Abstract EN:

Maintenance documents are traditionally provided in the form of hardcopy supports for different users like end-users, maintenance technicians and manufacturers. These documents are generally elaborated for a standard product configuration and for general utilization conditions. In the traditional maintenance document, maintenance requirements are often predesigned. Once it is published, it is difficult to be updated to follow the product versions changing or to be adapted according to various Utilization Conditions (UCs). Therefore, the end users or technicians need more time to decide what appropriate maintenance tasks should be accomplished on the new configuration product with a new UC. In order to solve the problems discussed above, a more intelligent maintenance document (dynamic, electronic, customized) is proposed in this thesis which is named Customized Maintenance Document (CMD). The goal of this thesis is to outline a methodology for developing a CMD which shows how to customize the maintenance document for complex mechanical products to be used in different typical contextual conditions, for users with different skill levels and tasks. Three steps are focused in this thesis : knowledge integration process, decision making system design, and document model adoption. After collection and analysis of historical data (such as : utilization condition, structural data and operational data), knowledge is elicited in the form of case and embedded in the system to predict the failure rate of each unit. A multi-model decision making system is designed aiming to help users to find the most ideal maintenance tasks by analyzing the environment in which the complex mechanical product is operating. In order to present information in a better form and take the contextual knowledge into consideration, a semantic document modeling is designed which includes three layers : customized data modules layer, task-oriented document layer, and semantic network layer.