WO2019062197A1 - 一种新能源车电机噪声信号提取方法及系统 - Google Patents

一种新能源车电机噪声信号提取方法及系统 Download PDF

Info

Publication number
WO2019062197A1
WO2019062197A1 PCT/CN2018/090214 CN2018090214W WO2019062197A1 WO 2019062197 A1 WO2019062197 A1 WO 2019062197A1 CN 2018090214 W CN2018090214 W CN 2018090214W WO 2019062197 A1 WO2019062197 A1 WO 2019062197A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
fourier transform
noise
fractional fourier
energy vehicle
Prior art date
Application number
PCT/CN2018/090214
Other languages
English (en)
French (fr)
Inventor
赵永吉
孙亚轩
Original Assignee
比亚迪股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 比亚迪股份有限公司 filed Critical 比亚迪股份有限公司
Publication of WO2019062197A1 publication Critical patent/WO2019062197A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01HMEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H17/00Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208Noise filtering
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208Noise filtering
    • G10L21/0216Noise filtering characterised by the method used for estimating noise
    • G10L21/0232Processing in the frequency domain
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0272Voice signal separating

Definitions

  • the invention relates to the field of noise signal extraction of new energy vehicles.
  • the high frequency noise of the motor is a result of noise synthesis, including mechanical noise, electromagnetic noise and air noise.
  • the frequency is from 1KHz to 12KHz or higher. This high frequency electromagnetic noise will bring people Strong discomfort. Therefore, the management of such noise is very necessary.
  • noise reduction schemes There are two main types of existing noise reduction schemes, one is passive noise reduction, also called physical noise reduction. Including structural optimization, eliminating resonance, damping materials for sound absorption and so on.
  • active noise reduction including active noise reduction and masking effects, as well as other sound compensation measures.
  • the target noise signal needs to be separately extracted.
  • the existing common signal extraction method can not achieve the extraction or extraction effect of the new energy vehicle motor noise signal is not good, and the special commercial software extraction method, although the effect is better, but The process is complicated and inconvenient for testing and research.
  • the present invention provides a method and system for extracting noise signals of new energy vehicles.
  • An aspect of the present invention provides a method for extracting a noise signal of a new energy vehicle motor, comprising the following steps:
  • the acoustic environment signal including a motor noise signal
  • the method for extracting noise signal of a new energy vehicle motor disclosed in the embodiment of the present invention is more convenient for processing the subsequent fractional Fourier transform by extracting the primary signal including the motor noise signal and the background signal; Fourier transform Analyze the characteristics of linear and stationary signals with better performance. Therefore, the P-order fractional Fourier transform is performed on the primary signal, and the time domain of the primary signal is transformed into the frequency domain by using the Fourier transform. The characteristic frequency and phase information are more easily seen, and the spectrum of the primary signal is extracted for analysis. Then, through the masking process, the spike signal is obtained.
  • the background signal is filtered by the P-order fractional Fourier transform signal, and only the signal of the motor noise signal after the P-order fractional Fourier transform is retained.
  • the P-order fractional Fourier inverse transform process is performed on the spike signal to obtain a motor noise signal as a target signal.
  • the method for extracting noise signal of the new energy vehicle motor disclosed in the present invention has a lower calculation amount and complexity, and is simpler to implement. The effect is very good enough to achieve the same effect as the commercial software extraction method, and can meet the standards of the applicant for research and testing. The cost is lower.
  • a second aspect of the present invention provides a new energy vehicle motor noise signal extraction system, comprising the following modules:
  • the acoustic environment signal acquisition module is configured to collect an acoustic environment signal of the new energy vehicle, and the acoustic environment signal includes a motor noise signal;
  • a primary signal extraction module configured to extract, from the acoustic environment signal, a primary signal including a motor noise signal and a background signal;
  • a fractional Fourier transform module configured to perform a P-order fractional Fourier transform process on the primary signal to obtain a secondary signal; wherein P is a time-frequency domain rotation coefficient;
  • a masking processing module configured to perform a masking process on the secondary signal to obtain a spike signal
  • a fractional-order inverse Fourier transform module configured to perform P-order fractional Fourier inverse transform processing on the spike signal to obtain a target signal.
  • the new energy vehicle motor noise signal extraction system disclosed in the embodiment of the present invention extracts the acoustic environment signal through the acoustic environment signal acquisition module, and extracts the motor noise signal and the background by the primary signal extraction module to facilitate subsequent fractional Fourier transform processing.
  • the primary signal of the signal; the Fourier transform analyzes the linear, stationary signal for better performance. Therefore, the P-order fractional Fourier transform is performed on the primary signal by the fractional Fourier transform module, and the time domain of the primary signal is transformed into the frequency domain by using the Fourier transform, and the characteristic frequency and phase information are more easily seen.
  • the spectrum from which the primary signal was extracted has been analyzed. Then, the occlusion processing is performed by the occlusion processing module to obtain a spike signal.
  • the background signal is filtered by the P-order fractional Fourier transform signal, and only the motor noise signal is retained by the P-order fractional Fourier transform. signal.
  • the P-order inverse Fourier transform process is performed on the spike signal by the fractional Fourier inverse transform module to obtain the motor noise signal as the target signal.
  • FIG. 1 is a flow chart for extracting a noise signal of a new energy vehicle motor provided in an embodiment of the present invention
  • FIG. 2 is a schematic diagram of a fractional Fourier domain formed by rotating a p ⁇ /2 counterclockwise rotation of a coordinate axis around an origin in a time-frequency domain according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of simulation of a chirp signal provided by a specific embodiment of the present invention in a MATLAB after fractional Fourier transform;
  • FIG. 4 is a schematic diagram of a time-frequency signal after the constant frequency howling noise extraction of the motor provided in the embodiment of the present invention
  • FIG. 5 is a schematic diagram of a time-frequency signal obtained by extracting a constant frequency howling noise of a motor provided by an existing commercial software
  • FIG. 6 is a schematic diagram of signal waveforms after extraction of constant frequency howling noise of a motor provided in an embodiment of the present invention
  • FIG. 7 is a schematic diagram of signal waveforms after extraction of a constant frequency howling noise extracted by a conventional commercial software
  • FIG. 8 is a schematic diagram of a time-frequency signal after the motor frequency conversion howling noise is extracted according to an embodiment of the present invention
