Building quality evaluation method based on data analysis

Abstract

The invention discloses a building quality evaluation method based on data analysis, and belongs to the field of building safety management, and the method comprises the steps: collecting monitoring data related to building quality monitoring, and segmenting a monitoring data set through the probability density corresponding to the monitoring data, so as to form a plurality of monitoring data segments; and according to the data volume in each monitoring data segment, fusing the monitoring data segments, deleting redundant monitoring data in the monitoring data segments, and performing interpolation on the monitoring data segments with insufficient data volume. A random forest network is constructed, the number of decision trees of the random forest network is the same as the number of all the monitoring data segments, an evaluation model for evaluating the building quality of child nodes corresponding to each decision tree is built, and the building quality is evaluated through monitoring data in all the monitoring data segments. Through high-precision evaluation of the building quality, early warning of problems and potential risks existing in the building quality is facilitated, the management level of the building quality can be effectively improved, and the decision-making efficiency of building safety risk early warning can be improved.

Description

Technical Field

[0001] This invention relates to the field of building safety management, and more specifically to a data analysis-based method for assessing building quality. Background Technology

[0002] In the field of construction engineering, building quality assessment is a crucial step in ensuring the safety, reliability, and durability of buildings. Traditional building quality assessment methods mainly rely on manual inspection and experience-based judgment, which suffers from problems such as high subjectivity, low efficiency, and poor accuracy. With the development of information technology, data analysis technology is being used more and more widely in construction engineering. How to utilize data analysis technology to improve the accuracy and efficiency of building quality assessment has become an urgent problem to be solved.

[0003] While some data-driven building quality assessment methods exist in the current technology, most are based on the analysis and assessment of single types of monitoring data. This leads to a lack of correlation between different types of data, resulting in poor comprehensive analysis. Furthermore, under the condition of monitoring data collected over a long period of time, there is a lack of preprocessing methods for large volumes of monitoring data, resulting in defects such as data missing or data redundancy, leading to low assessment accuracy. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this invention provides a data analysis-based building quality assessment method that uses a network model to comprehensively assess building quality based on the analysis of multiple types of monitoring data.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0006] A data analysis-based method for building quality assessment is provided, comprising:

[0007] Step S1: Collect monitoring data related to building quality, obtain a time-series-based monitoring dataset, calculate the probability density in the time distribution based on the time interval between monitoring datasets, and segment the monitoring dataset using the probability density corresponding to the monitoring data to form several monitoring data segments.

[0008] Step S2: Based on the amount of data in each monitoring data segment, merge the monitoring data segments with smaller data amounts into adjacent monitoring data segments. Based on the distribution density of the monitoring data in the monitoring data segment, calculate the amount of redundant data and the amount of data that needs to be interpolated, delete redundant monitoring data in the monitoring data segment, and interpolate monitoring data segments with insufficient data amounts.

[0009] Step S3: Output all monitoring data segments after deleting redundant monitoring data, completing interpolation, and meeting the distribution density requirements. Construct a random forest network with the same number of decision trees as the number of monitoring data segments. Build an evaluation model for the child nodes of each decision tree to assess building quality. Use the monitoring data from all monitoring data segments to evaluate building quality.

[0010] Further, step S1 includes:

[0011] Step S11: Determine the monitoring data related to the building quality. The types of monitoring data include foundation settlement, crack width, structural deformation, vibration data, etc., and obtain a time-series-based monitoring dataset. n represents the type of monitoring data, t m For the time of data collection, Let time t m The collected monitoring data, where m is the number of monitoring data;

[0012] Step S12: Calculate the probability density of the monitoring data in the time distribution based on the time intervals between monitoring data;

[0013]

[0014] Where i is the number of the monitoring data, p i Let be the probability density of the i-th monitoring data, and represent the time period t during which monitoring data i was acquired. i-1 ~t i+1 Total time period t1~t m The proportion;

[0015] Step S13: Obtain the probability density dataset P = {p2, p3, ... pm-1} corresponding to the 2nd monitoring data to the (m-1)th monitoring data. m-1}, p m-1 Let be the probability density corresponding to the (m-1)th monitoring data;

[0016] Step S14: Set the ideal probability density threshold p0, and divide the monitoring dataset A into several monitoring data segments.

