Intelligent Analysis and Assessment Methods and Systems for Risk Perception Bias among Residents in Post-Disaster Reconstruction Areas

Abstract

This invention relates to the field of big data analytics, specifically to a method and system for intelligent analysis and assessment of risk perception bias among residents in post-disaster reconstruction areas. The method involves fitting risk indicators using multi-source data and a collaborative model, and calculating risk assessment values ​​based on the fitting results. It obtains basic information data, subjective perception data, and resident behavior data through structured questionnaires. Initial correction coefficients are matched to the subjective perception data based on the basic information data, and error analysis is performed using resident behavior data to determine whether dynamic compensation of the initial correction coefficients is necessary. This significantly improves the accuracy of the final obtained true subjective perception data. By assessing residents' risk perception bias from two perspectives—true subjective perception data and risk assessment values—the method ensures the accuracy and comprehensiveness of the final results, achieving intelligent analysis and accurate assessment of risk perception bias among residents in post-disaster reconstruction areas.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, specifically to a method and system for intelligent analysis and assessment of risk perception bias among residents in post-disaster reconstruction areas. Background Technology

[0002] With the increasing frequency of natural disasters and public emergencies worldwide, post-disaster reconstruction has become a crucial link in restoring social order and ensuring people's safety. During the reconstruction process, residents' perception of potential risks directly affects their psychological state, decision-making behavior, and cooperation with reconstruction policies.

[0003] Traditional objective risk indicator analysis techniques tend to overlook the spatiotemporal evolution of objective risks, which leads to an inability to accurately analyze the objective risk assessment status of post-disaster reconstruction areas. At the same time, existing resident risk perception bias assessment techniques are unable to dynamically adjust and correct compensation parameters based on the discrepancies between residents' subjective perceptions and actual behaviors, resulting in low accuracy of the final assessment results. Summary of the Invention

[0004] To address the problems in related technologies, this invention provides an intelligent analysis and assessment method and system for risk perception bias among residents in post-disaster reconstruction areas, thereby overcoming the aforementioned technical problems existing in existing related technologies.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a method for intelligent analysis and assessment of risk perception bias among residents in post-disaster reconstruction areas, comprising the following steps:

[0006] S1. Set monitoring intervals for post-disaster areas and regularly monitor the objective risk indicators corresponding to the post-disaster reconstruction areas to obtain objective risk indicator data.

[0007] S2. Fit the risk indicators of the post-disaster reconstruction area to the pre-constructed risk indicator weighted fitting model to obtain the risk assessment curve of the post-disaster reconstruction area and calculate the corresponding risk assessment value.

[0008] S3. Collect subjective perception data and resident behavior data of residents in the post-disaster reconstruction area;

[0009] S4. Correct and compensate the subjective perception data using a preset initial correction coefficient to obtain the true subjective perception data, and perform consistency analysis in conjunction with resident behavior data to obtain the compensation error data.

[0010] The compensation error discrimination data is output as normal only when the compensation error data is less than or equal to the preset error threshold, and the process proceeds directly to S5;

[0011] Otherwise, the initial correction coefficient is optimized and adjusted, and S4 is executed again until the compensation error data is less than or equal to the preset error threshold.

[0012] S5. Based on the actual subjective perception data and the risk assessment value, conduct a risk perception deviation assessment of residents to obtain risk perception deviation assessment data of residents in the post-disaster reconstruction area.

[0013] Preferably, the specific steps for regularly monitoring objective risk indicators corresponding to the post-disaster reconstruction area at a set monitoring interval to obtain objective risk indicator data are as follows:

[0014] S11. Set the post-disaster area monitoring interval, and use IoT sensors to periodically monitor various objective risk indicators corresponding to the target post-disaster reconstruction area according to the post-disaster area monitoring interval to obtain objective risk indicator data.

[0015] S12. Collect historical objective risk indicator data and corresponding actual risk assessment values ​​of several post-disaster reconstruction areas through the Internet of Things to obtain historical objective risk indicator datasets and actual risk assessment value datasets, and construct the final risk indicator weighted fitting model.

[0016] Preferably, the specific steps for fitting the risk indicators of the post-disaster reconstruction area to the pre-constructed risk indicator weighted fitting model, obtaining the risk assessment curve of the post-disaster reconstruction area, and calculating the corresponding risk assessment value are as follows:

[0017] S21. By combining the pre-constructed risk index weighted fitting model with the objective risk index data, the risk index of the target post-disaster reconstruction area is fitted to obtain the risk assessment curve of the post-disaster reconstruction area.

[0018] S22. Calculate the risk assessment value of the target post-disaster reconstruction area based on the risk assessment curve. The calculation formula is as follows:

[0019] ,

[0020] in, and These represent the end time and start time of post-disaster area monitoring, respectively. This indicates the time-related information in the risk assessment curve. The function.

[0021] By combining multi-source data with a pre-set model, the intuitive quantification of multi-dimensional objective risk indicators is achieved, ensuring the accurate measurement of risk assessment values ​​and providing a reliable data foundation for the subsequent assessment of residents' risk perception bias.

[0022] Preferably, the specific steps for collecting subjective perception data and resident behavior data from residents in the post-disaster reconstruction area are as follows:

[0023] S31. Collect basic information data, subjective perception data, and resident behavior data of residents in the post-disaster reconstruction area through structured questionnaires to obtain a basic information dataset of residents. Subjective perception dataset and resident behavior dataset ,in, , as well as These represent the results collected through structured questionnaires. Data on residents' basic information, subjective perceptions, and behavior in the post-disaster reconstruction area. This indicates the number of people who participated in the structured questionnaire survey.

