Deformation Behavior Identification And Multi-monitoring Model Of Concrete Gravity Dam Based On Observation Data Restoration

As one of the important infrastructures for flood control,power generation and shipping,the healthy operation of concrete gravity dam is related to the safety of life,property and social stability of the people.However,in-service concrete gravity dam safety monitoring system is susceptible to many source correlation factors,such as monitoring instrument anomaly,external load disturbance and monitoring process omission,which often leads to missing prototype monitoring data and data anomaly.Establishing effective safety monitoring model for concrete gravity dam is an important means to perceive the safe service behavior of dam.Reliable monitoring data is the precondition for constructing dam safety monitoring model.In view of the incompleteness and data outliers existing in dam observation data,this paper comprehensively applies statistics,in-depth learning and signal processing methods to carry out reliability identification and pre-processing of prototype deformation monitoring data.Further,deformation monitoring data with inherent prototype observation information are obtained,and multi-point monitoring method for deformation behavior parameter identification and deformation behavior of concrete gravity dam is systematically put forward.The main research contents are as follows:(1)Aiming at the problems of abnormal value and missing value in prototype observation sequence of concrete gravity dam,through analyzing the effect of observation outliers and incompleteness of time series data,isolate forest method is used to identify the reliability of time series data of deformation observation,and VMD-Bi LSTM method is used to effectively fill in the incomplete time series data of deformation of concrete gravity dam.Furthermore,the reliability identification and pre-treatment method for deformation observation data of concrete dam are put forward.(2)Considering the tedious problems and inefficiencies in inversion of main mechanical parameters of concrete gravity dam,this paper constructs an agent model for static load response of concrete gravity dam by using BP neural network algorithm based on fruit fly optimization,and combines Latin Hypercubic sampling to extract mechanical parameters combination and COMSOL finite element platform calculation results,thus proposes a method for identifying deformation behavior parameters of concrete gravity dam based on the agent model.(3)Aiming at the problem that traditional single-point deformation monitoring model of dam is difficult to reasonably characterize the comprehensive response characteristics of different measuring points in the same dam section in the space deformation field,based on its single-point deformation monitoring statistical model,this paper constructs a multi-point deformation monitoring model of concrete gravity dam based on mathematical statistics,and uses Bayesian optimized Light GBM algorithm and safety monitoring hybrid model construction method.Furthermore,a mixed model for monitoring multi-point deformation of concrete gravity dam based on data and physical drive is proposed.