CN-122021174-A - Distributed sensor network deployment and data fusion method for high-rise steel structure

CN122021174ACN 122021174 ACN122021174 ACN 122021174ACN-122021174-A

Abstract

The invention discloses a distributed sensor network deployment and data fusion method for a high-rise steel structure. The method relates to the technical field of network deployment and data fusion, and comprises the following steps of key monitoring part identification and sensor deployment, monitoring data processing and self-adaptive fusion, digital twin driving and early warning emergency. The method comprises the steps of determining key monitoring parts of each stage based on a finite element model and construction simulation, deploying a heterogeneous sensor network capable of being dynamically adjusted, binding acquired data in the construction stage after edge pretreatment, generating monitoring characteristic values through three-stage self-adaptive fusion, finally constructing a digital twin model for updating and displaying in real time, and improving the accuracy of deployment and data fusion of the distributed sensor network according to whether the characteristic values exceed an early warning interval to match an emergency plan, so that the problem of low accuracy of deployment and data fusion of the distributed sensor network caused by insufficient construction environment interference and network topology optimization in the prior art is solved.

Inventors

SONG MIN
YANG CHAO
LV JIANNING
CAO XUE
BAO CHAO
DANG JINZHONG
MA QUAN
REN LEPING
SUN ZHIWEI
BAI YUNFEI

Assignees

中建三局集团西北有限公司
中建三局集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260210

Claims (10)