[0024] Preferably, the subjective perception data is corrected and compensated using a preset initial correction coefficient to obtain true subjective perception data, and then combined with resident behavior data for consistency analysis to obtain compensated error data. Only when the compensated error data is less than or equal to a preset error threshold is the compensated error discrimination data output as normal, and the process proceeds directly to S5; otherwise, the initial correction coefficient is optimized and adjusted, and S4 is re-executed until the compensated error data is less than or equal to the preset error threshold. The specific steps are as follows:

[0025] S41. Set the correction coefficient dataset ,in, Indicates the first Correction coefficients for various subjective perception data corresponding to the basic information of residents. The total number of categories representing residents' basic information;

[0026] S42. The BERT language model algorithm is used to sequentially process the... The basic information data of each resident in the data and the data mentioned above The correction coefficient data in the database is matched with characters to find correction coefficients that match the basic information data of each resident, and these correction coefficients are then labeled to obtain the initial set of correction coefficients. ,in, Indicates the first Correction coefficients for various subjective perception data corresponding to the basic information data of each resident;

[0027] S43, according to the above The initial correction coefficients in the above are for the The corresponding subjective perception data is corrected and compensated to obtain the true subjective perception dataset. ,in, Indicates the first Real subjective perception data corresponding to residents in the post-disaster reconstruction area;

[0028] S44, By combining the pre-built final compensation error mapping model with the above and stated Consistency analysis was performed to obtain the compensation error dataset. ,in, Indicates the first Compensation error data for the real subjective perception data of residents in the post-disaster reconstruction area;

[0029] Set an error threshold, and then... Each compensation error data in the data is sequentially compared with the error threshold;

[0030] If the compensation error data is greater than the error threshold, the output compensation error discrimination data is abnormal;

[0031] Otherwise, the output error compensation judgment data is normal;

[0032] Traversing the The compensation error discrimination dataset is obtained by collecting all the compensation error data. ,in, Indicates the first Compensation error discrimination data for the real subjective perception data of residents in the post-disaster reconstruction area;

[0033] S45, if the above If all the compensation error discrimination data output is normal, proceed directly to S5; otherwise, optimize and adjust the initial correction coefficients corresponding to the compensation error discrimination data with abnormal output using an intelligent optimization algorithm, and re-execute S4 until the process is complete. All the compensation error discrimination data in the output are normal.

[0034] The optimization and adjustment of the initial correction coefficients corresponding to the compensation error discrimination data with abnormal output using an intelligent optimization algorithm includes the following steps:

[0035] S451, Set the current iteration number to... The maximum number of iterations is And the search space dimension of the correction coefficient data is Randomly generated in the data search space for optimizing the correction coefficients. Each set of correction coefficients is used to optimize the data, and each set of correction coefficients corresponds to a set of correction coefficients, thus obtaining the correction coefficient optimization dataset.

[0036] S452. Calculate the fitness value of each modified coefficient optimized data in the modified coefficient optimized dataset according to the fitness function formula, sort each modified coefficient optimized data in the modified coefficient optimized dataset according to the fitness value from largest to smallest, and select the modified coefficient optimized data with the highest fitness value as the current optimal solution;

[0037] S453. Set update strategy parameters, and compare the current iteration number with the update strategy parameters.

[0038] S4531. If the current iteration number is less than or equal to the update strategy parameter, then each correction coefficient optimization data in the correction coefficient optimization dataset updates its own speed.

[0039] Each data point in the modified coefficient optimized dataset updates its position around its own location in the modified coefficient optimized data search space at the updated speed.

[0040] S4532. If the current iteration number is greater than the update strategy parameter, then the modified coefficient optimization dataset is updated, retaining the dataset with higher fitness values. Optimize the data with correction coefficients to obtain the updated optimized dataset with correction coefficients, and calculate the center position of the updated optimized dataset with correction coefficients.

[0041] In the updated modified coefficient optimized dataset, each modified coefficient optimized data point is updated in the search space of the modified coefficient optimized data according to its center position and its own position.

[0042] S454. Calculate the fitness value of each modified coefficient optimized data in the modified coefficient optimized dataset after position update according to the fitness function formula. If the fitness value of the modified coefficient optimized data after position update is greater than the original fitness value, replace the original position with the new position of the modified coefficient optimized data; otherwise, retain the original position.

[0043] S455, Determine the above Is it greater than or equal to the stated If the above Greater than or equal to the The correction coefficient with the highest fitness value is then output as the optimized correction coefficient.

[0044] By dynamically matching initial correction coefficients based on residents' basic information, subjective perception data is compensated and corrected according to the initial correction coefficients. At the same time, compensation error analysis is conducted using residents' actual behavior as a reference, which effectively reduces the random error in data acquisition and more accurately uncovers residents' true subjective perception data. In particular, the initial correction coefficients of data with large compensation errors are optimized and adjusted through intelligent optimization algorithms. Through multiple iterations of optimization, correction coefficients that conform to the actual situation are searched out, which improves the reliability and accuracy of data acquisition and ensures the long-term effectiveness and robustness of the evaluation method.

[0045] Preferably, the specific steps for assessing residents' risk perception bias based on the actual subjective perception data and the risk assessment value to obtain risk perception bias assessment data for residents in the post-disaster reconstruction area are as follows:

[0046] S51. Calculate the risk perception value of residents in the post-disaster reconstruction area based on the aforementioned real subjective perception dataset to obtain the risk perception value set. ,in, Indicates the first The risk perception value corresponding to residents in each post-disaster reconstruction area is calculated using the following formula:

[0047] ,

[0048] in, Indicates the first The residents of the post-disaster reconstruction area correspond to the first Normalized data resembling real subjective perception data. Indicates the first Preset weights for data resembling real subjective perception. The total number of categories representing real subjective perception data;

[0049] S52, regarding the above The risk perception values ​​in the data are averaged to obtain the residents' average risk perception value. ;

[0050] Set the risk perception deviation threshold as , and in conjunction with the above With the Assess the risk perception bias of residents in post-disaster reconstruction areas;

[0051] like Then it means the first The risk perception of residents in the post-disaster reconstruction area is too low, and the output risk perception bias assessment data of residents in the post-disaster reconstruction area is underestimated.