[0017] Furthermore, the method for segmenting the monitoring data segment is as follows:

[0018] Iterate through the probability density data in the probability density dataset P, and compare the probability density with the probability density threshold p0. If p i If ≤p0, it indicates that the monitoring data The corresponding probability density meets the requirements, and the monitoring data will be used. Corresponding acquisition time t i As a segmentation point; otherwise, monitoring data The corresponding probability density does not meet the requirements;

[0019] Based on the segment points in monitoring dataset A and the monitoring data at the beginning and end The monitoring data between two segment points is considered as a single monitoring data segment.

[0020] Further, step S2 includes:

[0021] Step S21: Obtain the data volume u within each monitoring data segment. v v is the number of the monitoring data segment. The monitoring data segment with smaller data volume is merged into the adjacent monitoring data segment to ensure the amount of data within the monitoring data segment.

[0022] Step S22: Repeat step S21 until the data volume in all monitoring data segments is greater than or equal to the minimum value u0. After that, output the monitoring data in each monitoring data segment. w is the monitoring data number within the v-th monitoring data segment, t w To obtain monitoring data The time of data collection;

[0023] Step S23: Calculate the distribution density of monitoring data within each monitoring data segment;

[0024]

[0025] Where, η v Let v be the distribution density of the monitoring data in the v-th monitoring data segment. Let v be the start time of the v-th monitoring data segment. This is the end time of the v-th monitoring data segment;

[0026] Step S24: Calculate the average distribution density η0 of the monitoring data based on the distribution density of the monitoring data within each monitoring data segment;

[0027]

[0028] Where V represents the number of monitoring data segments;

[0029] Step S25: Calculate the distribution density η of the monitoring data within each monitoring data segment. v Compare with the average distribution density η0;

[0030] If η v If η > 0, it is determined that the distribution density of the monitoring data in the vth monitoring data segment is high, and the redundant monitoring data in the monitoring data segment needs to be deleted.

[0031] If η vIf η < 0, it is determined that the distribution density of the monitoring data in the vth monitoring data segment is small, and the monitoring data segment is interpolated to supplement the amount of monitoring data in the monitoring data segment;

[0032] If η v If η = 0, then the distribution density of the monitoring data in the vth monitoring data segment meets the requirements, and step S3 is executed;

[0033] Furthermore, the method for integrating smaller monitoring data segments into adjacent monitoring data segments is as follows:

[0034] Set the minimum data volume requirement u0 within the monitoring data segment, and compare the data volume u. v The size between the minimum value u0 and the minimum value u0;

[0035] If u v If <u0, then compare the data volume u of the adjacent monitoring data segments v-1 and v+1 on both sides of the monitoring data segment v. v-1 u v+1 Size; if u v-1 ≤u v+1 Then the monitoring data in monitoring data segment v will be merged into monitoring data segment v-1, if u v-1 ≥u v+1 Then the monitoring data in monitoring data segment v will be integrated into monitoring data segment v+1 to form a new monitoring data segment;

[0036] If u v If u ≥ 0, then retain the monitoring data segment v.

[0037] Furthermore, the method for deleting redundant monitoring data within a monitoring data segment is as follows:

[0038] Step S26: Calculate the amount of redundant data u1 in the v-th monitoring data segment;

[0039] Indicates rounding down;

[0040] Sort the probability density of each monitoring data point within the v-th monitoring data segment from smallest to largest to form the sorted probability density dataset. It is the maximum value of the probability density, and satisfies

[0041] Based on the sorted probability density dataset P v The monitoring data corresponding to the first u1 probability densities are deleted from the vth monitoring data segment to obtain the vth monitoring data segment after deleting the redundant monitoring data.

[0042] Furthermore, the method for supplementing the amount of monitoring data within the monitoring data segment is as follows:

[0043] Step S27: Calculate the amount of data u2 that needs to be interpolated within the v-th monitoring data segment;

[0044] Indicates rounding up;

[0045] Sort the probability density data corresponding to each monitoring data point within the v-th monitoring data segment from largest to smallest to form the sorted probability density dataset. It is the minimum value of the probability density, and satisfies

[0046] Based on the sorted probability density dataset P v ′, from the time of collection of monitoring data corresponding to the highest probability density p1′ Initially, interpolation is performed sequentially within the time interval between two monitoring data points until the number of interpolated monitoring data points in the vth monitoring data segment reaches the data volume u2. Then, interpolation is stopped, and the interpolated vth monitoring data segment is output.

