ICAM-1998-A Computational Architecture for Lean Manufacturing Machine Systems

Presentations

This is a project I am now working on with The University of Southern California (L.A.)’s Industrial and Systems Engineering Department in my spare time.

We hope to bring the internet to the plant floor with this technology. We are moving the controllers that control equipment/tools on the plant floor to a remote location so that the same remote server can be used to control the operations of different machines all across the Internet/Intranet.

The project is completely Java based and it involves a lot of Enterprise JavaBeans (EJB), Jini and serial communications.

These controller/servers will bring push technology to the plant floor. Users can now subscribe to the appropriate services that they want to help them detect different kinds of anamolies and automatically rectify them. This also will help people control plant-floor tools from remote corners of the globe.

Currently researchers investigating specific anamolies tend to go more deeper into one particular type of anamoly that they do not have time to research other types of anamolies. This project helps to tie all the research together and help alleviate this problem.

We are developing a Lean Manufacturing Machine System (LMMS) that facilitates real-time process control and maintenance for a multitude of similar production equipment. The LMMS is based on combining nonlinear dynamics with multi-agent systems to integrate several signal-level and decision-level tasks. We have developed a formalism to model agents for LMMS and are implementing an architecture to implement computations and communications involved in LMMS.

The architecture uses a local server that gathers signals from an equipment, stores and transfers these to a remote host, receives control input-generating algorithms from a remote host server, updates control codes and executes these through several actuators.

The remote gateway is connected to several distributed well-known agents, each responsible for detecting a particular anomaly and synthesizing a remedial control policy to optimize a limited local objective. Global objectives are attained through a collaborative negotiation among these agents. The remote servers and agents are multiplexed to simultaneously cater to several machines.

We presented this software architecture in specific reference to controlling machining process under the presence of multiple anomalies occurring over a reasonably wide range of cutting conditions (process parameters). We considered only 3 anomalies, namely, tool wear, chatter and forces for performance assessment

We recently presented a paper on LMMS at The International Conference on Agile Manufacturing (ICAM 1998), held at the Radisson Hotel Metrodome in Minneapolis, Minnesotta and The University of Minnesota Campus, Minneapolis, Minnesota, USA from June 21-23 ‘98.

If you are really interested in learning more about this project, please send in a note to me or to Prof.Dr.Satish Bukkapatnam of USC.