[0052] like Then it means the first The risk perception of residents in the post-disaster reconstruction area is too high, and the output risk perception bias assessment data of residents in the post-disaster reconstruction area is an overestimation.

[0053] like Then it means the first The risk perception of residents in the post-disaster reconstruction area is normal, and the output risk perception deviation assessment data of residents in the post-disaster reconstruction area is normal.

[0054] If the risk perception bias assessment data output for residents in the post-disaster reconstruction area is normal, then proceed directly to S53;

[0055] Otherwise, the statistically estimated risk perception values ​​are concentrated in the interval. Number of risk perception values ​​within And calculate the proportion of residents with normal risk perception deviation. ;

[0056] like Then the risk perception bias assessment data of residents in the post-disaster reconstruction area will be corrected to normal.

[0057] like If so, no action will be taken;

[0058] in, This indicates a preset threshold for the percentage of risk perception.

[0059] S53. The risk perception deviation assessment data is pushed to the risk perception deviation assessment platform for display via a wireless communication network.

[0060] By assigning appropriate weights to different subjective perception data, the risk perception value of residents in the post-disaster reconstruction area is reasonably quantified by combining real subjective perception data. The risk perception deviation of residents in the post-disaster reconstruction area is initially assessed based on the average risk perception value of residents and the risk assessment value. At the same time, when the assessment result is not normal, the final result is corrected by counting the number of residents whose risk perception is within the normal range, so as to avoid the final result being greatly deviated due to the perception results of a small number of individuals.

[0061] The present invention also includes an intelligent analysis and assessment system for risk perception deviation of residents in post-disaster reconstruction areas, comprising a risk indicator acquisition module, a risk indicator assessment module, a resident information collection module, a subjective perception data compensation module, and a risk perception deviation assessment module.

[0062] The risk indicator acquisition module is used to set the monitoring interval for post-disaster areas and regularly monitor the objective risk indicators corresponding to the post-disaster reconstruction areas to obtain objective risk indicator data.

[0063] The risk indicator assessment module fits the risk indicators of the post-disaster reconstruction area to a pre-constructed risk indicator weighted fitting model, obtains the risk assessment curve of the post-disaster reconstruction area, and calculates the corresponding risk assessment value.

[0064] The resident information collection module is used to collect subjective perception data and resident behavior data of residents in the post-disaster reconstruction area;

[0065] The subjective perception data compensation module corrects and compensates the subjective perception data using a preset initial correction coefficient to obtain the true subjective perception data. It then performs consistency analysis based on resident behavior data to obtain compensation error data. Only when the compensation error data is less than or equal to a preset error threshold is the compensation error discrimination data output as normal, and the process proceeds directly to S5. Otherwise, the initial correction coefficient is optimized and adjusted, and S4 is re-executed until the compensation error data is less than or equal to the preset error threshold.

[0066] The risk perception deviation assessment module assesses residents' risk perception deviation using the real subjective perception data and the risk assessment value, thereby obtaining risk perception deviation assessment data for residents in the post-disaster reconstruction area.

[0067] By employing the above technical solution, the present invention provides an intelligent analysis and assessment method for risk perception bias among residents in post-disaster reconstruction areas, which has at least the following beneficial effects:

[0068] 1. This invention uses multi-source data and model collaboration to fit risk indicators and calculates risk assessment values ​​based on the fitting results. It obtains residents' basic information, subjective perception, and behavioral data through structured questionnaires. Initial correction coefficients are matched to the residents' basic information data to compensate for the subjective perception data, and error analysis is performed using the residents' behavioral data to determine whether dynamic compensation of the initial correction coefficients is necessary. This significantly improves the accuracy of the final obtained true subjective perception data. By comparing the true subjective perception data with the risk assessment values, the invention assesses residents' risk perception bias from two perspectives, ensuring the accuracy and comprehensiveness of the final results.

[0069] 2. This invention achieves intuitive quantification of multi-dimensional objective risk indicators through the synergy of multi-source data and preset models, ensuring accurate measurement of risk assessment values ​​and providing a reliable data foundation for the subsequent assessment of residents' risk perception bias.

[0070] 3. This invention dynamically matches initial correction coefficients based on residents' basic information, and compensates and corrects subjective perception data according to the initial correction coefficients. At the same time, it uses residents' actual behavior as a reference to conduct compensation error analysis, which effectively reduces the random error in data acquisition and more accurately mines residents' true subjective perception data. In particular, the intelligent optimization algorithm optimizes and adjusts the initial correction coefficients for data with large compensation errors. Through multiple iterations of optimization, it searches for correction coefficients that conform to the actual situation, which improves the reliability and accuracy of data acquisition and ensures the long-term effectiveness and robustness of the evaluation method.

[0071] 4. This invention sets corresponding weights for different subjective perception data, combines real subjective perception data to reasonably quantify the risk perception value of residents in the post-disaster reconstruction area, and conducts a preliminary assessment of the risk perception deviation of residents in the post-disaster reconstruction area based on the average risk perception value of residents and the risk assessment value. At the same time, when the assessment result is not normal, the final result is corrected by counting the number of residents whose risk perception is within the normal range, so as to avoid the final result being greatly deviated due to the perception results of a small number of individuals. Attached Figure Description

[0072] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, the drawings can be obtained from these drawings without creative effort.

[0073] Figure 1 A flowchart of the intelligent analysis and assessment method for risk perception bias of residents in post-disaster reconstruction areas provided by the present invention;

[0074] Figure 2 A schematic diagram of the modules of the intelligent analysis and assessment system for risk perception deviation of residents in post-disaster reconstruction areas provided by the present invention. Detailed Implementation

[0075] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0076] Example 1 is as follows:

[0077] To address the limitations of existing technologies in achieving accurate analysis of objective risk assessments, dynamic compensation of subjectively perceived data, and precise correlation between objective risk assessments and subjectively perceived data, this embodiment proposes an intelligent analysis and assessment method for risk perception bias among residents in post-disaster reconstruction areas. For example... Figure 1 As shown, the method includes the following steps:

[0078] S1. Set monitoring intervals for post-disaster areas and regularly monitor the objective risk indicators corresponding to the post-disaster reconstruction areas to obtain objective risk indicator data.

