Intelligent building monitoring method and system based on artificial intelligence, and medium

Abstract

The invention relates to the technical field of data processing, and discloses an intelligent building monitoring method and system based on artificial intelligence, and a medium. The method comprises the following steps: performing time sequence decomposition feature extraction, weight quantification and feature mapping unification on preprocessed data to obtain a feature matrix; building variables are decomposed, protection parameters are encrypted, a topological relation is established, and state indexes are obtained; inputting indexes to iteratively identify abnormity, enhancing rare categories, and performing confidence grading to obtain early warning data; based on the predicted use condition, a maintenance sequence is optimized in a multi-objective mode, and a maintenance strategy is generated; extracting experience, cross-domain migration features and group updating knowledge from strategy compression, and constructing a monitoring library; through the technical means of multi-source data fusion, abnormal mode recognition, knowledge migration and the like, the accuracy of building monitoring and the efficiency of maintenance management are improved, and intelligent evaluation of the building health state and optimization decision of the maintenance strategy are achieved.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to an artificial intelligence-based smart building monitoring method, system and medium. Background Art

[0002] In the field of building monitoring and maintenance management, traditional monitoring methods mainly rely on manual inspections and regular testing, and obtain information on the health status of buildings through on-site surveys, non-destructive testing and other means. With the development of the Internet of Things and sensor technology, building monitoring has gradually become automated and networked, and various sensors can be used to collect structural response data in real time, including physical quantities such as displacement, strain, and acceleration. In terms of data analysis, computer-aided analysis technology has begun to be applied to process and analyze the collected monitoring data to evaluate the status of the building. In terms of building maintenance management, a standardized maintenance system has also been gradually established, and corresponding maintenance strategies have been formulated for different types of damage, accumulating a large amount of engineering practice experience.

[0003] However, the existing building monitoring technology has some obvious shortcomings: first, the collection and analysis of monitoring data lacks intelligent means, the data processing efficiency is low, and it is difficult to detect potential safety hazards in a timely manner; second, the monitoring system has limited ability to identify abnormal conditions, especially for rare damage patterns, and it is often impossible to accurately judge the degree of danger; third, the extraction and application of maintenance experience are too dependent on manual experience and lack of systematic knowledge management methods, resulting in low maintenance efficiency and difficulty in technology promotion; fourth, it is difficult to effectively share monitoring and maintenance experience between different buildings, resulting in waste of resources and affecting the overall improvement of building operation and maintenance management level. Summary of the invention

[0004] The present application provides an artificial intelligence-based smart building monitoring method, system and medium, which are used to improve the accuracy of building monitoring and the efficiency of maintenance management through technical means such as multi-source data fusion, abnormal pattern recognition, and knowledge transfer, and to achieve intelligent assessment of the health status of buildings and optimal decision-making on maintenance strategies.

[0005] In the first aspect, the present application provides an artificial intelligence-based smart building monitoring method, which includes: importing building sensor data into edge computing nodes for distributed sampling, performing wavelet transform noise reduction filtering on the high-frequency signals of the sampled data, and removing abnormal points based on interval statistics to obtain a preprocessed data set; completing time series decomposition according to the preprocessed data set to extract data periodic features, quantifying each feature weight according to feature importance, using multi-source feature mapping to unify feature space conversion, and generating a building feature matrix; based on the building feature matrix, obtaining building state variables through multi-dimensional parameter decomposition, and taking data encryption transformation Sensitive parameter protection is achieved, and the topological structure between parameters is established through correlation calculation to generate building status indicators; the building status indicators are input into the gradient iteration process to identify abnormal patterns, data enhancement processing is performed on rare abnormal categories, and the degree of abnormality is graded in combination with confidence interval calculation, and building early warning data is output; based on the building early warning data, trend extrapolation is carried out to predict the use conditions, and the maintenance task priority is determined through multi-objective optimization, and the maintenance plan is screened according to the benefit ratio calculation to generate a building maintenance strategy; knowledge compression is applied to the building maintenance strategy to extract maintenance experience, and knowledge features are transferred using cross-domain mapping. Maintenance knowledge is updated through group collaboration to build an intelligent monitoring knowledge base.

[0006] In a second aspect, the present application provides an artificial intelligence-based smart building monitoring system, the artificial intelligence-based smart building monitoring system comprising:

[0007] The sampling module is used to import the building sensor data into the edge computing node for distributed sampling, perform wavelet transform noise reduction filtering on the high-frequency signal of the sampled data, and remove abnormal points based on the interval statistics method to obtain a preprocessed data set;

[0008] An extraction module is used to perform time series decomposition according to the preprocessed data set to extract data period features, quantify each feature weight according to feature importance, use multi-source feature mapping to unify feature space conversion, and generate a building feature matrix;

[0009] Establishing a module for obtaining building state variables through multi-dimensional parameter decomposition based on the building feature matrix, adopting data encryption transformation to realize sensitive parameter protection, establishing a topological structure between parameters through correlation calculation, and generating building state indicators;

[0010] An identification module is used to input the building status indicators into a gradient iteration process to identify abnormal patterns, perform data enhancement processing on rare abnormal categories, classify the degree of abnormality in combination with confidence interval operations, and output building early warning data;

[0011] A prediction module, which is used to predict the use condition based on the building early warning data by trend extrapolation, determine the maintenance task priority through multi-objective optimization, select the maintenance plan according to the benefit ratio calculation, and generate the building maintenance strategy;

[0012] The updating module is used to apply knowledge compression to the building maintenance strategy to extract maintenance experience, use cross-domain mapping to migrate knowledge features, update maintenance knowledge through group collaboration, and build an intelligent monitoring knowledge base.

[0013] The third aspect of the present application provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute the above-mentioned artificial intelligence-based smart building monitoring method.

[0014] In the technical solution provided by the present application, distributed sampling is performed through edge computing nodes to achieve on-site processing and transmission optimization of monitoring data, reduce the burden of data transmission, and improve real-time response capabilities; wavelet transform noise reduction filtering and interval statistics are used to eliminate abnormal points, which significantly improves the quality and reliability of original data; time series decomposition and feature extraction are performed based on the preprocessed data set, combined with feature importance quantification and multi-source feature mapping, to achieve accurate expression of building status characteristics; through multi-dimensional parameter decomposition and data encryption transformation, both the integrity of building status variables and the security of sensitive data are guaranteed; gradient iteration is used to identify abnormal patterns, and combined with data enhancement processing and confidence interval operations, the accuracy and reliability of abnormal state identification are improved; through trend extrapolation prediction and multi-objective optimization, scientific decision-making of maintenance tasks and optimal allocation of resources are achieved; knowledge compression is used to extract maintenance experience, and through cross-domain mapping and group collaborative updating, a sustainable and evolving intelligent monitoring knowledge base is established, which deeply combines artificial intelligence technology with building monitoring practice and improves monitoring efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0016] Figure 1 A schematic diagram of an embodiment of a smart building monitoring method based on artificial intelligence in an embodiment of the present application;

[0017] Figure 2 This is a schematic diagram of an embodiment of an artificial intelligence-based smart building monitoring system in an embodiment of the present application. DETAILED DESCRIPTION

[0018] Embodiments of the present application provide a method, system and medium for intelligent building monitoring based on artificial intelligence. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0019] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiments of the present application, an embodiment of the intelligent building monitoring method based on artificial intelligence includes:

[0020] Step S101, importing the building sensor data into the edge computing node for distributed sampling, performing wavelet transform noise reduction filtering on the high-frequency signal of the sampled data, removing abnormal points according to the interval statistics method, and obtaining a preprocessed data set;

[0021] Step S102, completing time series decomposition according to the preprocessed data set to extract data periodic features, quantifying each feature weight according to feature importance, using multi-source feature mapping to unify feature space conversion, and generating a building feature matrix;

[0022] Step S103: Based on the building feature matrix, the building state variables are obtained through multi-dimensional parameter decomposition, sensitive parameter protection is achieved through data encryption transformation, and the topological structure between parameters is established through correlation calculation to generate the building state index;

[0023] Step S104: input the building status index into the gradient iteration process to identify abnormal patterns, perform data enhancement processing on rare abnormal categories, classify the degree of abnormality in combination with confidence interval calculation, and output building early warning data;

[0024] Step S105: Based on the building early warning data, the trend is extrapolated and the usage condition is predicted, the maintenance task priority is determined through multi-objective optimization, the maintenance plan is selected according to the benefit ratio calculation, and the building maintenance strategy is generated;

[0025] Step S106: Apply knowledge compression to extract maintenance experience for building maintenance strategies, use cross-domain mapping to migrate knowledge features, update maintenance knowledge through group collaboration, and build an intelligent monitoring knowledge base.

[0026] It is understandable that the execution subject of the present application can be an intelligent building monitoring system based on artificial intelligence, or a terminal or a server, which is not limited here. The present application embodiment is described by taking the server as the execution subject as an example.

