefficient. If the result is significant (p<0.05), introduce the variable and proceed to the step of considering whether the variable can be eliminated. If it is not significant (p≥0.05), the process of gradually screening variables ends.

[0086] Consider whether to eliminate variables: Calculate the partial regression sum of squares of the independent variables in the regression equation, find the independent variable with the smallest impact on the dependent variable Y, and then perform a significance test on the regression coefficient. If it is not significant (p≥0.05), eliminate the variable and re-establish the regression equation of Y and the remaining variables. Then test whether the independent variable with the smallest impact on the dependent variable Y in the equation can be eliminated, until there are no variables that can be eliminated in the equation, and then proceed to consider whether new variables can be introduced. If it is significant (p<0.05), the variable cannot be eliminated, and proceed to consider whether new variables can be introduced.

[0087] Repeat the above two processes until no more variables are introduced or eliminated.

[0088] The above p-values ​​are all used to determine whether the results are significant, but the application scenarios and calculation methods are different.

[0089] Step 3.3: Based on the calculation results of steps 3.1 and 3.2 above, the retained variables are used to summarize the original variables as the key characteristic parameters of the status of old cultural and tourism building facilities, including the completeness of emergency plans, geological stability, humidity impact, daily crowd density, maximum number of people, building inclination, main building materials, building age, and building height.

[0090] Step 4: Build in Genie software Figure 4The Bayesian network model shown in the figure takes the key feature parameters obtained in step 3 as network nodes. Specifically, the key feature parameters are parent nodes, safety and usability are their child nodes, and the comprehensive status is the child node of safety and usability. Then, the network is drawn according to the correlation analysis results; the continuous variables in the key feature parameters obtained in step 3 are screened out, including building age, building height, maximum number of people, and daily pedestrian density, and the intervals are divided according to their value ranges and label encoding is performed; then the sample data set containing all the key feature parameters is sorted into a training set and a test set, wherein the test set is the 20% sample set divided from the original sample in step 2.4, and the training set is the collection of the 80% sample set divided from the original sample in step 2.4 and the synthetic sample.

[0091] The training set samples are imported into the model to perform parameter learning and obtain the probability distribution table of each node. The table lists the probability of each node occurring under all combinations of its parent node states, which serves as the basis for obtaining subsequent test results.

[0092] Step 5: Import the test set and enter the node status information with key feature parameters as nodes (for example, the completeness of the emergency plan is level B). Finally, the probability distribution results with each evaluation result indicator as the node are obtained. The results reflect the probability of usability, safety, and comprehensive status being rated as three levels of ABC respectively. The level with the largest probability value is taken as the final evaluation result to realize the evaluation of the status of old cultural and tourism building facilities.

[0093] According to the above method, this embodiment obtains the following after actual data testing: Figure 5 The sample true state and predicted results shown in the figure show that Figure 5 Among the 10 test sets, 5 samples with security level A were correctly predicted, 2 samples with security level B were predicted as level B, and 2 of the 3 samples with security level C were predicted as level A and 1 was predicted correctly; 1 of the 5 samples with usability level A was predicted as level B and 4 were predicted correctly, 3 samples with usability level B were predicted correctly, and 1 of the 2 samples with usability level C was predicted as level A and 1 was predicted correctly; 3 samples with comprehensive status level A were predicted correctly, 1 of the 2 samples with comprehensive status level B was predicted as level C and 1 was predicted correctly, and 1 of the 6 samples with comprehensive status level C was predicted as level B and 4 were predicted correctly.

[0094] This embodiment also provides a Bayesian network old cultural tourism building facility status evaluation system based on data imbalance processing improvement, including:

[0095] The data collection module is used to collect multi-dimensional influencing factors and indicator data of old cultural and tourism buildings and facilities;

[0096] The preprocessing module is used to preprocess the data collected by the data collection module. The preprocessing includes missing value processing, outlier processing, text data label encoding processing, and unbalanced data processing in sequence, and finally obtains synthetic samples;

[0097] The key characteristic parameter screening module determines the key characteristic parameters of old cultural and tourism buildings and facilities through correlation analysis and stepwise regression analysis based on original samples and synthetic samples;

[0098] The model building module is used to construct a Bayesian network model, import key feature parameters into the trained model, and obtain the evaluation results of the status of old cultural and tourism building facilities.