[0079] S1 includes the following steps:

[0080] S11. Set the post-disaster area monitoring interval, and use IoT sensors to periodically monitor various objective risk indicators corresponding to the target post-disaster reconstruction area according to the post-disaster area monitoring interval to obtain objective risk indicator data.

[0081] The IoT sensors include, but are not limited to, multispectral cameras, geological disaster monitoring equipment, and ultrasonic flaw detectors;

[0082] The objective risk indicators include building structure dimension, geological and environmental dimension, and infrastructure dimension;

[0083] The structural dimensions of the building include, but are not limited to, concrete compressive strength and structural crack density.

[0084] The geological and environmental dimensions include, but are not limited to, rainfall, soil moisture content, and soil erosion.

[0085] The infrastructure dimensions include, but are not limited to, water supply failure rate, drainage system operating load, and emergency facility availability.

[0086] S12. Collect historical objective risk indicator data and corresponding actual risk assessment values ​​of several post-disaster reconstruction areas through the Internet of Things to obtain historical objective risk indicator datasets and actual risk assessment value datasets, and construct the final risk indicator weighted fitting model.

[0087] The weighted fitting model for the final risk index described in S12 adopts the TCN attention hybrid model;

[0088] S12 includes the following steps:

[0089] S121. Construct an initial risk index weighted fitting model and set the first training data ratio, such as 8:2 or 7.5:2.5. The specific ratio can be adjusted reasonably according to the actual situation.

[0090] S122. Perform time alignment and data cleaning on the historical objective risk indicator dataset and the actual risk assessment value dataset, and divide the data according to the first training data ratio to obtain the first training dataset and the first test dataset.

[0091] S123. Set a first training error threshold and a first maximum number of training iterations. Input the training data in the first training dataset into the initial risk index weighted fitting model for training. Continuously adjust the parameters of the initial TCN attention hybrid model according to the training results until the training error is less than the first training error threshold or the number of training iterations is greater than the first maximum number of training iterations, and obtain the trained risk index weighted fitting model.

[0092] The first training error threshold can be set to 5%-10%, and can be adjusted reasonably according to the actual situation; the first maximum number of training iterations can be adjusted according to the amount of training data; the adjustment formula is as follows:

[0093] ,

[0094] in, Indicates the first maximum number of training iterations. Indicates the number of training data. This indicates a preset boundary threshold.

[0095] S124. Set the first test precision, such as 90%-95%, which can be reasonably adjusted according to the actual situation. Input the test data in the first test dataset into the trained risk index weighted fitting model for testing, calculate the accuracy of the test results. If the accuracy of the test results is greater than the first test precision, the final risk index weighted fitting model is obtained; otherwise, return to S123 until the accuracy of the test results is greater than the first test precision.

[0096] The structure of the initial risk index weighted fitting model can be seen in Table 1 below:

[0097] Table 1

[0098] Model Name Model type Model Structure Initial risk index weighted fitting model TCN attention hybrid model Base learner: Temporal convolutional network-attention hybrid model, with 4-6 TCN layers to balance feature extraction capability and overfitting risk; Number of attention heads: 4-8 heads, using multi-head scaling dot product attention to focus on key historical stages; Learning rate: Initial learning rate set to 3e-4, using cosine annealing scheduling combined with a warm-up strategy, and adaptive decay of validation set loss during plateau period; Feature importance assessment: Visualized analysis through temporal attention weight matrix, combined with TCN channel gradient weighted class activation mapping to automatically identify key temporal patterns and turning points.

[0099] S2. Fit the risk indicators of the post-disaster reconstruction area to the pre-constructed risk indicator weighted fitting model to obtain the risk assessment curve of the post-disaster reconstruction area and calculate the corresponding risk assessment value.

[0100] S2 includes the following steps:

[0101] S21. By combining the pre-constructed risk index weighted fitting model with the objective risk index data, the risk index of the target post-disaster reconstruction area is fitted to obtain the risk assessment curve of the post-disaster reconstruction area.

[0102] S22. Calculate the risk assessment value of the target post-disaster reconstruction area based on the risk assessment curve. The calculation formula is as follows:

[0103] ,

[0104] in, and These represent the end time and start time of post-disaster area monitoring, respectively. This indicates the time-related information in the risk assessment curve. The function.

[0105] The TCN attention hybrid model utilizes TCN to quickly extract multi-level local temporal features and dynamically focuses on key information points through an attention mechanism to achieve accurate capture of long and short-range dependencies.

[0106] By combining multi-source data with a pre-set model, the intuitive quantification of multi-dimensional objective risk indicators is achieved, ensuring the accurate measurement of risk assessment values ​​and providing a reliable data foundation for the subsequent assessment of residents' risk perception bias.

[0107] S3. Collect subjective perception data and resident behavior data of residents in the post-disaster reconstruction area;

[0108] S31. Collect basic information data, subjective perception data, and resident behavior data of residents in the post-disaster reconstruction area through structured questionnaires to obtain a basic information dataset of residents. Subjective perception dataset and resident behavior dataset ,in, , as well as These represent the results collected through structured questionnaires. Data on residents' basic information, subjective perceptions, and behavior in the post-disaster reconstruction area. This indicates the number of people who participated in the structured questionnaire survey;

[0109] The basic information data of residents includes, but is not limited to, age, gender, education level, income, and family structure;

[0110] The subjective perception data includes, but is not limited to, the degree of disaster concern, the level of post-disaster anxiety, the sense of security, and probability assessment indicators;

[0111] The resident behavior data includes, but is not limited to, information on material reserves, insurance purchases, and awareness of emergency measures.

