EMD时频分析拉曼光谱和近红外光谱

用时频方法分析拉曼光谱和近红外光谱。经验模态分解光谱成为特征模态分量,模态分量比重计算显示拉曼光谱能量均匀分布于各个分量,而近红外光谱的低阶特征模态分量只承载了较少的原光谱有效信息。真实光谱和数值实验均显示,经验模态分解视拉曼光谱为调幅信号,具有高频能量吸附特性; 视近红外光谱为调频信号,在一阶特征模态分量中可以较好实现高频窄带解调。一阶特征模态分量希尔伯特变换显示,经验模态分解拉曼光谱时易出现模态混叠现象。进一步在时频域分析玉米叶片近红外光谱,经验模态分解后截掉低能量的一、二阶分量,用剩余特征模态分量重构光谱信号,均方根误差为1.001 1,相关系数为0.981 3,两个指标反映出重构精度较高; 分解趋势项表明在近红外光波段,吸光度随着波长的减小呈现递增趋势; 特征模态分量的希尔伯特变换显示,657 cm-1是碱胁迫光谱特有频率,可作为碱胁迫光谱特征频率来辨识。     

基于滑动峭度相关性准则的局部特征尺度分解分量筛选方法

轴承故障振动信号具有非平稳、非线性特征,且可视为多个调幅-调频分量的叠加,单分量的包络蕴含了轴承的故障特征。局部特征尺度分解可将振动信号准确分解为多个内禀尺度分量之和,某些分量能清晰反映轴承的运行状态,根据包络谱可进行故障诊断。为了准确筛选有用分量,提出了基于滑动峭度相关性准则的分量筛选方法。首先,对信号进行局部特征尺度分解,得到若干个内禀尺度分量;然后,对分量和原始信号分别计算滑动峭度,生成时间序列;最后,依据分量滑动峭度序列与原始信号滑动峭度序列的互相关系数筛选有用分量。通过轴承内圈故障数据分析发现：有用分量与非有用分量之间的滑动峭度互相关系数比互相关系数差异明显,区分度更大,有益于分量的分类、筛选。     

变分模态分量降噪后重构的水声弱信号检测

针对能量检测法在低信噪比下对非合作水声探测信号的检测性能显著下降的问题,提出了一种组合变分模态分解和小波变换降噪重构的信号检测方法。以信号分解出的各个本征模态函数的近似熵与互相关系数比值作为分量分类参数,将所得分量分为信号分量、含噪信号分量与噪声分量,然后利用第二代小波变换对含噪信号分量降噪后与信号分量组成重构信号,最后对重构信号进行检测。数值仿真结果表明该方法可以在无先验信息的情况下对CW和LFM信号自适应降噪,信噪比0 dB以下时CW信号重构后信噪比提升约12 dB,宽带LFM信号信噪比提升约8～9 dB,有效提升了低虚警概率下信号的检测概率。湖试结果表明,虚警概率为0.1时检测概率可提升至0.9以上,验证了该方法的有效性。     

基于DE-LSSVM的滚动轴承故障诊断

针对滚动轴承故障诊断难以获得大量样本的问题以及LS-SVM 模型参数选择方法易陷入局部最优的缺点,提出了一种集合经验模态分解能量熵和差分进化算法（DE）优化最小二乘支持向量机相结合的轴承故障诊断方法。首先原始振动信号采用EEMD分解得到一组固有模态函数（IMF）,从有效本征模态函数IMF分量中提取的能量特征作为输入建立支持向量机,通过计算不同振动信号的能量熵值大小来判断轴承的故障损伤程度。为了提高模型的诊断精度,采用差分进化算法对LS-SVM的结构参数进行优化,并与LS-SVM和PSO-LSSVM模型相比较。结果表明,DE-LSSVM 模型的故障分类准确性得到了提高,可以有效应用于滚动轴承故障诊断中。     