[0027] Specifically, distributed sampling of building sensor data is performed through edge computing nodes. Edge computing nodes refer to data processing units distributed on each floor of a building. These nodes can process sensor data nearby to reduce the burden of data transmission. During the sampling process, edge computing nodes dynamically adjust the sampling frequency according to the degree of data change: when rapid data changes are detected, the sampling frequency is increased to capture the change process; when the data is relatively stable, the sampling frequency is reduced to save computing resources. The collected high-frequency signals are subjected to wavelet transform denoising and filtering. The wavelet transform can effectively separate the noise components in the signal by decomposing the signal into wavelet coefficients of different scales. Specifically, the wavelet transform first selects a suitable wavelet basis function (such as db4 wavelet) to decompose the original signal into coefficients of multiple scales, then thresholds the coefficients of each scale, removes the wavelet coefficients corresponding to the noise, and finally reconstructs the signal to obtain the denoised data. Taking building structure vibration monitoring as an example, the acceleration signal collected by the sensor often contains high-frequency noise. Wavelet transform can effectively filter out these interferences and retain the true structural vibration response.

[0028] Calculate the statistical characteristics of the data (mean, standard deviation), and then determine the outlier judgment threshold based on the 3σ criterion, that is, mark the data points that deviate from the mean by more than three times the standard deviation as outliers. For the marked outliers, replace them by the interpolation method of the adjacent data to ensure the continuity of the data. The processed data form a preprocessed data set, which more clearly reflects the real state of the building. In the time series decomposition stage, the preprocessed data set is analyzed for periodicity to extract the periodic characteristics of the data. The time domain signal is converted to the frequency domain by Fourier transform to identify the main periodic components. For building monitoring data, important periodic features include daily change cycles (such as temperature and load changes), weekly change cycles (such as usage patterns), etc. Based on the extracted periodic features, calculate the importance of each feature, that is, the contribution of the feature to the judgment of the building status. Feature importance is quantified by indicators such as information entropy or variance contribution rate, and features with high importance receive higher weights in subsequent analysis.

[0029] The multi-source feature mapping process converts features from different sensors and different physical quantities into a unified feature space. The various features are normalized to unify the numerical range, and then the mapping relationship between features is established through orthogonal transformation to generate a building feature matrix. Each row in the feature matrix represents a time point, each column corresponds to a feature dimension, and the matrix element value represents the strength or degree of the feature. The building feature matrix is ​​decomposed by multidimensional parameters to obtain variables reflecting the status of different aspects of the building. The decomposition process uses dimensionality reduction methods such as principal component analysis to extract the main state characteristics. For sensitive state parameters (such as stress data of key structural components), data encryption transformation is used to protect them to ensure data security. By calculating the correlation coefficient between parameters, the topological relationship between parameters is established to form a building status indicator system. In order to identify abnormal building states, the building status indicators are input into the gradient iteration process. Gradient iteration minimizes the prediction error by continuously adjusting parameters and gradually identifies abnormal patterns. For rare abnormal categories with a small number of them, data enhancement processing is used to expand the samples. Data enhancement includes adding random perturbations, feature combinations and other methods to generate new abnormal samples. The confidence interval operation is based on the statistical distribution characteristics, quantitatively classifies the degree of abnormality, and outputs the building early warning data. The trend extrapolation prediction analyzes the development trend based on the building early warning data in the working condition stage. Through the time series analysis method, the change trend of each state parameter is predicted, and combined with the influence of environmental factors, the working condition forecast for a period of time in the future is obtained. The determination of maintenance task priority takes into account multiple goals: maintenance urgency, resource constraints, maintenance effect, etc. Through the multi-objective optimization algorithm, the optimal balance is found between multiple goals to form a task priority ranking. The benefit ratio calculation further evaluates the input-output ratio of different maintenance plans and selects the optimal maintenance strategy.

[0030] In the process of knowledge compression and maintenance experience extraction, maintenance strategies are structured and analyzed to extract core maintenance rules and experience. Through cross-domain mapping, maintenance experience gained on one building is transformed into knowledge features that can be applied to other similar buildings. The group collaborative update mechanism allows monitoring systems of multiple buildings to share and exchange maintenance experience, continuously optimize maintenance knowledge, and build an intelligent monitoring knowledge base.

[0031] In the embodiment of the present application, distributed sampling is performed through edge computing nodes to achieve on-site processing and transmission optimization of monitoring data, reduce the data transmission burden, and improve real-time response capabilities; wavelet transform noise reduction filtering and interval statistics are used to eliminate abnormal points, which significantly improves the quality and reliability of the original data; time series decomposition and feature extraction are performed based on the preprocessed data set, combined with feature importance quantification and multi-source feature mapping, to achieve accurate expression of building status characteristics; through multi-dimensional parameter decomposition and data encryption transformation, both the integrity of building status variables and the security of sensitive data are guaranteed; gradient iteration is used to identify abnormal patterns, and combined with data enhancement processing and confidence interval operations, the accuracy and reliability of abnormal state identification are improved; through trend extrapolation prediction and multi-objective optimization, scientific decision-making of maintenance tasks and optimal allocation of resources are achieved; knowledge compression is used to extract maintenance experience, and through cross-domain mapping and group collaborative updating, a sustainable and evolving intelligent monitoring knowledge base is established, which deeply combines artificial intelligence technology with building monitoring practice and improves monitoring efficiency and accuracy.

[0032] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0033] (1) The building sensor data is hierarchically sorted by timestamp, the data is grouped and sliced ​​using a dynamic window length, and the slicing parameters are adjusted according to the data distribution characteristics to obtain the initial data fragments;

[0034] (2) The missing data points in the initial data segment are located and marked, and an interpolation benchmark is established based on the change trend of the previous and next data. The missing positions are supplemented by piecewise linear interpolation, and a complete data sequence is obtained by combining data continuity test;

[0035] (3) Based on the sampling density distribution of the complete data sequence, the data sampling threshold of the edge computing node is partitioned and set, and the sampling time interval is determined by adaptive adjustment of the data change rate. Combined with the cross-validation of the data between nodes, the distributed sampling data is obtained;

[0036] (4) Using the distributed sampling data as the basic signal, multi-scale wavelet basis function decomposition is performed, the energy proportion of the decomposed high-frequency signal is calculated, and the filtering threshold is set by evaluating the importance of the wavelet transform coefficients. The high-frequency signal is filtered to reduce noise and obtain the reduced-noise data;

[0037] (5) Sliding segment the denoised data according to fixed length, calculate the abnormality judgment benchmark based on interval statistical characteristics, classify the data points into abnormal levels according to the 3σ criterion, mark them according to the quantitative index of the abnormal degree, and obtain the abnormal data points;

[0038] (6) Perform correlation analysis on the abnormal data points and the adjacent data, select the optimal replacement value based on the data distribution law, repair and reconstruct the data breakpoints through piecewise smoothing function, and obtain the preprocessed data set by combining data consistency verification.

[0039] Specifically, the timestamp records the exact moment of data collection, including complete information of year, month, day, hour, minute, and second. The hierarchical sorting process first sorts the data progressively according to the different levels of the timestamp (year, month, day, etc.) to ensure the temporal integrity of the data. The dynamic window length refers to the time span of data analysis that is adaptively adjusted according to the characteristics of the data. Different window lengths are used for different monitoring parameters of the building: structural vibration data requires a shorter window (such as 1 minute) to capture the instantaneous response, while temperature changes require a longer window (such as 1 hour) to reflect the change process. Data grouping and slicing is to divide the continuous data stream into several interrelated data fragments in the time dimension. Slicing parameters include window length, overlap rate, and sampling interval, which are dynamically adjusted according to the data distribution characteristics. The data distribution characteristics are to describe the discrete degree and distribution shape of the data by calculating statistics such as the coefficient of variation, kurtosis, and skewness of the data. When the data fluctuates violently, the window length is reduced and the overlap rate is increased to capture the rapidly changing characteristics; when the data is relatively stable, the window length is increased to reduce redundant data.

[0040] Data missing point location marking is the process of identifying blank points or invalid values ​​in the data sequence. The causes of missing data include sensor failure, communication interruption or external interference. The location process identifies the location and duration of the missing point by checking the time continuity and validity of the data. The establishment of the interpolation benchmark is based on the change trend of the data before and after the missing point, and the local characteristics of the data (such as slope and curvature) are analyzed to determine the appropriate interpolation method. Piecewise linear interpolation is a commonly used interpolation method, and different interpolation strategies are used for different types of data: linear interpolation is used for slowly changing data (such as temperature); high-order interpolation methods are used for rapidly changing data (such as acceleration). The sampling density distribution of the complete data sequence reflects the distribution characteristics of the data in the time domain. The sampling density is measured by the number of data points per unit time, and a higher sampling density means more detailed data capture. The data sampling threshold of the edge computing node is the condition for triggering data collection, including time threshold and change threshold. The partition setting is to use different sampling strategies for different areas according to the importance and change characteristics of the data: a higher sampling density is used for key areas (such as major load-bearing components), and a relatively low sampling density is used for non-key areas.