1. The distributed sensor network deployment and data fusion method for the high-rise steel structure is characterized by comprising the following steps of: S100, identifying key monitoring parts of each construction stage based on a finite element model of a high-rise steel structure to be monitored and the whole construction process simulation, and carrying out distributed deployment on the key monitoring parts through a heterogeneous sensor network to adapt to construction dynamic adjustment so as to allow position migration and supplement along with the construction stage; S101, acquiring original parameters of a key monitoring part, preprocessing the original data at an edge computing gateway, binding the preprocessed data with a current construction stage, performing three-level self-adaptive fusion, and recording the value output after the self-adaptive fusion feature fusion as a monitoring feature value, wherein the three-level self-adaptive fusion comprises data level fusion, feature level fusion and decision level fusion; s102, constructing a digital twin model of the high-rise steel structure, driving the digital twin model to update the state and visually display in real time according to the self-adaptive fusion characteristics and the evaluation result, judging whether the monitoring characteristic value is in an early warning critical interval output by the digital twin model, and if so, providing a corresponding emergency plan according to the early warning level.
2. The method for deploying and data fusion of a distributed sensor network for a towering steel structure of claim 1 wherein said critical monitoring sites include, but are not limited to, main load bearing member mounting nodes, temporary support system connection points, overhead welding and bolting areas that dynamically change with construction progress; The heterogeneous sensor network comprises, but is not limited to, a fiber bragg grating sensor and a resistance strain gauge for monitoring component stress, a total station or electronic theodolite reference observation for monitoring three-dimensional deformation and perpendicularity, a piezoelectric acceleration sensor for monitoring structural vibration, and an environment sensor for collecting environment wind speed, temperature and humidity.
3. The method for deployment and data fusion of a distributed sensor network for high-rise steel structures according to claim 1, wherein the specific steps of allowing location migration and augmentation with construction stage are: The method comprises the steps of synchronously collecting original strain signals of all monitoring points at a predefined moment by a plurality of strain gages arranged on key monitoring positions of a structure at a preset sampling frequency, filtering and performing temperature compensation treatment on the original strain signals to obtain effective strain values, and obtaining real-time strain increment of the predefined moment in a preset time window; Traversing all monitoring points in the preset time window to obtain real-time strain increment with the maximum absolute value, calculating a strain increment normalization weight factor of each monitoring point, acquiring the real-time stress duty ratio of the monitoring point at a predefined moment, and dynamically adjusting the stress duty ratio early warning threshold of each monitoring point by the strain increment normalization weight factor to generate a dynamic stress duty ratio early warning threshold.
4. The method for deploying and fusing the data of the distributed sensor network for the high-rise steel structure according to claim 3, wherein the specific steps of generating the dynamic stress duty ratio early warning threshold value are as follows: If the strain increment normalization weight factor is larger than the strain increment weight reference value, indicating that the monitoring point is subjected to relatively severe stress change, relaxing a stress duty ratio early warning threshold of the monitoring point to obtain a dynamic stress duty ratio early warning threshold so as to avoid triggering false alarm due to normal load redistribution; if the strain increment normalization weight factor is smaller than or equal to the strain increment weight reference value, maintaining the current dynamic stress duty ratio early warning threshold; Comparing the real-time stress proportion of each monitoring point with a dynamic stress proportion early warning threshold, if the real-time stress proportion is larger than the dynamic stress proportion early warning threshold, immediately triggering an early warning signal, and identifying the number of the monitoring point and a strain increment normalization weight factor; if the real-time stress proportion is smaller than or equal to the dynamic stress proportion early warning threshold, no trigger signal is needed.
5. The distributed sensor network deployment and data fusion method for towering steel structures of claim 3, wherein said method allows for location migration and augmentation with construction stages, further comprising: Establishing a mapping relation between displacement variation and stress early warning threshold adjustment coefficient, acquiring a support displacement value of a predefined monitoring period in real time by acquiring an initial value of support displacement of each monitoring point and a stress initial value of a corresponding part, acquiring the displacement variation of the support displacement value relative to the initial value, and comparing the displacement variation with a variation reference interval, wherein the method specifically comprises the following steps: If the displacement variable quantity is in a variable quantity reference interval, judging that the displacement variable quantity is a plastic development area, inputting the current displacement variable quantity into a mapping relation between the displacement variable quantity and a stress early warning threshold value adjusting coefficient, outputting an early warning threshold value adjusting coefficient, and carrying out combined treatment on the early warning threshold value adjusting coefficient and the current stress early warning threshold value to obtain a target stress early warning threshold value so as to reduce the stress early warning threshold value and prevent missing report, wherein the variable quantity reference interval represents an opening interval formed by a variable quantity reference lower limit and a variable quantity reference upper limit; If the displacement variation is smaller than or equal to the variation reference lower limit, judging that the displacement variation is an elastic stable region, and maintaining the current stress early warning threshold; If the displacement variation is larger than or equal to the variation reference upper limit, judging that the displacement variation is a warning dangerous area, and sending early warning information to preset related personnel.
6. The method for deploying and fusing data of a distributed sensor network for a towering steel structure according to claim 1, wherein the specific steps of the data level fusion are as follows: And carrying out space-time alignment and data optimization on the multi-sensor data of the same key monitoring part by adopting a sliding window weighted average algorithm, and outputting a time sequence, wherein the specific steps of carrying out space-time alignment and data optimization are as follows: Presetting a three-level granularity standard of construction stage time coding, wherein the three-level granularity standard comprises macroscopic granularity, mesoscopic granularity and microscopic granularity, each level of granularity corresponds to a unique coding identification bit, a mapping relation between the three-level time coding granularity and an interpolation step reference value is established, when data loss occurs in the macroscopic granularity, the data loss is considered as short interruption in the same stable construction stage, and the interpolation step is set to be a preset first short step length so as to realize high-precision local waveform reconstruction; When the data loss occurs in the mesoscopic granularity, the data loss relates to different state transitions in the same construction stage cycle, and the interpolation step length is set to be a preset second mesoscopic step length so as to identify and avoid crossing key event boundaries in the sub-stages; when the data loss occurs in the microscopic granularity, the data loss means that the construction state and the load have changed significantly, and the interpolation step length is set to be a preset third length step length for maintaining the integrity of the form of the data sequence.