[0047] The interpolation method for monitoring data is as follows:

[0048]

[0049] in, For the data acquisition time t w With the acquisition time t w+1 Monitoring data interpolated between time periods This is the first monitoring data within the v-th monitoring data segment.

[0050] Further, step S3 includes:

[0051] Step S31: Output all monitoring data segments v′ after deleting redundant monitoring data, completing interpolation, and meeting the distribution density requirements, and obtain the monitoring data within monitoring data segment v′. e is the monitoring data number within the monitoring data segment v′, t e For monitoring data The corresponding data collection time;

[0052] S32: Construct a random forest network. The number of decision trees in the random forest network is the same as the number of monitoring data segments v′. Build an evaluation model for the child nodes corresponding to each decision tree to evaluate the building quality. Each child node outputs a building quality coefficient based on the evaluation model.

[0053]

[0054] Where, Δ d This is a difference operation, where d is the order of the difference. These are the autoregressive coefficients. For the data acquisition time t ePredicted monitoring data, To monitor the data collection time t within the data segment v′ e-1 Monitoring data, u v ' represents the number of monitoring data points within the monitoring data segment v'. It is white noise. The building quality coefficient output for child nodes;

[0055] S33: Input the monitoring data within each monitoring data segment v′ into the child nodes of the decision tree. Each decision tree outputs a building quality coefficient. Then, use a fusion model to fuse the building quality coefficients output by each decision tree and output the total building quality coefficient F obtained using the nth type of monitoring data. n ;

[0056]

[0057] Where, λ v′ λ1 is the weighting coefficient for the impact of monitoring data segment v′ on building quality assessment, k is the incrementing gradient of the weighting coefficient, and V′ is the number of monitoring data segments v′.

[0058] S34: Output the total building quality coefficient data {F} based on each type of monitoring data. 1 ,F 2 ,…,F N}; N represents the type of monitoring data, F N This is the total building quality coefficient output based on the Nth type of monitoring data;

[0059] S35: Combining the total building quality coefficient data {F 1 ,F 2 ,…,F N}, calculate the building quality assessment coefficient F0;

[0060]

[0061] S36: Set the standard value F′ for the building quality assessment coefficient. If F′ < F 0 If F′≥F, then the current building quality is determined to be poor, and a building quality risk warning needs to be issued. 0 If the current building quality is good, then there is no need to issue a building quality risk warning.

[0062] The beneficial effects of this invention are as follows: This invention performs data analysis and preprocessing on large volumes and multiple types of monitoring data, optimizes and supplements the large volumes of monitoring data based on time series, ensures the rationality of the collected monitoring data, reduces data collection errors, and outputs quantitative analysis data characterizing building quality by building a random forest network and integrating it with an autoregressive model, thereby achieving high-precision assessment of building quality. This facilitates early warning of problems and potential risks in building quality, effectively improving the management level of building quality and enhancing the decision-making efficiency of building safety risk early warning. Attached Figure Description

[0063] Figure 1 This is a flowchart of a data analysis-based building quality assessment method.

[0064] Figure 2 A diagram illustrating the segmentation of monitoring data.

[0065] Figure 3 This is a diagram of the structure of a random forest network. Detailed Implementation

[0066] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0067] like Figure 1 As shown, a data analysis-based building quality assessment method includes:

[0068] Step S1: Collect monitoring data related to building quality, obtain a time-series-based monitoring dataset, calculate the probability density in the time distribution based on the time intervals between monitoring datasets, and segment the monitoring dataset using the probability density corresponding to the monitoring data to form several monitoring data segments.

[0069] Step S1 specifically includes:

[0070] Step S11: Determine the monitoring data related to the building quality. The types of monitoring data include foundation settlement, crack width, structural deformation, vibration data, etc., and obtain a time-series-based monitoring dataset. n represents the type of monitoring data, t m For the time of data collection, Let time t m The collected monitoring data, where m is the number of monitoring data;

[0071] In this embodiment, the monitoring of building foundation settlement can be achieved using the FH-8003 online settlement monitoring system. This system integrates new MEMS sensor technology and wireless communication technology, offering advantages such as large measurement range, high accuracy, simple installation, small size, and convenient installation. It can achieve automated real-time monitoring of settlement in buildings, towers, subways, bridges, tunnels, railways, water conservancy projects, dams, and mining subsidence areas. The system incorporates an embedded high-precision aerospace-grade silicon voltage sensor chip and a 24-bit acquisition chip, ensuring high measurement accuracy. It employs LoRa wireless communication technology with a communication frequency of 433MHz, providing strong anti-interference capabilities and a long communication distance.