[0041] Adaptive adjustment of the data change rate is the core mechanism for dynamically optimizing the sampling interval. When the monitored data changes rapidly, the data accuracy is improved by shortening the sampling interval; when the data changes slowly, the data redundancy is reduced by increasing the sampling interval. Cross-validation of data between nodes ensures the reliability of sampled data by comparing the consistency of data from adjacent nodes. The data of adjacent nodes should show a certain degree of correlation and continuity, and data that deviates significantly from this characteristic needs to be specially marked and processed.

[0042] Multi-scale wavelet basis function decomposition can analyze the characteristics of signals at different scales. When performing wavelet decomposition on building monitoring data, first select a suitable wavelet basis function (such as Daubechies wavelet, Haar wavelet, etc.), and then decompose the signal into sub-signals of different frequency bands. High-frequency signals usually contain noise and interference components, and energy ratio analysis is required to distinguish effective signals from noise. Energy ratio calculation is to evaluate the importance of the signal by calculating the ratio of the energy of the signal in each frequency band to the total energy. The importance evaluation of wavelet transform coefficients is based on the amplitude and frequency of occurrence of the coefficients to judge their contribution to the original signal. By setting an appropriate threshold, important wavelet coefficients are retained and minor coefficients are suppressed, thereby achieving signal noise reduction. High-frequency signal noise reduction filtering not only removes random noise, but also retains the mutation characteristics of the signal.

[0043] Sliding segmentation is to divide the denoised data into fixed lengths, with a certain overlap between adjacent segments to ensure the continuity of the data. Interval statistical features include statistics such as mean, standard deviation, and extreme value, which are used to describe the central tendency and dispersion of data. The 3σ criterion is a commonly used anomaly detection method, which determines data points that deviate from the mean by more than three times the standard deviation as abnormal. The abnormal level classification further divides the abnormal points into different levels according to the degree of deviation. The processing of abnormal data points needs to consider the temporal and spatial correlation of the data. By analyzing the relationship between the abnormal points and the adjacent data points, the appropriate replacement strategy is determined. The data distribution law includes characteristics such as periodicity, trend and randomness, which guide the repair process of the abnormal points. The piecewise smoothing function is used to eliminate mutations and jumps in the data to ensure that the repaired data maintains the original change characteristics.

[0044] Taking the structural health monitoring of a high-rise building as an example, the specific application of the entire data preprocessing process is explained: First, the data from the acceleration sensors on each floor are hierarchically sorted by time stamp, and a 10-minute dynamic window is set to slice the vibration data. By analyzing the coefficient of variation of the data, it is found that when the building is subjected to wind load, the volatility of the high-rise vibration data increases. At this time, the window length is shortened to 5 minutes to capture more details. For the missing data points found, the interpolation benchmark is established according to the vibration characteristics of the previous and next moments, and the missing data is supplemented by the cubic spline interpolation method. After the complete data sequence is formed, the edge computing node dynamically adjusts the sampling frequency according to the rate of change of the vibration amplitude, and increases the sampling density when obvious vibration is detected. The db4 wavelet is used to decompose the data into 6 layers, analyze the energy distribution of each scale coefficient, and set a threshold to remove high-frequency noise. Finally, the processed data is subjected to anomaly detection, and it is found that individual data points exceed the 3σ range. These points often correspond to sudden external loads. By comparing and analyzing the data of adjacent sensors, the authenticity of these anomalies is confirmed, and valuable anomaly records are retained.

[0045] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0046] (1) Divide the preprocessed data set into periodic segments on the time axis, identify the decomposition points according to the data change trend, extract the main periodic components through frequency domain transformation, and obtain the data periodic sequence;

[0047] (2) Through multi-scale time division of the data periodic sequence, the phase calibration of the periodic component is completed, and the time series component characteristics are extracted by combining signal reconstruction technology to obtain periodic feature data;

[0048] (3) Quantify the importance based on the periodic characteristic data, calculate the contribution ratio of each feature through the fluctuation amplitude, calibrate the weight parameter by normalization, and obtain the feature weight;

[0049] (4) Performing spatial transformation on the periodic feature data according to the feature weights, aligning the multi-source data using feature projection, and obtaining a unified feature space by unifying the metric standard through coordinate mapping conversion;

[0050] (5) Simplify the redundant dimensions in the unified feature space, select the dominant feature components according to the information retention threshold, construct the feature expression through dimensional reorganization, and obtain the feature expression vector;

[0051] (6) The feature expression vectors are structured and organized according to the spatiotemporal topological relationship, and the correlation strength between features is described by combining matrix operations. The effective feature combinations are screened through correlation verification to obtain the architectural feature matrix.

[0052] Specifically, the intelligent assessment of building structural safety starts with the period division of the preprocessed data set. Intelligent structural monitoring first needs to identify the characteristic periods of different types of sensor data, including daily periodic deformation captured by structural deformation sensors (reflecting the thermal expansion and contraction of the structure caused by temperature stress), weekly periodic changes recorded by strain sensors (reflecting the cumulative strain caused by building use loads), and seasonal periods monitored by displacement sensors (reflecting foundation settlement and concrete creep). Intelligent decomposition point recognition is based on the mutation characteristics of structural response data. By intelligently calculating the first-order strain difference and the second-order stress difference, the moment when the structural performance changes significantly is accurately located. The intelligent frequency domain analysis of the structure uses fast Fourier transform to convert the time domain response signal of the building into the frequency domain feature space, extract the dominant periodic component, and establish a data period sequence. The multi-scale time analysis of the building intelligent monitoring system sets corresponding time windows for different structural response characteristics: the millisecond time window is used to monitor the structural vibration characteristics (identify the building's natural frequency and damping ratio), the minute time window is used to analyze the temperature stress changes (evaluate the impact of thermal stress on the structure), and the hourly time window is used to study the use load effect (monitor the cumulative deformation of the structure). Phase calibration in building intelligent monitoring ensures accurate alignment of various structural response data in time series by calculating the intelligent cross-correlation function between different sensor signals. Intelligent signal reconstruction technology organically combines the structural response characteristics of each scale after calibration to form a representation of the building's health status.

[0053] In the process of building intelligent safety assessment, the evaluation of the importance of periodic characteristic data involves multiple intelligent monitoring indicators: structural response energy, vibration amplitude, duration, etc. The structural intelligent monitoring system uses peak analysis and root mean square calculation to quantify the intensity of changes in structural response signals. Feature contribution intelligent analysis calculates the weight coefficients of each monitoring indicator in the overall structural performance evaluation, and uses intelligent normalization processing to convert the structural response parameters of different physical quantities into a unified evaluation space. Intelligent weight calibration comprehensively considers the impact of each monitoring indicator on structural safety and gives higher evaluation weights to the characteristics that reflect the key performance of the structure. Intelligent space transformation performs dimensional transformation on structural periodic characteristic data based on feature weights. Building intelligent monitoring uses principal component analysis to map high-dimensional structural response features to the main deformation mode directions. Multi-source data intelligent alignment solves the problem of asynchronous sampling of different types of structural sensors, and all monitoring data are strictly aligned in the time dimension through intelligent interpolation and resampling. Intelligent coordinate mapping unifies the structural monitoring results of various physical quantities (such as displacement, strain, acceleration) into a standardized feature space. The unified feature space of building intelligent monitoring has information redundancy and requires intelligent dimensionality reduction processing. The information retention threshold is intelligently calculated based on the cumulative variance contribution rate to ensure that the main characteristic information of the structural state is retained after dimensionality reduction. Intelligent feature selection takes into account both the physical meaning and engineering application value of the monitoring indicators. The dimension reorganization process reorganizes the selected main structural features according to the principles of structural mechanics to construct an intelligent feature expression with clear engineering meaning.

[0054] The building intelligent monitoring system organizes feature expression vectors according to the spatiotemporal topological relationship of the structure, taking into full account the spatial structure and stress characteristics of the building. Intelligent matrix operations describe the correlation strength between structural features, including calculating the correlation coefficient matrix and characteristic distance matrix of the structural response. Intelligent correlation verification sets a correlation threshold to screen out structural feature combinations with significant correlation and form a building feature matrix.

[0055] For example, the building structure is equipped with an intelligent sensor network, including structural strain sensors, displacement sensors, acceleration sensors and other monitoring equipment. The intelligent monitoring system performs periodic analysis on the preprocessed data and finds that the temperature strain of the concrete structure shows an obvious 24-hour periodicity, the basic period of the building structure is about 3 seconds, and the wind-induced structural response shows irregular random vibration characteristics. Through multi-scale intelligent analysis, the structural features of these different periods are extracted respectively: the structural vibration characteristics are extracted from the high-frequency data, and the temperature deformation characteristics are extracted from the low-frequency data.

[0056] In the displacement monitoring data, the intelligent fluctuation analysis found that the characteristic contribution of the horizontal displacement of the top floor was the largest, which is consistent with the deformation characteristics of high-rise buildings; the temperature strain monitoring showed that the temperature effect of the middle floors was more obvious, which reflects the thermal deformation law of the building. In the process of feature space conversion, the intelligent monitoring system uniformly maps the structural response data of different physical quantities (displacement, strain, acceleration) to the standardized feature space. Through the dimensionality reduction of intelligent principal component analysis, the main characteristic components that can fully describe the structural state are retained. Finally, based on the spatial layout of structural components and the mechanical transfer relationship, these features are reorganized into a feature matrix reflecting the overall state of the building.