  • FIG. 9 is a schematic diagram of a time-frequency signal after the motor frequency conversion howling noise is extracted by the existing commercial software
  • FIG. 10 is a schematic diagram of signal waveforms after motor frequency whistle noise extraction according to an embodiment of the present invention.
  • 11 is a schematic diagram of signal waveforms after motor frequency conversion howling noise extraction provided by the existing commercial software
  • FIG. 12 is a block diagram of an extraction system for a noise signal of a new energy vehicle motor provided in an embodiment of the present invention.
  • acoustic environment signal acquisition module 1, primary signal extraction module; 3, fractional Fourier transform module; 4, isolation processing module; 5, fractional Fourier inverse transformation module.
  • the embodiment provides a method for extracting noise signal of a new energy vehicle motor, as shown in FIG. 1 , comprising the following steps:
  • Step S1 Acoustic environment signal acquisition step: collecting an acoustic environment signal of a new energy vehicle, the acoustic environment signal including a motor noise signal;
  • Step S2 a primary signal extraction step: extracting a primary signal including a motor noise signal and a background signal from the acoustic environment signal;
  • Step S3 a fractional Fourier transform step: performing P-order fractional Fourier transform processing on the primary signal to obtain a secondary signal; wherein P is a time-frequency domain rotation coefficient;
  • Step S4 a masking processing step: performing a masking process on the secondary signal to obtain a spike signal;
  • Step S5 a fractional Fourier inverse transform step: performing P-order fractional Fourier inverse transform processing on the spike signal to obtain a target signal.
  • the step S1 specifically includes the following steps:
  • the so-called acoustic environment refers to the system composed of all sounds in a certain area.
  • the acoustic environment of the so-called new energy vehicle refers to the internal environment of the new energy vehicle.
  • the space in which the driver and the passenger are located such as in the cab (or in the passenger's cab, is equivalent), in the motor compartment, and the like.
  • the acoustic environment signals in this example include noise signals (target signals) and other signals in the acoustic environment; noise in the acoustic environment refers to noise that can be felt by the human body in the acoustic environment.
  • environmental noise includes high frequency noise originating from the motor, which is called motor noise.
  • motor noise high frequency noise originating from the motor
  • the frequency of the howling sound can be generally divided into two categories, one is frequency. Unchanged, we call it constant frequency howling, and the other type is frequency conversion. We call it frequency conversion howling. Both of the above whistle are high frequency motor noise.
  • the main target object is motor noise.
  • the purpose of the present application is to extract the motor noise signal of the new energy vehicle. In order to enable the applicant to carry out research and experiments on active noise reduction.
  • the collection of high-frequency noise of the motor must be precise.
  • other non-motor high-frequency noises such as body friction noise, horn sound, and exterior noise are collected. Therefore, it is preferred to collect the motor noise signal directly in the vicinity of the motor (in the motor compartment where the motor is arranged), thereby ensuring the integrity of the collected motor noise signal and eliminating uncertainties such as attenuation caused by noise during propagation. And to prevent other non-motor noise from interfering.
  • the noise generated by the motor can be collected in real time by a sound signal receiver such as a microphone or the like.
  • step S2 is specifically: extracting the primary signal from the acoustic environment signal by spectrum analysis; and the general extraction method is filtering.
  • the howling signal and frequency of the motor have a certain relationship, and the preprocessing (generally filtering) is performed according to the frequency of the howling signal.
  • the frequency of the howling signal is between 3500 Hz and 4500 Hz.
  • MATLAB is a commercial mathematics software produced by MathWorks, USA, advanced technical computing language and interactive environment for algorithm development, data visualization, data analysis, and numerical calculation; etc.; well known to those skilled in the art
  • the signal is simulated to extract the above primary signal including the motor noise signal.
  • the primary signal x(t) is represented by the following function:
  • the s(t) is a motor noise signal
  • the w(t) is a background signal.
  • the above motor noise signal s(t) is the target signal that we need to extract.
  • the background signal is a signal other than the other motor noise signals, which is the signal we need to filter out.
  • the background signal typically includes Gaussian white noise. Since the Fourier transform has excellent performance when analyzing linear and stationary signals. However, it is not effective for non-stationary signals, such as when the signal changes drastically. Therefore, the extracted primary signal x(t) is preferably a chirp signal mixed with Gaussian white noise.
  • the motor noise signal s(t) is a chirp signal, and its function is expressed as:
  • a 0 is the envelope function
  • m 0 is the linear modulation frequency
  • f 0 is the initial frequency
  • the Fourier transform is not seen in the prior art as a solution for motor noise signal extraction, and specific Fourier transform processing methods are well known in the art.
  • the fractional Fourier transform which is a class of Fourier transforms, is also not used as motor noise signal extraction, and its processing methods are also well known in the art.
  • the present application creatively combines a fractional Fourier transform, a masking process, and a fractional Fourier inverse transform to extract motor noise signals.
  • the so-called P-order fractional Fourier transform refers to the fractional Fourier transform using the time-frequency domain rotation coefficient P.
  • Fig. 3 is a schematic diagram of the simulation of the chirp signal in MATLAB after fractional Fourier transform. It can be seen that the vertical direction represents the signal amplitude, the u axis represents the signal frequency, and the p axis represents the p signal. value. It can be seen from the figure that the energy concentration of the signal is maximum when the p value is about 0.8, that is, when the rotation angle is At the time of the formed (v, u) plane, the focus of the target signal is the best. At this time, it is masked (bandpass filtered) on the (v, u) plane, and then the inverse Fourier transform is performed to complete the extraction of the target signal.
  • step S3 described in this example includes the following steps:
  • S p (u) is the signal obtained by the fractional Fourier transform of the motor noise signal s(t)
  • N p (u) is the signal obtained by the fractional Fourier transform of the background signal w(t).
  • the kernel function is known to the public. Different kernel functions can be selected according to different requirements.
  • the kernel function can be known as the rotation angle.
  • the basic characteristics of the kernel function are determined, and the P value determines the rotation angle.
  • the so-called occlusion processing refers to extracting the peak value of the secondary signal.
  • the peak value indicates the target signal; and the peak is extracted, that is, the target signal is extracted.
  • the manner in which the target signal is extracted is mathematically generally achieved by filtering.