[0112] The structured questionnaire was developed through literature review and qualitative interviews to determine the core dimensions of risk perception, and then through standardized psychometric procedures such as expert review, pre-survey item analysis, and large-sample formal survey structure validity and reliability testing.

[0113] S4. Correct and compensate the subjective perception data using a preset initial correction coefficient to obtain the true subjective perception data. Combine this with resident behavior data for consistency analysis to obtain compensation error data. Only when the compensation error data is less than or equal to a preset error threshold is the compensation error discrimination data output as normal, and proceed directly to S5. Otherwise, optimize and adjust the initial correction coefficient, and re-execute S4 until the compensation error data is less than or equal to the preset error threshold.

[0114] S4 includes the following steps:

[0115] S41. Set the correction coefficient dataset ,in, Indicates the first Correction coefficients for various subjective perception data corresponding to the basic information of residents. The total number of categories representing residents' basic information;

[0116] S42. Using the BERT language model algorithm, each resident's basic information data in the resident basic information dataset is sequentially matched with the correction coefficient data in the correction coefficient dataset. The correction coefficients that match each resident's basic information data are searched and identified, thus obtaining the initial correction coefficient set. ,in, Indicates the first Correction coefficients for various subjective perception data corresponding to the basic information data of each resident;

[0117] S43. Correct and compensate the corresponding subjective perception data in the subjective perception dataset according to the initial correction coefficients in the initial correction coefficient set to obtain the true subjective perception dataset. ,in, Indicates the first Real subjective perception data corresponding to residents in the post-disaster reconstruction area;

[0118] S44. By combining the pre-constructed final compensation error mapping model with the real subjective perception dataset and the resident behavior dataset, a consistency analysis is performed to obtain the compensation error dataset. ,in, Indicates the first Compensation error data for the real subjective perception data of residents in the post-disaster reconstruction area;

[0119] Set an error threshold, and compare each compensation error data in the compensation error dataset with the error threshold in turn;

[0120] If the compensation error data is greater than the error threshold, the output compensation error discrimination data is abnormal;

[0121] Otherwise, the output error compensation judgment data is normal;

[0122] After traversing all the compensation error data in the compensation error dataset, the compensation error discrimination dataset is obtained. ,in, Indicates the first Compensation error discrimination data for the real subjective perception data of residents in the post-disaster reconstruction area;

[0123] The final compensation error mapping model in S44 adopts a quantile regression forest model; the construction process of the final compensation error mapping model includes the following steps:

[0124] S441. Collect questionnaire survey data on risk perception bias among residents in post-disaster reconstruction areas through a big data platform.

[0125] The big data platform refers to any one or more of Wenjuanxing, SurveyMonkey, Alibaba Cloud, and Snowflake.

[0126] S442. Construct the initial compensation error mapping model and set the second training data ratio, such as 8:2 or 7.5:2.5. The specific ratio can be adjusted reasonably according to the actual situation.

[0127] S443. The questionnaire survey information data is cleaned and divided according to the proportion of the second training data to obtain the second training dataset and the second test dataset.

[0128] S444. Set a second training error threshold, such as 5%-10%, which can be reasonably adjusted according to the actual situation. Input the training data in the second training dataset into the initial compensation error mapping model for training. Continuously adjust the parameters of the initial compensation error mapping model according to the training results until the training error is less than the second training error threshold, and obtain the trained compensation error mapping model.

[0129] S445. Set a second test precision, such as 90%-95%, which can be reasonably adjusted according to the actual situation. Input the test data in the second test dataset into the trained compensation error mapping model for testing, and calculate the accuracy of the test results. If the accuracy of the test results is greater than the second test precision, the final compensation error mapping model is obtained; otherwise, return to S444 until the accuracy of the test results is greater than the second test precision.

[0130] The structure of the initial compensation error mapping model can be seen in Table 2 below:

[0131] Table 2

[0132] Model Name Model type Model Structure Initial compensation error mapping model Quantile Regression Forest Model Base learner: Spatial adaptive split regression forest, with a tree depth of 8-12 layers to balance local fitting and generalization ability; Number of iterations: 500-1000 rounds, with an early stopping mechanism that terminates when the validation set loss does not decrease for 10 consecutive rounds; Parameter configuration: Number of trees set to 200-500, minimum number of leaf node samples to 3-5, and each tree is forced to use only a random subset of features to enhance diversity.

[0133] S45. If all the compensation error discrimination data in the compensation error discrimination dataset are output as normal, proceed directly to S5; otherwise, optimize and adjust the initial correction coefficients corresponding to the compensation error discrimination data with abnormal outputs through the intelligent optimization algorithm, and re-execute S4 until all the compensation error discrimination data in the compensation error discrimination dataset are output as normal.

[0134] The optimization and adjustment of the initial correction coefficients corresponding to the compensation error discrimination data with abnormal output using an intelligent optimization algorithm includes the following steps:

[0135] S451 sets the current iteration number to... The maximum number of iterations is And the search space dimension of the correction coefficient data is Randomly generated in the data search space for optimizing the correction coefficients. Each set of correction coefficients is used to optimize the data, and each set of correction coefficients corresponds to a set of correction coefficients, thus obtaining the correction coefficient optimization dataset.

[0136] S452. Calculate the fitness value of each modified coefficient optimized data in the modified coefficient optimized dataset according to the fitness function formula. Sort each modified coefficient optimized data in the modified coefficient optimized dataset according to its fitness value from largest to smallest, and select the modified coefficient optimized data with the highest fitness value as the current optimal solution. The fitness function formula is as follows:

[0137] ,

[0138] in, Indicates the first The fitness value of the data is optimized using a correction coefficient. Indicates passing through the first The actual subjective perception data is obtained after optimizing the data with correction coefficients and correcting the subjective perception data corresponding to the data. The compensation error data is obtained after compensation error analysis. This indicates the preset correction value;

[0139] S453. Set update strategy parameters, and compare the current iteration number with the update strategy parameters.