[0057] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0058] (1) Extract the structural displacement, deformation stress value and vibration acceleration from the building feature matrix and perform multi-dimensional decomposition. Reconstruct the building load-bearing parameters through primary and secondary stress analysis, obtain the building mechanical characteristics through inter-layer response decomposition, and obtain the parameter decomposition sequence;

[0059] (2) Extract local features of the material stiffness coefficient, structural damping ratio and natural frequency in the parameter decomposition sequence, calibrate the core indicators according to the deformation law of reinforced concrete structure, and obtain the building state variables through static and dynamic coupling analysis;

[0060] (3) The crack width, deflection change and strain data in the building state variables are divided into different levels of sensitivity, and the structural stress and strain data are scrambled using a key, and sensitive parameters are protected through layered encryption;

[0061] (4) Analyze the foundation settlement, wall cracking degree and floor crack data from the sensitive parameter protection, build the connection relationship with the structural stress and strain as the node, select the key association according to the safety threshold, and obtain the topological basic data;

[0062] (5) Extract the transfer characteristics of beam-column deformation, node displacement value and structural stiffness coefficient in the topological basic data, construct the damage propagation network through the correlation strength calculation, characterize the variable dependency through the topological structure between parameters, and obtain the correlation topological map;

[0063] (6) The structural safety, component integrity and durability indicators in the associated topological map are reconstructed and combined, the building structure condition is described after bearing capacity verification, and the building status indicator is obtained through hierarchical analysis.

[0064] Specifically, key structural parameters are extracted from the characteristic matrix collected by the building intelligent sensor network. The structural displacement reflects the deformation state of the building under the action of external loads. The intelligent monitoring system captures the displacement between each layer in real time through the displacement sensor; the deformation stress value measures the stress state of the structural components through the strain sensor, reflecting the stress condition of the building's load-bearing system; the vibration acceleration monitors the dynamic response characteristics of the building through the acceleration sensor. Intelligent multi-dimensional splitting decomposes these physical quantities according to spatial distribution and time series, and determines the stress state of key load-bearing components through primary and secondary stress analysis. The inter-layer response decomposition technology decomposes the overall response of the building into the relative response of each floor, thereby obtaining a description of the building's mechanical characteristics. The parameter decomposition sequence in the intelligent monitoring of buildings contains the key mechanical parameters of materials and structures. The material stiffness coefficient reflects the deformation characteristics of the structural material and is calculated through intelligent strain analysis; the structural damping ratio characterizes the ability of the building to consume vibration energy and is intelligently identified through the free vibration attenuation curve; the natural frequency is the inherent dynamic characteristic of the building, which is intelligently extracted through the structural response under environmental excitation. The local feature extraction technology establishes a mapping relationship between deformation law and material properties for the nonlinear deformation characteristics of reinforced concrete structures. Static-dynamic coupling analysis takes into account static performance (such as bearing capacity) and dynamic characteristics (such as damping characteristics) to form a set of building state variables.

[0065] Sensitivity analysis of building state variables is extremely important. Crack width is a direct indicator for assessing the degree of structural damage, and intelligent monitoring tracks it in real time through crack monitoring sensors; the change in deflection reflects the degree of deformation of the component, which is accurately measured by laser displacement sensors; and strain data reflects the stress state inside the material. The sensitivity level is divided based on the degree of influence of these parameters on structural safety. Intelligent encryption algorithms are used to protect sensitive data, and data security is ensured through layered encryption strategies. Key structural performance indicators are extracted from sensitive parameters. Foundation settlement reflects the stability of the foundation and is intelligently monitored through precise leveling measurements; wall cracking represents the integrity of the load-bearing wall and is quantified through intelligent image recognition technology; floor crack data reflects changes in the bearing capacity of the floor. Taking structural stress and strain as the basic nodes, a mechanical association network between components is established, and important structural response associations are screened by setting safety thresholds.

[0066] The topological basic data of intelligent building monitoring includes structural deformation characteristics. The deformation of beams and columns reflects the stress state of the main structure, the node displacement value represents the deformation of key connection parts, and the structural stiffness coefficient reflects the overall anti-deformation ability. Through intelligent correlation analysis, a damage propagation network is established to reveal the development law and propagation path of structural damage. The topological structure between parameters describes the mutual influence relationship between each monitoring indicator and forms a correlation topological map. By analyzing the correlation topological map, the health status of the building structure is evaluated. Structural safety is a comprehensive indicator to measure the reliability of the overall structure. The integrity of the component reflects the damage state of each component, and the durability index predicts the long-term performance of the structure. The current status of the structure is quantitatively evaluated through bearing capacity verification, and the building status index is obtained by combining the hierarchical analysis method.

[0067] For example, the building adopts reinforced concrete frame shear wall structure, and the building intelligent monitoring system has deployed a multi-level sensor network. In daily monitoring, the intelligent analysis system first extracts the interlayer displacement angle from the displacement sensor data, combines the concrete strain value measured by the strain sensor and the structural vibration response collected by the acceleration sensor, and forms a preliminary parameter decomposition sequence through multi-dimensional data fusion. Subsequently, based on the concrete elastic modulus and steel bar strain hardening characteristics determined by the material test data, combined with the measured structural damping ratio (obtained through environmental vibration testing) and natural frequency (obtained through intelligent spectrum analysis), a building state variable system is established. When the intelligent monitoring system finds cracks in a shear wall, it immediately starts local encrypted monitoring: the development of cracks is continuously tracked through the intelligent crack monitor, and the wall strain monitoring points are increased. These sensitive data are stored in the database after multiple encryption. When the intelligent analysis finds that the foundation has an uneven settlement trend, the multi-source data association analysis is started: the settlement observation data is intelligently associated with the upper structure deformation data, the stress redistribution law caused by settlement is discovered, and the possible damage development path is predicted.

[0068] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0069] (1) The crack propagation rate, foundation settlement, and concrete strength change rate in the building status indicators are subjected to gradient optimization iteration, and the numerical range is limited by the foundation bearing capacity boundary constraint. The parameter update amount is calculated through structural response analysis to obtain the iterative optimization result;

[0070] (2) The distribution characteristics of shear wall cracking degree, steel bar corrosion rate, and concrete carbonization depth were extracted from the iterative optimization results. The damage types were divided through cluster analysis. Rare structural defects were screened by structural deformation distance to obtain rare anomaly categories.

[0071] (3) Structural crack patterns, deformation mutations, and vibration responses in rare anomaly categories are sampled according to the damage probability distribution, and typical damage samples are generated through component stress perturbations. Abnormal working conditions are supplemented in the structural feature space to obtain an enhanced data set;

[0072] (4) Statistically analyze the material strength degradation, anchoring performance loss, and structural stiffness degradation rate in the enhanced data set, calculate the confidence limit based on the structural reliability, obtain the damage threshold through failure probability density estimation, and obtain the confidence interval boundary;

[0073] (5) Quantify the structural bearing capacity loss, seismic performance degradation and durability attenuation based on the confidence interval boundary, determine the damage level by hazard level classification, calibrate the abnormality level according to the degree of structural response deviation, and obtain the abnormality degree index;

[0074] (6) The abnormality index is associated with the dynamic characteristics of the building structure, and warning information is generated through structural safety judgment. The impact range is analyzed based on the damage extension mechanism to obtain building warning data.

[0075] It should be noted that the crack expansion rate is the rate of change of the crack width over time continuously tracked by intelligent crack monitoring equipment, the foundation settlement is measured by high-precision settlement monitoring instruments to measure the vertical displacement of the foundation, and the concrete strength change rate is evaluated by intelligent non-destructive testing equipment to assess the degradation of concrete strength. During the intelligent gradient optimization process, the foundation bearing capacity boundary constraints ensure that the calculation results are consistent with the actual project, and the parameter values ​​are continuously updated through structural response analysis to obtain iterative optimization results that meet the actual project.

[0076] Damage identification in building intelligent monitoring extracts key damage features from iterative optimization results. The cracking degree of shear walls is monitored in real time by intelligent image recognition technology. The corrosion rate of steel bars is evaluated by electrochemical impedance spectroscopy. The carbonization depth of concrete is measured by an intelligent carbonization detector to measure the advancement depth of the carbonization front. Intelligent clustering analysis groups these damage features according to similarity, and identifies abnormal types that are significantly different from common damage patterns through structural deformation distance calculation, thereby obtaining rare abnormal categories. For the identified rare abnormal categories, the intelligent monitoring system performs data enhancement processing. The structural crack pattern includes characteristics such as the direction, width and depth of the crack. The deformation mutation reflects the instantaneous deformation of the structural components, and the vibration response describes the changes in the dynamic characteristics of the structure. Sampling is performed based on the damage probability distribution, and new damage conditions are generated by changing the stress state of the components, enriching the sample library of abnormal conditions and forming an enhanced data set.

[0077] In structural reliability analysis, it is necessary to accurately calculate the confidence limit. First, define the calculation formula for the probability of structural failure:

[0078]

[0079] Among them, R f represents the probability of structural failure, is the joint probability density function of the structural parameters, ψ(g(x)) is the limit state function, and g(x) is the structural performance function.

