Research 

[ Home ] [ Personnel ] [ Research ] [ Publications ] [ Alumnus ] [ Useful Links ] [ Contact Us ]

The current research activities of the laboratory, under the direction of Dr. He, cover a wide range of theoretical and applied areas in Manufacturing Scheduling, Quality Control, System Reliability, Diagnosis and Prognosis, etc. These areas are broadly characterized as follows:

 

Alphabetical listing

Intelligent Equipment Health Diagnosis and Prognosis

 

 

Back to Top

 

Multi-Agent Based Manufacturing/Production Systems Scheduling

Continuously emerging global competition and rapidly changing customer requirements are forcing major changes in the production styles and configuration of production systems. Increasingly, traditional centralized and sequential manufacturing planning, scheduling, and control mechanisms are being found insufficiently flexible to respond to changing production styles and highly dynamic variations in market requirements. The traditional approaches limit the expandability and reconfiguration capabilities of the manufacturing systems. The traditional centralized hierarchical organization may also result in much of the system being shut down by a single point of failure, as well as plan fragility and increased response overheads. Agent based technology provides a natural way to overcome such problems, and to design and implement distributed intelligent manufacturing environments. Agent based technology reveals a new way of thinking about many complex systems. In the past ten years, researchers have been applying agent technology to manufacturing enterprise integration and supply chain management, manufacturing planning, scheduling and execution control, materials handling, inventory management, etc. 

 

A methodology of agent-based agile manufacturing scheduling incorporating game theoretic analysis of agent cooperation is developed in this research. The developed model integrates data and decisions associated with several entities within a scheduling system. This approach can be used to model and solve large-scale scheduling problems for any manufacturing system configuration. The manufacturing system is of a general type consisting of multiple machining and assembly stages each containing number of machines. The product assembly structure is represented by a directed graph where nodes represent assembly and machining operations and arcs represent precedence relations between operations. Agent identification and assignment of scheduling tasks to agents are achieved by digraph decomposition strategy of product’s assembly structure. The cooperation between the entities in the system is achieved by the solution concepts developed in cooperative game theory. The power index of an agent corresponds to the Shapley value for that agent which is associated to the schedule completion time in the system achieved by an agent. The overall scheduling process is an integration of defined inner and outer games where outer game is among the agents eligible to schedule their jobs in the system and inner game is among the agents eligible to re-schedule their jobs in the system. 

 

Back to Top

 

Reliability and Safety Analysis and Assessment of Complex Systems

Failure analysis with multiple failures and reliability assessment of complex systems is a very important task for many industries.  The objective of the research is to develop a new integrated approach for systems failure analysis and enhance the existing sequential failure analysis methods.

 

An extensive research is conducted to investigate the computational methods and their implementation for complex systems reliability and sequential failure logic.  To overcome the limitations in sequential failure analysis and the failure analysis of complex systems with multiple failures we have developed three new methods and hardware-based implementation.

 

Existing methods for assessing the reliability and safety of a system with sequential failures assume that the sequences of the failures are given, which is not the case in real life situations.  Therefore, the sequences of the failures should be identified and the probability of their occurrence should be determined.  The identification and analysis of sequential failures method identifies the possible sequences of failures and computes the probability of their occurrence.

 

Current Petri net–based failure analysis methods first, imply that the times between failures of different components are the same and constant; and, second, do not take into the consideration the complex Petri net structures, for example, inhibitor arcs.  The failure analysis based on the markings of Petri net method is developed to deal with complex Petri net structures and different failure rates of basic events.

 

It is discovered that the marking-based method is not flexible for sequential failure analysis and cannot be applied directly to the Timed Petri net.  Therefore, the method of failure analysis based on the concept of counters of Petri net simulation is developed.  The counters-based method is designed to deal with complex Petri net structures and compute sequential failure probabilities.

 

The hardware-based implementation of counters-based method is developed.  Different structures of Petri net are mapped into the circuits.  The prototype of a simple manufacturing assembly system is developed.  The testing of the prototype demonstrated the feasibility and accuracy of the method.

 

Back to Top

 

Statistical Quality/Process Control for Agile Manufacturing

As today’s manufacturing firms are moving towards agile manufacturing, quick detection of the shifts in manufacturing process are in high demand. Multiple sampling control charts are one of the promising alternatives alongside the cumulative-sum (CUSUM), and exponentially weighted moving averages (EWMA) charts. Multiple sampling techniques, when the decision about the lot under certain circumstances is delayed until next lot is inspected is a well known statistical quality control technique and have long been applied in acceptance sampling. The application of multiple sampling to control charts was studied recently. In its simplest form the, univariate double sampling (DS) X-bar control charts appear to be a good technique for statistical process control in an agile manufacturing environment.

 

Given the promising results of the comparison of the DS X-bar charts with existing ones, the further research in the area of multiple sampling charts is expected and encouraged. The existing CUSUM and EWMA charts can provide good detection of small shifts, but the investigation of the other alternative control schemes still continue. It is known that both CUSUM and EWMA schemes utilize information from previously collected samples by giving to them certain weight.. Once the process changes behavior again then these charts might not detect it.

 

The objective of this research is to develop multiple sampling charts for both univariate and multivariate cases, for monitoring and controlling process mean and variability. To investigate the statistical properties and performance of the developed multiples sampling control charts and to develop automated design methods and tools to aid practitioners in statistical process control implementation

 

Back to Top
 
© 2002 Intelligent Systems Modeling & Development Laboratory, UIC
 For problems or questions regarding this web contact WEBMASTER
Last updated: 十一月 27, 2002.