[0072] In this embodiment, crack width monitoring can be performed using an LVDT-type crack gauge, which can be used for online monitoring of crack depth, width, and opening degree in buildings. Structural deformation can be measured using a MAS-HLS series differential pressure deformation measurement sensor. Vibration data monitoring can be performed using an accelerometer.

[0073] Step S12: Calculate the probability density of the monitoring data in the time distribution based on the time intervals between monitoring data;

[0074]

[0075] Where i is the number of the monitoring data, p i Let be the probability density of the i-th monitoring data, and represent the time period t during which monitoring data i was acquired. i-1 ~t i+1 Total time period t1~t m The proportion;

[0076] Step S13: Obtain the probability density dataset P = {p2, p3, ... pm-1} corresponding to the 2nd monitoring data to the (m-1)th monitoring data. m-1}, p m-1 Let be the probability density corresponding to the (m-1)th monitoring data;

[0077] Step S14: Set the ideal probability density threshold p0, and segment the monitoring dataset A to form several monitoring data segments. The specific method is as follows:

[0078] Iterate through the probability density data in the probability density dataset P, and compare the probability density with the probability density threshold p0. If p i If ≤p0, it indicates that the monitoring data The corresponding probability density meets the requirements, and the monitoring data will be used. Corresponding acquisition time t i As a segmentation point; otherwise, monitoring data The corresponding probability density does not meet the requirements;

[0079] Based on the segment points in monitoring dataset A and the monitoring data at the beginning and end The monitoring data between two segment points is considered as a single monitoring data segment; for example... Figure 2 As shown in the figure, an example of segmentation points in this embodiment is given, and the monitoring data between segmentation points is a monitoring data segment.

[0080] This embodiment segments the monitoring data based on the probability density corresponding to the monitoring data. The probability density represents the density of monitoring data acquired in the time series. Monitoring data with higher distribution density are separated as segmentation points. The higher the data distribution density, the more continuous the monitoring data collected by the sensor is within that time period, and the lower the probability of sensor failure. This ensures that the data at both ends of the segmented monitoring data are within a relatively normal range, without data loss or delay. This provides a basis for subsequent monitoring data interpolation and error correction, improving the accuracy of the monitoring data within the output monitoring data segment.

[0081] Step S2: Based on the amount of data in each monitoring data segment, merge the monitoring data segments with smaller data amounts into adjacent monitoring data segments. Based on the distribution density of the monitoring data in the monitoring data segment, calculate the amount of redundant data and the amount of data that needs to be interpolated, delete redundant monitoring data in the monitoring data segment, and interpolate monitoring data segments with insufficient data amounts.

[0082] Step S2 specifically includes:

[0083] Step S21: Obtain the data volume u within each monitoring data segment. v v is the number of the monitoring data segment. The monitoring data segment with smaller data volume is merged into the adjacent monitoring data segment to ensure the amount of data within the monitoring data segment.

[0084] The method for merging small monitoring data segments into adjacent monitoring data segments is as follows:

[0085] Set the minimum data volume requirement u0 within the monitoring data segment, and compare the data volume u. v The size between the minimum value u0 and the minimum value u0;

[0086] If u v If <u0, then compare the data volume u of the adjacent monitoring data segments v-1 and v+1 on both sides of the monitoring data segment v. v-1 u v+1 Size; if u v-1 ≤u v+1 Then the monitoring data in monitoring data segment v will be merged into monitoring data segment v-1, if u v-1 ≥u v+1 Then the monitoring data in monitoring data segment v will be integrated into monitoring data segment v+1 to form a new monitoring data segment;

[0087] If u vIf u ≥ 0, then retain the monitoring data segment v;

[0088] In this embodiment, when performing data fusion of monitoring data segments, when the data volume u of the monitoring data segments on both sides... v-1 u v+1 When data are equal, priority should be given to incorporating the monitoring data within monitoring data segment v into the preceding monitoring data segment. The closer to the initial moment of data acquisition, the shorter the usage time of the relevant sensors, the fewer interference factors, the lower the probability of error, and the higher the accuracy of the acquired data. During the interpolation process of subsequent monitoring data, the error of the interpolated monitoring data should be minimized.

