Massive Data Analysis Theory And Application In Bridge Monitoring System

The conventional bridge load experiment examination, in regard to the safety evaluation of the bridges newly put into use or damaged by accident, has the positive significance. But it has shortcomings of discontinuities, traffic interruptions, and double-spending massive human, material, financial resources. The bridge remote monitoring which has such characteristics as safe, real-time, continuous, timely warning and forecast, is the international research focus. To the bridges especially the large span bridges, setting up the health monitoring system, to take the place of the traditional bridge periodical or irregular safety examination, is of paramount importance to the bridge safety assessment, the design verification.This paper is aimed at dealing with the massive data problem in bridge monitoring system, proposed the K Line Index Method, the K Line Random Sequence Prediction Method and the Fast Computation Method Based on Limited Sensors. Meanwhile, this paper is a part of the researches in the Chinese National Program for Science and Technology Development "the Development and Demonstration of Large Span Bridge's Health Remote Monitoring Complete Technology"(2002BA105C). This paper was completed on the base of reorganizing the related research results and the relevant running data.The main contents include:1. The plan design of the bridge remote intelligent cluster monitoring system. As one of the main researchers of the project, the author has taken part in the plan design of the three bridges'monitoring systems, determined each monitoring system's sensor types, locations, counts, accuracy and measuring range. In this paper, the subsystems in the three bridges'monitoring system were introduced, as well as the newly invented sensors in this project. The hardware and software schemes of cluster monitoring center, the structures, functions and requirements of the on-site monitoring system and the remote transmission system were presented in detail.2. In order to solve the real-time calculation problem in the bridge health monitoring, the Fast Computation Method Based on Limited Sensors was proposed. The theory and the related software were elaborated and developed, which were successfully applied to the three bridges'health monitoring, realizing a fast calculation and real-time evaluation in bridge health monitoring. 3. For the first time, the K-line graph from the securities analysis was used in the bridge health monitoring to deal with the massive data problem in bridge health monitoring system. The method and process for the K-line application in bridge health monitoring was in detail discussed.4. On the base of the K-line graph, according to the characteristics of bridge health monitoring, the author created the Directional Movement Index, the Bearing Degree Index, the Irregularity Index, the Diversity Index and etc, which were used in the safety assessment of the three bridges' monitoring. These indexes well described the changes and trends of the monitored parameters (such as deflection, strain, temperature and etc).5. On the basis of the alteration of the existing gray model, grey predictions were successfully carried out for forecasts in regard to the time series generated from the K-line technology, especially the time interval sequences of maximum value and minimum value. Satisfactory prediction accuracy was achieved. A more comprehensive method of non-equal interval data modeling method was proposed to deal with the non-uniform spacing K-line sequence in bridge health monitoring. Meanwhile, the detail programming flow charts of the main forecast processes were presented.The Fast Computation Method Based On Limited Sensors provides the way to get all the bridge displacements, strains and stresses through limited displacements while the surface loads are unknown. The fast computation requirement from the bridge health monitoring was achieved. The K-Line Index Method and the K-Line Random Sequence Prediction Method have the advantages of a small amount of data storage. They are applicable to different kinds of bridges and monitored parameters, and are suitable to develop general bridge health monitoring software. This brought new ideas and contents, and has wide range of applications.