Integrated Prognostics, Maintenance and Life Extending Control of Mechanical and Production Systems

Project: Research project

Project Details

Description

9424587 This proposal supports research in integrated prognostics, maintenance, and life-extending control of continuous time production processes. The research includes three tasks. The first task involves the development of stochastic fatigue damage dynamics models with data determined by experiment and obtained from published literature. A specific fossil fuel power plant model will be used as an example case. The second task focuses on the synthesis of decision and control policies based on both continuously varying and discrete-event models. This formulation will include a feed-forward policy over a finite time horizon using nonlinear programming, a constrained parameter optimization for nonlinear decision and control in a feedback loop, and interactive discrete event strategies for maintenance decisions and life-extending control at the coordination and supervisory levels using the concept of finite-state automata. The third task is the simulation of the decision and control system on a simulation testbed. If successful, this research will lead to enhanced plant safety, operational reliability, and availability of continuous-time production processes such as chemical and processing plants, pulp and paper mills, and large rolling mills. The results should also be extendable to discrete production processes such as the manufacture of glass containers and light bulbs, semi-conductors, ceramic-enamel coatings on appliances, and food products. This research comprises a step toward the notion that control and diagnostic systems based on sensory and operational information can be used to improve plant operational characteristics.

StatusFinished
Effective start/end date7/1/956/30/99

Funding

  • National Science Foundation: $180,000.00

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.