Search

CN-122022460-A - Peripheral monitoring method for shield construction

CN122022460ACN 122022460 ACN122022460 ACN 122022460ACN-122022460-A

Abstract

The invention relates to the field of shield construction, in particular to a peripheral monitoring method for shield construction, which realizes the focus of a core risk area by grading and distributing geological-structure coupling risk matrixes, combines geological-working condition dual-drive dynamic adjustment of monitoring frequency, improves risk capture timeliness, adopts an attribute-LSTM model to conduct AI time sequence prediction and multi-source data fusion three-stage early warning, reduces false alarm rate, further establishes a structure multi-dimensional deformation verification system to ensure data reliability, and realizes intelligent closed-loop management of emergency response through a digital twin platform. The method can improve the monitoring efficiency by 40%, reduce the cost by 25%, improve the risk capture aging by 80%, control the data error within +/-0.1 mm in advance by 24-48 hours, shorten the emergency treatment time to 1 hour, reduce the accident rate by 70%, and promote the monitoring to intelligently drive the data.

Inventors

  • ZHAO HAOXIANG
  • LI XIAOHUI
  • HE PEIJI
  • ZENG FANYU
  • PAN TIANJIA

Assignees

  • 中国水利水电第十一工程局有限公司

Dates

Publication Date
20260512
Application Date
20260123

Claims (7)

  1. 1. The peripheral monitoring method for shield construction is characterized by comprising the following steps: (1) Establishing a coupling matrix of a geological risk index and a structural sensitivity index based on an engineering geological longitudinal section map and a structure questionnaire, and implementing differential monitoring and point distribution according to the coupling risk level; (2) The geological-working condition dual-drive dynamic monitoring frequency is adaptively adjusted, wherein the monitoring frequency is dynamically bound with real-time geological parameters and construction working conditions, key geological parameter threshold values are set, and normal period, early warning period and emergency period are divided according to the threshold values and the distance between the shield and the structure, so that different monitoring frequencies are corresponding; (3) Performing three-stage early warning by combining AI time sequence prediction and multi-source data, namely performing deformation trend pre-judgment by adopting an LSTM model combined with an Attention mechanism, triggering monitoring early warning according to grades, generating patrol early warning by linking AI image recognition, and finally comprehensively judging early warning grades through a digital twin platform; (4) Building a 'main monitoring and auxiliary checking' redundant system aiming at different types of sensitive structures, and mutually checking and ensuring data reliability through multiple physical quantities; The bridge adopts pile foundation settlement and beam body inclination as main monitoring, bridge pier stress and bridge deck crack as verification, more than 3 layers of buildings adopts a cast-in-place method to directly measure the inclination as main monitoring, differential settlement calculation as verification, underground pipelines adopt a direct method displacement rod to measure as main monitoring, and soil mass around the pipe is subjected to layered settlement as verification; (5) The digital twin driving emergency response intelligent closed loop comprises the steps of mapping engineering scenes through a digital twin model, synchronizing multi-source data in real time, sending early warning trigger to responsibilities of parties involved in construction, automatically recommending and adapting an emergency plan by a system, optimizing model parameters and a plan strategy based on historical data, realizing self-evolution of the twin model, automatically generating an alarm elimination application and completing on-line closed loop approval after monitoring data are continuously stabilized within 50% of a control value and simulation risks are relieved.
  2. 2. The method for monitoring the periphery of shield construction according to claim 1, wherein in the geological risk index calculation model, the values of all parameters are verified by at least 3 sets of parallel investigation data.
  3. 3. The method for monitoring the periphery of shield construction according to claim 2, wherein the weight coefficient of the geological risk index calculation model is determined by combining an AHP (advanced high performance) analytic hierarchy process with inversion of engineering investigation borehole data, and the weight can be calibrated and adjusted according to different engineering geological conditions.
  4. 4. The method for monitoring the periphery of shield construction according to claim 3, wherein in the geological-working condition dual-drive dynamic monitoring, the construction working condition parameters comprise shield tunneling speed, thrust force, grouting quantity and cutter head rotating speed, and when tunneling speed fluctuates by +/-20% and thrust force suddenly changes by +/-15%, the monitoring frequency adjustment is triggered synchronously.
  5. 5. The shield construction periphery monitoring method according to claim 4 is characterized in that the differential distribution rules are as follows, pile foundation settlement, pile body stress, pile periphery soil body inclination and bridge deck crack four-dimensional monitoring points are distributed in a high risk coupling area, a middle risk coupling area is monitored by adopting a direct method and an indirect method, and at least 3 settlement points are distributed in a low risk coupling area according to a structure type.
  6. 6. The method for monitoring the periphery of shield construction according to claim 5, wherein training data of the Attention and LSTM models comprises historical subsidence/inclination sequences and geological parameter sequences, and the models are retrained every week by adopting newly-added measured data, so that feature capturing capacity and prediction accuracy are optimized.
  7. 7. The method for monitoring the periphery of shield construction according to claim 6, wherein the on-line closed loop approval process comprises four links of construction unit reporting, supervision unit auditing, design unit rechecking and construction unit approval, the whole process takes no more than 2 hours, and approval records are automatically filed to a digital twin platform.

