CN-121599236-B - Intelligent building monitoring management method based on multidimensional data and data analysis

CN121599236BCN 121599236 BCN121599236 BCN 121599236BCN-121599236-B

Abstract

The invention relates to the technical field of intelligent building monitoring and discloses an intelligent building monitoring and management method based on multidimensional data and data analysis. According to the method, original heterogeneous data of a plurality of sensor networks in a building are synchronously collected, and a building digital mirror image body with space-time consistency is constructed through data fusion and structure reduction processing. And on the basis, the dynamic response characteristic and the static intrinsic characteristic of the structure are separated and extracted, so that two data sets of time-varying and steady state are obtained. By progressively matching the time-varying features with the historical failure feature patterns in the knowledge base, an abnormal evolution mode is identified, and then performance attenuation tracks of the building structure in a plurality of future maintenance periods are deduced, and an active intervention scheme accurately corresponding to the time point is generated according to the performance attenuation tracks. The method realizes the whole process from unified modeling of multi-source heterogeneous data to intelligent prediction of structural performance degradation, and supports the transition of building monitoring from post alarming to pre prediction and accurate maintenance.

Inventors

DONG ZHONGSHU
PENG JING
HUANG SONG
YU PENGXIANG
XIE YIPING

Assignees

四川琪汇新材料有限责任公司

Dates

Publication Date: 20260512
Application Date: 20260127

Claims (7)

