摘要
为缩短自通风牵引电机风扇噪声的预测周期,同时保持较高的预测精度,引入代理模型技术,基于电机稳态流场结果快速预测电机风扇噪声的声功率级。首先通过网格无关性验证和风量测试校核自通风电机稳态流场仿真方法,并开展15种电机/风扇结构、多转速下共68组稳态流场计算;然后依据GB/T 25123.2-2018和GB/T 25123.4-2015测试电机相应工况下的噪声声功率级;并通过相关性分析和显著性检验从8种稳态流场数据结果中选出7种参数作为输入,以电机噪声声功率级作为输出,建立初始样本集;最后建立样本筛选规则,选出61组样本建立组合代理模型。经分析,建立的代理模型决定系数为0.993,交叉验证结果显示最大绝对误差为3.0 dB(A),平均绝对误差为1.0 dB(A),满足工程应用要求,可大幅降低电机离心风扇噪声预测周期。
In order to shorten the prediction period of fan noise with self-ventilated traction motor and maintain high prediction accuracy,surrogate model technology is introduced to predict the sound power level of the motor fan noise based on the steady-state flow field of the motor.Firstly,the steady-state flow field simulation method of self-ventilated motor is verified by grid independence verification and air flow test,and a totally 68 groups of steady-state flow fields of 15 sorts of motor/fan structures are calculated under multiple speeds.Then,according to the standards of GB/T 25123.2-2018 and GB/T 25123.4-2015,the sound power level of the motor noise under corresponding working conditions is tested and measured.Through correlation analysis and significance test,seven parameters are selected from 8 steady-state flow field data results and used as the input of the model,and the sound power level of motor noise is taken as output to establish the initial sample set.Finally,the sample selection rules are established,and 61 groups of samples are selected to establish the ensemble surrogate model.After the analysis,the determination coefficient of the established surrogate model is found to be 0.993.The cross-validation results show that the maximum absolute error is 3.0 dB(A),and the average absolute error is 1.0 dB(A),which meets the requirements of engineering application,while the noise prediction cycle of the motor centrifugal fan is greatly reduced.
作者
张伟
康炜
庞聪
刘永强
贾喜勤
王文庆
ZHANG Wei;KANG Wei;PANG Cong;LIU Yongqiang;JIA Xiqin;WANG Wenqing(CRRC Yongji Electric Co.,Ltd.,Xi′an 710016,China)
出处
《噪声与振动控制》
CSCD
北大核心
2023年第6期173-177,195,共6页
Noise and Vibration Control
基金
中国中车重点项目资助(2020CYB064)。
关键词
声学
自通风牵引电机
离心风扇
气动噪声
代理模型
快速预测
acoustics
self-ventilated traction motor
centrifugal fan
aerodynamic noise
surrogate model
fast prediction