Description

Peripheral monitoring method for shield construction Technical Field The invention relates to the field of shield construction, in particular to a peripheral monitoring method for shield construction. Background In shield tunnel underpass sensitive structure engineering, construction disturbance and geological mutation coupling effect are easy to cause risks such as structure deformation, leakage and even collapse, and the traditional monitoring method has various limitations, and particularly has outstanding performance in Guangzhou rail transit No. 8 North line extension Duan Yayao station-Xiuquan park station intervals: 1. The distribution lack of pertinence is that differences of geological conditions and vulnerability of structures are not fully considered, the coverage range of the measuring points is not matched with the risk level, so that a core risk area cannot be monitored in place, and a resource waste phenomenon occurs in a non-critical area. 2. The response is delayed from geological mutation, namely the monitoring frequency is only related to the construction progress, the response to fluctuation of geological parameters such as the groundwater level, the karst fracture water pressure, the sand layer pore water pressure and the like is not achieved, and the sudden geological risk cannot be captured timely. 3. The early warning has isolation and high false alarm rate, only depends on a single accumulated value and rate threshold value to alarm, is not combined with deformation trend to perform pre-judgment, is disjointed with an emergency disposal flow, cannot link a hexagonal participating unit to respond rapidly, and is easy to cause potential safety hazards or construction stagnation due to false alarm and missing alarm. 4. The single monitoring technology has larger error, for example, for the inclination condition of more than 3 layers of buildings, the inclination condition is calculated only by relying on differential settlement, the error caused by uneven basic rigidity due to geological characteristics is ignored, and the reliability of data is difficult to be ensured. In recent years, although research attempts are made to introduce an automatic monitoring or BIM technology, the research is focused on data acquisition visualization, the real-time coupling analysis of geological risk dynamic parameters and construction disturbance working conditions is not realized, an early warning model still stays at a static threshold stage, the trend prejudging capability based on machine learning is lacking, and an intelligent closed-loop management system penetrating through monitoring, studying, disposing and verifying is not formed. Disclosure of Invention The invention provides a peripheral monitoring method for shield construction, which aims to solve the problems in the prior art. In order to achieve the above object, the present invention adopts the following technical scheme: a peripheral monitoring method for shield construction comprises the following steps: (1) Establishing a coupling matrix of a geological risk index and a structural sensitivity index based on an engineering geological longitudinal section map and a structure questionnaire, and implementing differential monitoring and point distribution according to the coupling risk level; (2) The geological-working condition dual-drive dynamic monitoring frequency is adaptively adjusted, wherein the monitoring frequency is dynamically bound with real-time geological parameters and construction working conditions, key geological parameter threshold values are set, and normal period, early warning period and emergency period are divided according to the threshold values and the distance between the shield and the structure, so that different monitoring frequencies are corresponding; (3) The AI time sequence prediction and multi-source data fusion three-level early warning is implemented by constructing a four-dimensional fusion early warning system of 'real-time monitoring data + geological parameters + AI trend prediction + patrol information', adopting a hybrid neural network model of fusion time convolutional neural network (TCN), multi-head sparse self-attention (SMHA), long short-time memory network (LSTM) and Time Perception Attention (TPA) to conduct deformation trend pre-judgment, triggering monitoring early warning according to grades, linking AI image recognition to generate patrol early warning, and finally comprehensively judging early warning grades through a digital twin platform; (4) Building a 'main monitoring and auxiliary checking' redundant system aiming at different types of sensitive structures, and mutually checking and ensuring data reliability through multiple physical quantities; The bridge adopts pile foundation settlement and beam body inclination as main monitoring, bridge pier stress and bridge deck crack as verification, more than 3 layers of buildings adopts a cast-in-place method to directly measure th