[0080] The structural damage threshold is determined by failure probability density estimation:

[0081]

[0082] Among them, D t is the damage threshold, λ is the correction factor, w i is the weight factor, ξ i is the damage characteristic value, θ i is the damage sensitivity, γ i is the uncertainty coefficient and n is the number of features.

[0083] Structural performance is evaluated based on the confidence interval boundary. The loss of structural bearing capacity is determined through load tests and finite element analysis. The degradation of seismic performance is evaluated based on the change in the structural period extension rate and damping ratio. The durability attenuation takes into account the degradation of material properties and the influence of environmental factors. The intelligent hazard level classification takes various indicators into consideration, determines the abnormality level according to the degree of deviation between the structural response and the normal state, and obtains the abnormality degree index. The building intelligent early warning combines the abnormality degree index with the structural dynamic characteristics. The dynamic characteristics include parameters such as natural frequency, vibration mode and damping ratio. The corresponding relationship between abnormal state and dynamic characteristic changes is established through intelligent correlation analysis. The weak parts of the structure are analyzed based on the damage extension mechanism, the potential impact range of the damage is evaluated, and the building early warning data is formed.

[0084] For example, the intelligent monitoring equipment detected cracks in a shear wall. Through continuous monitoring, it was found that the crack expansion rate showed an accelerating trend. At the same time, the foundation settlement monitoring showed that the uneven settlement exceeded the alarm value, and the local concrete strength decreased significantly. The intelligent analysis system immediately started the optimization iterative calculation, considering the foundation bearing capacity constraint conditions and analyzing the structural response characteristics. Data analysis found that the cracking pattern of the shear wall was abnormal, which was significantly different from the common temperature cracks and shrinkage cracks. At the same time, a composite damage mode of steel corrosion and concrete carbonization was detected. Intelligent clustering analysis marked this damage mode as a rare abnormal category, and then simulated similar damage states by changing the load conditions to expand the abnormal sample library. After reliability analysis and failure probability calculation, the critical threshold of damage was determined, and a graded warning standard was established accordingly. As the monitoring continued, the intelligent analysis system found that the local stiffness of the structure was significantly reduced. Through dynamic characteristics comparison analysis, the potential expansion path of the damage was predicted, and early warning information was issued in time.

[0085] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0086] (1) The structural safety, crack propagation rate, and component damage degree in the building early warning data are reconstructed into time series, extracted in sections according to the concrete strength degradation cycle, and the component degradation trend is calibrated by displacement deformation spectrum analysis to obtain a trend prediction benchmark;

[0087] (2) Carry out extrapolation calculation of beam-column deformation, reinforcement stress level and crack depth in the trend prediction benchmark, correct for environmental factors such as earthquake and wind load, and obtain the service condition through structural displacement correlation analysis;

[0088] (3) The structural damage level distribution, concrete spalling range and component bearing capacity changes in the use conditions are screened according to the reinforcement resource constraints and sorted according to the principle of structural force balance to obtain a task sequence list;

[0089] (4) Match the task sequence list with the historical reinforcement records, select the optimal reinforcement combination through structural reinforcement effect evaluation, and obtain the maintenance task priority through bearing capacity recovery comparison;

[0090] (5) Allocate resources for the structural reinforcement scheme with maintenance task priority, determine the construction time by calculating the concrete repair cycle, select the optimal scheme by calculating the benefit ratio, and obtain the maintenance resource allocation;

[0091] (6) The maintenance resource allocation and structural warning level are combined, and a maintenance plan is generated according to the structural reinforcement specification constraints. The maintenance process is constructed according to the construction sequence to obtain the building maintenance strategy.

[0092] Specifically, the structural safety index reflects the overall safety status of the building, which is obtained through a comprehensive evaluation of continuously monitored stress, strain, displacement and other data; the crack expansion rate describes the development speed of structural cracks over time, which is obtained by continuously recording the change in crack width through intelligent crack monitoring equipment; the component damage degree quantifies the degree of damage to the structural components, which is obtained based on non-destructive testing data evaluation. The intelligent time series reconstruction technology first segments the data according to the natural cycle of concrete strength degradation (usually 28 days, 90 days, 180 days and other key time nodes), and the data in each time period reflects the degradation characteristics of different stages. The displacement deformation spectrum analysis converts the time domain data to the frequency domain through Fourier transform, identifies the main deformation modes and frequency components, and thus calibrates the degradation development trend of the components. The trend prediction benchmark includes the core indicators of the evolution of structural performance. The deformation of beams and columns is obtained by real-time monitoring of displacement sensors, reflecting the deformation state of the main load-bearing components; the stress level of steel bars is obtained by measuring strain sensors, which characterizes the stress level of steel bars; the crack depth is determined by non-destructive testing methods such as ultrasonic testing. Intelligent extrapolation calculations are based on these monitoring data, taking into account the impact of environmental factors: seismic loads are corrected by ground motion parameters, and wind loads are calibrated based on wind pressure coefficients and wind vibration effects. Structural displacement correlation analysis establishes the correlation between the displacements of each monitoring point and obtains a description of the operating conditions.

[0093] Based on the analysis of the use conditions, intelligent maintenance management evaluates the distribution of structural damage levels and reflects the damage status of each component through a damage map; the extent of concrete spalling is determined by high-precision three-dimensional scanning to determine the damaged area; and the change in component bearing capacity is calculated based on load test data. Reinforcement resource constraints include factors such as material supply, construction equipment, and technicians. The principle of structural force balance requires that the reinforcement sequence must ensure the overall stability of the structure during the maintenance process. The task sequence table takes these factors into consideration and sorts the maintenance tasks. The historical reinforcement records contain detailed information on previous maintenance projects, such as reinforcement plans, construction processes, and material selection. The intelligent matching process compares and analyzes the current task sequence table with historical records to evaluate the applicability of different reinforcement plans. The evaluation of the structural reinforcement effect is based on indicators such as the degree of bearing capacity recovery and durability improvement after reinforcement, and the optimal reinforcement combination plan is determined through comprehensive scoring. The determination of maintenance task priorities takes into account both the urgency of the project and the efficiency of resource utilization.

[0094] Maintenance resource allocation is a multi-objective optimization process. For structural reinforcement schemes with different priorities, first determine the resource requirements such as materials, equipment, and manpower, and arrange the construction sequence in combination with the maintenance cycle of concrete repair (usually including stages such as demolding, maintenance, and strength compliance). The benefit ratio calculation comprehensively considers the maintenance cost investment and performance improvement effect, and selects the implementation plan with the best economic and technical indicators. The maintenance resource allocation plan needs to meet the project quality requirements and construction period constraints. The formulation of the building intelligent maintenance strategy organically combines the maintenance resource allocation with the structural warning level. The structural reinforcement specification has clear requirements for material selection, structural requirements, construction technology, etc., and the maintenance plan must strictly follow these technical specifications. The arrangement of the construction sequence needs to consider the structural force transmission path to ensure the structural safety during the reinforcement process. The formation of a building maintenance strategy includes implementation plans and quality control measures.

[0095] For example, intelligent monitoring found that the main load-bearing frame of the building had different degrees of damage, and a scientific repair and reinforcement plan needed to be formulated. Time series analysis of building early warning data showed that the crack expansion rate of the frame beam showed an accelerating trend, and the concrete strength of some columns showed obvious degradation. Through displacement deformation spectrum analysis, the deformation development trend of the damaged components was determined, and a trend prediction benchmark was established. Subsequently, the structural performance was extrapolated and predicted based on the monitoring data, and the prediction results were corrected for environmental factors considering the earthquake intensity and basic wind pressure in the area. The analysis found that the displacement correlation in the frame node area was strong, indicating that the damage had affected the overall performance of the structure. Combined with the results of the on-site inspection, the concrete spalling area and the degree of steel corrosion were determined, and a preliminary repair plan was formulated. By reviewing the historical maintenance records, it was found that similar frame structures had adopted solutions such as bonding steel plates and steel cladding reinforcement. After comparative analysis, a more suitable carbon fiber reinforcement technology was selected. The maintenance plan determines the construction sequence according to the principle of "first the main, then the secondary, first the top, then the bottom, first the inside, then the outside", and arranges the construction period according to the curing time and strength development law of the carbon fiber material. The maintenance strategy formed not only ensures the reinforcement effect, but also minimizes the impact on the building's use function.

[0096] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0097] (1) Extract the processing flow of the building maintenance strategy in sections, obtain the operation sequence by disassembling the maintenance steps, and organize the core processes in time sequence to obtain maintenance operation data;

[0098] (2) Compress the maintenance operation data to remove redundancy, extract key feature points through data dimension reduction, filter and retain the core content according to the information weight, and obtain maintenance experience;

[0099] (3) Discretely sample maintenance experience, establish cross-domain mapping relationships through feature space transformation, align feature distribution through coordinate transformation, and obtain knowledge features;

[0100] (4) The knowledge features are verified in different building scenarios, the optimal migration path is matched through similarity calculation, and the migration content is selected according to the feature importance to obtain the migration mapping result;

[0101] (5) Perform group distribution analysis on the migration mapping results, select effective knowledge through maintenance effect evaluation, determine the update content according to the group voting mechanism, and obtain maintenance knowledge;

[0102] (6) The maintenance knowledge is organized in a hierarchical structure, a knowledge graph is constructed through association rule mining, and a retrieval index is established through semantic annotation to obtain an intelligent monitoring knowledge base.

