A Memetic Algorithm for Multi Level Redundancy Allocation

A Memetic algorithmic rule for Multi Level circumlocution Allocation pilferReliability Redundancy allocation task determines the portion reliability and tautology accept for each sub brass in order to improve the schema reliability. Redundancy allocation problem is NP-hard problem and hence various Heuristic and meta-heuristic courtes be beingness employ.1 This paper proposes a memetic algorithm to solve the pick apart problem. Although the veritable engineering experience deals with multi- take systems, in this paper prior assumptions be do such that the algorithm has been applied to single take systems. This proposed MA is so comparisond against the HGA of a both multi-level systems and the proposed MA has outperformed the HGA of 2 multi-level system.INTRODUCTION of ArticlesIn this member, a refreshed Memetic algorithm has been proposed to solve the Redundancy Allocation Problem which has succeed great attention in recent years.Prior assumptions has been mad e in order to investigate the Multi-Level Redundancy Allocation problem and the three crucial assumptions atomic number 18 If a whole is not at its lowest level then its child units are assumed to be serial and they are fixed. The second assumption is that the redundancy plunder be allocated to the units at whatever level and the final assumption is that the t angiotensin converting enzyme of each component is predefined and the hail and the reliability are calculated based on child units if and totally if the unit is not a component. From the various literature review which has been carried out, it is fix that on that point are few approaches to MLRAP and these are rarely being investigated. inheritable Algorithm and the change Genetic Algorithm which is the Hierarchical Genetic Algorithm is although considered to be the significant approach for MLRAP, the effectiveness of these approaches can be improved. thus for the improvement in these approaches, the hot Memetic Algorithm has been proposed in this article. Memetic algorithm is a population based eta-heuristic try method which uses the combination of global chase engines along with the topical anaesthetic search heuristics. According to the article MA is more(prenominal) successful than the GA because of ii key issues. One is the appropriate match between the global and local search engines and the other is the speak to effectiveness. In this article two new patrimonial slatterns and a new problem specific local search means are incorporated in the MA frame locomote and a new MA has been proposed for approaching the MLRAP.In this article, the parts of a multi-level serial system has been defined hierarchically at the topmost level and the sub system has been defined at the get off level and the components in the lowest level. As per the assumption made previously, in that location is fixed number of child units for each unit except a component. The redundancy allocation procedur e always starts from the system level and moves to the component level for a multi-level serial system. The reliability of the multi-level serial system can be calculated using,While the reliability of the units at the lower level can be calculated on the basis of the components using,Also the greet for the multi-level serial system is calculated using,When applying the Conventional GA for solve MLRAP, the purpose variables becomes in fixed number during problem solving. Whereas in MLRAP problem solving the decision variable changes due to the change in redundancy allocation to a unit. Hence to overcome this problem, a hierarchical structure has been proposed in the article which is capable of changing the decision variables.NomenclatureExplanation of the work presented in journal articlesThe proposed MA in the article has two types of operators which consists of the genetic operator for global explorations and the local search operator for exploitation. This section explains the articles you reviewed.In the proposed model, the quality of the origins should be evaluated and from the literature the author of the article has pointed out both(prenominal) techniques to measure the quality of the dissolver out of which a penalty work on which has been proposed by Gen and Cheng has been utilize to derive the fitness accountability to evaluate the quality of the solution during the search march. This fitness function is disposed(p) by,(x) is the penalty function which measures the extent of the solution violating the coldnesss.The initialization of the MA is generally done at the system level and it starts from generating the random population of solutions. For a system of multi-level series, K integer is generated randomly for n child unit n X k redundancy units have to be generated at the second level and goes on until an individual is obtained.Of the two proposed genetic operators, before being applied the hierarchical representation need to be select ed. And hence the solution extremely depends on type of the hierarchical representation being selected. thus the two genetic operators will be applied to the iteration and both the operator treat the unit or the system at the akin level.Of the two genetic operators, the crossover between two individuals occurs in three steps. ab initio an intermediate is selected from the system and component level and the high level is being assigned higher probability. Finally the selected levels exchange their lower structures to give two new individuals. Whereas the mutation also occurs in three steps alone is applied to individual level. The first two steps remains the same as the crossover and the third step is replacing the redundancy of the selected unit by a randomly generated integer. Also when changing the redundancy equivalent decision variable o the parent unit should also be updated.The other operator of the proposed MA is the local search operator. In the new MA proposed in the article, local search operator is implemented in three steps. The population solution are evaluated through the metric which is given by,Then a single individual is chosen from the population solution. The local search is carried out based on it. Then the individual with the higher metric is selected for local search. For an MLRAP, it is desired to have an optimal system for a higher reliability and the