Dottorato

Dottori di Ricerca in Sistemi di Produzione

MICHELE AMBRICO

Dottorato in Ingegneria Industriale e dell’Innovazione XXV – ciclo nell’ambito della progettazione robusta di sistemi di produzione cellulari e relatore della tesi di dottorato

dal titolo: “Cellular Manufacturing System in Dynamic Market Environment: robust design and reconfiguration models”.

ROCCO PADALINO

Dottorato di Ricerca in “Ingegneria Industriale e dell’Innovazione” XXII ciclo e relatore della tesi di dottorato

dal titolo: “Scheduling in an agent based manufacturing environment: new adaptive dynamic approach in a market like system”.

Abstract

The thesis is concerning with scheduling in Flexible Manufacturing System (FMS) according to Multi-Agent System (MAS) paradigm. One important tool for manufacturing and engineering isthe Scheduling, where it can have a major impact on the process productivity. The purpose of scheduling, in manufacturing environment, is to minimize the production time and costs; it is possible by telling a production facility what to make, when, with which staff, and on which equipment. So, a good production scheduling allows to maximize the efficiency of the operation and reduce costs. The scheduling problem is an NP-hard problem. This kind of problem is not possible to solve in a centralized way; so the idea is to solve the scheduling problems in adecentralized way.

This thesis idea is to seek eective modelling and ecient solution techniques that can help increase the productivity and the service level of an enterprise, together with reducing production costs, by bringing down the control complexity of the system. A flexible and adaptive scheduling is essential to achieve the system production goals; the complexity, uncertainty and dynamics in the modern manufacturing environment require this scheduling features. To obtain it the Multi Agent Systems appear the best solution. The agent-based computation is the emerging paradigm has revolutionized the building of intelligent and decentralized systems. This new paradigm and its technological implementation met well the requirements in all domains of manufacturing where problems of uncertainty and temporal dynamics, information sharing and distributed operation, or coordination and cooperation of autonomous entities had to be reached.A Multi Agent System architecture have demonstrated their ability to develop flexible and agile manufacturing, able to guaranteefault tolerance, scalability, hardware and software integration, agility. Further on, the effi- ciency is a very important issue in agile manufacturing systems management and the adaptability todifferent exception that can lead to the degradation of the system performance. The counter arguments against agents are varying. First of all it is not sure that agents could reach the final target for which they are implemented. It could happen because agents do not have a global vision of the system. In the most of cases the target of the single agent is really far from system

one. So the choice of the right agent target assures to aim the system in the right direction. But it can be not enough. In fact another problem is how agents relate each others. It is not important if they collaborate or are in competition, it is important that the protocol which allows them to communicate is the appropriate one. The communication implies a big exchange of information that can be considered another disadvantage against agent technology. If it is possible to arm that a Multi-Agent System works as it is required, there are other problems. For example, its behavior is guaranteed in each work condition? The solution is good? As regards the

rst question, it depends how the Multi-Agent System works, it is linked to the stability and robustness of the system itself. As regards the second question, we know that it is quite impossible to obtain an optimal solution from a Multi-Agent System, but the advantage is to obtain a sub-optimal solution near to the optimal one in usually less time. Another problem is linked to the distributed structure of the system that brings sometime to leave uncompleted job or task.

This thesis proposes innovative decision making strategies for autonomous agents in cellular manufacturing systems able to maintain such an elevate level of eciency. The proposed Multi-Agent System controverts these considerations, and results show that the proposed approach is robust and gives good performances.

The proposed scheduling is based on a multi-agent system in a very dynamic environment. It is an innovative decision making strategies for autonomous agents in cellular manufacturing systems based on a global cell effiiciency level. The focus of this research is on the negotiation protocol and decision-making strategies that keep adequate level of system performance under different kinds of uncertainty.

The decision strategy is the budget approach; it is able to schedule the production of coming jobs taking care of the workload of the system, the distinctive characteristics of each work cell and some peculiarities of the same

jobs. The approach has a high level of versatility and it able to operate in extended manufacturing environments keeping a high level of performance.

The budget approach is compared with others scheduling approach proposed in literature. The implementation shows how the proposed model leads to better performance in different conditions of manufacturing system

workload, number of cells, dynamic cell characteristics. The reduction of complexity is due to the reduction to only one parameter to be calculated to decide how system has to work.

The armed scheduling is implemented in the ARENA software, which is able to implement the Multi Agent Architecture and simulate the dynamic environment.