利用经验模态分解的高频水声信号滤波方法*

研究了一种高频水声信号的滤波问题,提出了一种改进的经验模态分解加小波阈值滤波方法。首先对信号进行带通滤波处理,再进行经验模态分解,将分解得到的各个模态转换为频域信号,采用小波软阈值方法对这些频域信号进行滤波,最后对信号进行重构,并转换为时域信号。经数值仿真与试验数据验证表明此方法是可行有效的,与原基于经验模态分解的小波阈值滤波方法相比,本方法滤波效果较好:对不同输入信噪比的仿真信号进行滤波后,本方法的输出信噪比最大提高17.41 d B,滤波后所得信号与加噪前纯信号的相关系数最大提高0.90;对实验数据进行滤波后,不同时间段信号的相关系数最大提高0.62。     

基于经验模态分解消噪的光纤光栅解调系统

为了提高光纤Bragg光栅解调系统的解调准确度,提出利用经验模态分解对信号进行滤波分析和降噪处理的方法.该方法将经验模态分解得到的固有模态函数,分为信号分量起主导作用模态与噪音分量起主导作用模态,并利用反映信号主要结构的模态对信号进行重构实现去噪.实验表明,解调系统输出信号能够识别出中心波长的位置,精确得到Bragg波长的漂移量,输出谱失真小、信噪比高.对温度实验数据进行曲线拟合,拟合线性度为0.998,提高了系统的解调准确度.     

混沌海杂波背景下的微弱信号检测混合算法

基于经验模态分解理论, 提出了一种基于粒子群算法的支持向量机预测方法. 采用总体平均经验模式分解法将混沌信号分解为若干固有模态函数和趋势分量, 将复杂的非线性信号转化为具有不同尺度特征的平稳分量. 利用粒子群算法对支持向量机的惩罚系数和核函数进行优化, 结合支持向量机建立混沌序列的单步预测模型. 从预测误差中检测淹没在混沌背景中的微弱信号(包括瞬态信号和周期信号). 对Lorenz系统和实测IPIX雷达数据进行仿真实验, 结果表明, 该方法能够有效地从混沌背景噪声中检测出微弱目标信号, Lorenz系统得到的均方根误差0.000000339 (-102.8225 dB时)比传统支持向量机方法的均方根误差0.049 (-54.60 dB时)降低了5个数量级, 从海杂波中检测出具有谐波特性的微弱信号, 表明预测模型具有更低的门限和误差.     

核爆与雷电电磁脉冲识别

用经验模态分解和分形分析相结合的方法对核爆和雷电电磁脉冲信号进行了识别研究。计算了核爆和雷电电磁脉冲原始信号的盒维数,以及原始信号经验模态分解后前4阶固有模态分量（IMF）的盒维数。用最近邻法对核爆和雷电进行了识别,实验结果表明：原始信号经验模态分解后一阶、二阶IMF盒维数的识别率要略高于三阶、四阶IMF盒维数的识别率;原始信号盒维数的识别率高于经验模态分解后各阶IMF盒维数的识别率;二维和三维特征的识别率要高于一维特征的识别率,另外二维和三维特征的识别率更加趋于稳定,并且三维特征的识别率都在90%以上。     

基于IFD与DE-ELM的轴承故障诊断

针对局部均值分解（LMD）实现过程中存在的模态混淆现象和端点效应,影响识别精度的问题,提出了一种基于本征频率尺度分解（IFD）与差分进化极限学习机（DE-ELM）的方法。该方法将数字图像处理的频率分辨率概念与LMD结合起来;首先确定原始振动信号中的所有局部极值点的频率分辨率,将振动信号分为高频率分辨率区域和低频率分辨率区域;然后,构造本征频率尺度函数,将本征频率尺度函数添加到局部极值点低频率分辨率区域;最后,对添加本征频率尺度函数的原始振动信号进行LMD分解,在得到的乘积函数（PF）分量中剔除本征频率尺度函数,就可以得到突出原始信号振动特征的不同频率分辨率的PF分量,提取PF分量的特征参数构建特征向量作为DE-ELM的输入,进行故障类型识别。将该方法应用于轴承故障诊断,与LMD相比,故障识别精度提高了8.33%,表明了该方法的有效性与可行性。     

