论文部分内容阅读
为了提高电路系统在高电压、强电磁等复杂电磁环境中的容错抗扰能力,有效降低故障率,增强系统的鲁棒性,提出了一种特征映射和异构评估相结合的电路冗余容错设计方法。该方法基于硬件演化设计思路,结合了神经网络特征映射算法,通过分析演化得到优秀个体电路特征,并输入到神经网络进行映射聚类学习,然后通过聚类得到几类不同子电路的组集;设计了异构电路评估选择策略,研究了类间的异构度相关性,并在异构度大的类中选择最佳电路个体,作为子模块构建冗余电路系统。研究结果表明:构建的新系统在受扰或局部损伤的情况下,仍能保持较高的稳定性和容错能力;在单点短路、单点断路和随机故障仿真实验中,所设计的新异构冗余系统的最佳无故障率为92.5%,系统的平均无故障率为90.5%,比随机的冗余电路系统平均无故障率提高了6.5%,比同构冗余电路系统平均无故障率提高了26.5%,实验证明异构度高的冗余电路系统具有更好的容错抗扰能力和可靠性。研究结果为复杂电磁环境下鲁棒电路的研究提供了一种新的解决思路。
In order to improve the fault-tolerant immunity of the system in complex electromagnetic environment such as high voltage and strong electromagnetic, effectively reduce the failure rate and enhance the robustness of the system, a circuit redundancy fault tolerance combined with feature mapping and heterogeneous evaluation Design method Based on the idea of hardware evolution design and combining neural network feature mapping algorithm, this method obtains the excellent individual circuit characteristics through the analysis and evolution, and then inputs them into the neural network for mapping and clustering learning. Then, the clustering is used to obtain the group sets of several different subcircuits. Heterogeneous circuit evaluation and selection strategy is designed, the correlation between the heterogeneous degree of the class is studied, and the best circuit individual is selected in the class with large degree of heterogeneity as the sub-module to construct the redundant circuit system. The results show that the new system can still maintain high stability and fault tolerance under disturbed or localized damage. In the single point short circuit, single point open circuit and stochastic fault simulation experiments, the new system designed The best non-failure rate of the redundant system is 92.5%, the average failure-free rate of the system is 90.5%, which is 6.5% more than the average failure-free rate of the redundant system, Rate increased by 26.5%, experiments show that high degree of heterogeneous redundant circuit system has better fault tolerance and reliability. The results provide a new solution to the research of robust circuits in complex electromagnetic environment.