[0103] Specifically, segmented processing divides the maintenance strategy into the preparation stage (including site investigation and scheme design), the implementation stage (including structural reinforcement and material repair) and the acceptance stage (including quality inspection and effect evaluation) according to the characteristics of structural maintenance projects. Intelligent maintenance step decomposition refines the work content of each stage into specific operation units, such as crack repair steps including crack cleaning, grouting material preparation, pressure grouting, etc. According to the construction process and quality control requirements, these operation units are organized in time sequence to form standardized maintenance operation data. Intelligent compression processing of maintenance operation data is achieved by removing redundant operations and optimizing processes. Key feature points such as control parameters of key processes and quality inspection standards are extracted through data dimensionality reduction technology. The calculation of information weight is based on the degree of influence of each operation step on the maintenance effect. The content with higher weight includes structural reinforcement process, material ratio parameters, maintenance conditions, etc. The maintenance experience obtained after screening and sorting reflects the most valuable technical points in engineering practice.

[0104] Discrete sampling of maintenance experience is the basis of knowledge transfer. By selecting representative sample points in maintenance experience, feature space transformation is performed. The establishment of cross-domain mapping relationship can be expressed as:

[0105]

[0106] Among them, M(x) is the mapping function, α i is the feature importance coefficient, ω i is the weight factor, h i (x) is the feature conversion function, σ is the activation function, β iis the bias parameter, η is the mapping error, and n is the feature dimension. The feature alignment between different maintenance scenarios is achieved through coordinate transformation to obtain standardized knowledge features. The verification process of knowledge features involves multiple building maintenance scenarios. The target scenarios suitable for knowledge transfer are identified through similarity calculation, considering the matching degree of factors such as building type, structural form, and damage characteristics. The feature importance evaluation is based on the verification results of the migration effect, and the migration content that performs well in multiple scenarios is given priority to form the migration mapping result.

[0107] Group distribution analysis focuses on the applicability of knowledge between different buildings. Maintenance effect evaluation uses multiple indicators, including the degree of bearing capacity recovery, durability improvement effect, construction quality compliance rate, etc. The group voting mechanism determines the maintenance knowledge with the most promotion value through statistical analysis, and selects technical content with good practical verification effect and wide application range. The construction of the intelligent knowledge base adopts a hierarchical organizational structure. The first layer is the basic knowledge layer, which contains basic information such as material properties and structural requirements; the middle layer is the technical process layer, which includes detailed processes of various maintenance and reinforcement methods; the upper layer is the experience decision-making layer, which provides maintenance plan selection suggestions under different damage conditions. The connection between knowledge units is established through association rule mining technology to form a knowledge map. Semantic annotation uses professional terms to index knowledge content and establish multi-dimensional retrieval channels.

[0108] For example, a frame structure building has concrete cracks and steel bar corrosion problems, and structural repair is required. First, extract the relevant maintenance processes from the maintenance strategy library, including crack repair, steel bar anti-corrosion, concrete repair and other processes. Through intelligent analysis, these processes are refined into specific operation steps, such as crack repair including crack detection and positioning, crack cleaning, epoxy resin preparation, pressure grouting, maintenance, etc., to form an operation sequence. After data compression processing, the most critical technical parameters are retained: the correspondence between crack width and grouting material selection, pressure control range of pressure grouting, maintenance temperature and humidity requirements, etc. These experiences are transformed through feature space to establish knowledge mapping relationships with other similar buildings. The applicability of these experiences is verified in the maintenance projects of other similar buildings, and the maintenance knowledge is continuously optimized through effect evaluation. These verified maintenance knowledge is sorted and summarized into the intelligent monitoring knowledge base to provide technical support for subsequent building maintenance projects.

[0109] In a specific embodiment, the step of discretely sampling the maintenance experience includes:

[0110] (1) The structural crack repair process, concrete reinforcement methods, and component reinforcement measures in maintenance experience are selected for sampling points in a time sequence, and the maintenance process is disassembled to obtain discrete data nodes. The sampling positions are calibrated according to the requirements of the construction specifications to obtain a discrete sampling sequence;

[0111] (2) The crack sealing depth, concrete strength improvement value and steel corrosion treatment degree in the discrete sampling sequence are analyzed, and the parameter mapping relationship is established through the evaluation of the structural reinforcement effect. The characteristic space is calibrated by the bearing capacity recovery rate to obtain the space conversion benchmark;

[0112] (3) The component reinforcement process, structural reinforcement method and durability improvement measures in the space conversion benchmark are mapped across domains, and the characteristic correspondence is established through damage repair degree calculation. The mapping parameters are determined according to the requirements of the reinforcement specification to obtain the cross-domain characteristic relationship;

[0113] (4) The coordinate system of the concrete repair process, crack treatment method and steel protection measures in the cross-domain feature relationship is transformed, and a feature description system is established through quantitative analysis of the reinforcement effect. The coordinate benchmark is calibrated through the structural performance improvement index to obtain an aligned coordinate system;

[0114] (5) The structural reinforcement parameters, repair process indicators and reinforcement effect evaluation under the aligned coordinate system are reorganized, and the feature verification standard is established through the repair quality evaluation. The effective features are screened according to the requirements of the engineering specifications to obtain the feature distribution map;

[0115] (6) Extract knowledge from the reinforcement process flow, repair quality indicators, and durability evaluation parameters in the characteristic distribution map, construct a characteristic knowledge system based on maintenance experience, and obtain knowledge characteristics through repair effect verification.

[0116] Among them, the structural crack repair process formulates specific repair plans for different types of cracks (such as temperature cracks, load cracks, and settlement cracks). Each crack has its own specific treatment process, such as surface coating for fine cracks, low-pressure grouting for small cracks, and pressure grouting for through cracks. Concrete reinforcement methods include increasing the cross-section method, pasting steel plates, and outsourcing steel methods. Each method has its applicable conditions and construction points. Component reinforcement measures use corresponding reinforcement technologies for different component types such as beams, columns, and plates, such as rebar reinforcement for beams and carbon fiber winding for columns. These process flows are decomposed into key nodes in chronological order, such as material preparation, surface treatment, reinforcement construction, and maintenance, to form a discrete sampling sequence. In the discrete sampling sequence, the crack sealing depth is an important indicator for evaluating the crack repair effect. The actual filling depth of the injection material is measured by ultrasonic testing and other means; the concrete strength improvement value is determined by rebound method, core drilling method and other detection methods to determine the strength increment after reinforcement; the degree of steel bar corrosion treatment is evaluated by electrochemical detection. The evaluation of the structural reinforcement effect adopts load tests, dynamic tests and other means to verify the improvement of the bearing capacity of the reinforced components. Through these characteristic analyses, parameter mapping relationships are established, and qualitative process parameters are converted into quantitative evaluation indicators to form a standardized feature space.

[0117] The various technical measures in the spatial conversion benchmark need to be mapped across domains to make them applicable to different building maintenance scenarios. The mapping of component reinforcement technology considers factors such as material properties, construction technology, and environmental conditions; the mapping of structural reinforcement methods needs to evaluate structural types, force characteristics, and usage requirements; and durability improvement measures focus on indicators such as environmental effects, service life, and maintenance costs. Through the quantitative calculation of the degree of damage repair, the characteristic correspondence between different maintenance scenarios is established to ensure the effective transformation of technical measures. The coordinate system conversion of cross-domain characteristic relationships is a key step in achieving knowledge standardization. Concrete repair processes include surface repair, crack grouting, and local replacement, and a unified evaluation standard needs to be established; crack treatment methods involve technical parameters such as grouting material selection, pressure control, and maintenance requirements, which need to be standardized; steel protection measures include cathodic protection, anti-corrosion coatings, epoxy resin sealing, and other technologies, and the effects need to be quantitatively evaluated. Through quantitative analysis of reinforcement effects, a unified measurement standard for various technical indicators is established to form a standardized alignment coordinate system.

[0118] In the process of feature reorganization under the aligned coordinate system, the structural reinforcement parameters include technical indicators such as material strength, structural requirements, and construction technology; the repair process indicators cover key parameters such as material ratio, construction temperature, and maintenance conditions; and the reinforcement effect evaluation includes aspects such as increased bearing capacity, improved durability, and construction quality. Through the maintenance quality assessment, the feature verification standard is established, and the technical feasibility, economic rationality, and construction operability are comprehensively considered to screen out the most valuable technical features. The knowledge extraction of the feature distribution map is the last link in forming standardized maintenance experience. The knowledge extraction of the reinforcement process includes the basis for process selection, key points of construction operation, quality control measures, etc.; the extraction of repair quality indicators focuses on detection methods, acceptance standards, and evaluation rules; and the durability evaluation parameters involve environmental factors, service life, maintenance cycle, and other requirements. Through the systematic summary of maintenance experience, a feature knowledge system is constructed to provide a scientific basis for intelligent monitoring and maintenance of buildings.