reliability to cost ratio serves as a measure to determine the potential of the individual solution. The local search operator randomly selects ten pairs of components and for each pair the redundancy of the components are modified and during this search newly generated individual will be stored in the catalogue if and only if it donot violate the cost constraint. Finally, the preserved inividual will be merge with the population solution and arranged descendingly and the top most individuals will be used discarding others.Discussion of ContributionsTwo systems has been taken into consideration. One system with three levels and the other system with four levels. With this systems into consideration, the capital punishment of the proposed MA is to be evaluated and compared with the performance of conventional HGA.Since there are a number of control parameters in both HGA and the proposed MA, there are some values which are to be preset before beginning the experiment. For fair process, the values which has been set for HGA is used as same for the proposed MA for problem A. In this experiment the aim is to study the crossroad behavior of the proposed MA in order to compare it with HGA. The best solution obtained in each generation is recorded and the similar system reliability is calculated. It is seen that both the methods converged fast whereas the convergence of the proposed MA is significantly better than the convergence of the HGA which is shown in Fig 1.Fig1 relation of convergence between HGA and MA for Problem A.1Furthermore the comparisons has be en carried by change the numerous cost constraints. 20 cost constraint values are varied between the intervals 150 to 340 leaving the system parameters unchanged. For every constraint values, MA and HGA were applied 10 times to each cost constraint. The observations are made for each and every run and the reliability and the cost constraints are calculated and compared between MA and HGA. From the calculative comparison, it is evident that the proposed new MA has outperformed the conventional HGA for problem A that is the system with three levels.The same grade of experiment is carried out for problem B to examine whether the advantage of MA holds a variety of system parameters. Similar to the problem A, ten tryout instances are obtained and the MA and HGA are applied to each instances for ten times. The convergence of MA is significantly better than the convergence of HGA which is shown below.Fig2 carrefour of MA and HGA for problem 21Since the same kind of experiment was done on problem B, the results were also quite the same. The proposed MA has outperformed the conventional HGA.Discussion of Dificiency and capability ImprovementsThe article being reviewed here discusses only about the multi-level serial system and the experimentation has been done considering this system alone. Changing the condition and the structure of the system changes the reliability and the cost function of the corresponding solution changes. Also the proposed MA gives significant reliability for the multi-level serial system, the proposed MA should also be extensive to multi-level serial systems of complex structures. Also the problems which is being explicate in the article are single objective or are of only one goal of increasing the reliability of the system by having the cost simplification as only one constraint. Instead, in the future research, the problems can be formulated with twofold objective and multiple constraint along with the cost constraint. This type of ap proach with multiple objective problems will yield multiple solutions which has trade-off between the system reliability and cost constraint.SummaryThe RAP which has attained a global attention among the researchers motivating them to find the solution for the RAP. Though many algorithms, techniques and approaches have been proposed by many researchers around the world, there is something which can be improved in each and every approach proposed by researchers to solve the MLRAP. In this article, the author has given a slender approach on how does a reliability problem works and formulated in a way that the problem deals with the multi-level serial system of honest structure. Upon formulating a problem, the author tries to make proper assumptions to advocate the formulated problem. Then the author just solves it with the conventional GA approach which yields a levelheaded system reliability. Then the problem is again solved with the new method. This new method is being proposed b y the author is the novel Memetic Algorithm where some preconditioning is being done to the solution that is being selected form the population solution. The solution is checked for the quality by using a fitness function. like a shot the population solution will be initialized and the two search operators of the MA is applied to the selected and preserved solution from the population samples of solution. The solutions which are being obtained from the search operators are then being combined with the population and the best among them is selected. The process is repeated for various generations and the best individual will be selected as the solution for the MLRAP. Then the approaches are compared and it is found that the proposed MA has outperformed the conventional HGA disregarding of the type of multi-level serial system of same structure.AcknowledgmentsI would like to thank Dr. Wang, fellow, IEEE, and Dr. Tang, fellow, IEEE and Dr. Yao, fellow, IEEE for their research study that was done by them to prepare the article, which helped me to understand the fantasy behind the RAP and MLRAP and the techniques used to solve or approach MLRAP. I would also like to express my Thanks to Dr. Pingfeng Wang, Graduate Coordinator in Wichita assign University for his valuable advice which guided me through this project and helped me to complete this review successfully.References1 Wang, Z., Tang, K., Yao, X. (2010). A memetic algorithm for multi-level redundancy allocation. IEEE Transactions on reliability, 59(4), 754-765.2 Sharifi, M., Cheragh, G., Maljaii, K. D., Zaretalab, A., Daei, A. V. F., Vahid, A. (2015). RELIABILITY OPTIMIZATION OF A SERIES-PARALLEL K-OUT-OF-N SYSTEM WITH FAILURE RATE DEPENDS ON WORKING COMPONENTS OF SYSTEM. International ledger of Industrial Engineering, 22(4), 438-453.