【作者】 王颖；

【导师】 刘正先；

【作者基本信息】 天津大学 ， 力学， 2022， 硕士

【摘要】 以轴流压气机为代表的流体机械是航空发动机、燃气轮机等旋转动力装置的核心部件。随着动力装置的不断更新升级,设计高性能、高稳定性压气机的需求也日益迫切。然而,以旋转失速为代表的压气机流动失稳问题已成为制约高性能、高稳定性压气机研制的主要障碍。当旋转失速发生时,压气机的气动性能(压比和效率)迅速下降,并极易引发低频大振幅的喘振现象,使得整个旋转动力装置的性能和稳定性急剧降低、进而导致动力装置无法正常工作甚至损毁。事实上,压气机在邻近旋转失速时,内部已经呈现出三维强非线性流动特性,理论模型与数值模拟通常不容易客观真实地还原流动现象,而实验测量是直接获取近失速流动特征最可靠的方法。深入分析压气机节流过程静压信号演化规律,不仅能够加深对轴流压气机失速诱发机制的认识,同时为失速先兆实时监测、失速预警和主动控制等方法的构建提供理论基础。本文针对某低速轴流压气机节流过程静压实验数据(共八组),从以下三个方面逐步深入研究了前失速信号的特征。首先,利用经典频谱分析方法(傅里叶分析和小波谱方法)探究了压气机失速阶段流动特征,并对比分析了不同方法的特点。发现,傅里叶分析可以获得压气机的基本流动结构并识别宽频多尺度扰动结构,但无法描述突尖失速的演化过程。而小波谱方法可以刻画失速瞬时演化过程,并识别失速先兆起始位置、失速团个数、传播速度、及其覆盖通道范围等关键失速特征。上述方法虽然成功挖掘了失速阶段物理特征,但以失速演化特征为依据实施失速主动控制为时已晚,必须开展前失速阶段流动不稳定性演化规律研究。进一步,基于时延互相关分析方法和扰动相位度量法研究了前失速阶段流动不稳定性演化规律。传统的互相关系数分析法应用于前失速阶段流动不稳定性探究时,存在初始相位不一致的问题,因而无法准确描述真实的物理流动特征。为此,发展了时延互相关分析方法,有效地消除了周向空间位置造成的信号时间滞后相位差,避免了周向空间位置的非物理干扰,更好地呈现压气机前失速流动状态的微观物理特征。为了监测信号间的相位变化,提出了扰动相位度量法,依据扰动相位的时空特征图像,发现压气机几何存在周向不均现象。基于最大扰动相位持续时间,分析了前失速阶段互相关系数阶段性变化机理和流动不稳定性特征,进一步确认相位差是导致互相关系数变化的原因,发现前失速流动稳定性呈现阶段性突变的特征。最后,基于感应学习算法对压气机失速信号进行预测,以期发展一种既能衡量前失速流动稳定性,还能阐明突尖失速时空演化机制的方法。利用感应学习算法对解析函数进行预测,发现传统时间序列预测方法存在误差叠加和虚假高精度问题。为此,提出了一种离散映射方法,大幅提高了预测精度。但该方法仍难以适用于强非线性的失速信号时间维度预测。因此,考虑建立一种可行的空间失速信号预测方法,并成功预测了失速信号演化趋势,证实了在近失速阶段就已经存在具有周向传播特征的小扰动结构。利用全时域权值系数时空图像研究了前失速流动演化规律,发现权值系数极值特征存在阶段性变化的性质。这一现象与扰动相位结果一致,表明前失速流动具有与突尖失速起始类似的阶跃式突变特征。更多还原