[0140] S4531. If the current iteration number is less than or equal to the update strategy parameter, then each corrected coefficient optimized data in the corrected coefficient optimized dataset updates its own speed; the update formula is as follows:

[0141] ,

[0142] in, Indicates the first The speed of data updates is optimized by adjusting the coefficients. Indicates the first The current search speed of the data is optimized using a correction factor. Represents map operators. This represents a random number that follows a uniform distribution between (0,1). Indicates the position of the current optimal solution. Indicates the first The current position of the data is optimized using each correction factor;

[0143] Each data point in the modified coefficient optimized dataset updates its position within the modified coefficient optimized data search space according to the updated velocity around its own location; the position update formula is as follows:

[0144] ,

[0145] in, Indicates the first The position is updated after optimizing the data with correction coefficients;

[0146] S4532. If the current iteration number is greater than the update strategy parameter, then the modified coefficient optimization dataset is updated, retaining the dataset with higher fitness values. We optimize the data using correction coefficients to obtain an updated dataset with corrected coefficients, and then calculate the center position of the updated dataset. The formula for calculating the center position is as follows:

[0147] ,

[0148] in, This indicates the center position of the dataset after the updated correction coefficients are optimized. This indicates the total number of data points with corrected coefficient optimization in the updated corrected coefficient optimization dataset;

[0149] In the updated modified coefficient optimized dataset, each modified coefficient optimized data point is updated in the modified coefficient optimized data search space based on its center position and its own position; the position update formula is as follows:

[0150] ,

[0151] in, This represents a random number that follows a uniform distribution between (0,1);

[0152] S454. Calculate the fitness value of each modified coefficient optimized data in the modified coefficient optimized dataset after position update according to the fitness function formula. If the fitness value of the modified coefficient optimized data after position update is greater than the original fitness value, replace the original position with the new position of the modified coefficient optimized data; otherwise, retain the original position.

[0153] S455, Determine the above Is it greater than or equal to the stated If the above Greater than or equal to the The correction coefficient with the highest fitness value is then output as the optimized correction coefficient.

[0154] By employing a dual mechanism of dynamic compensation and correction of subjectively perceived data and assessment of the accuracy of the compensation and correction, combined with intelligent optimization algorithms to rationally adjust the correction coefficients required in the process, scientific decision-making for acquiring authentic subjectively perceived data is achieved. Initial correction coefficients are dynamically matched to residents' basic information, and subjectively perceived data is compensated and corrected based on these initial coefficients. Simultaneously, compensation error analysis is conducted using residents' actual behavior as a reference, effectively reducing random errors in data acquisition and more accurately uncovering residents' true subjectively perceived data. Specifically, intelligent optimization algorithms optimize and adjust the initial correction coefficients for data with large compensation errors. Through multiple iterations, correction coefficients that conform to the actual situation are searched, improving the reliability and accuracy of the acquired data and ensuring the long-term effectiveness and robustness of the evaluation method.

[0155] S5. Based on the real subjective perception data and the risk assessment value, conduct a risk perception deviation assessment of residents to obtain risk perception deviation assessment data of residents in the post-disaster reconstruction area.

[0156] S5 includes the following steps:

[0157] S51. Calculate the risk perception value of residents in the post-disaster reconstruction area based on the aforementioned real subjective perception dataset to obtain the risk perception value set. ,in, Indicates the first The risk perception value corresponding to residents in each post-disaster reconstruction area is calculated using the following formula:

[0158] ,

[0159] in, Indicates the first The residents of the post-disaster reconstruction area correspond to the first Normalized data resembling real subjective perception data. Indicates the first Preset weights for data resembling real subjective perception. The total number of categories representing real subjective perception data;

[0160] S52. The risk perception values ​​in the risk perception value set are averaged to obtain the average risk perception value of residents. ;

[0161] Set the risk perception deviation threshold as For example, 0.1, the specific value can be reasonably adjusted according to the actual situation, and combined with the average risk perception value of residents. With the aforementioned risk assessment value Assess the risk perception bias of residents in post-disaster reconstruction areas;

[0162] like Then it means the first The risk perception of residents in the post-disaster reconstruction area is too high, and the output risk perception bias assessment data of residents in the post-disaster reconstruction area is an overestimation.

[0163] like Then it means the first The risk perception of residents in the post-disaster reconstruction area is normal, and the output risk perception deviation assessment data of residents in the post-disaster reconstruction area is normal.

[0164] If the risk perception bias assessment data output for residents in the post-disaster reconstruction area is normal, then proceed directly to S53;

[0165] Otherwise, the statistically estimated risk perception values ​​are concentrated in the interval. Number of risk perception values ​​within And calculate the proportion of residents with normal risk perception deviation. ;

[0166] like Then the risk perception bias assessment data of residents in the post-disaster reconstruction area will be corrected to normal.

[0167] like If so, no action will be taken;

[0168] in, This indicates the preset risk perception threshold, such as 80%, which can be adjusted reasonably according to the actual situation.

[0169] S53. The risk perception deviation assessment data is pushed to the risk perception deviation assessment platform for display via a wireless communication network.

[0170] By assigning appropriate weights to different subjective perception data, the risk perception value of residents in the post-disaster reconstruction area is reasonably quantified by combining real subjective perception data. The risk perception deviation of residents in the post-disaster reconstruction area is initially assessed based on the average risk perception value of residents and the risk assessment value. At the same time, when the assessment result is not normal, the final result is corrected by counting the number of residents whose risk perception is within the normal range, so as to avoid the final result being greatly deviated due to the perception results of a small number of individuals.

[0171] Example 2 is as follows:

[0172] Please see Figure 2 A smart analysis and assessment system for risk perception deviation of residents in post-disaster reconstruction areas includes a risk indicator acquisition module, a risk indicator assessment module, a resident information collection module, a subjective perception data compensation module, and a risk perception deviation assessment module.

[0173] The risk indicator acquisition module is used to set the monitoring interval for post-disaster areas and regularly monitor the objective risk indicators corresponding to the post-disaster reconstruction areas to obtain objective risk indicator data.