  • the step S4 specifically includes the following steps: filtering the secondary signal with a band pass filter to obtain the spike signal.
  • the spike signal is represented as X_p ⁇ '(u), and the peak blocking processing is specifically implemented by the following function:
  • the M_p(u) is the center frequency Bandpass filter.
  • the bandpass filter Through the bandpass filter, the flat portion of the secondary signal can be filtered out, and the proper bandwidth can be selected to effectively filter out most of the noise energy.
  • the larger the bandwidth the smoother the signal at both ends, and the larger bandwidth filtering will form a bell pulse. Therefore, to filter the equal-amplitude signal, it is better to restore the waveform, and the bandwidth selection is not too large.
  • the target signal and the background signal are separated, but the result of the separation is the signal after the P-order fractional Fourier transform; for this, it is necessary to perform reverse reduction to obtain Target signal. Due to the order additivity (rotation additivity) of the fractional Fourier transform, the P-order inverse Fourier transform is performed on it to obtain the desired target signal.
  • step S5 is specifically implemented by the following function:
  • x'(t) is the target signal and K -p (t, u) is the kernel function of the P-order inverse Fourier transform.
  • the applicant carried out simulation analysis on the extraction method provided by the method, and compared with the extraction method provided by the existing commercial software (LEA software product of Beijing Langdifeng Technology Co., Ltd.), as shown in Figure 4 and Figure 5,
  • the current commercial software analyzes the time-frequency signal of the motor constant-frequency howling noise.
  • the horizontal axis is time and the vertical axis is frequency.
  • the signal is extracted from the constant-frequency motor noise, it can be extracted relatively clearly.
  • Figure 4 and Figure 5 are equivalent. As shown in FIG. 6 and FIG.
  • the signal waveforms of the present application and the existing commercial software for extracting the constant frequency howling noise of the motor are as follows: the horizontal axis is time and the vertical axis is amplitude, and the howling extracted by the two methods can be seen.
  • the signal results are basically the same, although there is a slight difference in amplitude, the information contained in the two signals is consistent, and the extracted two signal sounds are completely inaudible in subjective perception.
  • the time-frequency signal of the present application and the commercial software for extracting the frequency conversion whistle noise of the motor is characterized in that the horizontal axis is time and the vertical axis is frequency, and signal extraction is performed on the frequency conversion howling noise.
  • FIG. 10 and FIG. 11 the signal waveforms of the present application and the existing commercial software for extracting the frequency conversion whistle noise of the motor are as follows: the horizontal axis is time and the vertical axis is amplitude, and the howling extracted by the two methods can be seen. The signal results are basically the same, although there is a slight difference in amplitude, the information contained in the two signals is consistent, and the extracted two signal sounds can not hear the difference in subjective feelings.
  • the noise extraction method provided by the present application has almost no difference between the extracted target signal and the target signal extracted by the commercial software, and can meet the standards for research and testing.
  • the method for extracting noise signal of a new energy vehicle motor disclosed in this example after acquiring the acoustic environment signal, extracts a primary signal including a motor noise signal and a background signal, which is more convenient for subsequent fractional Fourier transform processing; linearity is analyzed by Fourier transform
  • the smooth signal has better performance characteristics. Therefore, the P-order fractional Fourier transform is performed on the primary signal, and the time domain of the primary signal is transformed into the frequency domain by using the Fourier transform. The characteristic frequency and phase information are more easily seen, and the spectrum of the primary signal is extracted for analysis. Then, through the masking process, the spike signal is obtained.
  • the background signal is filtered by the P-order fractional Fourier transform signal, and only the signal of the motor noise signal after the P-order fractional Fourier transform is retained.
  • the P-order fractional Fourier inverse transform process is performed on the spike signal to obtain a motor noise signal as a target signal.
  • the method for extracting noise signal of the new energy vehicle motor disclosed in the present invention has a lower calculation amount and complexity, and is simpler to implement. The effect is very good enough to achieve the same effect as the commercial software extraction method, and can meet the standards of the applicant for research and testing. The cost is lower.
  • This embodiment provides a new energy vehicle motor noise signal extraction system, as shown in FIG. 12, including the following modules:
  • the acoustic environment signal acquisition module 1 is configured to collect an acoustic environment signal of the new energy vehicle, and the acoustic environment signal includes a motor noise signal;
  • a primary signal extraction module 2 configured to extract a primary signal including a motor noise signal and a background signal from the acoustic environment signal
  • a fractional Fourier transform module 3 configured to perform a P-order fractional Fourier transform process on the primary signal to obtain a secondary signal; wherein P is a time-frequency domain rotation coefficient;
  • the mask processing module 4 is configured to perform a masking process on the secondary signal to obtain a spike signal
  • the fractional-order inverse Fourier transform module 5 is configured to perform P-order fractional Fourier inverse transform processing on the spike signal to obtain a target signal.
  • the new energy vehicle motor noise signal extraction system disclosed in the embodiment of the present invention extracts the acoustic environment signal through the acoustic environment signal acquisition module, and extracts the motor noise signal and the background by the primary signal extraction module to facilitate subsequent fractional Fourier transform processing.
  • the primary signal of the signal; the Fourier transform analyzes the linear, stationary signal for better performance. Therefore, the P-order fractional Fourier transform is performed on the primary signal by the fractional Fourier transform module, and the time domain of the primary signal is transformed into the frequency domain by using the Fourier transform, and the characteristic frequency and phase information are more easily seen.
  • the spectrum of the primary signal is extracted for analysis. Then, the occlusion processing is performed by the occlusion processing module to obtain a spike signal.
  • the background signal is filtered by the P-order fractional Fourier transform signal, and only the motor noise signal is retained by the P-order fractional Fourier transform. signal.
  • the P-order inverse Fourier transform process is performed on the spike signal by the fractional Fourier inverse transform module to obtain the motor noise signal as the target signal.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Quality & Reliability (AREA)
  • Computational Linguistics (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • General Physics & Mathematics (AREA)
  • Electric Propulsion And Braking For Vehicles (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)