【Abstract】 Fluid Machinery represented by axial compressors is the core components of rotating power plants such as aero-engines and gas turbines.With the continuous updating and upgrading of power plants,the demand for designing high-performance and high-stability compressors is becoming more and more urgent.However,the problem of compressor flow instability represented by rotating stall has become the main obstacle restricting the development of high-performance and high-stability compressors.When rotating stall occurs,the aerodynamic performances(pressure ratio and efficiency)of the compressor decrease rapidly.This is easy to cause a surge phenomenon with low frequency and large amplitude,which drastically reduces the performance and stability of the entire rotating power plant,and even be damaged.In fact,when the compressor is close to the rotating stall,the internal flow of the compressor already exhibits three-dimensional strong nonlinear characteristics.The theoretical model and numerical simulation are usually not easy to objectively and truly restore the flow phenomenon,while the experimental measurement is the most reliable way to directly obtain the flow characteristics near the stall.In-depth analysis of the static pressure signal evolution in throttling process can not only deepen the understanding of stall inception mechanism of axial flow compressor,but also provide a theoretical basis for the construction of stall inception real-time monitoring,stall warning and active control methods.According to the static pressure experimental data of a low-speed axial compressor during throttling process,the characteristics of the front stall signal are gradually and deeply studied from the following three aspects.Firstly,the classical spectral analysis methods(Fourier analysis and wavelet spectral method)were used to explore the flow characteristics of the compressor in the stall stage,and the characteristics of different methods were compared and analyzed.It is found that Fourier analysis can obtain the basic flow structures of the compressor and identify the broadband multi-scale disturbance structures,but cannot describe the evolution process of the spike stall.The wavelet spectrum method can describe the instantaneous evolution process of stall,and identify the key stall characteristics such as the moment of stall inception,the number of stall cells,the propagation speed,and the channels’ range of their coverages.Although the above methods have successfully excavated the physical characteristics of the stall stage,it is too late to implement the active stall control based on the stall evolution characteristics,and it is necessary to carry out research on the evolution law of flow instability in the front stall stage.Furthermore,the evolution law of flow instability in the front stall stage is studied based on the time-delay cross-correlation analysis method and the disturbance phase measurement method.When the traditional cross-correlation coefficient analysis is applied to the investigation of flow instability in the front stall stage,there is a problem of inconsistency in the initial phase,and it cannot accurately describe the real physical flow characteristics.Therefore,a time-delay cross-correlation analysis method is developed.It effectively eliminates the time lag phase difference of the signal caused by the circumferential spatial position.It also avoids the non-physical interference of the circumferential spatial position,and better presents the microphysical characteristics of the compressor’s flow before entering the stall state.In order to monitor the phase change between the signals,a perturbation phase measurement method is proposed.According to the spatiotemporal characteristic image of the perturbation phase,it is found that the compressor’s geometry has circumferential unevenness.Based on the maximum disturbance phase duration,the phase change mechanism of the cross-correlation coefficient and the characteristics of flow instability in the front stall stage are analyzed.It is further confirmed that the phase difference is the cause of the change of the cross-correlation coefficient,and it is found that the stability of the flow in the front stall state is characterized by a sudden stage change.Finally,the compressor stall signal is predicted based on the induction learning algorithm,to develop a method that can not only measure the stability of the front stall flow,but also elucidate the spatiotemporal evolution mechanism of the spike stall.Using the GMDH algorithm to predict the analytic function,it is found that the traditional time series prediction method has the problems of error superposition and false high precision.Therefore,a discrete mapping method is proposed,which greatly improves the prediction accuracy.However,this method is still difficult to apply to the prediction of the time dimension of the stall signal with strong nonlinearity.Therefore,a feasible method for spatial stall signal prediction is considered,and the evolution trend of the stall signal is successfully predicted,which confirms the existence of small perturbations with circumferential propagation characteristics in the near-stall stage.The evolution law of the front stall flow is studied by using the spatiotemporal images of the weight coefficients in the full-time domain,and it is found that the extreme value characteristics of the weight coefficients have the nature of stage changes.This phenomenon is consistent with the perturbation phase results,indicating that the front stall flow has a stage change characteristic,which is similar to the spike stall inception.更多还原