| Summary: | 碩士 === 國立雲林科技大學 === 營建工程系碩士班 === 95 === Due to environments, more and more important structures such as highway bridges appear to be deteriorated in Taiwan; therefore, maintenance of all components of bridges becomes one of the most important issues. Currently, a lot of types of bridge expansion joints are used in Taiwan, and are renewed frequently. Therefore, investigation on the benefits of all expansion joints would help transportation agencies to have a detailed understanding about them. On the side, prediction of the service lives of all expansion joints in different conditions would be useful for the maintenance stratigies and the budgets planning.
This research make analyzed the four major types of expansion joints (i.e., “finger plate joint”, “modular joint”, “angle expansion joint”, and “finger expansion joint”) in highway bridges. From literature reviews and data collection, twenty-one possible factors that may influence the service lives of expansion joints were collected. Analysis of variance (ANOVA) was conducted with SPSS 12.0 to find out the significant factors of service lives of different types of expansion joints. Artificial neural network (ANN) was used to establish the service lives prediction model for each type of expansion joint to predict service life of different types of expansion joints in different conditions. The predictions of the models can be a basis for maintenance strategies in the future.
As the results, this research found that there are eight factors significantly influencing the service life of the finger plate joint; and fourteen, eighteen, and twelve factors for the modular joint, and the angle expansion joint,and the finger expansion joint, respectively. Among the factors analyzed, only the factor of “crossover object” is not a signiciant factor for all four types of expansion joints, and the factors of “designed horizontal acceleration”, “total spans”, and “annual average daily traffic (AADT)” are the common signicant factors for all of them. Moreover, this research established the service life prediction models for three types of expansion joints (“finger plate joint”, “angle expansion joint”, and “finger expansion joint”). The root mean squared error (RMSE) of training for these models are 7.108%, 13.6135%, and 4.441% respectively, and RMSE for testing are 6.240%, 8.948%, and 3.287% respectively. The results show that these models obtain good prediction results. These models don’t only predict service life of various types of expansion joints on different conditions, but also be a basis to transportation angencies for maintenance and management.
|
|---|