Abstract

一种新能源车电机噪声信号提取方法及系统,包括如下步骤:采集新能源车的声环境信号,所述声环境信号包括电机噪声信号;从所述声环境信号中提取包括电机噪声信号和背景信号在内的初级信号;对所述初级信号进行P阶分数傅里叶变换处理获得次级信号,P为时频域旋转系数;对所述次级信号进行隔遮处理,获得尖峰脉冲信号;对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,获得目标信号。解决了现有普通的信号提取方法不能很好的实现电机噪声信号的提取或者提取效果不佳的问题。

Description

一种新能源车电机噪声信号提取方法及系统
相关申请的交叉引用
本申请要求比亚迪股份有限公司于2017年9月29日提交的、发明名称为“一种新能源车电机噪声信号提取方法及系统”的、中国专利申请号“201710905357.8”的优先权。
技术领域
本发明涉及新能源车电机噪声信号提取领域。
背景技术
随着新能源产业的迅速发展,也同样给我们带来了电机噪声的问题。尤其是新能源汽车,电机的高频噪声是一种噪声综合的结果,包括机械噪声,电磁噪声以及空气噪声,频率从1KHz到12KHz或者更高,这种高频的电磁噪声会给人带来强烈的不适感。所以对这种噪声的治理是非常有必要的。
现有的降噪方案主要有两种,一种是被动降噪,也叫做物理降噪。包括结构优化,消除共振,阻尼材料进行吸隔声等。另外一种是主动降噪,包括有源降噪和掩蔽效应以及其它一些声音补偿措施等手段。
对于主动降噪方式,对电机的高频信号进行主被动降噪时,都需要单独的把目标噪声信号提取处理,现有语音信号的提取方法有很多,但由于电机噪声信号的特殊性(1KHz到12KHz或者更高频率),现有普通的信号提取方法并不能很好的实现新能源车电机噪声信号的提取或者提取效果不佳,而专用的商业软件的提取方法,虽然效果较好,但是其过程复杂,不方便用来进行试验和研究。
发明内容
为解决现有普通的信号提取方法并不能很好的实现电机噪声信号的提取或者提取效果不佳,而专用的商业软件提取方法,虽然效果较好,但是其过程复杂,不方便用来进行试验和研究的问题,本发明提供了一种新能源车电机噪声信号提取方法及系统。
本发明一方面提供了一种新能源车电机噪声信号提取方法,包括如下步骤:
采集新能源车的声环境信号,所述声环境信号包括电机噪声信号;
从所述声环境信号中提取包括电机噪声信号和背景信号在内的初级信号;
对所述初级信号进行P阶分数傅里叶变换处理获得次级信号;其中,P为时频域旋转系数;
对所述次级信号进行隔遮处理,获得尖峰脉冲信号;
对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,获得目标信号。
本发明实施例公开的新能源车电机噪声信号提取方法,在采集声环境信号后,通过提取包括电机噪声信号和背景信号的初级信号,更便于后续分数傅里叶变换的处理;傅里叶变换分析线性、平稳信号更具优良性能的特性。因此,再对该初级信号进行P阶分数傅里叶变换,利用傅里叶变换将初级信号的时域变换为频域,更容易看出特征频率和相位信息,提取初级信号的频谱进行分析。然后通过隔遮处理,获得尖峰脉冲信号,如此,将背景信号经过P阶分数傅里叶变换后的信号滤除,仅保留电机噪声信号经过P阶分数傅里叶变换后的信号。如此,再对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,即可获得作为目标信号的电机噪声信号。本发明公开的上述新能源车电机噪声信号提取方法,此方法计算量和复杂程度更低,实现更简单。其效果非常好,足以达到商业软件提取方法相同的效果,能够达到申请人进行研究和试验的标准。成本更低。
本发明第二方面提供了一种新能源车电机噪声信号提取系统,包括如下模块:
声环境信号采集模块,用于采集新能源车的声环境信号,所述声环境信号包括电机噪声信号;
初级信号提取模块,用于从所述声环境信号中提取包括电机噪声信号和背景信号在内的初级信号;
分数阶傅里叶变换模块,用于对所述初级信号进行P阶分数傅里叶变换处理获得次级信号;其中,P为时频域旋转系数;
隔遮处理模块,用于对所述次级信号进行隔遮处理,获得尖峰脉冲信号;
分数阶傅里叶反变换模块,用于对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,获得目标信号。
本发明实施例公开的新能源车电机噪声信号提取系统,在通过声环境信号采集模块采集声环境信号后,通过初级信号提取模块提取更便于后续分数傅里叶变换处理的包括电机噪声信号和背景信号的初级信号;傅里叶变换分析线性、平稳信号更具优良性能的特性。因此,再通过分数阶傅里叶变换模块对该初级信号进行P阶分数傅里叶变换,利用傅里叶变换将初级信号的时域变换为频域,更容易看出特征频率和相位信息,提取初级信号的频谱已进行分析。然后通过隔遮处理模块进行隔遮处理,获得尖峰脉冲信号,如此,将背景信 号经过P阶分数傅里叶变换后的信号滤除,仅保留电机噪声信号经过P阶分数傅里叶变换后的信号。如此,再通过分数阶傅里叶反变换模块对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,即可获得作为目标信号的电机噪声信号。本发明公开的上述新能源车电机噪声信号提取系统,此系统计算量和复杂程度更低,实现更简单。且提取效果非常好,足以达到商业软件提取软件相同的效果,能够达到申请人进行研究和试验的标准。成本更低。
附图说明
图1是本发明具体实施方式中提供的新能源车电机噪声信号的提取流程图;
图2是本发明具体实施方式中提供的信号在时频域上坐标轴绕原点逆时针旋转pπ/2后构成的分数阶傅里叶域的示意图;
图3是本发明具体实施方式中提供的线性调频信号经分数阶傅里叶变换后在MATLAB中仿真示意图;
图4是本发明具体实施方式中提供的电机恒频啸叫噪声提取后的时频信号示意图;
图5是现有商业软件提供的电机恒频啸叫噪声提取后的时频信号示意图;
图6是本发明具体实施方式中提供的电机恒频啸叫噪声提取后的信号波形示意图;
图7是现有商业软件提取的电机恒频啸叫噪声提取后的信号波形示意图;
图8是本发明具体实施方式中提供的电机变频啸叫噪声提取后的时频信号示意图;
图9是现有商业软件提供的电机变频啸叫噪声提取后的时频信号示意图;
图10是本发明具体实施方式中提供的电机变频啸叫噪声提取后的信号波形示意图;
图11是现有商业软件提供的电机变频啸叫噪声提取后的信号波形示意图;
图12是本发明具体实施方式中提供的新能源车电机噪声信号的提取系统框图。
其中,1、声环境信号获取模块;2、初级信号提取模块;3、分数阶傅里叶变换模块;4、隔遮处理模块;5、分数阶傅里叶反变换模块。
具体实施方式