[0174] The risk indicator assessment module fits the risk indicators of the post-disaster reconstruction area to a pre-constructed risk indicator weighted fitting model, obtains the risk assessment curve of the post-disaster reconstruction area, and calculates the corresponding risk assessment value.

[0175] The resident information collection module is used to collect subjective perception data and resident behavior data of residents in the post-disaster reconstruction area;

[0176] The subjective perception data compensation module corrects and compensates the subjective perception data using a preset initial correction coefficient to obtain the true subjective perception data. It then performs consistency analysis based on resident behavior data to obtain compensation error data. Only when the compensation error data is less than or equal to a preset error threshold is the compensation error discrimination data output as normal, and the process proceeds directly to S5. Otherwise, the initial correction coefficient is optimized and adjusted, and S4 is re-executed until the compensation error data is less than or equal to the preset error threshold.

[0177] The risk perception deviation assessment module assesses residents' risk perception deviation using the real subjective perception data and the risk assessment value, thereby obtaining risk perception deviation assessment data for residents in the post-disaster reconstruction area.

[0178] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0179] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0180] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method for intelligent analysis and assessment of risk perception bias among residents in post-disaster reconstruction areas, characterized in that, Includes the following steps: S1. Set monitoring intervals for post-disaster areas and regularly monitor the objective risk indicators corresponding to the post-disaster reconstruction areas to obtain objective risk indicator data. S2. Fit the risk indicators of the post-disaster reconstruction area to the pre-constructed risk indicator weighted fitting model to obtain the risk assessment curve of the post-disaster reconstruction area and calculate the corresponding risk assessment value. S3. Collect subjective perception data and resident behavior data of residents in the post-disaster reconstruction area; S4. Correct and compensate the subjective perception data using a preset initial correction coefficient to obtain the true subjective perception data, and perform consistency analysis in conjunction with resident behavior data to obtain the compensation error data. The compensation error discrimination data is output as normal only when the compensation error data is less than or equal to the preset error threshold, and the process proceeds directly to S5; Otherwise, the initial correction coefficient is optimized and adjusted, and S4 is executed again until the compensation error data is less than or equal to the preset error threshold. S5. Based on the actual subjective perception data and the risk assessment value, conduct a risk perception deviation assessment of residents to obtain risk perception deviation assessment data of residents in the post-disaster reconstruction area.

2. The intelligent analysis and assessment method for risk perception bias of residents in post-disaster reconstruction areas according to claim 1, characterized in that, S1 includes the following steps: S11. Set the post-disaster area monitoring interval, and use IoT sensors to periodically monitor various objective risk indicators corresponding to the target post-disaster reconstruction area according to the post-disaster area monitoring interval to obtain objective risk indicator data. S12. Collect historical objective risk indicator data and corresponding actual risk assessment values ​​of several post-disaster reconstruction areas through the Internet of Things to obtain historical objective risk indicator datasets and actual risk assessment value datasets, and construct the final risk indicator weighted fitting model.

3. The intelligent analysis and assessment method for risk perception bias of residents in post-disaster reconstruction areas according to claim 2, characterized in that, The weighted fitting model for the final risk index described in S12 adopts the TCN attention hybrid model.

4. The intelligent analysis and assessment method for risk perception bias of residents in post-disaster reconstruction areas according to claim 3, characterized in that, S2 includes the following steps: S21. By combining the pre-constructed risk index weighted fitting model with the objective risk index data, the risk index of the target post-disaster reconstruction area is fitted to obtain the risk assessment curve of the post-disaster reconstruction area. S22. Calculate the risk assessment value of the target post-disaster reconstruction area based on the risk assessment curve. The calculation formula is as follows: , in, and These represent the end time and start time of post-disaster area monitoring, respectively. This indicates the time-related information in the risk assessment curve. The function.

5. The intelligent analysis and assessment method for risk perception bias of residents in post-disaster reconstruction areas according to claim 4, characterized in that, S3 includes the following steps: S31. Collect basic information data, subjective perception data, and resident behavior data of residents in the post-disaster reconstruction area through structured questionnaires to obtain a basic information dataset of residents. Subjective perception dataset and resident behavior dataset ,in, , as well as These represent the results collected through structured questionnaires. Data on residents' basic information, subjective perceptions, and behavior in the post-disaster reconstruction area. This indicates the number of people who participated in the structured questionnaire survey.

6. The intelligent analysis and assessment method for risk perception bias of residents in post-disaster reconstruction areas according to claim 5, characterized in that, S4 includes the following steps: S41. Set the correction coefficient dataset ,in, Indicates the first Correction coefficients for various subjective perception data corresponding to the basic information of residents. The total number of categories representing residents' basic information; S42. The BERT language model algorithm is used to sequentially process the... The basic information data of each resident in the data and the data mentioned above The correction coefficient data in the database is matched with characters to find correction coefficients that match the basic information data of each resident, and these correction coefficients are then labeled to obtain the initial set of correction coefficients. ,in, Indicates the first Correction coefficients for various subjective perception data corresponding to the basic information data of each resident; S43, according to the above The initial correction coefficients in the above are for the The corresponding subjective perception data is corrected and compensated to obtain the true subjective perception dataset. ,in, Indicates the first Real subjective perception data corresponding to residents in the post-disaster reconstruction area; S44, By combining the pre-built final compensation error mapping model with the above and stated Consistency analysis was performed to obtain the compensation error dataset. ,in, Indicates the first Compensation error data for the real subjective perception data of residents in the post-disaster reconstruction area; Set an error threshold, and then... Each compensation error data in the data is sequentially compared with the error threshold; If the compensation error data is greater than the error threshold, the output compensation error discrimination data is abnormal; Otherwise, the output error compensation judgment data is normal; Traversing the The compensation error discrimination dataset is obtained by collecting all the compensation error data. ,in, Indicates the first Compensation error discrimination data for the real subjective perception data of residents in the post-disaster reconstruction area; S45, if the above If all the compensation error discrimination data output is normal, proceed directly to S5; otherwise, optimize and adjust the initial correction coefficients corresponding to the compensation error discrimination data with abnormal output using an intelligent optimization algorithm, and re-execute S4 until the process is complete. All the compensation error discrimination data in the output are normal.

