基于SA-GA自适应压缩感知的风力机齿轮振动信号压缩与重构方法  

Vibration Signal Compression and Reconstruction Method for Wind Turbine Gearboxes Based on SA-MPGA

作　　者：周爽 孟凡勇 许璇 吴越 Shuang Zhou;Fan-yong Meng;Xuan Xu;Yue Wu(School of Instrument Science and Opto-Electronics Engineering,Beijing Information Science and Technology University)

机构地区：[1]北京信息科技大学仪器科学与光电工程学院

出　　处：《风机技术》2024年第2期74-80,共7页Chinese Journal of Turbomachinery

基　　金：北京市教育委员会科学研究计划项目资助(KM202211232014)。

摘　　要：为应对风力机齿轮箱振动信号压缩与重构过程存在复杂的参数设置问题,提出了基于模拟退火多种群遗传算法(Simulating Annealing and Multiple Population Genetic Algorithm,SA-MPGA)自适应设置过完备学习字典生成、振动信号压缩、压缩信号重构过程所需参数集。在传统遗传算法基础上引入多种群思想,增加了遗传算法对解空间的覆盖。在种群繁衍时个体选择引入模拟退火策略在种群进化过程中以不同概率接受一定程度的劣解,从而有助于遗传算法跳出局部最优解的缺陷。基于SA-MPGA的多参数自适应选择降低了传统遗传算法容易收敛到局部最优解的概率。应用实际工程数据验证基于SA-MPGA多参数优化问题,实验结果表明,在保持压缩率的前提下,基于模拟退火多种群算法比基于遗传算法重构信号与原始信号的峰值信噪比提升了16.5%,相关性提升了12.5%,均方根误差降低了13.4%。In response to the complex parameter configuration challenges encountered in the compression and reconstruction process of vibration signals from wind turbine gearbox,an adaptive parameter set is proposed based on the Simulated Annealing and Multiple Population Genetic Algorithm(SA-MPGA).This set encompasses the requirements for overcomplete dictionary generation,signal compression,and compressed signal reconstruction.Building upon the foundation of traditional genetic algorithms,the introduction of a multi-population concept enhances the genetic algorithm's coverage of the solution space.During population evolution,individuals are selected with the incorporation of simulated annealing strategies,enabling the acceptance of suboptimal solutions to varying degrees.This feature contributes to the ability of the genetic algorithm to overcome the limitations associated with local optima.The self-adaptive parameter selection,facilitated by SA-MPGA,reduces the likelihood of traditional genetic algorithms converging to local optima.Experimental validation with real-world engineering data demonstrates that,while maintaining a consistent compression rate,the SA-MPGA-based approach outperforms the Genetic Algorithm-based approach by improving the peak signal-to-noise ratio of the reconstructed signal compared to the original signal by 16.5%,increasing correlation by 12.5%,and reducing the root-mean-square error by 13.4%.

关 键 词：遗传算法 超参数优化 压缩感知 退火算法 振动信号压缩 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] TM315[自动化与计算机技术—控制科学与工程]