[0119] For example, the building's exterior wall has large cracks and concrete spalling problems, which require systematic repair and reinforcement. First, the repair process of similar cases is extracted from the maintenance experience database, the crack type and damage degree are analyzed, and the appropriate repair process is selected. Through discretization processing, the repair process is decomposed into key nodes such as cleaning, grouting, repair, and maintenance, and the control parameters of each node are recorded. Then the repair effect is quantitatively evaluated, the fullness of the crack grouting is confirmed by ultrasonic testing, and the concrete strength recovery of the repair area is detected by rebound method. These test data are standardized to form measurable characteristic indicators. Through comparative analysis with other similar engineering cases, a mapping relationship between repair technology and effect evaluation is established. The successful repair experience is transformed into standardized process specifications, including material selection standards, construction process requirements, quality control measures, etc. These experiences are adapted to different building maintenance scenarios through feature space conversion, and are continuously improved and optimized through practical verification to form a maintenance knowledge system.

[0120] The above describes the smart building monitoring method based on artificial intelligence in the embodiment of the present application. The following describes the smart building monitoring system based on artificial intelligence in the embodiment of the present application. Figure 2 In the embodiments of the present application, an embodiment of the smart building monitoring system based on artificial intelligence includes:

[0121] Sampling module 201, used to import building sensor data into edge computing nodes for distributed sampling, perform wavelet transform noise reduction filtering on high-frequency signals of the sampled data, remove abnormal points according to interval statistics, and obtain a preprocessed data set;

[0122] Extraction module 202, used to complete time series decomposition according to the preprocessed data set to extract data period features, quantify each feature weight according to feature importance, use multi-source feature mapping to unify feature space conversion, and generate a building feature matrix;

[0123] Establishing module 203, which is used to obtain building state variables through multi-dimensional parameter decomposition based on the building feature matrix, adopt data encryption transformation to realize sensitive parameter protection, establish a topological structure between parameters through correlation calculation, and generate building state indicators;

[0124] Identification module 204, used to input the building status indicators into the gradient iteration process to identify abnormal patterns, perform data enhancement processing on rare abnormal categories, classify the degree of abnormality in combination with confidence interval calculation, and output building warning data;

[0125] Prediction module 205, used to predict the use condition based on the building early warning data by trend extrapolation, determine the maintenance task priority through multi-objective optimization, select the maintenance plan according to the benefit ratio calculation, and generate the building maintenance strategy;

[0126] The updating module 206 is used to extract maintenance experience from the building maintenance strategy application knowledge compression, migrate knowledge features using cross-domain mapping, update maintenance knowledge through group collaboration, and build an intelligent monitoring knowledge base

[0127] Through the collaborative cooperation of the above-mentioned components, distributed sampling is carried out through edge computing nodes to achieve on-site processing and transmission optimization of monitoring data, reduce the burden of data transmission, and improve real-time response capabilities; wavelet transform noise reduction filtering and interval statistics are used to eliminate abnormal points, which significantly improves the quality and reliability of original data; time series decomposition and feature extraction are carried out based on the preprocessed data set, combined with feature importance quantification and multi-source feature mapping, to achieve accurate expression of building status characteristics; through multi-dimensional parameter decomposition and data encryption transformation, both the integrity of building status variables and the security of sensitive data are guaranteed; gradient iteration is used to identify abnormal patterns, and combined with data enhancement processing and confidence interval operations, the accuracy and reliability of abnormal state identification are improved; through trend extrapolation prediction and multi-objective optimization, scientific decision-making of maintenance tasks and optimal allocation of resources are achieved; knowledge compression is used to extract maintenance experience, and through cross-domain mapping and group collaborative updating, a sustainable and evolving intelligent monitoring knowledge base is established, which deeply combines artificial intelligence technology with building monitoring practice and improves monitoring efficiency and accuracy.

[0128] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the artificial intelligence-based smart building monitoring method.

[0129] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0130] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0131] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A smart building monitoring method based on artificial intelligence, characterized in that: The artificial intelligence-based smart building monitoring method includes: Import building sensor data into edge computing nodes for distributed sampling, perform wavelet transform noise reduction and filtering on the high-frequency signals of the sampled data, remove abnormal points based on interval statistics, and obtain a preprocessed data set; According to the preprocessed data set, time series decomposition is performed to extract data periodic features, each feature weight is quantified according to feature importance, multi-source feature mapping is used to unify feature space conversion, and a building feature matrix is ​​generated; Based on the building characteristic matrix, the building state variables are obtained through multi-dimensional parameter decomposition, sensitive parameter protection is achieved through data encryption transformation, the topological structure between parameters is established through correlation calculation, and the building state index is generated; Input the building status indicators into the gradient iteration process to identify abnormal patterns, perform data enhancement processing on rare abnormal categories, classify the degree of abnormality in combination with confidence interval calculation, and output building early warning data; Based on the building early warning data, the trend is extrapolated to predict the use condition, the maintenance task priority is determined through multi-objective optimization, the maintenance plan is selected according to the benefit ratio calculation, and the building maintenance strategy is generated; The building maintenance strategy is applied with knowledge compression to extract maintenance experience, knowledge features are transferred using cross-domain mapping, maintenance knowledge is updated through group collaboration, and an intelligent monitoring knowledge base is constructed.

2. The method for monitoring smart buildings based on artificial intelligence according to claim 1, characterized in that: The building sensor data is imported into the edge computing node for distributed sampling, the high-frequency signal of the sampled data is subjected to wavelet transform noise reduction filtering, and abnormal points are removed according to the interval statistics method to obtain a preprocessed data set, including: The initial data segments are obtained by sorting the building sensor data by timestamp, grouping and slicing the data using dynamic window length, and adjusting the slicing parameters according to the data distribution characteristics. The data missing points in the initial data segment are located and marked, an interpolation benchmark is established according to the change trend of the previous and next data, the missing positions are supplemented by piecewise linear interpolation, and a complete data sequence is obtained by combining data continuity test; According to the sampling density distribution of the complete data sequence, the data sampling threshold of the edge computing node is partitioned and set, the sampling time interval is determined by adaptive adjustment of the data change rate, and the distributed sampling data is obtained by combining the data cross-validation between nodes; Using the distributed sampling data as the basic signal, multi-scale wavelet basis function decomposition is performed, energy proportion of the decomposed high-frequency signal is calculated, filtering threshold is set by evaluating the importance of wavelet transform coefficients, and noise reduction data is obtained by filtering the high-frequency signal for noise reduction; The noise reduction data is segmented according to a fixed length, and the abnormality judgment benchmark is calculated in combination with the interval statistical characteristics. The data points are classified into abnormal levels according to the 3σ criterion, and are marked according to the quantitative index of the abnormal degree to obtain abnormal data points; The abnormal data point is subjected to correlation analysis with adjacent data, the optimal replacement value is selected based on the data distribution law, the data breakpoints are repaired and reconstructed through a piecewise smoothing function, and the preprocessed data set is obtained by combining data consistency verification.

3. The method for monitoring a smart building based on artificial intelligence according to claim 1, characterized in that: The method of performing time series decomposition according to the preprocessed data set to extract data periodic features, quantifying each feature weight according to feature importance, using multi-source feature mapping to unify feature space conversion, and generating a building feature matrix includes: Divide the preprocessed data set into periodic segments on the time axis, identify decomposition points according to the data change trend, extract the main periodic components through frequency domain transformation, and obtain a data periodic sequence; Through multi-scale time division of the data periodic sequence, phase calibration is completed for the periodic component, and the time series component characteristics are extracted in combination with signal reconstruction technology to obtain periodic characteristic data; The importance is quantified based on the periodic characteristic data, the contribution ratio of each characteristic is calculated by the fluctuation amplitude, and the weight parameter is calibrated by normalization to obtain the characteristic weight; According to the feature weights, the periodic feature data is spatially transformed, multi-source data is aligned using feature projection, and a unified metric standard is converted through coordinate mapping to obtain a unified feature space; Simplifying redundant dimensions in the unified feature space, selecting dominant feature components according to an information retention threshold, constructing feature expressions through dimensional reorganization, and obtaining feature expression vectors; The feature expression vectors are structured and organized according to the spatiotemporal topological relationship, and the strength of the association between features is described by combining matrix operations. The effective feature combinations are screened through correlation verification to obtain the building feature matrix.

