Refine
Document Type
- Article (14)
- Conference Proceeding (2)
Keywords
- Lernendes System (16) (remove)
Year of publication
- 2004 (16) (remove)
This paper presents an agent-based software, Virtual Administrator System (VAS) for the smallscale maintenance of school buildings. VAS is capable of handling a heavy load of routine, lowtech maintenance jobs. It assigns a different priority to each job application according to its significance and urgency, and automatically adjusts schedules for maintenance engineers when on-site supervision is needed. The system can help ease off the burden of routine small-scale maintenance work, making it more cost-effective and efficient in the overall management of school building maintenance. VAS posts jobs on the Web in a multi-media format and classified all applications into four categories: the on-call maintenance contract, the term maintenance contract, the guaranty maintenance contract, and the regular maintenance contract. It then estimates their urgency level and passes the information to maintenance engineers who will decide whether on-site inspection is needed. Based on the engineers’ feedback, VAS automatically implements the scheduling for inspection as well as sends out real-time or batch notifications to contractors. All these activities are recorded in a database to allow continuous research and data mining and the analysis and diagnosis of specific jobs for followup maintenance plans.
Expert systems integrating fuzzy reasoning techniques represent a powerful tool to support practicing engineers during the early stages of structural design. In this context fuzzy models have proved themselves to be very suitable for the representation of complex design knowledge. However their definition is a laborious task. This paper introduces an approach for the design and the optimization of fuzzy systems based upon Genetic Programming. To keep the emerging fuzzy systems transparent a new framework for the definition of linguistic variables is also introduced.
This paper presents a new design environment based on Multi-Agents and Virtual Reality (VR). In this research, a design system with a virtual reality function was developed. The virtual world was realized by using GL4Java, liquid crystal shutter glasses, sensor systems, etc. And the Multi-Agent CAD system with product models, which had been developed before, was integrated with the VR design system. A prototype system was developed for highway steel plate girder bridges, and was applied to a design problem. The application verified the effectiveness of the developed system.
A Multi-objective Model for Optimizing Construction Planning of Repetitive Infrastructure Projects
(2004)
This paper presents the development of a model for optimizing resource utilization in repetitive infrastructure projects. The model provides the capability of simultaneous minimization of both project duration and work interruptions for construction crews. The model provides in a single run, a set of nondominated solutions that represent the tradeoff between these two objectives. The model incorporates a multiobjective genetic algorithm and scheduling algorithm. The model initially generates a randomly selected set of solutions that evolves to a near optimal set of tradeoff solutions in subsequent generations. Each solution represents a unique scheduling solution that is associated with certain project duration and a number of interruption days for utilized construction crews. As such, the model provides project planners with alternative schedules along with their expected duration and resource utilization efficiency.
A comprehensive framework of information management system for construction projects in China has been established through extensive literature survey and field investigation. It utilizes the potential information technologies and covers the practical management patterns as well as the major aspects of construction project management. It can be used to guide and evaluate the design of the information management systems for construction projects in order to make the system to be applicable to a wide variety of construction projects and survive the changes in project management.
The purpose of this research is to develop the method to retrieve a building name from the impression of the building. First, the images of the building are registered as database by the questionnaire. Next, the images of the objective building are compared with the degree of matching in image databases, and the building with high synthetic matching degree is retrieved. This system could get a good retrieval result. Moreover, image processing was done, and image databases are trained by neural network from the amount of characteristics of the image, and the retrieval system by image processing was examined.