[0089] Step S22: Repeat step S21 until the data volume in all monitoring data segments is greater than or equal to the minimum value u0. After that, output the monitoring data in each monitoring data segment. w is the monitoring data number within the v-th monitoring data segment, t w To obtain monitoring data The time of data collection;

[0090] Step S23: Calculate the distribution density of monitoring data within each monitoring data segment;

[0091]

[0092] Where, η v Let v be the distribution density of the monitoring data in the v-th monitoring data segment. Let v be the start time of the v-th monitoring data segment. This is the end time of the v-th monitoring data segment;

[0093] Step S24: Calculate the average distribution density η0 of the monitoring data based on the distribution density of the monitoring data within each monitoring data segment;

[0094]

[0095] Where V represents the number of monitoring data segments;

[0096] Step S25: Calculate the distribution density η of the monitoring data within each monitoring data segment. v Compare with the average distribution density η0;

[0097] If η v If η > 0, then it is determined that the distribution density of the monitoring data in the vth monitoring data segment is large, and step S26 is executed;

[0098] If η v If η < 0, then the distribution density of the monitoring data in the vth monitoring data segment is small, and step S27 is executed;

[0099] If ηv If η = 0, then the distribution density of the monitoring data in the vth monitoring data segment meets the requirements, and step S3 is executed;

[0100] Step S26: Calculate the amount of redundant data u1 in the v-th monitoring data segment;

[0101] Indicates rounding down;

[0102] Sort the probability density of each monitoring data point within the v-th monitoring data segment from smallest to largest to form the sorted probability density dataset. It is the maximum value of the probability density, and satisfies

[0103] Based on the sorted probability density dataset P v The monitoring data corresponding to the first u1 probability densities are deleted from the vth monitoring data segment to obtain the vth monitoring data segment after deleting the redundant monitoring data.

[0104] This embodiment assesses whether data is anomalous by monitoring the probability density data within the monitoring data segment. If the probability density of the monitoring data is low in the current time period, it indicates that the data may be anomalous monitoring data. These anomalous monitoring data may be caused by clock errors, data transmission delays or errors, or other reasons. Outliers can affect the accuracy of data analysis results, so they need to be deleted to reduce the possibility of anomalous data affecting the data analysis results.

[0105] Step S27: Calculate the amount of data u2 that needs to be interpolated within the v-th monitoring data segment;

[0106] Indicates rounding up;

[0107] Sort the probability density data corresponding to each monitoring data point within the v-th monitoring data segment from largest to smallest to form the sorted probability density dataset. It is the minimum value of the probability density, and satisfies

[0108] Based on the sorted probability density dataset P v ′, from the time of collection of monitoring data corresponding to the highest probability density p1′ Initially, interpolation is performed sequentially within the time interval between two monitoring data points until the number of interpolated monitoring data points in the vth monitoring data segment reaches the data volume u2. Then, interpolation is stopped, and the interpolated vth monitoring data segment is output.

[0109] The interpolation method for monitoring data is as follows:

[0110]

[0111] in, For the data acquisition time t w With the acquisition time t w+1 Monitoring data interpolated between time periods This is the first monitoring data within the v-th monitoring data segment.

[0112] This embodiment calculates the rate of change of monitoring data over time within the vth monitoring data segment as the basis for interpolation, and uses the most recent monitoring data as the basis for interpolation calculation to calculate the monitoring data interpolated in the middle of the time period between adjacent collection times.

[0113] For example, when interpolating monitoring data for the time interval between acquisition time t1 and acquisition time t2 within the v-th monitoring data segment, the rate of change of the monitoring data with time within the v-th monitoring data segment is first calculated. Then use the monitoring data corresponding to the acquisition time t1 The monitoring data interpolated during the calculation period t1 to t2

[0114]

[0115] Similarly, when calculating the interpolated monitoring data within the time period t2 to t3, the monitoring data is used as the basis. As the basis for interpolation calculations, this process is repeated from the acquisition time corresponding to the monitoring data with the highest probability density. To begin, iterate through the first u2 time periods and interpolate u2 monitoring data points.

