The Data Mining Technology Of Small Bridge Monitoring Based On Time Series Theory Research

Currently,at home and abroad in the field of bridge health monitoring technology research is focused on large bridges. However Due to the small bridge with a huge base, frequent accidents in small bridge. So for System health monitoring should increase attention to the small bridge. This paper relies on "long-term safety studies based on small bridges Things"(973 research projects). Through the bridge monitoring data analysis, bridge monitoring data processing provides a new thinking. Future research and practice have some reference value. The main results of the work have made:â‘ It introduces the basic theory of time series analysis methods, analyzes the influencing factors of each data time series, provide a theoretical basis for the mathematical model.â‘¡Bridge long-term monitoring system based on data collected features, the feasibility of a time series as the theoretical basis for the establishment of ARMA model to analyze data to predict. ARMA model building process and is described in detail, including the initial data collection, data smoothing, normalization processing, the estimated model parameters, the model order is determined so on. Estimation and model order to determine the model parameters is the key to establishing ARMA model, the usual method of parameter estimation and model order selection were compared, ARMA model to estimate the parameters of the final selection of the least squares estimation method, model order to determine the criteria for selecting the AIC.â‘¢In one of the bridge- the bridge to rely on PianYan Zi. The time series modeling to predict the theory is applied to engineering practice. Using MATLAB software, the crack-point bridge deflection measurement data point, tilt measuring point, strain measurement points established ARMA model, proper order and adjust parameters, every outwardly extending short-term forecast of five steps, the predicted value and the actual detected values are compared, to verify the feasibility and effectiveness of the bridge monitoring data analysis methods based on ARMA model.