为了使本发明所解决的技术问题、技术方案及有益效果更加清楚明白,以下结合附图及实施例,对本发明进行进一步详细说明。应当理解,此处所描述的具体实施例仅仅用以解释本发明,并不用于限定本发明。
实施例1
本实施例提供了一种新能源车电机噪声信号提取方法,如图1所示,包括如下步骤:
步骤S1、声环境信号采集步骤:采集新能源车的声环境信号,所述声环境信号包括电 机噪声信号;
步骤S2、初级信号提取步骤:从所述声环境信号中提取包括电机噪声信号和背景信号在内的初级信号;
步骤S3、分数阶傅里叶变换步骤:对所述初级信号进行P阶分数傅里叶变换处理获得次级信号;其中,P为时频域旋转系数;
步骤S4、隔遮处理步骤:对所述次级信号进行隔遮处理,获得尖峰脉冲信号;
步骤S5、分数阶傅里叶反变换步骤:对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,获得目标信号。
下面对各步骤进行具体解释说明。
其中,所述步骤S1具体包括如下步骤:
在新能源车中,电机噪声的频率一般都会很高,其中,所谓声环境指在一定的区域中,所有声音组成的系统,所谓新能源车的声环境指新能源车内部环境,具体的,本例中指驾驶员和乘客所处的空间中,比如驾驶室内(或者置于副驾驶室内,效果也是等效的)、电机仓内等。
本例中的声环境的信号包括声环境内的噪声信号(目标信号)和其他信号;声环境的噪声,指在声环境中为人体可以感受到的噪声,对于新能源车而言,其声环境噪声一方面包括来源于电机的高频噪声,将其称为电机噪声。通过对新能源车的高频电机噪声的研究,我们发现,当电机转速达到一定的值时,会产生啸叫声,这种啸叫声的频率总体上可以分为两类,一类是频率不变的,我们称之为恒频啸叫,另一类是变频的,我们称之为变频啸叫。上述两种啸叫均是高频的电机噪声。另一方面,还包括其他非电机噪声,比如路噪、胎噪、结构振动噪声等,这些非电机噪声的频率相对较低。本申请中,主要采集的目标对象为电机噪声,换句话说,本申请的目的是提取该新能源车的电机噪声信号。以便申请人后续进行主动降噪的研究和试验。
由于本申请要采集电机噪声信号,因此,最好的方式是直接采集新能源车电机附近的声音。当然,根据需要,可以使用本方法采用任意需要检测位置的噪声信号。
对电机高频噪声的采集必须要精确一点,在进行电机噪声采集时,同时会采集到其它非电机高频噪声,比如车体摩擦噪声、喇叭声、车外噪声等。因此,优选直接在电机附近(布置电机的电机仓中)采集该电机噪声信号,由此可以保证采集到的电机噪声信号的完整性,排除噪声在传播过程中产生的衰减等不确定因素。并防止其他非电机噪声进行干扰。具体地,可以通过声音信号接收器(如麦克风等)实时采集电机产生的噪声。
其中,步骤S2具体为:通过频谱分析从所述声环境信号中提取获所述初级信号;一 般提取方法为滤波。一般电机的啸叫信号和频率是有一定的关系,根据啸叫信号的频率进行预处理(一般为滤波)。比如如图8所示,啸叫信号的频率是3500Hz到4500Hz之间,如此,我们提取3500Hz~4500Hz之间的频段的信号,即为初级信号。一般的,其在MATLAB(MATLAB是美国MathWorks公司出品的商业数学软件,用于算法开发、数据可视化、数据分析以及数值计算的高级技术计算语言和交互式环境等;为本领域技术人员所公知)中对信号进行仿真,可以提取出上述包括电机噪声信号在内的初级信号。
其中,所述初级信号x(t)通过如下函数表示:
x(t)=s(t)+w(t)
其中,所述s(t)为电机噪声信号,所述w(t)为背景信号。上述电机噪声信号s(t)是我们需要提取的目标信号,该背景信号为其他电机噪声信号以外的信号,是我们需要滤除的信号。比如,背景信号一般包括高斯白噪声。由于傅里叶变换在分析线性、平稳信号时,有优良的性能。但其对非平稳信号,比如信号变化剧烈时,则效果不佳。因此,该提取的初级信号x(t)最好为混有高斯白噪声的线性调频信号。
比如,所述电机噪声信号s(t)为线性调频信号,其函数表示为:
s(t)=a 0exp(jπm 0t 2+j2πf 0)
其中,a 0是包络函数,m 0是线性调频率,f 0是初始频率。
现有技术中没有见到将傅里叶变换用作电机噪声信号提取的方案,具体的傅里叶变换处理方法为本领域公知。分数阶傅里叶变换作为傅里叶变换的一类,也没有被用作电机噪声信号提取,其处理方法也是本领域公知的。本申请创造性的将分数阶傅里叶变换、隔遮处理和分数阶傅里叶反变换结合以提取电机噪声信号。所谓的P阶分数傅里叶变换即指采用时频域旋转系数P对其进行分数阶傅里叶变换。
事实上,P阶分数傅里叶变换是本领域技术人员所公知的。所谓傅里叶变换如图2所示,其实质为将时频面旋转的方法,将信号在时频平面(或时频域,图中所示(ω,t平面))上旋转特定的旋转角(如图中所示的
Figure PCTCN2018090214-appb-000001
),使得信号在新的二维平面(图中所示(v,u平面)) 上退化为单频的正弦信号;或可以理解成时频域上坐标轴绕原点逆时针旋转
Figure PCTCN2018090214-appb-000002
角度后构成的分数阶傅里叶域(v,u平面)上的表示。此时,不同的P值表现出不同的旋转角。不同的旋转角旋转后的结果其聚集性(或称聚焦性)不一样,选择聚集性最优的结果即可获得不同的想要的目标信号。如图3所示,图3是线性调频信号经分数阶傅里叶变换后在MATLAB中仿真示意图,可以看出,该图中纵向表示信号幅度,u轴表示信号频率,p轴表示p的取值。从图中可以看出,当p值为0.8左右时该信号的能量聚集最大,也就是说当旋转角为
Figure PCTCN2018090214-appb-000003
时,形成的(v,u)平面上,目标信号的聚焦性最好。此时在(v,u)平面上对其进行隔遮(带通滤波)处理,然后进行傅里叶反变换就可以完成目标信号的提取。
具体的,关于聚集性最佳的P值的提取方法,并不限定具体的方式,本领域技术人员可以根据经验获取,或者在某个取值范围内离散取样,逐个比较,已选取最优结果,比如,本例中所述步骤S3包括如下步骤:
从预设的【p1,p2】范围内取若干离散值作为P,对所述初级信号进行分数傅里叶变换获得分数阶傅里叶变换域,从所述分数阶傅里叶变换域中取其中能量聚集性最优的信号作为次级信号。
比如,仍以图3为例,设p1为0.5,p2为1.5;以0.1或者0.05等为间隔做离散取样,获得若干P值,然后将其进行P阶分数傅里叶变换,将变换结果在MATLAB中进行仿真,然后从中选择出聚集性最佳的P值。以此该具体性最佳的P值仿真的信号即为该次级信号。此外,在选择P值时,优选还考虑目标信号和背景信号之间尽量没有耦合。
虽然P阶分数傅里叶变换为本领域技术人员所公知,为便于普通公众了解,申请人以下进一步进行具体解释说明。
比如,假设所述P阶分数傅里叶变换后的次级信号表示为X p(u),其通过如下函数表示:
X p(u)=S p(u)+N p(u);
