Reliability optimisation for flexible manufacturing system by using intelligent agents.

Ciotea, Valentin Florin ; Ciotea, Margareta ; Rahau, Loredana 等

1. INTRODUCTION

Flexible Manufacturing System (FMS) represents a new manufacturing system, which determines a well manufacturing process economically. It is orientated in sense of real products from human society, means that diversify large goods which are produced in low quantity. In system manufacturing the principal conception problem are centred on reliability and automation (Abrudan, 1996). FMS are in general high level investment and in that condition, its reliability problem is very important. The particularity FMS's functioning will generate any particularity in determining our Flexible Manufacturing System. We will make some consideration about FMS Flexible Manufacturing System. Our contribution is determining SFM Flexible Manufacturing System use analytic method, a program from analytic method and simulation (Monte Carlo) method (Ciotea, V. 1997).

2. RELIABILITY OF FSM

FSM functioning can be see on to the many point of view:

--Possibility that one good may out from SFM on one moment (Ciotea, 2003)..

--Possibility that all goods out from SFM on a given moment;

--Possibility that a specified good to came out from SFM on a given moment. We consider here the first point within tree, because it implies a big volume of processing.

[FIGURE 1 OMITTED]

3. THE EMPLOYMENT OF INTELLIGENT AGENTS IN THE OPTIMIZATION OF FLEXIBLE MANUFAC-TURING SYSTEMS RELIABILITY USING FLOW DYNAMICS

Flow dynamics represent a method witch identified any alternative fabrication flows. This flow are using in case of a flow is down and is necessarily its replacement. I identify two alternative flows (Ciotea, 2006).:

1. Total redundant flows--it was introduced in system to increase the reliability of a fabrication flow. Total redundant flows can be beneficiate by a new working module, transportation module, specially introduced for increasing flow reliability.

Costs to involve total redundant flows are:

-Implementation cost--with introducing a new working module, transportation module

-Functional costs--with flexibility of manufacturing system, supplementary stock

Implementation of total redundant flows is justified when it was respected condition as sum of implementation costs and functional costs must be less than costs determine when flow is down.

2. Partial redundant flows--it was introduced in system to increase the reliability of a fabrication flow without introducing a new working or transportation modules. Partial redundant flows can be beneficiate only by flow witch exist in system. In this case implementation costs are null. This method increases the reliability of flexible manufacturing system less than first method, but also cost it is less than first method.

The number of factors that can influence flow dynamics in flexible manufacturing systems reliability is relatively high. Thus, one has to take into account all types of operations that can be performed by installations, the load of installations, the type of markers that are being processed in distinct installations, and the benefits that they bring to the system, the processing stages, module reliability, system flexibility costs, storage costs, loses determined by product failure, etc. the high number of factors that may influence flow dynamics and relationships among these makes the use of classic methods of optimization even more complicated. One of the most flexible and elegant method to obtain optimal reliability using flow dynamics, is the use of intelligent agents.

[FIGURE 2 OMITTED]

By using negotiation and a minimum of artificial intelligence-based on neuronal networks, fuzzy systems, genetic algorithms or hybrids--these can be satisfactorily used to determine the optimal reliability.

The main advantage of this method is that the introduction of new parameters and of transformations does not trigger alterations of the method, but only lets the agents know, teaches them, about the changes that have taken place.

In principle the method consists of defining the module for each marker that is processed. It is also necessary to define one agent for the entire manufacturing system. The roles of different agent types are as follows:

--The installation agent represents an installation or a group of installations (a module). He may possess knowledge about the operations that the installation can perform, the processing stages for different operations, processing costs, processing quality, degree of occupation, etc.

--The marker agent represents a marker that is being processed. It possesses knowledge about the technologic stages that have to be followed, the type of technological stages, the processing stages for different operations, processing quality, etc.

--The system or subsystem agent is the most important in the system. It must supervise the negotiation procedure between he marker agent and the module agent, so that negotiation is clear and optimal, and the result of the negotiation is as close as possible to the optimal result.

The main duty of the system agent is to trigger negotiation inside the system.