7. The distributed sensor network deployment and data fusion method for a high-rise steel structure of claim 6, wherein said data level fusion further comprises: Establishing a mapping relation between the filtering cut-off frequency and the interpolation error tolerance window adjustment coefficient, if the filtering cut-off frequency is smaller than or equal to the cut-off frequency critical lower limit, inputting the current filtering cut-off frequency into the mapping relation between the filtering cut-off frequency and the interpolation error tolerance window adjustment coefficient, outputting the corresponding tolerance window adjustment coefficient, and carrying out combination treatment on the tolerance window adjustment coefficient and the current interpolation error tolerance window to obtain a target tolerance window so as to ensure that the allowable instantaneous absolute error can be relatively larger and the trend and the low-frequency phase characteristic of the signal after interpolation are kept correct; if the filtering cut-off frequency is in a cut-off frequency critical interval, maintaining a current interpolation error tolerance window, wherein the cut-off frequency critical interval represents an open interval formed by a cut-off frequency critical lower limit and a cut-off frequency critical upper limit; If the filter cut-off frequency is greater than or equal to the cut-off frequency critical upper limit, the current filter cut-off frequency is input into a mapping relation between the filter cut-off frequency and the interpolation error tolerance window adjustment coefficient, the corresponding tolerance window adjustment coefficient is output, the tolerance window adjustment coefficient and the current interpolation error tolerance window are combined to obtain a target tolerance window, so that instantaneous absolute errors are limited, and transient characteristics and peak information of signals are reserved.
8. The method for deploying and fusing data of a distributed sensor network for a towering steel structure according to claim 1, wherein the specific steps of feature level fusion are as follows: Extracting characteristic parameters related to construction quality and safety state from various time series data, including but not limited to extracting characteristic frequency and damping ratio of construction stage from vibration signals through fast Fourier transform, obtaining real-time stress ratio and stress concentration coefficient from strain signals, obtaining verticality deviation, interlayer displacement and accumulated settlement from deformation data, and adopting principal component analysis to perform dimension reduction and redundancy elimination on high-dimensional characteristic vector; obtaining a perpendicularity deviation reference vector through the perpendicularity deviation of each vertical component control point, calculating a weighted average value of the perpendicularity deviation of all monitoring points on a structure integral coordinate system, and recording the weighted average value as an integral inclination synthetic vector, wherein the integral inclination synthetic vector is used for representing the geometric deflection state and direction of the structure integral, and establishing a mapping relation between the integral inclination synthetic vector and a load path abnormality judgment threshold; If the integral inclination composite vector is greater than the standard vector modular length, inputting the current integral inclination composite vector into a mapping relation between the integral inclination composite vector and a judgment threshold adjustment coefficient, outputting the judgment threshold adjustment coefficient, carrying out combination processing on the judgment threshold adjustment coefficient and the current load path abnormality judgment threshold to obtain a target load path abnormality judgment threshold so as to relax the load path abnormality judgment threshold, and keeping high sensitivity to abnormality caused by dangerous deformation modes such as local torsion, thereby reducing invalid interference on the premise of ensuring early warning accuracy; If the overall inclination composite vector is smaller than or equal to the standard vector modulo length, the current overall inclination composite vector is input into the mapping relation between the overall inclination composite vector and the judging threshold adjustment coefficient, the judging threshold adjustment coefficient is output, the judging threshold adjustment coefficient and the current load path abnormality judging threshold are combined to obtain a target load path abnormality judging threshold, so that the load path abnormality judging threshold is tightened, and the sensitivity to the load path local abnormality is improved.
9. The method for deploying and fusing data of a distributed sensor network for a towering steel structure according to claim 1, wherein the specific steps of the decision-level fusion are as follows: the feature vector after dimension reduction and the label of the current construction stage are input into a pre-trained deep learning model together, and the safety evaluation grade of the current construction state is output; Aiming at different construction stages, analyzing time sequence evolution characteristics of the low-dimensional feature vector after dimension reduction, establishing a mapping relation between a state evolution rate and an optimal serialization window length, and if the state evolution rate is larger than an evolution rate reference value, indicating that a structure is in a rapid change and transient response stage, inputting the current state evolution rate into the mapping relation between the state evolution rate and the optimal serialization window length, and outputting the adjusted optimal serialization window length so as to enhance real-time capturing capability and early warning timeliness of transient risks and emergencies; If the state evolution rate is less than or equal to the evolution rate reference value, indicating that the structure is in a relatively steady state, maintaining the current optimal serialization window length.
10. The method for deploying and fusing the data of the distributed sensor network for the high-rise steel structure according to claim 1, wherein the specific steps for judging whether the monitoring characteristic value is in the early warning critical interval output by the digital twin model are as follows: If the monitoring characteristic value is in an early warning critical interval output by the digital twin model, determining that primary early warning is performed, prompting attention through a data platform, wherein the early warning critical interval represents a closed interval formed by an early warning critical lower limit and an early warning critical upper limit; if the monitoring characteristic value is smaller than the early warning critical lower limit output by the digital twin model, storing the self-adaptive fusion characteristic and the evaluation result in a time sequence database to form a traceable construction digital file; If the monitoring characteristic value is larger than the critical upper limit of the early warning output by the digital twin model, judging that the digital twin model is a secondary early warning, automatically pushing warning information to related construction managers, and connecting the digital twin model to highlight the risk part.