其中S p(u)为电机噪声信号s(t)的分数傅里叶变换后得到的信号,N p(u)为背景信号w(t)的分数傅里叶变换后得到的信号。
数学上,所述P阶分数傅里叶变换可以理解成通过如下函数实现:
Figure PCTCN2018090214-appb-000004
本例中,其中,
Figure PCTCN2018090214-appb-000005
为核函数:
Figure PCTCN2018090214-appb-000006
Figure PCTCN2018090214-appb-000007
核函数为公众所知,根据不同的需求,可以选择不同的核函数,本例中,如上核函数可知,旋转角
Figure PCTCN2018090214-appb-000008
决定了核函数的基本特性,P值决定了旋转角大小。
具体的,所谓的隔遮处理,是指对次级信号的峰值进行提取,如图3所示,因其峰值表示的是目标信号;提取峰值也即提取了目标信号。提取该目标信号的方式在数学上一般通过滤波实现。
所述步骤S4具体包括如下步骤:用带通滤波器对次级信号进行滤波,获得所述尖峰脉冲信号。
所述尖峰脉冲信号表示为X_p^′(u),所述尖峰隔遮处理具体通过如下函数实现:
X_p^′(u)=X_p(u)M_p(u)=S_p(u)M_p(u)+N_p(u)M_p(u);
其中,所述M_p(u)是中心频率为
Figure PCTCN2018090214-appb-000009
的带通滤波器。通过该带通滤波器,即可滤除次级信号的平坦部分,选择适当的带宽可以有效的滤除大部分噪声能量。但由于带宽越大信号两端越平滑,较大带宽滤波会形成一个钟形脉冲。因此对等幅信号进行滤波要想较好还原波形,带宽选择不易过大。
如此,通过上述隔遮处理后,则对目标信号和背景信号进行了分离,但是其分离的结果是经P阶分数傅里叶变换后的信号;为此,还需要对其进行逆向还原才能获得目标信号。由于分数阶傅里叶变换的阶数可加性(旋转可加性)特性,对其进行P阶分数傅里叶反变换,即可获得所需的目标信号。
其中,步骤S5具体通过如下函数实现:
Figure PCTCN2018090214-appb-000010
其中,x′(t)为目标信号,K -p(t,u)为P阶分数傅里叶反变换的核函数。
申请人对本方法提供的提取方法进行了仿真分析,并与现有商业软件(北京郎迪锋科技有限公司的LEA软件产品)提供的提取方法进行了对比,如图4、图5所示的本申请及现有商业软件对电机恒频啸叫噪声提取后的时频信号示意图,其横轴为时间,纵轴为频率,当对恒频电机噪声进行信号提取时,都可以相对较清晰的提取到多个高频噪声信号,且背景干扰很小。图4、图5效果相当。如图6、图7所示本申请及现有商业软件对电机恒频啸叫噪声提取后的信号波形示意图,其横轴为时间,纵轴为幅度,可看出两种方法提取的啸叫信号结果基本完全一致,虽然在幅度上有少许差异,但两个信号中包含的信息是一致的,而且提取的两个信号声音在主观感受上完全听不出差异。如图8、图9所示的本申请及现有商业软件对电机变频啸叫噪声提取后的时频信号示意图,其横轴为时间,纵轴为频率,当对变频啸叫噪声进行信号提取时,可以相对较清晰的提取到多个高频噪声信号,且背景干扰很小。图8、图9效果相当。如图10、图11所示的本申请及现有商业软件对电机变频啸叫噪声提取后的信号波形示意图,其横轴为时间,纵轴为幅度,可看出两种方法提取的啸叫信号结果基本也完全一致,虽然在幅度上有少许差异,但两个信号中包含的信息是一致的,而且提取的两个信号声音在主观感受上完全听不出差异。
综上可知,本申请提供的噪声提取方法,提取的目标信号与商业软件提取的目标信号几乎没有差异,能够达到我们进行研究和试验的标准。
本例公开的新能源车电机噪声信号提取方法,在采集声环境信号后,通过提取更便于后续分数傅里叶变换处理的包括电机噪声信号和背景信号的初级信号;由于傅里叶变换分析线性、平稳信号更具优良性能的特性。因此,再对该初级信号进行P阶分数傅里叶变换,利用傅里叶变换将初级信号的时域变换为频域,更容易看出特征频率和相位信息,提取初级信号的频谱进行分析。然后通过隔遮处理,获得尖峰脉冲信号,如此,将背景信号经过P阶分数傅里叶变换后的信号滤除,仅保留电机噪声信号经过P阶分数傅里叶变换后的信号。如此,再对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,即可获得作为目标信号的电机噪声信号。本发明公开的上述新能源车电机噪声信号提取方法,此方法计算量和复杂程度更低,实现更简单。其效果非常好,足以达到商业软件提取方法相同的效果,能够达到申请人进行研究和试验的标准。成本更低。
实施例2
本实施例提供了一种新能源车电机噪声信号提取系统,如图12所示,包括如下模块:
声环境信号采集模块1,用于采集新能源车的声环境信号,所述声环境信号包括电机噪声信号;
初级信号提取模块2,用于从所述声环境信号中提取包括电机噪声信号和背景信号在内的初级信号;
分数阶傅里叶变换模块3,用于对所述初级信号进行P阶分数傅里叶变换处理获得次级信号;其中,P为时频域旋转系数;
隔遮处理模块4,用于对所述次级信号进行隔遮处理,获得尖峰脉冲信号;
分数阶傅里叶反变换模块5,用于对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,获得目标信号。
本例公开的新能源车电机噪声信号提取系统已在上述实施例1中对应做了详细说明,为免重复,不再赘述。
本发明实施例公开的新能源车电机噪声信号提取系统,在通过声环境信号采集模块采集声环境信号后,通过初级信号提取模块提取更便于后续分数傅里叶变换处理的包括电机噪声信号和背景信号的初级信号;傅里叶变换分析线性、平稳信号更具优良性能的特性。因此,再通过分数阶傅里叶变换模块对该初级信号进行P阶分数傅里叶变换,利用傅里叶变换将初级信号的时域变换为频域,更容易看出特征频率和相位信息,提取初级信号的频谱进行分析。然后通过隔遮处理模块进行隔遮处理,获得尖峰脉冲信号,如此,将背景信号经过P阶分数傅里叶变换后的信号滤除,仅保留电机噪声信号经过P阶分数傅里叶变换后的信号。如此,再通过分数阶傅里叶反变换模块对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,即可获得作为目标信号的电机噪声信号。本发明公开的上述新能源车电机噪声信号提取系统,此系统计算量和复杂程度更低,实现更简单。且提取效果非常好,足以达到商业软件提取软件相同的效果,能够达到申请人进行研究和试验的标准。成本更低。
以上所述仅为本发明的较佳实施例而已,并不用以限制本发明,凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等,均应包含在本发明的保护范围之内。

