基于RBF神经网络和MIGA的液压锥阀降噪研究  

Noise reduction of hydraulic cone valve based on RBF neural network and MIGA

作　　者：王华伟 周鑫 王博[1] 胡溧[1] WANG Hua-wei;ZHOU Xin;WANG Bo;HU Li(School of Automotive and Traffic Engineering,Wuhan University of Science and Technology,Wuhan 43000,China)

机构地区：[1]武汉科技大学汽车与交通工程学院,湖北武汉430000

出　　处：《机电工程》2022年第11期1527-1534,共8页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(51905389);湖北省教育厅科学技术研究项目(B2020002)。

摘　　要：液压锥阀在气液两相流状态下工作时会产生剧烈的噪声,严重影响锥阀的工作性能及其工作环境,针对这一问题,提出了一种基于径向基函数(RBF)神经网络和多岛遗传算法(MIGA)的方法,对液压锥阀的结构参数进行了优化。首先,采用有限元软件分析了影响锥阀流场及声场的结构参数;然后,以阀芯半锥角角度、喉部长度和阀芯入口角度这3个参数为优化变量,以加权平均噪声最小和加权最大噪声最小为优化目标,通过最优拉丁超立方设计方法确定了样本数据;最后,采用了RBF神经网络方法,建立了锥阀结构参数与噪声关系的近似模型,利用多岛遗传算法对近似模型进行了优化;根据得到的最优参数建立了锥阀优化模型,并进行了声学特性分析。研究结果表明:与原模型相比,优化模型的平均噪声降低23.846 dB,最大噪声降低5.092 dB;该结果验证了基于RBF神经网络和MIGA优化方法的有效性,可为液压锥阀的进一步降噪研究提供理论支持。Hydraulic cone valve will produce severe noise when it works in the state of gas-liquid two-phase flow,which seriously affects the working performance and working environment of the cone valve.Aiming at this problem,a method based on radial basis function(RBF)neural network and multi-island genetic algorithm(MIGA)was proposed to optimize the structural parameters of hydraulic cone valves.Firstly,the structural parameters affecting the flow field and sound field of the cone valve were analyzed by the finite element analysis software.Then,the half cone angle,throat length and inlet angle of the valve core were taken as the optimization variables,and the noise including weighted average noise and the weighted maximum noise were used as the optimization objectives.The optimal Latin hypercube test method was used to design the sample points.Finally,RBF neural network method was used to establish the approximate model of the relationship between cone valve structural parameters and noise,and the approximate model was optimized by the multi-island genetic algorithm.The optimal parameters were used to establish the optimization model of cone valve,and the acoustic characteristics of that model were analyzed.The results show that compared with the original model,the average noise of the optimized model is reduced by 23.846 dB and the maximum noise is reduced by 5.092 dB,thus the effectiveness of the proposed optimization method is verified.The research results can provide theoretical support for the noise reduction research of hydraulic cone valve.

关 键 词：液压控制阀 锥阀噪声抑制 径向基函数神经网络 多岛遗传算法 锥阀结构参数 声学特性分析 最优拉丁超立方 

分 类 号：TH137.52[机械工程—机械制造及自动化]