[0116] Step S3: Output all monitoring data segments after deleting redundant monitoring data, completing interpolation, and meeting the distribution density requirements. Construct a random forest network with the same number of decision trees as the number of monitoring data segments. Build an evaluation model for the child nodes of each decision tree to assess building quality. Use the monitoring data from all monitoring data segments to evaluate building quality.

[0117] Step S3 specifically includes:

[0118] Step S31: Output all monitoring data segments v′ after deleting redundant monitoring data, completing interpolation, and meeting the distribution density requirements, and obtain the monitoring data within monitoring data segment v′. e is the monitoring data number within the monitoring data segment v′, t e For monitoring data The corresponding data collection time;

[0119] S32: As Figure 3As shown, a random forest network is constructed, with the number of decision trees in the random forest network being the same as the number of monitoring data segments v′. An evaluation model is built for each child node corresponding to each decision tree to evaluate the building quality, and each child node outputs a building quality coefficient based on the evaluation model.

[0120]

[0121] Where, Δ d This is a difference operation, where d is the order of the difference. These are the autoregressive coefficients. For the data acquisition time t e Predicted monitoring data, To monitor the data collection time t within the data segment v′ e-1 Monitoring data, u v ' represents the number of monitoring data points within the monitoring data segment v'. It is white noise. The building quality coefficient output for child nodes;

[0122] In this embodiment, the evaluation model is constructed based on an autoregressive model. By performing differential operations on the monitoring data within the monitoring data segment, the ideal monitoring data at the corresponding collection time is predicted. The building quality coefficient is obtained by calculating the difference rate between the ideal monitoring data and the actual monitoring data. The larger the building quality coefficient, the greater the degree of abnormality in the monitoring data, and the more likely the building has quality problems.

[0123] S33: Input the monitoring data within each monitoring data segment v′ into the child nodes of the decision tree. Each decision tree outputs a building quality coefficient. Then, use a fusion model to fuse the building quality coefficients output by each decision tree and output the total building quality coefficient F obtained using the nth type of monitoring data. n ;

[0124]

[0125] Where, λ v′ λ1 is the weighting coefficient for the impact of monitoring data segment v′ on building quality assessment, k is the incrementing gradient of the weighting coefficient, and V′ is the number of monitoring data segments v′.

[0126] In this embodiment, the importance of the weighting coefficients is assessed based on time sequence. The later the data collection time, the more factors affecting building quality accumulate, and the more important the collected monitoring data is for building quality assessment. This scheme introduces an increasing gradient k so that the weighting coefficients increase with time, with the unit of increase being the time period corresponding to the monitoring data segment. For example, an increase in the number of earthquakes, floods, and mudslides will accelerate building damage, leading to faster foundation settlement, widening cracks, and structural deformation.

[0127] S34: Output the total building quality coefficient data {F} based on each type of monitoring data. 1 ,F 2 ,…,F N}; N represents the type of monitoring data, F N This is the total building quality coefficient output based on the Nth type of monitoring data;

[0128] S35: Combining the total building quality coefficient data {F 1 ,F 2 ,…,F N}, calculate the building quality assessment coefficient F0;

[0129]

[0130] S36: Set the standard value F′ for the building quality assessment coefficient. If F′ < F 0 If F′≥F, then the current building quality is determined to be poor, and a building quality risk warning needs to be issued. 0 If the current building quality is good, then there is no need to issue a building quality risk warning.

[0131] This invention analyzes and preprocesses large volumes of multi-type monitoring data, optimizes and supplements the large volumes of monitoring data based on time series, ensures the rationality of the collected monitoring data, reduces data collection errors, and outputs quantitative analysis data characterizing building quality by building a random forest network and an autoregressive model. This enables high-precision assessment of building quality, facilitates early warning of existing problems and potential risks in building quality, effectively improves the management level of building quality, and enhances the decision-making efficiency of building safety risk early warning.