[0099] This embodiment also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the above-mentioned Bayesian network old cultural and tourism building facility status evaluation method improved based on data imbalance processing are implemented.

[0100] The embodiments are preferred implementations of the present invention, but the present invention is not limited to the above-mentioned implementations. Any obvious improvements, substitutions or modifications that can be made by those skilled in the art without departing from the essential content of the present invention belong to the protection scope of the present invention.

Claims

1. A Bayesian network old cultural and tourism building facility status evaluation method based on improved data imbalance processing, characterized in that: The process includes the following: Step 1: Extract multi-dimensional influencing factors and indicator data of the old cultural and tourism buildings and facilities to be tested; Step 2: For the data extracted in step 1, obtain synthetic samples by performing missing value processing, outlier processing, encoding processing, and unbalanced data processing; Step 3: Based on the original samples collected in step 1 and the synthetic samples generated in step 2, the key characteristic parameters of old cultural and tourism buildings and facilities are screened and determined through correlation analysis and stepwise regression analysis; Step 4: Build a Bayesian network model; Step 5: Input node status information into the trained model to evaluate the status of old cultural and tourism building facilities.

2. The improved Bayesian network old cultural and tourism building facility status evaluation method based on data imbalance processing according to claim 1 is characterized in that: The data extracted in step 1 include building condition indicators, environmental condition indicators, operation condition indicators, facility management condition indicators, and evaluation result indicators; Building condition indicators include seven secondary indicators: main building materials, structure type, building age, building area, building height, building cracks, and building tilt; environmental condition indicators include five secondary indicators: geological stability, humidity impact, temperature impact, corrosion risk, and traffic vibration impact; operating condition indicators include three secondary indicators: maximum number of people, daily passenger density, and peak passenger density; facility management condition indicators include three secondary indicators: facility inspection frequency, facility maintenance frequency, and completeness of emergency plans; evaluation result indicators include three indicators: safety, usability, and comprehensive status; main building materials, structure type, building cracks, building tilt, geological stability, humidity impact, temperature impact, corrosion risk, traffic vibration impact, completeness of emergency plans, safety, usability, and comprehensive status are categorical variables, and the rest are continuous variables.

3. The improved Bayesian network old cultural and tourism building facility status evaluation method based on data imbalance processing according to claim 1 is characterized in that: In step 2, the specific process of unbalanced data processing is as follows: 80% of the samples are taken as the training set, and the remaining 20% ​​are used as the test set. In the face of the imbalanced data categories in the training set, the oversampling technology based on k-means clustering is used to balance the category distribution. First, the k-means clustering algorithm is applied to the minority class and majority class samples respectively, and small clusters are filtered. Samples are generated for large clusters to avoid expanding noise, so that the number of majority class samples is increased to 1.2 times the original number, so that the number of minority class samples is balanced with the number of majority class samples; Set the number of majority class samples to D and the number of large clusters after clustering to d k , each large cluster contains X samples d1 ,X d2 ,…,X dk , the number of samples in the small cluster is X d′ ; The number of minority class samples is E, and the number of large clusters after clustering is e k , each large cluster contains X samples e1 ,X e2 ,…,X ek , the number of samples in the small cluster is X e′ Then, the synthetic minority class oversampling algorithm is used to generate synthetic samples. The specific algorithm flow is as follows: x new =x+rand(0,1)×(x′-x) In the formula, x new is a synthetic sample, x is an original sample, and x′ is one of the q nearest neighbors to sample x; For the majority class samples, the number of new samples synthesized for each large cluster is For each large cluster of minority class samples, the number of new samples synthesized is 4. The improved Bayesian network old cultural and tourism building facility status evaluation method based on data imbalance processing according to claim 2 is characterized in that: In step 3, the specific process of correlation analysis is as follows: Categorical variables are divided into unordered categorical variables and ordered categorical variables. Unordered categorical variables include the main building materials and structural types, and the rest are ordered categorical variables. The Spearman correlation coefficient of continuous variables and ordered categorical variables is calculated, and the indicators with weak correlation of old cultural and tourism building facilities are eliminated according to the calculation results. Then, the Cramer correlation coefficient was used to determine the degree of correlation between the unordered categorical variables and the evaluation result indicators in the sample of old cultural and tourism building facilities: first, a chi-square test was performed to determine whether there was a correlation between the unordered categorical variables and the evaluation result indicators in the status indicators of old cultural and tourism building facilities, and then the chi-square distribution table was queried to obtain the P value to determine whether the difference between the actual data and the expected data was statistically significant; based on the above results, the Cramer correlation coefficient was calculated to obtain the relationship degree information. The Cramer correlation coefficient V ranged from [0, 1]. When V = 0, it means that the unordered categorical variables have no correlation with the evaluation result indicators; when V = 1, it means that the unordered categorical variables are completely correlated with the evaluation result indicators; finally, according to the rules, variables with less correlation with the facility status are eliminated.

