CN-121998420-A - Construction risk analysis and emergency method

Abstract

The invention discloses a construction risk analysis and emergency method, and relates to the field of hydraulic engineering construction. The method comprises the steps of constructing a digital twin body, initializing stratum loss transfer function parameters and a time sequence rule base based on a geological survey report and an indoor test, collecting real-time monitoring data, inverting stratum loss distribution, correcting model parameters through a Kalman filtering data assimilation technology, calculating a dynamic threshold value based on the updated model, matching time sequence association rules, rolling to predict engineering states, calling a physical evolution model to reversely deduce optimal treatment parameters and issuing instructions, collecting whole-flow data, and realizing accurate risk prevention and control through constructing the digital twin body, the data assimilation correction parameters, dynamic threshold value calculation and state prediction, calling the physical evolution model to generate treatment instructions and the self-learning optimization parameters and rule base.

Inventors

• ZENG FANYU
• CAO YILE
• HAN SIJIA
• LI XIAOHUI
• SI JIAXUAN

Assignees

• 中国水利水电第十一工程局有限公司

Dates

Publication Date

20260508

Application Date

20260123

Claims (5)

1. 1. The construction risk analysis and emergency method is characterized by comprising the following steps of: s1, constructing a digital twin body integrating a three-dimensional geological model, a tunnel model, an environment model and a sensor network, and initializing a stratum loss transfer function based on a geological survey report and an indoor test Parameter and time sequence rule base initial version; S2, collecting Inversion of actual formation loss distribution Modifying model parameters by a set Kalman filtering data assimilation technology; S3, based on updated Calculating a dynamic threshold value, matching time sequence association rules, and rolling and predicting the engineering state of one time step in the future; S4, invoking a physical evolution model matched with the corresponding risk type, deducing optimal treatment parameters reversely, and synchronously transmitting instructions to equipment and an external unit; S5, collecting whole-flow data, and optimizing stratum loss transfer function parameters and a time sequence rule base through a self-learning unit to complete learning circulation.
2. 2. The construction risk analysis and emergency method according to claim 1, wherein the digital twin body sensor network in step 1 includes sensor arrays of different depths of the stratum.
3. 3. The construction risk analysis and emergency method according to claim 1, wherein the implementation process of the integrated kalman filter data assimilation technology in step 2 includes inversion of stratum loss distribution based on real-time data of a PLC of a shield tunneling machine, and correction of model parameters through 50-100 iterations.
4. 4. The construction risk analysis and emergency method according to claim 3, wherein in step S3, calculation of the dynamic early warning threshold predicts a future preset time period according to the stratum loss state at the current time t The theoretical risk index value of the rear monitoring point is taken as the future time t+ Dynamic pre-warning threshold of (2).
5. 5. The construction risk analysis and emergency method according to claim 4, wherein in step S4, the self-learning unit adjusts the safety factor and the time-series rule time window of the stratum loss transfer function by using a reinforcement learning framework.

Description

Construction risk analysis and emergency method Technical Field The invention relates to the field of hydraulic engineering construction, in particular to a construction risk analysis and emergency method. Background In shield tunnel engineering such as urban rail transit and utility tunnel, the working conditions of sensitive building structures such as underpass high-speed railways, dense building groups and important municipal pipe networks are becoming more common. The core technical challenge faced by such engineering is that the traditional experience-based static risk management and control mode is difficult to effectively cope with the dynamic risk evolution process of 'strong time variability of geological conditions, complex construction disturbance propagation path, lagged and linked environmental response'. The prior art scheme mainly has the following systematic technical bottlenecks: The statics and experience of the risk criteria-risk early warning generally depends on a preset, single, fixed threshold (e.g. sedimentation rate >2 mm/d). However, stratum loss caused by shield construction disturbance and transmission of pore water pressure change to the earth surface are complex space-time processes. The actual risk level for subsurface structures (e.g., pile foundations) and surface facilities (e.g., high-speed rail tracks) is quite different at the same surface subsidence rate, with different combinations of formations (e.g., hard-under-hard, pressurized water-rich) or different depths of burial. The prior art completely ignores the time-space hysteresis effect and attenuation characteristic of stratum loss transmission, so that the early warning and real risk state are seriously disjointed, or the construction is frequently interfered by false alarm or accidents are caused by missing report. The isolation and one-sided performance of risk perception are remarkable, most of the existing monitoring systems are simple stacking of sensors, and depth correlation analysis based on a physical model is absent between data. For example, if only the earth surface subsidence is monitored and the deep soil displacement and the change of pore water pressure are ignored, whether the subsidence is caused by improper shield posture or the erosion caused by instability of a karst cave in front cannot be accurately identified. The perception mode of only seeing trees and not seeing forests cannot capture the chain evolution precursor of risks. The hysteresis and discretization problems of the treatment decisions are highlighted in that emergency plans are mostly text clauses only and are seriously disjointed with real-time and multi-source risk state information. When dangerous cases occur, comprehensive judgment, layer-by-layer request and telephone scheduling still need to be performed manually, the time span from sensing dangerous cases to implementing effective treatment actions is too long, and the optimal intervention time is often missed. The lack of effective synergy between treatment actions (such as grouting, shield parameter adjustment, and traffic control) based on the same risk model may result in mutual elbow detent and influence on treatment effect. The system lacks self-adaptation and evolution capability, the existing system parameters (such as early warning threshold values) are solidified after being set, self-tuning is difficult to carry out according to the geological conditions of specific engineering and the differences of construction processes, experience cannot be drawn from the dangerous case, and therefore the system is poor in applicability and low in intelligent degree at a new working point. Therefore, the integrated intelligent prevention and control technology capable of deeply fusing geomechanical mechanism, construction disturbance theory and real-time monitoring data to achieve risk advanced perception, accurate quantification, cooperative intervention and autonomous learning is urgently needed in the field. Disclosure of Invention The invention provides a construction risk analysis and emergency method for solving the problems in the prior art. In order to achieve the above purpose, the present invention adopts the following technical scheme: a construction risk analysis and emergency method comprises the following steps: s1, constructing a digital twin body integrating a three-dimensional geological model, a tunnel model, an environment model and a sensor network, and initializing a stratum loss transfer function based on a geological survey report and an indoor test Parameter and time sequence rule base initial version; S2, collecting Inversion of actual formation loss distributionModifying model parameters by a set Kalman filtering data assimilation technology; S3, based on updated Calculating a dynamic threshold value, matching time sequence association rules, and rolling and predicting the engineering state of one time step in the future; S4, calling a physical evolu