Stages of intelligent agent's method, used in flow dynamics:

1. It starts from a system initially designed as satisfactory in terms of reliability. Then the working modules, the markers that are being processed, the existent flows, the current costs and benefits are defined.

2. Supposing an installation or a module breaks down the system agent is warned. The system agent calculates potential losses determined by the event.

3. The system agent warns the agents of critical markers that are currently on the same flow with the defective installation or module.

4. Marker agents negotiate their distribution on working modules, going from priorities to the less vital. Negotiation with modules of the respective flows is done according to criteria settled for each marker: type of processing, processing quality, processing costs, time, etc. These criteria, as well as the order of negotiation are established by the system agent.

5. The markers' agent designated to negotiate with modules' agent creates a negotiation priority order according to its experience acquired in other negotiations, or to the distance to as certain installation, or simply at random (if none of the previous conditions can be applied).

6. The marker and the system agents establish whether the chosen module interrupts the technologic flow of the procedure, that is, if there is a way to return to the normal manufacturing flow, or whether the marker can be terminated using the new flow. If this condition is met, one can go to the next step.

7. The marker and the module agents establish whether the marker can technically be processed on the respective module. If it is possible one passes to the next step.

8. The marker agent inquires the module agent about the markers that are already being processed. If the marker cannot be taken over, then, with the system agent's approval, those markers that do not represent a priority will be temporarily eliminated.

9. The system agent is informed about the result of the negotiation. In case of a positive result the system agent compares the costs of system status changes. If the costs are higher than the losses determined by the breaking down of the module, then negotiation is considered to be a failure and one must go back to stage 6.

10. The marker agent that put an end to negotiation conveys to the system agent the new technologic flow, so that the later is able to direct marker agents of identical markers, on the new technologic flow.

11. The agents of markers that were suspended from processing will demand the system agent to put the respective markers on the markers-being-negotiated list, in a position corresponding to their priority. The algorithm is resumed from stage 4, until all markers on the list are dealt with.

12. The system agent calculates cost and benefits once the negotiation is over. If the new benefits are higher than those of the previous situation the new situation is considered to be the most propitious.

Agents' involvement is done in real time, as events occur in the system--defaults or changes in the system structure-, which is, called system status. The knowledge possessed by the agents can be ensured through:

--An initial loading of the system with information. In this case each type of agent must posses a certain amount of knowledge. Module agents must know the parameters of the modules they represent, such as the number of installations, types of processing that they can perform, processing costs, markers that are initially processed, initial occupation times, etc. Marker agents must know the stages of the technological flow of the markers they represent, processing costs, potential benefits, etc. The system agent must know the structure of the system, initial and potential manufacturing flows and posses a top of markers' importance.

--Previous answers of the system to different real changes of the system status.

--Previous answers of the system to different simulated changes of the system status.

4. CONCLUSION

After enterprise study of SFM reliability, as probability that one product to leave the system, we can conclusion:

--The last reliability module of system has a substantial influence in system's reliability than other module from system.

--Intelligent agents in the optimization of flexible manufacturing systems reliability method involve a maximum flexibility, low costs for implementation and dynamic structures of system.

In the next step of this project we want to extend area above economical systems by researching reliability optimisation for these systems by using intelligent agents. This method is possible to extend to social systems, where it can possible to simulate social group behaviour, by using one agent for any person or subgroup from studied social group (Ciotea, V. & Ciotea, M., 2007).

5. REFERENCES

Abrudan, I. (1996). Flexible Manufacturing System, Publisher Editura Dacia, ISBN 973-35-0568-4, Cluj-Napoca.

Ciotea, V. (1997). About reliability of FSM, Petru Maior University, Tg. Mures.

Ciotea, V. (2003). Intelligent agents and FSM, in Revista Inginerului Economist, no. 1, page 5-21, Cluj Napoca.

Ciotea, V. (2006). Research by reliability of FSM--doctoral thesis, Publisher Tehnical University of Cluj Napoca.

Ciotea, V. & Ciotea, M. (2007). Application for computations of the reliability flexible manufacturing system, In: InterIng 2007, Petru Maior University, page 151-154, Targu Mures.