1. The intelligent building monitoring management method based on multidimensional data and data analysis is characterized by comprising the following processing flows: synchronously acquiring original heterogeneous data streams in a monitoring time window from a plurality of appointed sensing networks of a building; performing data fusion and structure restoration processing on the original heterogeneous data stream to construct a building digital mirror image body with complete space-time consistency; On the basis of the building digital mirror image body, the separation and extraction processing of dynamic load response characteristics and static structure intrinsic characteristics is executed, and a time-varying characteristic data set and a steady-state characteristic data set are separated, and the method comprises the following steps: Applying a group of preset virtual excitation signals to the building digital mirror body, and obtaining a theoretical response field of the building digital mirror body under the virtual excitation signals through calculation; extracting actual response field data corresponding to an actual environmental load from the continuous space-time field data; Calculating a residual field between the theoretical response field and the actual response field data, the residual field characterizing a response bias caused by a structural local property change; Carrying out multi-scale space decomposition on the residual field, and separating a macro-scale residual component related to the rigidity of the whole structure and a micro-scale residual component related to the local damage and the connection state; Respectively extracting time domain features of the macro-scale residual components and the micro-scale residual components to form the time-varying feature data set describing the integral time-varying characteristics of the structure; extracting material attribute parameters and geometric connection parameters which remain relatively stable in a monitoring time window from the structural parameter distribution of the building digital mirror image body to form the steady-state characteristic data set; performing progressive pattern matching on the time-varying feature data set and a historical failure feature pattern in a knowledge base, and identifying an abnormal evolution pattern of structural response, wherein the method comprises the following steps of: Retrieving historical monitoring cases which are the same as or similar to the type of the target building structure from the knowledge base, extracting complete characteristic evolution sequences which are developed from a normal state to a failure state in each case, and forming a historical failure characteristic map; Arranging the feature vectors in the time-varying feature data set into a feature evolution sequence to be identified according to a time sequence; calculating similarity distances between the feature evolution sequence to be identified and early, middle and late stage features of each historical failure feature map in the knowledge base by adopting a dynamic time warping algorithm; according to the similarity distance, screening a plurality of target historical failure feature maps with highest matching degree with the feature evolution sequence to be identified, and extracting subsequent evolution paths; Based on the difference between the evolution path of the target historical failure characteristic spectrum and the current characteristic evolution sequence to be identified, predicting the abnormal evolution mode possibly occurring in the future of the current structure through interpolation and extrapolation calculation, wherein the abnormal evolution mode comprises a damage type, a position and a diffusion direction; According to the abnormal evolution mode and the steady-state characteristic data set, a performance attenuation track of the building structure in a plurality of future maintenance periods is deduced, and the method comprises the following steps: establishing a performance degradation prediction model taking the steady-state characteristic data set as an initial state and the abnormal evolution mode as a driving rule; in the performance degradation prediction model, dividing a time axis into a plurality of continuous tiny time steps, and updating parameters of corresponding structure parts in each tiny time step according to a damage expansion rule described by the abnormal evolution mode; introducing random load disturbance conforming to probability distribution into the performance degradation prediction model by combining statistical distribution of environmental load, and simulating uncertainty in actual use; operating the performance degradation prediction model, performing iterative computation until a preset performance failure threshold is reached, and recording a complete curve of the overall performance index of the structure along with the time change in the whole simulation process; Intercepting performance index predicted values corresponding to different future maintenance period time points from the complete curve, and marking time points when the performance index is lower than the early warning threshold value of each level for the first time, wherein the performance attenuation track is defined by the time points and the corresponding performance index predicted values; And generating an active intervention scheme which is accurately matched with the future time point according to the performance attenuation track.
2. The intelligent building monitoring management method based on multidimensional data and data analysis according to claim 1, wherein the performing data fusion and structure restoration processing on the original heterogeneous data stream constructs a building digital mirror with complete space-time consistency, comprising the following steps: Establishing unified time stamp alignment and space registration references aiming at sensing data of different sources in the original heterogeneous data stream, and mapping all data to a space-time coordinate system of the same building information model; Constructing a multi-source data incidence matrix according to the physical attribute and the space topological relation of the sensing data, wherein the multi-source data incidence matrix is used for describing the coupling and transmission relation among vibration, stress, temperature and humidity data; Based on the multi-source data association matrix, carrying out missing value filling and abnormal value correction processing on the data with space-time association to generate cleaned continuous space-time field data; Driving a preset building physical characteristic calculation model by taking the continuous space-time field data as boundary conditions and initial conditions, and solving the distribution of structural parameters which cannot be directly measured in the building in an inverse way; And carrying out depth fusion on the structural parameter distribution obtained by solving and the building information model to form a three-dimensional computable model capable of reflecting the internal state of the structure in real time, wherein the three-dimensional computable model is the building digital mirror image body.
3. The intelligent building monitoring management method based on multidimensional data and data analysis according to claim 2, wherein the generating an active intervention plan accurately matched with a future point in time according to the performance decay trajectory comprises the following steps: analyzing the performance attenuation track, and determining a specific future time point of which the predicted value of the performance index is lower than the early warning threshold value of each level; reversely analyzing a key component set and parameter degradation amounts thereof which lead to performance degradation in the performance degradation prediction model for each specific future time point; matching executable interventions from a library of interventions including, but not limited to, prestressing, increasing temporary support, replacing local components, applying protective coatings, according to the set of critical components and their amount of parameter degradation; Evaluating the implementation effect of each matched intervention measure in the performance degradation prediction model, namely the recovery degree and the maintenance time of the structural performance index after implementation; Combining engineering cost and building use constraint, performing multi-objective optimization on the intervention measures capable of effectively maintaining performance at each future time point, and designating an accurate implementation time window, specific operation parameters and expected performance maintenance period for each optimal intervention measure to form a series of time-triggered active intervention schemes.
4. The intelligent building monitoring and managing method based on multidimensional data and data analysis according to claim 3, wherein the driving a preset building physical characteristic calculation model, solving the distribution of the structural parameters inside the building in a reverse direction, comprises the following steps: Constructing a physical equation inverse problem model which takes the structural parameter distribution as a variable to be solved and the continuous space-time field data as a known observation value; Introducing regularization constraint terms reflecting the structural parameter spatial distribution smoothness into the physical equation inverse problem model; minimizing the difference between the predicted field data generated by the building physical characteristic calculation model and the continuous space-time field data by adopting an iterative optimization algorithm, and simultaneously meeting the regularization constraint; in each iteration, calculating the gradient of an objective function to the variable to be solved by a concomitant state method according to the difference between prediction and actual measurement, and updating the structural parameter distribution along the gradient descending direction; And stopping calculation when the difference between the predicted field data and the actually measured continuous space-time field data is smaller than a preset tolerance or the iteration number reaches an upper limit, and taking the finally obtained optimized parameter distribution as the structural parameter distribution.
5. The intelligent building monitoring and management method based on multidimensional data and data analysis according to claim 4, wherein the construction of a physical equation inverse problem model with the structural parameter distribution as variables to be solved and the continuous spatio-temporal field data as known observations comprises the following steps: Geometrically dispersing the building digital mirror image body into a limited number of units, distributing a group of structural parameters to be identified to each unit, and forming the variables to be solved by the structural parameters of all the units; Establishing a control equation describing building mechanics response, and taking the variable to be solved as a spatial distribution coefficient in the control equation; defining an objective function, wherein the objective function is a norm of a difference between a theoretical response calculated by the control equation under a given boundary condition and an observed value of a corresponding physical quantity in the continuous space-time field data; adding the regularization constraint term in the objective function, wherein the regularization constraint term is a gradient of the variable to be solved in space or a norm of a Laplacian, and is used for inhibiting discomfort of a solution; The inverse problem model of the physical equation is expressed as finding the distribution of the structural parameters that minimizes the objective function under the regularization constraint.
6. The intelligent building monitoring and management method based on multidimensional data and data analysis according to claim 5, wherein the method further comprises a process of incremental learning update to the knowledge base, specifically comprising: in the subsequent monitoring of the target building structure, recording the structure state evolution data which actually occurs and the effect feedback data of the implemented active intervention scheme; When the actual structural state evolution and the predicted abnormal evolution mode or the performance attenuation track have significant deviation, taking the complete monitoring, predicting, intervening and feedback sequence as a new case; performing feature extraction and standardization processing on the new case to generate a new feature evolution sequence and a corresponding actual evolution end point label; storing the processed new cases into the knowledge base, and forming an updated knowledge base together with the original historical failure characteristic patterns; And when the next execution mode is matched, the updated knowledge base is used for realizing continuous evolution of the model prediction capability.
7. The intelligent building monitoring and management method based on multidimensional data and data analysis according to claim 6, wherein the calculating the similarity distance between the feature evolution sequence to be identified and the early, middle and late stage features of each historical failure feature map in the knowledge base by adopting a dynamic time warping algorithm comprises the following steps: Three characteristic subsequences representing early, middle and late stages of failure development are respectively cut out from each historical failure characteristic map of the knowledge base according to a preset time division ratio; Taking the characteristic evolution sequence to be identified as a sequence to be matched, and taking early, middle and late characteristic subsequences of the historical failure characteristic spectrum as three reference sequences respectively; for the sequence to be matched and each reference sequence, executing a dynamic time warping algorithm, and calculating the minimum accumulated cost required for the nonlinear stretching or compression of the sequence to be matched on a time axis to optimally match the reference sequence by constructing an accumulated cost matrix and searching an optimal warping path; and quantifying the minimum accumulated cost as a similarity distance between the two, wherein the sequence to be matched and the early, middle and late characteristic subsequences of the historical failure characteristic spectrum are respectively calculated to finally obtain three similarity distance values respectively representing the matching degree with the early, middle and late stages.