Description

Distributed sensor network deployment and data fusion method for high-rise steel structure Technical Field The invention relates to the technical field of network deployment and data fusion, in particular to a distributed sensor network deployment and data fusion method for a high-rise steel structure. Background The high-rise steel structure has large ratio of height to section size, strong overall flexibility and concentrated weak links (such as a bottom stress area, a top displacement sensitive area and a variable section stress concentration area), the existing deployment is insufficient, and the dynamic characteristics and the modal analysis results of the structure are combined, so that the sensor is not fully distributed at a key part to form a monitoring blind area, or is redundant distributed on a symmetrical or similar rod piece to cause resource waste, and the root cause of the problem is that a systematic layout strategy which takes structural characteristics, a monitoring target and cost effectiveness into consideration is lacking, and the contribution of a measuring point to vibration mode identification cannot be quantified through a scientific method such as a modal confidence criterion. Meanwhile, strong wind, high temperature, electromagnetic interference and shielding of reinforced concrete members in a high-rise steel structure construction environment are combined with inherent limitations that sensor nodes are powered by batteries, so that signal attenuation, delay and even packet loss easily occur in a data transmission process, the root cause is that the influence of environment interference on a communication link is not fully considered in deployment, a data transmission path and a network topology are not optimized, overload energy consumption of partial nodes is too fast, and network stability is further damaged. In the data fusion layer, because the sensor type selection lacks pertinence (the sensor is matched and adapted according to different monitoring parameters such as vibration and displacement), and the structure dynamic response presents obvious nonlinear characteristics in the construction process, the existing fusion algorithm cannot effectively integrate multi-source heterogeneous data, and is difficult to distinguish environmental noise from structure real response, the root cause is that the fusion model does not fully incorporate the structure nonlinear effect and multi-load coupling influence, the accurate mapping of the monitoring data and the structure health state cannot be realized, the data redundancy and the information loss are finally caused to coexist, and the problem that the distribution type sensor network deployment and the data fusion accuracy are low due to the insufficient optimization of construction environment interference and network topology exists. Disclosure of Invention In order to solve the technical problems of low accuracy of deployment and data fusion of a distributed sensor network due to construction environment interference and insufficient network topology optimization in the prior art, the embodiment of the invention provides a method for deployment and data fusion of a distributed sensor network for a high-rise steel structure. The technical scheme is as follows: On the one hand, the distributed sensor network deployment and data fusion method for the high-rise steel structure comprises the steps of S100, identifying key monitoring parts of each construction stage based on a finite element model of the high-rise steel structure to be monitored and the whole construction process simulation, carrying out distributed deployment on the key monitoring parts through a heterogeneous sensor network, adapting to construction dynamic adjustment to allow position migration and supplementation along with the construction stages, S101, obtaining original parameters of the key monitoring parts, preprocessing original data at an edge computing gateway, binding the preprocessed data with the current construction stage, carrying out three-level self-adaptive fusion, recording values output after the self-adaptive fusion features are fused as monitoring feature values, and constructing a digital twin model of the high-rise steel structure, carrying out state update and visual display on the self-adaptive fusion features and evaluation results, and judging whether the monitoring feature values are in a critical pre-warning section output by the digital twin model or not, and providing corresponding pre-warning levels according to emergency pre-warning levels if the monitoring feature values are in the critical section. One or more technical solutions provided in the embodiments of the present application at least have the following technical effects or advantages: 1. through realizing the dynamic adaptation and the accurate coverage of the key monitoring of construction full cycle, effectively solved traditional monitoring and deployed fixed, the pa