A 3D visualized expert system for maintenance and management of existing building facilities using reliability-based method

Facility maintenance and management (FMM) is an emerging issue in civil engineering. Decisions involving maintenance-related tasks are generally made based on various sources of accumulated historical data, such as design drawings, inspection records, and sensing data. Systems are developed for storing and maintaining such maintenance-related data electronically in a database. However, the data-accessing method of these systems is based mainly on text input in Web form, which is occasionally insufficiently intuitive to interpret retrieved information for decision making. Besides simple data management practices, the feasibility of implementing analysis on FMM-related data to provide estimated or predictive information for decision making should be examined. This paper presents an expert system model for the maintenance and management of existing facilities. A prototype system was developed for concept proofing. A 3D facility model is introduced in the system as the interface for accessing various maintenance-related data intuitively. Various maintenance-related data and analysis results should be presented visually on the model as much as possible to provide users with an intuitive understanding of the facility status in many aspects. Behind the 3D visualized interface is a database that integrates and stores various maintenance-related data systematically. This database information should be accumulated continuously via input from users and sensors in appropriate formats. Moreover, a reliability-based module should analyze the accumulated data periodically to provide predictive forecast information, subsequently facilitating decision making during maintenance.

► We propose an expert system for the maintenance and management of facilities.
► A 3D facility model is introduced as the system interface for intuitive data access.
► The system includes a database that integrates various maintenance-related data.
► A reliability-based method is developed for predictive forecast information.
► Predictive information from data analysis can facilitate decision making during FMM.