EEMD-ICA在功能性近红外光谱特征信号提取中的应用

近年来,功能性近红外光谱技术(fNIRS)广泛应用于神经影像学领域。为解决fNIRS特征信号提取中的信噪频谱混叠问题,依据近红外光谱脑功能成像信号非线性与非平稳特点,提出一种结合集合经验模态分解法和独立成分分析的多分辨率联合信号提取方法EEMD-ICA。在脑功能成像仪器平台上采集多通道多波长脑功能成像近红外光密度信号,先对该信号进行集合经验模态分解将其按频率成分分解为多层本征模态函数,之后将独立成分分析应用于目标频率分量函数进行自适应去噪,最后将处理后的分量累加、重构获得近红外光谱脑功能成像的特征信号。将Valsalva氏实验测试数据作为研究对象进行滤噪处理,与经验模态分解法和集合经验模态分解法对fNIRS特征信号的提取效果对比。对实测数据的处理结果进行信噪比和误差参数分析,结果表明,该方法能够有效解决去噪过程中丢失原始信号有用信息及由于信噪频谱混叠不能完整去除噪声的问题,信号处理效果理想,对比另外两种信号提取方法更为优化。     

超声振动对光学玻璃加工的锯切力特征研究

超声振动辅助方法已在各种硬脆性材料的加工工艺中得以应用,其优异的加工能力和效果已得到广泛证明。本研究中通过采集有无超声振动条件下锯切光学玻璃的平均锯切力以及单颗金刚石磨粒划擦实验下的力信号,对不同工艺条件下的平均锯切力、单颗磨粒受力特征进行分析。同时通过扫描电镜观察对应力信号下工件与工具加工后表面形貌,进一步通过超声振动下材料去除机理解释超声振动对锯切力影响。结果表明：与传统锯切工艺相比,超声振动辅助使得单颗磨粒划擦过程中的受力降低引起平均锯切力的降低;超声振动改变普通锯切下材料的去除方式;同时可使工具保持良好的锯切状态,降低光学玻璃材料的锯切力比,改善其可加工性。     

Autoresonant control of nonlinear mode in ultrasonic transducer for machining applications

Experiments conducted in several countries have shown that the improvement of machining quality can be promoted through conversion of the cutting process into one involving controllable high-frequency vibration at the cutting zone. This is achieved through the generation and maintenance of ultrasonic vibration of the cutting tool to alter the fracture process of work-piece material cutting to one in which loading of the materials at the tool tip is incremental, repetitive and controlled. It was shown that excitation of the high-frequency vibro-impact mode of the tool-workpiece interaction is the most effective way of ultrasonic influence on the dynamic characteristics of machining. The exploitation of this nonlinear mode needs a new method of adaptive control for excitation and stabilisation of ultrasonic vibration known as autoresonance. An approach has been developed to design an autoresonant ultrasonic cutting unit as an oscillating system with an intelligent electronic feedback controlling self-excitation in the entire mechatronic system. The feedback produces the exciting force by means of transformation and amplification of the motion signal. This allows realisation for robust control of fine resonant tuning to bring the nonlinear high Q-factor systems into technological application. The autoresonant control provides the possibility of self-tuning and self-adaptation mechanisms for the system to keep the nonlinear resonant mode of oscillation under unpredictable variation of load, structure and parameters. This allows simple regulation of intensity of the process whilst keeping maximum efficiency at all times. An autoresonant system with supervisory computer control was developed, tested and used for the control of the piezoelectric transducer during ultrasonically assisted cutting. The system has been developed as combined analog-digital, where analog devices process the control signal, and parameters of the devices are controlled digitally by computer. The system was applied for advanced machining of aviation materials.     

Theoretical and experimental investigations on rotary ultrasonic surface micro-machining of brittle materials