Claims (11)

  1. 一种新能源车电机噪声信号提取方法,其特征在于,包括如下步骤:
    采集新能源车的声环境信号,所述声环境信号包括电机噪声信号;
    从所述声环境信号中提取包括电机噪声信号和背景信号在内的初级信号;
    对所述初级信号进行P阶分数傅里叶变换处理获得次级信号;其中,P为时频域旋转系数;
    对所述次级信号进行隔遮处理,获得尖峰脉冲信号;
    对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,获得目标信号。
  2. 根据权利要求1所述的新能源车电机噪声信号提取方法,其特征在于,所述“采集新能源车的声环境信号”具体包括如下步骤:
    将声音信号接收器置于电机附近,采集新能源车的声环境信号。
  3. 根据权利要求1或2所述的新能源车电机噪声信号提取方法,其特征在于,所述“从所述声环境信号中提取包括电机噪声信号和背景信号在内的初级信号”具体为:通过频谱分析从所述声环境信号中提取获所述初级信号;
    其中,所述初级信号x(t)通过如下函数表示:
    x(t)=s(t)+w(t)
    其中,所述s(t)为电机噪声信号,所述w(t)为背景信号。
  4. 根据权利要求1至3中任意一项所述的新能源车电机噪声信号提取方法,其特征在于,所述s(t)为线性调频信号,其函数表示为:
    s(t)=a 0exp(jπm 0t 2+j2πf 0)
    其中,a 0是包络函数,m 0是线性调频率,f 0是初始频率;
    所述背景信号w(t)为高斯白噪声。
  5. 根据权利要求1至4中任意一项所述的新能源车电机噪声信号提取方法,其特征在于,所述“对所述初级信号进行P阶分数傅里叶变换处理获得次级信号”具体包括如下步骤:
    从预设的【p1,p2】范围内取若干离散值作为P,对所述初级信号进行分数傅里叶变换获得分数阶傅里叶变换域,从所述分数阶傅里叶变换域中取其中能量聚集性最优的信号作为次级信号。
  6. 根据权利要求1至5中任意一项所述的新能源车电机噪声信号提取方法,其特征在于,所述P阶分数傅里叶变换后的次级信号表示为X p(u),其通过如下函数表示:
    X p(u)=S p(u)+N p(u);
    其中,其中S p(u)为电机噪声信号的分数傅里叶变换后得到的信号,N p(u)为背景信号的分数傅里叶变换后得到的信号。
  7. 根据权利要求1至6中任意一项所述的新能源车电机噪声信号提取方法,其特征在于,
    所述P阶分数傅里叶变换通过如下函数实现:
    Figure PCTCN2018090214-appb-100001
    其中,其中
    Figure PCTCN2018090214-appb-100002
    为核函数:
    Figure PCTCN2018090214-appb-100003
    Figure PCTCN2018090214-appb-100004
  8. 根据权利要求1至7中任意一项所述的新能源车电机噪声信号提取方法,其特征在于,所述“对所述次级信号进行隔遮处理,获得尖峰脉冲信号”具体包括如下步骤:用带通滤波器滤对次级信号进行滤波,获得所述尖峰脉冲信号。
  9. 根据权利要求1至8中任意一项所述的新能源车电机噪声信号提取方法,其特征在于,所述尖峰脉冲信号表示为
    Figure PCTCN2018090214-appb-100005
    所述尖峰隔遮处理具体通过如下函数实现:
    X_p^′(u)=X_p(u)M_p(u)=S_p(u)M_p(u)+N_p(u)M_p(u);
    其中,所述M_p(u)是中心频率为u0的带通滤波器。
  10. 根据权利要求1至9中任意一项所述的新能源车电机噪声信号提取方法,其特征在于,所述“对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,获得目标信号”具体通过如下函数实现:
    Figure PCTCN2018090214-appb-100006
    其中,X ′(t)为目标信号,K -p(t,u)为P阶分数傅里叶反变换的核函数。
  11. 一种新能源车电机噪声信号提取系统,其特征在于,包括如下模块:
    声环境信号采集模块,用于采集新能源车的声环境信号,所述声环境信号包括电机噪声信号;
    初级信号提取模块,所述初级信号提取模块与声环境信号采集模块相连,用于从所述声环境信号中提取包括电机噪声信号和背景信号在内的初级信号;
    分数阶傅里叶变换模块,所述分数阶傅里叶变换模块与初级信号提取模块相连,用于对所述初级信号进行P阶分数傅里叶变换处理获得次级信号;其中,P为时频域旋转系数;
    隔遮处理模块,所述隔遮处理模块与分数阶傅里叶变换模块相连,用于对所述次级信号进行隔遮处理,获得尖峰脉冲信号;
    分数阶傅里叶反变换模块,所述分数阶傅里叶反变换模块与隔遮处理模块相连,用于对所述尖峰脉冲信号进行P阶分数傅里叶反变换处理,获得目标信号。
PCT/CN2018/090214 2017-09-29 2018-06-07 一种新能源车电机噪声信号提取方法及系统 WO2019062197A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710905357.8A CN109579979A (zh) 2017-09-29 2017-09-29 一种新能源车电机噪声信号提取方法及系统
CN201710905357.8 2017-09-29