Claims

1. A data analysis-based method for assessing building quality, characterized in that, include: Step S1: Collect monitoring data related to building quality, obtain a time-series-based monitoring dataset, calculate the probability density in the time distribution based on the time interval between monitoring datasets, and segment the monitoring dataset using the probability density corresponding to the monitoring data to form several monitoring data segments. Step S2: Based on the amount of data in each monitoring data segment, merge the monitoring data segments with smaller data amounts into adjacent monitoring data segments. Based on the distribution density of the monitoring data in the monitoring data segment, calculate the amount of redundant data and the amount of data that needs to be interpolated, delete redundant monitoring data in the monitoring data segment, and interpolate monitoring data segments with insufficient data amounts. Step S3: Output all monitoring data segments after deleting redundant monitoring data, completing interpolation, and meeting the distribution density requirements. Construct a random forest network with the same number of decision trees as the number of monitoring data segments. Build an evaluation model for the child nodes of each decision tree to assess building quality. Use the monitoring data from all monitoring data segments to evaluate building quality.

2. The building quality assessment method based on data analysis according to claim 1, characterized in that, Step S1 includes: Step S11: Determine the monitoring data related to the building quality. The types of monitoring data include foundation settlement, crack width, structural deformation, vibration data, etc., and obtain a time-series-based monitoring dataset. n represents the type of monitoring data, t m For the time of data collection, Let time t m The collected monitoring data, where m is the number of monitoring data; Step S12: Calculate the probability density of the monitoring data in the time distribution based on the time intervals between monitoring data; Where i is the number of the monitoring data, p i Let be the probability density of the i-th monitoring data, and represent the time period t during which monitoring data i was acquired. i-1 ~t i+1 Total time period t1~t m The proportion; Step S13: Obtain the probability density dataset P = {p2, p3, ... pm-1} corresponding to the 2nd monitoring data to the (m-1)th monitoring data. m-1 }, p m-1 Let be the probability density corresponding to the (m-1)th monitoring data; Step S14: Set the ideal probability density threshold p0, and divide the monitoring dataset A into several monitoring data segments.

3. The building quality assessment method based on data analysis according to claim 2, characterized in that, The segmentation method for the monitoring data segment is as follows: Iterate through the probability density data in the probability density dataset P, and compare the probability density with the probability density threshold p0. If p i If ≤p0, it indicates that the monitoring data The corresponding probability density meets the requirements, and the monitoring data will be used. Corresponding acquisition time t i As a segmentation point; Otherwise, monitoring data The corresponding probability density does not meet the requirements; Based on the segment points in monitoring dataset A and the monitoring data at the beginning and end The monitoring data between two segment points is considered as a single monitoring data segment.

4. The building quality assessment method based on data analysis according to claim 3, characterized in that, Step S2 includes: Step S21: Obtain the data volume u within each monitoring data segment. v v is the number of the monitoring data segment. The monitoring data segment with smaller data volume is merged into the adjacent monitoring data segment to ensure the amount of data within the monitoring data segment. Step S22: Repeat step S21 until the data volume in all monitoring data segments is greater than or equal to the minimum value u0. After that, output the monitoring data in each monitoring data segment. w is the monitoring data number within the v-th monitoring data segment, t w To obtain monitoring data The time of data collection; Step S23: Calculate the distribution density of monitoring data within each monitoring data segment; Where, η v Let v be the distribution density of the monitoring data in the v-th monitoring data segment. Let v be the start time of the v-th monitoring data segment. This is the end time of the v-th monitoring data segment; Step S24: Calculate the average distribution density η0 of the monitoring data based on the distribution density of the monitoring data within each monitoring data segment; Where V represents the number of monitoring data segments; Step S25: Calculate the distribution density η of the monitoring data within each monitoring data segment. v Compare with the average distribution density η0; If η v If η > 0, it is determined that the distribution density of the monitoring data in the vth monitoring data segment is high, and the redundant monitoring data in the monitoring data segment needs to be deleted. If η v If η < 0, it is determined that the distribution density of the monitoring data in the vth monitoring data segment is small, and the monitoring data segment is interpolated to supplement the amount of monitoring data in the monitoring data segment; If η v If η = 0, then the distribution density of the monitoring data in the vth monitoring data segment meets the requirements, and step S3 is executed.