Many brittle materials, such as single-crystal materials, amorphous materials, and ceramics, are widely used in many industries such as the energy industry, aerospace industry, and biomedical industry. In recent years, there is an increasing demand for high-precision micro-machining of these brittle materials to produce precision functional parts. Traditional ultra-precision micro-machining can lead to workpiece cracking, low machined surface quality, and reduced tool life. To reduce and further solve these problems, a new micro-machining process is needed. As one of the nontraditional machining processes, rotary ultrasonic machining is an effective method to reduce the issues generated by traditional machining processes of brittle materials. Therefore, rotary ultrasonic micro-machining (RUμM) is investigated to conduct the surface micro-machining of brittle materials. Due to the small diameter cutting tool (<500 μm) and high accuracy requirements, the impact of input parameters in the rotary ultrasonic surface micro-machining (RUSμM) process on tool deformation and cutting quality is extremely different from that in rotary ultrasonic surface machining (RUSM) with relatively large diameter cutting tool (∼10 mm). Up till now, there is still no investigation on the effects of ultrasonic vibration (UV) and input variables (such as tool rotation speed and depth of cut) on cutting force and machined surface quality in RUSμM of brittle materials. To fill this knowledge gap, rotary ultrasonic surface micro-machining of the silicon wafer (one of the most versatile brittle materials) was conducted in this study. The effects of ultrasonic vibration, tool rotation speed, and depth of cut on tool trajectory, material removal rate (MRR), cutting force, cutting surface quality, and residual stress were investigated. Results show that the ultrasonic vibration could reduce the cutting force, improve the cutting surface quality, and suppress the residual compressive stress, especially under conditions with high tool rotation speed.     

Ultrasonically assisted turning of aviation materials: simulations and experimental study

Ultrasonically assisted turning of modern aviation materials is conducted with ultrasonic vibration (frequency f approximately 20 kHz, amplitude a approximately 15 microm) superimposed on the cutting tool movement. An autoresonant control system is used to maintain the stable nonlinear resonant mode of vibration throughout the cutting process. Experimental comparison of roughness and roundness for workpieces machined conventionally and with the superimposed ultrasonic vibration, results of high-speed filming of the turning process and nanoindentation analyses of the microstructure of the machined material are presented. The suggested finite-element model provides numerical comparison between conventional and ultrasonic turning of Inconel 718 in terms of stress/strain state, cutting forces and contact conditions at the workpiece/tool interface.     

Longitudinal ultrasonic vibration assisted guillotining of stacked paper

Ultrasonic vibration assisted cutting is a complex process with high dynamics. The interaction between cutting tool and workpiece is of key interest to understand the entire process. Experimental investigations are limited by the dynamics of the measurement system, and thus appropriately modeling of the ultrasonic vibration assisted cutting process is essential. In this investigation, a dynamic model regarding the ultrasonic vibration assisted guillotining of stacked paper sheets is developed. A Kelvin–Voigt material model, representing the individual sheets, is chosen, with its stiffness and damping parameters being empirically determined. A novel measurement strategy for studying the contact time and interaction between cutting tool and workpiece is introduced. It allows the verification of the highly dynamic behavior of the developed model. With the dynamic model, the experimentally observed cutting forces can be calculated. It is found that the dynamic forces cause a quicker failure of the material, which leads to a lower compression of the stack prior to reaching the critical cutting force.     

阵列式压电振子超声复合喷丸强化*

为了进一步改善超声喷丸强化质量,提升工件表面残余应力和显微硬度的均匀性,对压电振子阵列型超声喷丸强化加工系统进行了改进,提出了在强化加工过程中辅助工件往复运动的方法。对改进后的压电振子阵列型超声喷丸强化加工原理进行了阐述,分析了激振片的振动特性并进行了结构尺寸优化,应用Abaqus对工件往复运动下超声喷丸强化过程进行了仿真研究,分析了在不同的工件往复运动频率下,工件表面残余应力分布情况。实验测试了在不同喷丸强化时间及工件往复运动频率下超声喷丸强化加工质量,探究了喷丸时间和工件往复运动频率对工件表面显微硬度的影响。结果表明：在压电振子阵列型超声喷丸强化过程中辅助工件往返运动,可以有效提高工件表面残余应力和显微硬度分布的均匀性,工件表面强化质量得到提升。     

噪声模态单元预判的经验模态分解脉冲星信号消噪

为了改善脉冲星辐射脉冲信号的消噪效果, 提出了一种基于噪声模态单元预判的经验模态分解(EMD) 消噪声方法. 该方法首先利用EMD将含噪辐射脉冲信号分解为一组内蕴模态函数(IMF), 根据IMF系数的统计特性采用局部均方误差准则进行噪声模态单元预判, 并将噪声模态单元置零; 然后对噪声模态单元预判处理后的IMF以模态单元为基本单位进行最优比例萎缩消噪, 从而达到抑制噪声、保留信号的目的. 实验结果表明: 与Sure Shrink小波阈值法、Bayes Shrink小波阈值法和EMD模态单元比例萎缩法相比, 基于噪声模态单元预判的EMD消噪方法可以更有效地去除脉冲辐射信号中的噪声, 同时更好地保留信号突变处的细节信息特征, 在信噪比、 均方误差、峰值相对误差、峰位误差和相位误差等方面都有一定程度的改善. 关键词： 脉冲星信号消噪 经验模态分解 噪声模态单元预判 局部均方误差     

