基于tsPSO算法的阵列自适应随机共振方法研究

RESEARCH ON ARRAY-ENHANCED SELF-ADAPTIVE STOCHASTIC RESONANCE BASED ON PARTICLE SWARM OPTIMIZATION

针对传统随机共振方法存在的单级自适应随机共振方法输出响应信噪比低、参数自适应时间长且阵列随机共振方法参数设置困难等不足,提出了一种基于带极值扰动的简化粒子群(Extremum Disturbed and Simple Particle Swarm Optimization,tsPSO)算法的阵列自适应随机共振方法,实现了强噪声背景下大参数微弱信号的有效、快速检测。首先,采用并联随机共振系统,通过对各子系统的输出响应进行自相关分析并合成提高最终输出响应的信噪比;其次,在每个并联子系统中,通过随机共振系统级联的方式进一步提高输出响应的信噪比;最后,以信噪比为适应度函数,对每个子系统的参数进行自适应选择,并通过变换尺度分段搜索和采用ts PSO算法缩短参数自适应的时间。仿真实验和工程应用结果验证了该方法的有效性。

Aiming at the disadvantages of traditional method of stochastic resonance（SR）,for example,the SNR of output response is low and the self-adaptive time of parameters is long for single self-adaptive SR,and the parameters are hard to set for array-enhanced SR,a array-enhanced self-adaptive SR based on extremum disturbed and simple particle swarm optimization（tsPSO） algorithm was proposed,which has realized an effective and fast detection of the weak signals under conditions of large parameters under high strong noise background. Firstly,the parallel SR system was adopted to enhance the SNR of the final output response by analyzing the output responses of sub-systems with the theory of auto correlation analysis and composing them.Secondly,in each sub-system in parallel,the cascaded SR system was adopted to enhance the SNR of output responses further.Finally,the parameters in each sub-systems in parallel were optimized with the SNR of output response as the fitness function,and the sectioning search and tsPSO algorithm were used to shorten the self-adaptive time of the parameters at the same time. The effectiveness of the method in the paper were proved by the results of simulation experiment and engineering application.