Task 4: Predictive models

Description and objectives

In order to meet a growing need fro complex systems to adapt maintenance dynamically by integrating uncertainties and data inaccuracies, we propose to develop predictive models based on hidden Markov models used for Learning process. These models are well adapted to represent our problem since they are able to reveal at the same time the states of the environment, the possible actions for an agent, and the transitions between states due to these actions. this will help to estimate the residual life of the components and of the overall system.

Start date

01/02/2018

Duration

36 months

Task leader

Heudiasyc

Sub-tasks and delivrables

T.4.1. State of the art on predictive maintenance under uncertainties

D.4.1. state of the art report

T.4.2. Expression and analysis of needs

D.4.2. Specifications of the end user

T.4.3. Methodology development and predictive model design

D.4.3. Report defining the methodology
D.4.4. Report on the integration and implementation 
of the methodology

T.4.4. Implementation on concrete cases defined by the end user

D.4.4. Test report on use cases

Milestone

Month 36