Based On The Data Driven Of Bridge Structure State Analysis

With the rapid popularization of bridge health monitoring systems,the research on the processing methods based on response signals of bridge structures has increasingly become the focus of attention.Based on the characteristics of abundant data and lack of information commonly found in the field of bridge health monitoring,this paper applies the thoughts and techniques of the data-driven field to symbolize modeling of b ridge response signals based on symbolic modeling.The segmentation and reconstruction of the causality state,the structure vector extraction,and then using the obtained structure vector to train the hidden Markov model to diagnose and evaluate the state of the bridge structure,combined with the real bridge scale model data for experimental verification,formed a set of data-driven bridge structure system state diagnosis evaluation algorithm system.Its main contents include:(1)Analysis of the Complexity of Bridge Structure System and Its Response Signals.Through the analysis of the material nonlinearity and geometric nonlinearity of the bridge structure,the mechanism of dynamic behavior of the complex system of the bridge structure is investigated.Taking the two-degree-of-freedom Hamiltonian system with square nonlinearity as an example,the structural system is further analyzed in non-linear chaotic dynamic behavior evolution model.At the same time,based on this,the complex characteristics of the no n-stationary and non-linear response signals of the bridge structural system are analyzed.(2)Build bridge structure state feature indicators with symbolic model.For the structural response signals collected directly by sensor technology,it has strong feat ures such as randomness,instability,and nonlinearity.It introduces the symbolic modeling method for solving complex problems in the field of artificial intelligence and.Deal with the removal of non-essential features by coarse-grained signal processing and keep key information as much as possible.Based on this,reconstruct the sequence machine and mine the implicit model of the structural system to achieve the extraction of the damage information of the reaction structure.(3)Constructing State Diagnosis Method of Bridge Structure System Based on Hidden Markov Model.Combining hidden Markov models has the advantages of analyzing and processing variable-length feature sequences.The structural vectors of response structure system features extracted by symbolic models are used as observation sequences,and the damage states of each structure are used as implicit sequences.They are trained using the Bambi Welch algorithm.Hidden Markov model,at the same time,through the comparative analysis of the evolutio nary mode of the log-likelihood rate calculated by the model under different states of the structural system,to realize the structural system state diagnosis.(4)Relying on the midas simulation model and the experimental analysis of the scale model of the real bridge of the Black Gully Bridge.Symbolic modeling was used to extract the structure vector of the vibration acceleration response signal at different measuring points and four different damage conditions.Then the initial hidden Markov model was trained on a large number of structural vectors to obtain the bridge structure.A set of hidden Markov models for the diagnosis of the system state is used to verify the feasibility of the algorithmic process of symbolic modelling to extract structure vectors and then perform hidden Markov model diagnosis.