5. The improved Bayesian network old cultural and tourism building facility status evaluation method based on data imbalance processing according to claim 2 is characterized in that: In step 3, the specific process of stepwise regression analysis is as follows: Set the dependent variable Y and p independent variables X1, X2, ..., X j , ..., X p , consider the two processes of introducing new variables into the regression model and removing insignificant variables from it; Consider the steps to introduce new variables: First calculate the independent variable X that is not in the regression equation. j The partial regression sum of squares P j·(1，2，...，p) ;P j·(1，2，...，p) The larger the independent variable is, the greater the contribution of the corresponding independent variable to the dependent variable is. After finding the independent variable that has the greatest impact on the dependent variable Y, perform a significance test on the regression coefficient. If the result is significant, introduce the variable and proceed to the step of considering whether the variable can be eliminated. If it is not significant, the process of gradually screening variables ends. Consider whether to eliminate variables: Calculate the partial regression sum of squares of the independent variables in the regression equation, find the independent variable with the smallest impact on the dependent variable Y, and then perform a significance test on the regression coefficient. If it is not significant, eliminate the variable, re-establish the regression equation of Y and the remaining variables, and then test whether the independent variable with the smallest impact on the dependent variable Y in the equation can be eliminated, until there are no variables that can be eliminated in the equation, and then proceed to consider whether to introduce new variables; if it is significant, the variable cannot be eliminated, and proceed to consider whether to introduce new variables; Repeat the above two steps until no more variables are introduced or eliminated.

6. The improved Bayesian network old cultural and tourism building facility status evaluation method based on data imbalance processing according to claim 2 is characterized in that: In step 4, after constructing the Bayesian network model, the key feature parameters obtained in step 3 are used as parent nodes, security and usability are used as child nodes, and the comprehensive status is used as the child nodes of security and usability. Then, the network is drawn according to the correlation analysis results; Filter out the continuous variables in the key feature parameters obtained in step 3, including building age, building height, maximum number of passengers, and daily crowd density, divide the intervals according to their value ranges and perform label encoding; then organize the sample data set containing all key feature parameters into training sets and test sets, import the training set samples into the model, perform parameter learning, and obtain the probability distribution table of each node. The table lists the probability of each node occurring under all combinations of its parent node states, which serves as the basis for obtaining subsequent test results.

7. The improved Bayesian network old cultural tourism building facility status evaluation method based on data imbalance processing according to claim 6 is characterized in that: In step 5, the test set is imported, and the node status information with key feature parameters as nodes is entered, and finally the probability distribution result with each evaluation result indicator as the node is obtained. The result reflects the probability of usability, safety, and comprehensive status being rated as three levels of ABC respectively. The level with the largest probability value is taken as the final evaluation result to realize the evaluation of the status of old cultural and tourism building facilities.

8. A Bayesian network old cultural and tourism building facility status evaluation system based on improved data imbalance processing, characterized in that: include: Data collection module, used to collect multi-dimensional influencing factors and indicator data of old cultural and tourism buildings and facilities; The preprocessing module is used to preprocess the data collected by the data collection module. The preprocessing includes missing value processing, outlier processing, text data label encoding processing, and unbalanced data processing in sequence, and finally obtains a synthetic sample; The key characteristic parameter screening module, based on the original samples and synthetic samples, screens and determines the key characteristic parameters of old cultural and tourism buildings and facilities through correlation analysis and stepwise regression analysis; The model building module is used to build a Bayesian network model, import key feature parameters into the trained model, and obtain the evaluation results of the status of old cultural and tourism building facilities.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the Bayesian network method for evaluating the status of old cultural and tourism building facilities based on improved data imbalance processing as described in claim 1.

Images