SOLEUROPE  United Kingdom

Optimisation of System Resources in RAM Problems Using Genetic Algorithms

Aviv Gruber - Andy J. Keane
University of Southampton

the paper was selected by the program committee as one of the most representative and best communications that were given at the 16th symposium. Thanks to Mirce Akademy and to the authors for having authorized the publication on the site.

. General  Mail 
.
16 thMirce Symposium
.Mirce 16 Follow-up
.
...RAM... ...Using Genetic Algorithms
.Abstract
.Introduction
.Optimisation Definition and Difficulty
.#.
Sufficiency
.the RAM System Model
.a Proposed Solution: the Hybrid Approach
.Obtaining the Density of the Resources Space by Random Samples
.Search Process Using Genetic Algorithms
.Summary & Conclusion
.References
Prior Meetings and Documents
......

Sufficiency

 Optimisation 

Every system comprises eleven LRUs selected from six different types. The distinction among different types is a matter of the design of the model concept. Every type may have unique properties such as failure data, repair data etc.; on the other hand, there can be different types of LRUs with similar inherent properties and other attributes such as identical cost or, there can be two LRUs of the same type which may have different inherent properties (e.g. two engines operating at different loads). However, the attribute according to which a type will unify a set of different LRUs will be whether or not all members of that type share the same spare parts. Hence, the number of types in a problem will be the different number of spare parts types. In this problem, eleven LRUs in a system "consume" six different types of spare parts. The system composition along with the failure data, repair/replacement data, repair at level B depot and shipment times are given in  table 1 . Note that only mean values are shown in the table, however, essentially each process should have a descriptive distribution as well. In the current model all the failure distributions are Exponential where the mean appears as the Mean Time To Failures (MTTF - sometimes appears as MTBF for Mean Time Between Failures), all Replacement distributions are Normal where the mean appears as Mean Time To Repair (MTTR) with relatively small standard deviations and all Repairs at the depot and the shipment times are assumed Constants.

 RAM Model 

 References 



Download the document
 ACROBAT PDF (257 Ko) 



 General  Mail 

 Mirce 16  Prior Meetings and Documents 

 Mirce 16 Follow-up  ...RAM... ...Using Genetic Algorithms 

 Abstract  Introduction  Optimisation Definition and Difficulty  Sufficiency  RAM Model  Proposed Solution  Resources Density  Genetic  Summary  References 


last update:  December 12, 2006

.
.
.

Webhost: DRIM Technologies