4. The method for monitoring a smart building based on artificial intelligence according to claim 1, characterized in that: The method is based on the building feature matrix, obtains building state variables through multi-dimensional parameter decomposition, adopts data encryption transformation to realize sensitive parameter protection, establishes the topological structure between parameters through correlation calculation, and generates building state indicators, including: Extracting structural displacement, deformation stress value and vibration acceleration from the building feature matrix for multi-dimensional decomposition, reconstructing the building load-bearing parameters through primary and secondary stress analysis, obtaining the building mechanical characteristics through inter-layer response decomposition, and obtaining a parameter decomposition sequence; Extract local features of the material stiffness coefficient, structural damping ratio and natural frequency in the parameter decomposition sequence, calibrate core indicators according to the deformation law of reinforced concrete structure, and obtain the building state variables through static and dynamic coupling analysis; The crack width, deflection change and strain data in the building state variables are divided into different levels of sensitivity, the structural stress and strain data are scrambled using a key, and the sensitive parameter protection is obtained through hierarchical encryption; The foundation settlement, wall cracking degree and floor crack data are parsed from the sensitive parameter protection, the structural stress and strain are used as nodes to build a connection relationship, and the key associations are screened according to the safety threshold to obtain the topological basic data; Extract the transfer characteristics of beam-column deformation, node displacement value and structural stiffness coefficient in the topological basic data, construct a damage propagation network through correlation strength calculation, characterize the variable dependency through the topological structure between parameters, and obtain the correlation topological map; The structural safety, component integrity and durability indicators in the associated topological map are reconstructed and combined, the building structure status is described through bearing capacity verification, and the building status indicator is obtained through hierarchical analysis.

5. The method for monitoring smart buildings based on artificial intelligence according to claim 1, characterized in that: The method of inputting the building status indicator into a gradient iteration process to identify abnormal patterns, performing data enhancement processing on rare abnormal categories, grading the degree of abnormality in combination with confidence interval calculation, and outputting building early warning data includes: The crack propagation rate, foundation settlement and concrete strength change rate in the building status index are subjected to gradient optimization iteration, the numerical range is limited by the foundation bearing capacity boundary constraint, the parameter update amount is calculated through structural response analysis, and an iterative optimization result is obtained; Extracting the shear wall cracking degree, steel bar corrosion rate and concrete carbonization depth distribution characteristics from the iterative optimization results, dividing the damage types through cluster analysis, screening rare structural defects through structural deformation distance, and obtaining the rare anomaly category; Structural crack patterns, deformation mutations and vibration responses in the rare anomaly categories are sampled according to the damage probability distribution, typical damage samples are generated through component stress disturbance, abnormal working conditions are supplemented in the structural feature space, and an enhanced data set is obtained; Statistically analyzing the material strength degradation, anchoring performance loss and structural stiffness degradation rate in the enhanced data set, calculating the confidence limit based on the structural reliability, obtaining the damage threshold through failure probability density estimation, and obtaining the confidence interval boundary; The structural bearing capacity loss, seismic performance degradation and durability attenuation are quantified based on the confidence interval boundary, the damage level is determined by the hazard level classification, the abnormality level is calibrated according to the deviation degree of the structural response, and the abnormality degree index is obtained; The abnormality index is associated with the dynamic characteristics of the building structure, and warning information is generated through structural safety judgment. The impact range is analyzed based on the damage extension mechanism to obtain the building warning data.

6. The method for monitoring a smart building based on artificial intelligence according to claim 1, characterized in that: The method of predicting the use condition based on the building early warning data by trend extrapolation, determining the maintenance task priority by multi-objective optimization, selecting the maintenance plan according to the benefit ratio calculation, and generating the building maintenance strategy includes: Reconstruct the structural safety, crack propagation rate and component damage degree in the building early warning data in time series, extract them in sections according to the concrete strength degradation cycle, calibrate the component degradation trend through displacement deformation spectrum analysis, and obtain a trend prediction benchmark; The deformation of beams and columns, the stress level of steel bars and the crack depth in the trend prediction benchmark are extrapolated and calculated, and the use condition is obtained through structural displacement correlation analysis after correction by environmental factors such as earthquakes and wind loads; The structural damage level distribution, concrete spalling range and component bearing capacity changes in the use conditions are screened by reinforcement resource constraints and sorted according to the principle of structural force balance to obtain a task sequence table; Match the task sequence table with historical reinforcement records, select the optimal reinforcement combination through structural reinforcement effect evaluation, and obtain the maintenance task priority through bearing capacity recovery comparison; Allocate resources for the structural reinforcement scheme with the maintenance task priority, determine the construction time by calculating the concrete repair cycle, select the optimal scheme by calculating the benefit ratio, and obtain the maintenance resource allocation; The maintenance resource configuration is combined with the structural warning level, a maintenance plan is generated according to the structural reinforcement specification constraints, and a maintenance process is constructed according to the construction sequence to obtain the building maintenance strategy.

7. The method for monitoring smart buildings based on artificial intelligence according to claim 1, characterized in that: The method of applying knowledge compression to the building maintenance strategy to extract maintenance experience, utilizing cross-domain mapping to migrate knowledge features, updating maintenance knowledge through group collaboration, and constructing an intelligent monitoring knowledge base includes: Extract the processing flow of the building maintenance strategy in sections, obtain the operation sequence by disassembling the maintenance steps, and organize the core processes in time sequence to obtain maintenance operation data; The maintenance operation data is compressed to remove redundancy, key feature points are extracted through data dimension reduction, and core content is retained by screening according to information weight to obtain the maintenance experience; Discrete sampling is performed on the maintenance experience, a cross-domain mapping relationship is established through feature space conversion, and feature distribution is aligned through coordinate transformation to obtain the knowledge feature; The knowledge features are verified in different building scenarios, the optimal migration path is matched through similarity calculation, and the migration content is selected according to the feature importance to obtain the migration mapping result; Performing group distribution analysis on the migration mapping results, screening effective knowledge through maintenance effect evaluation, determining update content according to a group voting mechanism, and obtaining the maintenance knowledge; The maintenance knowledge is organized according to a hierarchical structure, a knowledge graph is constructed through association rule mining, and a retrieval index is established through semantic annotation to obtain the intelligent monitoring knowledge base.

8. The method for monitoring a smart building based on artificial intelligence according to claim 7, characterized in that: The discrete sampling of the maintenance experience, establishing a cross-domain mapping relationship through feature space conversion, aligning feature distribution through coordinate transformation, and obtaining the knowledge feature include: The structural crack repair process, concrete reinforcement means and component reinforcement measures in the maintenance experience are selected for sampling points in a time sequence, and the maintenance process is disassembled to obtain discrete data nodes. The sampling positions are calibrated according to the construction specification requirements to obtain a discrete sampling sequence; The crack sealing depth, concrete strength improvement value and steel corrosion treatment degree in the discrete sampling sequence are analyzed, a parameter mapping relationship is established through structural reinforcement effect evaluation, and the characteristic space is calibrated through the bearing capacity recovery rate to obtain a space conversion benchmark; The component reinforcement process, structural reinforcement method and durability improvement measures in the space conversion benchmark are cross-domain mapped, the characteristic correspondence is established through damage repair degree calculation, the mapping parameters are determined according to the reinforcement specification requirements, and the cross-domain characteristic relationship is obtained; The concrete repair process, crack treatment method and steel protection measures in the cross-domain characteristic relationship are converted into coordinate systems, a characteristic description system is established through quantitative analysis of the reinforcement effect, and the coordinate reference is calibrated through the structural performance improvement index to obtain an aligned coordinate system; The structural reinforcement parameters, repair process indicators and reinforcement effect evaluation under the alignment coordinate system are reorganized, and feature verification standards are established through repair quality evaluation. Effective features are screened according to engineering specifications to obtain a feature distribution map; Knowledge is extracted from the reinforcement process flow, repair quality indicators and durability evaluation parameters in the characteristic distribution map, a characteristic knowledge system is constructed through maintenance experience summary, and the knowledge characteristics are obtained through repair effect verification.

9. An artificial intelligence-based smart building monitoring system, used to implement the artificial intelligence-based smart building monitoring method as described in any one of claims 1 to 8, characterized in that: The artificial intelligence-based smart building monitoring system includes: The sampling module is used to import the building sensor data into the edge computing node for distributed sampling, perform wavelet transform noise reduction filtering on the high-frequency signal of the sampled data, and remove abnormal points based on the interval statistics method to obtain a preprocessed data set; An extraction module is used to perform time series decomposition according to the preprocessed data set to extract data period features, quantify each feature weight according to feature importance, use multi-source feature mapping to unify feature space conversion, and generate a building feature matrix; Establishing a module for obtaining building state variables through multi-dimensional parameter decomposition based on the building feature matrix, adopting data encryption transformation to realize sensitive parameter protection, establishing a topological structure between parameters through correlation calculation, and generating building state indicators; An identification module is used to input the building status indicators into a gradient iteration process to identify abnormal patterns, perform data enhancement processing on rare abnormal categories, classify the degree of abnormality in combination with confidence interval operations, and output building early warning data; A prediction module, which is used to predict the use condition based on the building early warning data by trend extrapolation, determine the maintenance task priority through multi-objective optimization, select the maintenance plan according to the benefit ratio calculation, and generate the building maintenance strategy; The updating module is used to apply knowledge compression to the building maintenance strategy to extract maintenance experience, use cross-domain mapping to migrate knowledge features, update maintenance knowledge through group collaboration, and build an intelligent monitoring knowledge base.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the artificial intelligence-based smart building monitoring method as described in any one of claims 1-8 is implemented.

Images