Scalable Web-Based Data Management System in Long Term Structure Health Monitoring Ping Lu Iowa State University 2901 S.Loop Dr.,Suite 3100 Ames, IA 50010 515-296-6686 luping@iastate.edu [Abstract] Recently, there has been increasing interest in developing long term Structure Health Monitoring (SHM) system for civil infrastructure. The extremely large amount of data collected during the long term structural monitoring is a great challenge for data processing and interpretation. In this paper, a web based scalable data management system is proposed, which allows for automatic database updating, advanced statistical and structural data interpretation, and web-based data searching. Open source tools, Apache web server, PHP, and MySQL are employed to implement the system. Matlab is integrated as the graphic and data processing engine. The system is developed dedicatedly for the E-12 th Street Bridge project on I-235 in Des Moines, IA. However, it could be extended to other projects very conveniently. [Keywords] long term Structure Health Monitoring (SHM) system; scalable web-based data management system; open source tools; PHP; MySQL; Apache; MatLab 1
1. Introduction Bridge deficient is the most critical challenge to sustain the nation s highway service level. As of 2003, 27.1% of the nation's bridges (160,570) were structurally deficient or functionally obsolete. The annual investment required to prevent the bridge investment backlog from increasing is estimated at $7.3 billion [1]. Limited budget calls for performance based highway-bridge maintenance scheduling to optimize the overall highway service. Obviously, precise and in time structural condition information is essential to achieve the scheduling optimization. Currently, bridge structural condition information is obtained mainly through regular visual inspection. A study funded by Federal Highway Administration (FHWA) revealed that this kind of inspection is not reliable and simply doing more intense visual inspection provided little additional useful information on structure condition. Therefore, increasing attentions are attracted to long term Structure Health Monitoring (SHM) systems which have shown to be promising in evaluating structural condition for highway bridges. More and more SHM data acquisition systems are installed in highway bridges all over the world and lots of researches are focused on the system selection, installation and data interpretation [2]. However, relatively fewer researches could be found in the literature addressing the large volume data storage, retrieve and sharing problem. Experiences learned from the East 12 th Street Bridge project show that efficient and effective huge volume data management system is very important for successful long term SHM data interpretation. Since the long term SHM system was installed on the East 12 th street bridge over I-235 in early 2004, approximately 120GB of continues data has been collected and the data will be accumulated to more than 450GB by the end of 2005. Conventional approaches to data interpretation simply do not work due to the volume and constant flow of data. To overcome the difficulty, a web based scalable data storage/retrieving, processing and sharing system is proposed in this paper, which allows for automatic database updating, web-based viewing of desired history data, and advanced statistical and structural interpretation of the data. This system is developed dedicatedly for E-12 th street bridge project. However, it could be easily extended to other long term SHM projects. The rest of the paper is organized as follows. Section two is a brief summary for E-12 bridge project and the installed SHM system. The details of the proposed web-based data management system are described in section three. Finally, the paper is concluded by conclusions and future work. 2. SHM System of E-12 High Performance Steel (HPS) Bridge Project [3] 2.1 introduction of E-12 HPS bridge project 2
In early 2004, the Iowa DOT completed construction of Iowa s first HPS (High Performance Steel) bridge, The East 12 th Street bridge over I-235 in Des Moines, IA, through the Federal Highway Administration s (FHWA) Innovative Bridge Research and Construction (IBRC) program. It is a 298 ft - 6 in. continuous two-span HPS girder bridge with a cast-in-place concrete deck and integral abutments. Six girders, transversely spaced at approximately 8 ft 8 in., utilize HPS 50W and 70W in the positive and negative moment regions, respectively. The two-lane bridge, which is approximately 50 ft 3 in. wide, features pedestrian walkways on both outside edges of the bridge. The picture of the bridge is shown in Fig.1. The Bridge Engineering Center (BEC) at Iowa State University has developed a long term SHM system to monitor and record the performance of the HPS bridge for a two-year period as required by IBPC program. With this system, the