Description

Intelligent building monitoring management method based on multidimensional data and data analysis Technical Field The invention relates to the technical field of intelligent building monitoring, in particular to an intelligent building monitoring management method based on multidimensional data and data analysis. Background Current building structure health monitoring systems generally rely on deployment of various sensor networks, such as vibration, strain, temperature and humidity, video monitoring, etc., to collect operational state data of a building. The main practice mode of the system is to display, store and send out over-limit alarm based on simple threshold value in real time on independent data of each sensor network. The mode regards data of different physical quantities, different sampling rates and different time space references as isolated information flow, and can only capture transient anomalies of specific parameters. This conventional solution has drawbacks. Due to the lack of deep fusion of multi-source heterogeneous data and construction of a unified space-time reference, the monitoring result presents fragmentation, and the real state of the building structure as a complete system is difficult to comprehensively and accurately reflect. The analysis stays on the surface layer, and the integral action mode and evolution rule of the structure hidden behind the data cannot be revealed. Such systems are passive in nature and only alarm after an abnormality or fault occurs or when the severity is reached, and potential risk evolution trends cannot be identified in advance, which cannot provide a decision basis for predictive maintenance. The method is used for solving the two core problems of how to construct a unified model capable of faithfully reflecting the overall appearance of the physical entity of the building from discrete heterogeneous data and how to realize early identification and prospective prediction of the structural performance degradation process based on the model. Disclosure of Invention The invention aims to provide an intelligent building monitoring management method based on multidimensional data and data analysis, which aims to solve the problems in the background technology. In order to achieve the above object, the present invention provides an intelligent building monitoring management method based on multidimensional data and data analysis, the method comprising: synchronously acquiring original heterogeneous data streams in a monitoring time window from a plurality of appointed sensing networks of a building; performing data fusion and structure restoration processing on the original heterogeneous data stream to construct a building digital mirror image body with complete space-time consistency; On the basis of the building digital mirror image body, performing separation and extraction processing of dynamic load response characteristics and static structure intrinsic characteristics, and separating a time-varying characteristic data set and a steady-state characteristic data set; Performing progressive pattern matching on the time-varying characteristic data set and a historical failure characteristic map in a knowledge base, and identifying an abnormal evolution mode of structural response; according to the abnormal evolution mode and the steady-state characteristic data set, deducing a performance attenuation track of the building structure in a plurality of future maintenance periods; And generating an active intervention scheme which is accurately matched with the future time point according to the performance attenuation track. Preferably, the data fusion and structure restoration processing are performed on the original heterogeneous data stream to construct a building digital mirror body with complete space-time consistency, which comprises the following steps: Establishing unified time stamp alignment and space registration references aiming at sensing data of different sources in the original heterogeneous data stream, and mapping all data to a space-time coordinate system of the same building information model; Constructing a multi-source data incidence matrix according to the physical attribute and the space topological relation of the sensing data, wherein the multi-source data incidence matrix is used for describing the coupling and transmission relation among vibration, stress, temperature and humidity data; Based on the multi-source data association matrix, carrying out missing value filling and abnormal value correction processing on the data with space-time association to generate cleaned continuous space-time field data; Driving a preset building physical characteristic calculation model by taking the continuous space-time field data as boundary conditions and initial conditions, and solving the distribution of structural parameters which cannot be directly measured in the building in an inverse way; And carrying out depth fusion on the stru