5. The building quality assessment method based on data analysis according to claim 4, characterized in that, The method for incorporating the small monitoring data segment into the adjacent monitoring data segment is as follows: Set the minimum data volume requirement u0 within the monitoring data segment, and compare the data volume u. v The size between the minimum value u0 and the minimum value u0; If u v If <u0, then compare the data volume u of the adjacent monitoring data segments v-1 and v+1 on both sides of the monitoring data segment v. v-1 u v+1 Size; if u v-1 ≤u v+1 Then the monitoring data in monitoring data segment v will be merged into monitoring data segment v-1, if u v-1 ≥u v+1 Then the monitoring data in monitoring data segment v will be integrated into monitoring data segment v+1 to form a new monitoring data segment; If u v If u ≥ 0, then retain the monitoring data segment v.

6. The building quality assessment method based on data analysis according to claim 5, characterized in that, The method for deleting redundant monitoring data within a monitoring data segment is as follows: Step S26: Calculate the amount of redundant data u1 in the v-th monitoring data segment; Indicates rounding down; Sort the probability density of each monitoring data point within the v-th monitoring data segment from smallest to largest to form the sorted probability density dataset. It is the maximum value of the probability density, and satisfies Based on the sorted probability density dataset P v The monitoring data corresponding to the first u1 probability densities are deleted from the vth monitoring data segment to obtain the vth monitoring data segment after deleting the redundant monitoring data.

7. The building quality assessment method based on data analysis according to claim 6, characterized in that, The method for determining the amount of monitoring data within the supplementary monitoring data segment is as follows: Step S27: Calculate the amount of data u2 that needs to be interpolated within the v-th monitoring data segment; Indicates rounding up; Sort the probability density data corresponding to each monitoring data point within the v-th monitoring data segment from largest to smallest to form the sorted probability density dataset. It is the minimum value of the probability density, and satisfies Based on the sorted probability density dataset P v ′, from the time of collection of monitoring data corresponding to the highest probability density p1′ Initially, interpolation is performed sequentially within the time interval between two monitoring data points until the number of interpolated monitoring data points in the vth monitoring data segment reaches the data volume u2. Then, interpolation is stopped, and the interpolated vth monitoring data segment is output. The interpolation method for monitoring data is as follows: in, For the data acquisition time t w With the acquisition time t w+1 Monitoring data interpolated between time periods This is the first monitoring data within the v-th monitoring data segment.

8. The building quality assessment method based on data analysis according to claim 7, characterized in that, Step S3 includes: Step S31: Output all monitoring data segments v′ after deleting redundant monitoring data, completing interpolation, and meeting the distribution density requirements, and obtain the monitoring data within monitoring data segment v′. e is the monitoring data number within the monitoring data segment v′, t e For monitoring data The corresponding data collection time; S32: Construct a random forest network. The number of decision trees in the random forest network is the same as the number of monitoring data segments v′. Build an evaluation model for the child nodes corresponding to each decision tree to evaluate the building quality. Each child node outputs a building quality coefficient based on the evaluation model. Where Δd represents the difference operation, and d is the order of the difference. These are the autoregressive coefficients. For the data acquisition time t e Predicted monitoring data, To monitor the data collection time t within the data segment v′ e-1 Monitoring data, u v ' represents the number of monitoring data points within the monitoring data segment v'. It is white noise. The building quality coefficient output for child nodes; S33: Input the monitoring data within each monitoring data segment v′ into the child nodes of the decision tree. Each decision tree outputs a building quality coefficient. Then, use a fusion model to fuse the building quality coefficients output by each decision tree and output the total building quality coefficient F obtained using the nth type of monitoring data. n ; Where, λ v′ λ1 is the weighting coefficient for the impact of monitoring data segment v′ on building quality assessment, k is the incrementing gradient of the weighting coefficient, and V′ is the number of monitoring data segments v′. S34: Output the total building quality coefficient data {F} based on each type of monitoring data. 1 ,F 2 ,…,F N }; N represents the type of monitoring data, F N This is the total building quality coefficient output based on the Nth type of monitoring data; S35: Combining the total building quality coefficient data {F 1 ,F 2 ,…,F N }, calculate the building quality assessment coefficient F0; S36: Set the standard value F′ for the building quality assessment coefficient. If F′ < F 0 If F′≥F, then the current building quality is determined to be poor, and a building quality risk warning needs to be issued. 0 If the current building quality is good, then there is no need to issue a building quality risk warning.

Images