ESTIMATED TEMPERATURE ON A MACHINED SURFACE USING AN INVERSE APPROACH

An experiment devoted to the heat flux estimation in a workpiece during a machining process by turning is presented. The method is based on temperature measurement from thermocouples embedded in the workpiece, close to the heated surface. A model that expresses the heat flux according to the temperature at the sensors is developed. The stationary and linearity assumptions are used in order to decompose the three-dimensional original problem into two bi-dimensional problems. This decomposition can be realized given the difference between the cutting speed and feed velocity in two orthogonal directions. The temperature on the machined surface is calculated from the estimated heat flux and the heat transfer model in the workpiece. The application concerns hard steel machining, using a CBN insert tool. Three parameters are placed into evidence from this application: the temperature magnitude on the machined surface, the thermal gradient in the workpiece, and the `thermal persistence' that represent the heating time of the machined surface. This study leads to a better understanding of the influence of temperature during a hard steel turning process.     

Ultrasonic excitation affects friction interactions between food materials and cutting tools

In the food industry, ultrasonic cutting is used to improve separation by a reduction of the cutting force. This reduction can be attributed to the modification of tool–workpiece interactions at the cutting edge and along the tool flanks because of the superposition of the cutting movement with ultrasonic vibration of the cutting tool. In this study, model experiments were used to analyze friction between the flanks of a cutting tool and the material to be cut. Friction force at a commercial cutting sonotrode was quantified using combined cutting–friction experiments, and sliding friction tests were carried out by adapting a standard draw-off assembly and using an ultrasonic welding sonotrode as sliding surface. The impact of material parameters, ultrasonic amplitude, and the texture of the contacting food surface on friction force was investigated. The results show that ultrasonic vibration significantly reduces the sliding friction force. While the amplitude showed no influence within the tested range, the texture of the contact surface of the food affects the intensity of ultrasonic transportation effects. These effects are a result of mechanical interactions and of changes in material properties of the contact layer, which are induced by the deformation of contact points, friction heating and absorption heating because of the dissipation of mechanical vibration energy.     

Numerical study on the splitting of a vapor bubble in the ultrasonic assisted EDM process with the curved tool and workpiece

Electrical discharge machining (EDM) is a powerful and modern method of machining. In the EDM process, a vapor bubble is generated between the tool and the workpiece in the dielectric liquid due to an electrical discharge. In this process dynamic behavior of the vapor bubble affects machining process. Vibration of the tool surface affects bubble behavior and consequently affects material removal rate (MRR). In this paper, dynamic behavior of the vapor bubble in an ultrasonic assisted EDM process after the appearance of the necking phenomenon is investigated. It is noteworthy that necking phenomenon occurs when the bubble takes the shape of an hour-glass. After the appearance of the necking phenomenon, the vapor bubble splits into two parts and two liquid jets are developed on the boundaries of the upper and lower parts of the vapor bubble. The liquid jet developed on the upper part of the bubble impinges to the tool and the liquid jet developed on the lower part of the bubble impinges to the workpiece. These liquid jets cause evacuation of debris from the gap between the tool and the workpiece and also cause erosion of the workpiece and the tool. Curved tool and workpiece affect the shape and the velocity of the liquid jets during splitting of the vapor bubble. In this paper dynamics of the vapor bubble after its splitting near the curved tool and workpiece is investigated in three cases. In the first case surfaces of the tool and the workpiece are flat, in the second case surfaces of the tool and the workpiece are convex and in the third case surfaces of the tool and workpiece are concave. Numerical results show that in the third case, the velocity of liquid jets which are developed on the boundaries of the upper and lower parts of the vapor bubble after its splitting have the highest magnitude and their shape are broader than the other cases.