7. The intelligent analysis and assessment method for risk perception bias of residents in post-disaster reconstruction areas according to claim 5, characterized in that, S45 includes the following steps: S451, Set the current iteration number to... The maximum number of iterations is And the search space dimension of the correction coefficient data is Randomly generated in the data search space for optimizing the correction coefficients. Each set of correction coefficients is used to optimize the data, and each set of correction coefficients corresponds to a set of correction coefficients, thus obtaining the correction coefficient optimization dataset. S452. Calculate the fitness value of each correction coefficient optimization data according to the fitness function formula, sort each correction coefficient optimization data in the correction coefficient optimization data according to the fitness value from largest to smallest, and select the correction coefficient optimization data with the highest fitness value as the current optimal solution; S453. Set update strategy parameters, and compare the current iteration number with the update strategy parameters. S4531. If the current iteration number is less than or equal to the update strategy parameter, then each correction coefficient optimizes its own speed. Each correction coefficient optimized data is updated in the correction coefficient optimized data search space around its own position at the updated speed; S4532. If the current iteration number is greater than the update strategy parameter, then the modified coefficient optimization dataset is updated, retaining the dataset with higher fitness values. Optimize the data with correction coefficients to obtain the updated optimized dataset with correction coefficients, and calculate the center position of the updated optimized dataset with correction coefficients. In the updated modified coefficient optimized dataset, each modified coefficient optimized data point is updated in the search space of the modified coefficient optimized data according to its center position and its own position. S454. Calculate the fitness value after updating the position using the data optimized by each correction coefficient according to the fitness function formula. If the fitness value after updating the position using the data optimized by the correction coefficient is greater than the original fitness value, replace the original position with the new position; otherwise, retain the original position. S455, Determine the above Is it greater than or equal to the stated If the above Greater than or equal to the The correction coefficient with the highest fitness value is then output as the optimized correction coefficient.

8. The intelligent analysis and assessment method for risk perception bias of residents in post-disaster reconstruction areas according to claim 7, characterized in that, S5 includes the following steps: S51. Calculate the risk perception value of residents in the post-disaster reconstruction area based on the aforementioned real subjective perception dataset to obtain the risk perception value set. ,in, Indicates the first The risk perception value corresponding to residents in each post-disaster reconstruction area is calculated using the following formula: ， in, Indicates the first The residents of the post-disaster reconstruction area correspond to the first Normalized data resembling real subjective perception data. Indicates the first Preset weights for data resembling real subjective perception. The total number of categories representing real subjective perception data; S52, regarding the above The risk perception values ​​in the data are averaged to obtain the residents' average risk perception value. ; Set the risk perception deviation threshold as , and in conjunction with the above With the Assess the risk perception bias of residents in post-disaster reconstruction areas; like Then it means the first The risk perception of residents in the post-disaster reconstruction area is too low, and the output risk perception bias assessment data of residents in the post-disaster reconstruction area is underestimated. like Then it means the first The risk perception of residents in the post-disaster reconstruction area is too high, and the output risk perception bias assessment data of residents in the post-disaster reconstruction area is an overestimation. like Then it means the first The risk perception of residents in the post-disaster reconstruction area is normal, and the output risk perception deviation assessment data of residents in the post-disaster reconstruction area is normal. If the risk perception bias assessment data output for residents in the post-disaster reconstruction area is normal, then proceed directly to S53; Otherwise, the statistically estimated risk perception values ​​are concentrated in the interval. Number of risk perception values ​​within And calculate the proportion of residents with normal risk perception deviation. ; like Then the risk perception bias assessment data of residents in the post-disaster reconstruction area will be corrected to normal. like If so, no action will be taken; in, This indicates a preset threshold for the percentage of risk perception. S53. The risk perception deviation assessment data is pushed to the risk perception deviation assessment platform for display via a wireless communication network.

9. A computer device, characterized in that, It includes a processor and a memory, the memory being used to store a computer program, which, when executed by the processor, implements the intelligent analysis and assessment method for risk perception deviation of residents in post-disaster reconstruction areas as described in any one of claims 1-8.

10. A system for implementing the intelligent analysis and assessment method for risk perception deviation of residents in post-disaster reconstruction areas as described in any one of claims 1-8, characterized in that: It includes a risk indicator acquisition module, a risk indicator assessment module, a resident information collection module, a subjective perception data compensation module, and a risk perception bias assessment module; The risk indicator acquisition module is used to set the monitoring interval for post-disaster areas and regularly monitor the objective risk indicators corresponding to the post-disaster reconstruction areas to obtain objective risk indicator data. The risk indicator assessment module fits the risk indicators of the post-disaster reconstruction area to a pre-constructed risk indicator weighted fitting model, obtains the risk assessment curve of the post-disaster reconstruction area, and calculates the corresponding risk assessment value. The resident information collection module is used to collect subjective perception data and resident behavior data of residents in the post-disaster reconstruction area; The subjective perception data compensation module corrects and compensates the subjective perception data using a preset initial correction coefficient to obtain the true subjective perception data. It then performs consistency analysis based on resident behavior data to obtain compensation error data. Only when the compensation error data is less than or equal to a preset error threshold is the compensation error discrimination data output as normal, and the process proceeds directly to S5. Otherwise, the initial correction coefficient is optimized and adjusted, and S4 is re-executed until the compensation error data is less than or equal to the preset error threshold. The risk perception deviation assessment module assesses residents' risk perception deviation using the real subjective perception data and the risk assessment value, thereby obtaining risk perception deviation assessment data for residents in the post-disaster reconstruction area.

Images