bridge performance can be evaluated at any point in time. 2.2 SHM System Installed on E-12 th Street Bridge Project The SHM system installed on E-12 bridge can be divided into three subsystems: the Data Acquisition Sub-System(DASS), the Gateway Sub-System(GSS), and the Data Storage/Processing Sub-System(DSPSS). The system is shown schematically in Fig 2. The Data Acquisition Sub-system, located at the bridge pier, collects and transfers data via wireless communication to the Gateway Sub-system, which is located in a secure facility adjacent to the bridge. The Gateway Sub-system uploads the strain data to the internet where it can be viewed from anywhere in the world in real time (http://www.ctre.iastate.edu/bec/structural_health/hps/index.htm) while it compresses and temporarily stores the strain data simultaneously. The data packets are automatically retrieved by the DSPSS at the BEC. However, with existing DSPSS component, the collected data and preliminary analysis results are stored as text file in BEC server. The data storage scheme is not scalable and introduces tremendous time overhead for data retrieving. Without data searching function nor web access interface provided, it is hard for end users, such as the bridge owners, to access the data. Therefore a scalable web-based data management system is proposed to allow for convenient and efficient remote data accessing. Due to the advanced data management functions and the scalability provided by the background database system, the overall data management component could accommodate extremely huge volume of data while furnish high accessibility. 3
Fig.1 East 12 th St. HPS Bridge over I-235 in Des Moines, IA Data Storage/Processing Sub-system Storage Server Client Processing: Elementary Data Checks Temperature Elimination Global Behavior Rainflow Counting Gateway Sub-system Linksys WAP Data Acquisition Sub-system Linksys WAP and Router Internet Cisco Modem Si425 Interrogator FBG Sensors Data Collection Server & Web Server Canon Network Camera Fig.2 E-12 th Street Bridge SHM System 4
3. Proposed Scalable Web-based Data Management System The ultimate purpose of long term SHM system is to identify structural damage and deterioration as early and precisely as possible. Fig. 3 illustrates the framework of a typical knowledge based decision support SHM system. Not until raw sensing data are abstracted into information and knowledge, they could not support decision making directly. However, successful information abstracting (data interpretation) strongly depends on the efficient data accessibility. Therefore, efficient data management system is a key component for long term SHM system. Data Information Knowledge Decision Web-based Database System Raw Data Storage Temperary Data Storage Advanced Query processing Data Mining, Modeling,Analysis Statitical Approach pattern Recognition Probability Analysis Finiate Element Approach FE based simulation Structure Health Assesment Fatigue Lifetime Evaluation Structure reliability analysis FE model validation & updating System Identification Emergency Reponse Permit-vehicle policy Maintenace, repair, rehabilitation and replacment of the structure Fig.3 Framework of the SHM system With current data storage/processing system in E-12 bridge project, the file server retrieves data from gateway system periodically, and stores them as text files. Each file has the size around 18.1 Mega bites and contains records for 45 minutes. Both the big file size and large number of files bring great time overhead for data searching and processing. In the proposed web-based data management system, database system is introduced to provide a more organized and efficient way for data storage and retrieving. Web interface is developed to ensure convenient remote data accessing and sharing. After exploring and comparing many possible solutions, an Apache-PHP-MySQL system was implemented. The open source tools have been proved to be very effective platform for web applications development. Due to the extremely large data volume, the web database MySQL Server will be running in a distributed fashion. The entire system is a typical client-server system. As illustrated in Fig.4, the client side is composed by end users web browsers, such as Windows Internet Explorer, Netscape and so on. The server side is more complicated and includes all necessary servers. As in the implemented system, the server side is composed by Apache web server, PHP server, file server, application server/servers and distributed MySQL database servers. The configuration and operation of the servers are transparent to end 5
