Reaearch Of Interconnection Network Fault Diagnosis Based On Improved Fireworks Algorithm

Nowadays,with the rapid development of science and technology,people are increasingly demanding large-scale and even ultra-large-scale multiprocessor systems.These large-scale multiprocessor systems interconnect nodes in the system through a certain organizational structure.Therefore,the interconnection network is the internal coordination mechanism and the main communication channel,and is an important part of multiprocessor computers.The increasing number of nodes in a multiprocessor system is followed by the stability of the system.The more nodes in the system,the higher the probability of failure.It is especially important to diagnose the faulty node quickly and efficiently.The traditional method of fault diagnosis is to use a special fault detection system.The main disadvantage is high cost and high overhead.In this case,interconnection network fault diagnosis comes into being.The basic principle is that by using the computing power and communication capability of the node machine in the system,the nodes can be tested or compared with each other to diagnose the system fault set.Under the concentrated research of many experts and scholars,interconnection network fault diagnosis has made a big breakthrough and has been well applied.At present,many intelligent diagnosis algorithms have obvious premature convergence problems.And the fireworks algorithm has good self-adjusting ability in local search and global search,which can effectively sovle this problem.Therefore,based on the characteristics of different fault diagnosis models,the fireworks algorithm is used to design efficient interconnection network fault diagnosis algorithm.The main research work in this paper on interconnection network fault diagnosis is as follows:(1)A fir-eworks algorithm for interconnection network fault diagnosis based on Malek model(FAINFM)is proposed.The algorithm introduces the specified fault-free node method to initialize the fireworks population,and designs the constraint equations and fitness evaluation function based on the Malek model.The improved fireworks algorithm is used to solve the fault diagnosis problem.Finally,the simulation experiment is carried out to verify that it is highly efficient on diagnosis problem.(2)A fireworks algorithm with genetic operators for interconnection network fault diagnosis(FAGOINF)is proposed.The algorithm is imed at the problem that there are many identical individuals in the population during the middle or late stages of the evolution in fireworks algorithm.It leads to the slower search speed of the algorithm and the tendency to fall into the local extremum problem.After the fireworks explosion operator and the Gaussian mutation operator,the elimination of duplicate individuals operation is added.And considering after the deletion of duplicate individuals,the population size may be reduced.So the cross-replenishment population method is proposed.This method draws on the cross-operation principle of genetic algorithm,and uses the existing individuals of the population to generate new individuals to supplement the number of individuals.At the same time,in order to strengthen the information exchange between individuals,the algorithm introduces the crossover operator and mutation operator of genetic algorithm.Finally,a series of simulation experiments are carried out to verify the efficiency and stability of the algorithm in the fault diagnosis.