Improved Fireworks Algorithm For System-Level Fault Diagnosis

Based on the background of big data era,the application of large-scale multiprocessor system is becoming more and more common.In order to obtain better operating efficiency,the scale of multiprocessor system is expanding,the complexity of the system is increasing,and the possibility of processor node failure is also increasing.Because this kind of system integrates many processors,each of which is equivalent to a host of the system,and each processor needs to communicate with each other to realize resource sharing.The system needs to deal with many real-time tasks,and the real-time communication between processors will be very heavy.Once some processors in the system fail and fail to be handled effectively in time,it will lead to serious losses and even disastrous consequences.Therefore,how to get an effective diagnosis technology to quickly and accurately locate the fault processor in the system and repair or replace it is a very meaningful research direction.As a powerful measure and effective method to improve the reliability of multiprocessor systems and reduce the risk of accidents,the research on system-level fault diagnosis is becoming more and more important.With the wide application of swarm intelligence algorithms,a variety of swarm intelligence diagnosis algorithms have emerged.However,a lot of intelligent group diagnosis algorithms have the problem of precocity and instability.Considering that fireworks algorithm has a strong global search capability and self-regulating mechanism for local search capabilities,it can solve these problems effectively.Therefore,according to the characteristics of different fault diagnosis models,this paper respectively improves the traditional fireworks algorithm.The improved firework algorithm is used to design the intelligent,efficient and high accurate system-level fault diagnosis algorithm under different models.In this paper,the research content and work on system-level fault diagnosis include the following three aspects:1.Introduce the two-population strategy into the fireworks algorithm,and propose a system-level fault diagnosis algorithm based on MM~* model.In this algorithm,two populations operate independently in parallel,and the cooperative operator and the optimal operator are cross-executed in the iterative process.The cooperative operator enable the two populations to exchange information effectively,avoiding excessive prematurity of the algorithm,while the optimal operator helps to strengthen the global search ability and improve the convergence rate of the algorithm.At the same time,the constraint equation is designed,a new fitness function is proposed,and the mutation operator and selection strategy are optimized.The simulation results show that the algorithm can effectively improve the efficiency and accuracy of system-level fault diagnosis,and has better practicability.Finally,the correctness of the algorithm is proved by theory,and the time complexity of the algorithm is analyzed.2.Combining fireworks algorithm with BP neural network,and propose a system-level fault diagnosis algorithm based on PMC model.Firstly,we introduce two population strategy,cooperative operator and optimal operator into fireworks algorithm,and design new fitness function,optimize mutation operator,mapping rule and selection strategy.Then,the improved fireworks algorithm is used to optimize the weight and bias of BP neural network.The simulation results show that compared with other algorithms,the proposed algorithm effectively reduces the number of iterations and training time,and also improves the diagnosis accuracy significantly.3.The organization structure of wireless sensor network is similar to that of multiprocessor system,so combining fireworks algorithm with convolutional neural network,and propose a fault sensor recognition algorithm of wireless sensor network based on MM~* diagnosis model.Firstly,the traditional fireworks algorithm is improved according to the characteristics of MM~* model.Then,the improved fireworks algorithm is used to optimize the weight and bias of the convolutional neural network,so that the problems of the convolutional neural network in extreme value judgment and convergence speed limitation can be solved,so as to effectively realize the fault diagnosis of the wireless sensor network.The simulation results show that the algorithm has high fault diagnosis accuracy.