users. In other words, the server side works as a black box to users. Users could access history sensing data and data processing results by inputting their request through web browsers. Three modules are implemented so far. They are 1) database remote updating module; 2) web-based data searching module; 3) data processing module. More modules will be developed and integrated into the system later to fulfill more advanced data analysis and interpretation tasks. Fig.4 Prototype of the web-based data management system 3.1 Database Update Module This module is designed for transferring history sensing data from current text files into database. Database updating is initiated by the end user through the web interface shown in Fig. 5. When a file is selected, all records in the file will be inserted into database. It is not necessary to run the database server at same machine in which the raw data files are stored. This module could also be implemented as a scheduled Windows Service, and then the database updating could be completed automatically. 6
Fig.5 Web interface for remote database updating 3.2 Web Search Module This module is designed for remote data retrieving. It allows user to select a set of interested sensors and time period through the web interface shown in Fig.6. To be user friendly, two selection approaches are provided. First is to select through checkboxes on the page. Each checkbox corresponds to a sensor. Selection is done by simply ticking the checkboxes. Obviously, one needs to know the sensor ID before his searching. Another way allows user select sensor through the sensor deployment map. Both overview map and cross-section zoom in map are displayed on the page. When user clicks on a cross section on the overview map a big size zoom in map is displayed in below. In the zoom in map, sensor positions are displayed more precisely and selection from the big size map is more convenient. Javascript is applied to implement this function, so the requirement of installing extra plug-ins in end users browsers is avoided. Fig.7 and Fig.8 show the typical search results displayed in graphical interfaces implemented by JpGraph and MatLab respectively. JpGraph is a fully OO (Object-Oriented) graph creating class library for PHP. It works very well with MySQL database and provides more graphical options than MatLab. However, considering the powerful data analysis ability of MatLab, the Matlab based result 7
display function is also implemented here. Although this approach incurs data communication between MySQL servers and MatLab server, it provides an easy way to implement and integrate more advanced data analysis and interpretation modules. Fig. 6 Web interface for remote searching 8
Fig 7. Search results displayed with JpGraph Fig 8. Search results displayed by MatLab 9
Fig 9. Search results displayed in table format The end user could also choose to display the search results in tabular format (shown in Fig.9), by which, exact readings of records are listed. Page flip strategy is employed to ensure fast query response. Data are displayed page by page such avoid the long waiting for retrieving all data from database. 3.3 Data Processing Module Some basic data processing functions has been developed that allows user get the most important statistics of the history data through web. Besides sensing records, the histogram, maximum, minimum, mean, mode and medium values for records are also provided for each search. Fig.10 is an example of the histogram for a sensor record during the time period specified by user. More data analysis functions will be developed as necessary. 10
Fig 10. Histogram for a specific sensor 3 Conclusions and Future Work 4.1 Advantages of the System The proposed web based data management system ensures efficient and convenient remote web data access. Users could update database, retrieve history data and explore data analysis results remotely through their web browsers. The design of the system promises great scalability in terms of both functionality and storage ability. The distributed database system supports virtually no limitation for data volume. Furthermore, additional modules, such as more advanced data interpretation components, could be integrated into the system conveniently. Compared with storing data with text files, database system is an organized data storage approach. It is very efficient since it avoids the overhead of file opening, which incurs long time for big files. With the indexing technique in MySQL and common database systems, searching for an entry takes only at most O(lgn) time. 4.2 Future Work In the future, more data analysis and interpretation functions will be implemented and integrated into this system to support more sophisticated structural condition evaluation. 11
Reference [1] U.S. Department of Transportation, Bureau of Transportation Statistics, Transportation Statistics Annual Report, 2003. [2] Hoon Sohn, Charles R.Farrar, Francois M. Hemez, and Etc. A Review of Structural Health Mornitoring Literature:1996-2001. LANL Report, LA-13976-MS,2003. [3] Dereck Hemphill, Structural Health Monitoring System for the East 12 th Street Bridge 12