激波风洞测力信号的频域数据深度学习建模分析方法

Deep learning modeling analysis method of frequency-domain data ofshock wind tunnel force measurement signals

高精准度气动力测量是激波风洞试验中的关键技术。在开展测力试验时,测力系统在风洞流场起动瞬间的冲击激励下产生振动,但振动信号无法在较短的有效试验时间内快速衰减,导致天平输出信号中耦合了惯性干扰。基于深度学习技术,对激波风洞天平信号在频域内开展数据处理,并针对动态信号的频域特征进行卷积神经网络建模分析,旨在消除测力信号中的惯性干扰。在频域模型训练样本和验证样本的结果分析中,天平信号的大幅惯性振动干扰被消除,达到预期的结果,验证频域建模分析方法的有效性和可靠性。此外,对处理结果进行误差分析,进一步验证该方法在激波风洞天平数据处理中具有较好的工程应用价值。

High accuracy aerodynamic measurement is key technology in shock wind tunnel tests.During force-measuring tests,a force-measuring system vibrates under impact excitation at the moment to start wind tunnel flow field,this vibration signal can’t rapidly decay in shorter effective test time to cause coupling of inertial interference in output signal of shock wind tunnel balance.Here,based on deep learning technology,data processing was performed for shock wind tunnel balance signals in frequency domain,and the convolution neural network modeling analysis was performed for frequency domain characteristics of dynamic sample signals to eliminate inertial interference in force measurement signals.In results analysis of frequency domain model training samples and validation samples,large amplitude inertial vibration interference of balance signals was eliminated to obtain the expected results,and verify the effectiveness and reliability of frequency domain modeling analysis method.In addition,the error analysis for the processed results further verified that the proposed method has larger engineering application value in data processing of shock wind tunnel balance.