Publications (1)

Publication Number Publication Date
WO2019062197A1 true WO2019062197A1 (zh) 2019-04-04

Family

ID=65900523

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/090214 WO2019062197A1 (zh) 2017-09-29 2018-06-07 一种新能源车电机噪声信号提取方法及系统

Country Status (2)

Country Link
CN (1) CN109579979A (zh)
WO (1) WO2019062197A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110687523B (zh) * 2019-08-29 2023-07-11 中国科学技术大学 一种障碍物检测系统、方法及存储介质
CN112946082A (zh) * 2021-01-22 2021-06-11 天津大学 一种基于激光声表面波的时频信息和频散曲线的获取方法
CN115295003A (zh) * 2022-10-08 2022-11-04 青岛民航凯亚系统集成有限公司 对民航机务维修领域的语音降噪方法及系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7203321B1 (en) * 1999-09-21 2007-04-10 Bayerische Motoren Werke Aktiengesellschaft Device for electroacoustic sound generation in a motor vehicle
CN101860512A (zh) * 2010-05-07 2010-10-13 深圳国微技术有限公司 多载波系统中脉冲噪声抑制与解映射软判决方法及系统
CN103743470A (zh) * 2013-12-23 2014-04-23 广西科技大学 一种汽车噪声频谱分析方法
CN106249208A (zh) * 2016-07-11 2016-12-21 西安电子科技大学 基于分数阶傅里叶变换的噪声调幅干扰下信号检测方法
CN106840378A (zh) * 2017-02-23 2017-06-13 江铃汽车股份有限公司 汽车啸叫噪声的评价方法及系统

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2013084019A (ja) * 2010-01-14 2013-05-09 Toyama Univ データ外挿方法、データ外挿装置、及びデータ外挿プログラム
CN104714925B (zh) * 2015-02-02 2017-07-07 北京工业大学 一种基于分数阶傅里叶变换和支持向量机的齿轮传动噪声分析方法
US9804999B2 (en) * 2015-12-09 2017-10-31 The Aerospace Corporation Signal/noise separation using FrFT rotational parameter obtained in relation to Wigner Distribution
CN106019243B (zh) * 2016-06-14 2019-01-04 南京理工大学 一种基于三次初相和frft的drfm干扰检测和对抗方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7203321B1 (en) * 1999-09-21 2007-04-10 Bayerische Motoren Werke Aktiengesellschaft Device for electroacoustic sound generation in a motor vehicle
CN101860512A (zh) * 2010-05-07 2010-10-13 深圳国微技术有限公司 多载波系统中脉冲噪声抑制与解映射软判决方法及系统
CN103743470A (zh) * 2013-12-23 2014-04-23 广西科技大学 一种汽车噪声频谱分析方法
CN106249208A (zh) * 2016-07-11 2016-12-21 西安电子科技大学 基于分数阶傅里叶变换的噪声调幅干扰下信号检测方法
CN106840378A (zh) * 2017-02-23 2017-06-13 江铃汽车股份有限公司 汽车啸叫噪声的评价方法及系统

Also Published As

Publication number Publication date
CN109579979A (zh) 2019-04-05

Similar Documents

Publication Publication Date Title
Chi et al. Multiresolution spectrotemporal analysis of complex sounds
CN103456312B (zh) 一种基于计算听觉场景分析的单通道语音盲分离方法
CN101430882B (zh) 一种抑制风噪声的方法及装置
WO2019062197A1 (zh) 一种新能源车电机噪声信号提取方法及系统
CN111261189B (zh) 一种车辆声音信号特征提取方法
CN108198545B (zh) 一种基于小波变换的语音识别方法
CN109147798B (zh) 语音识别方法、装置、电子设备及可读存储介质
CN109523999A (zh) 一种提升远场语音识别的前端处理方法和系统
Mourjopoulos et al. Modelling and enhancement of reverberant speech using an envelope convolution method
CN113168843A (zh) 音频处理方法、装置、存储介质及电子设备
Hosseini et al. Time difference of arrival estimation of sound source using cross correlation and modified maximum likelihood weighting function
Mesgarani et al. Speech enhancement based on filtering the spectrotemporal modulations
Wang et al. Auditory analysis of spectro-temporal information in acoustic signals
Tchorz et al. Combination of amplitude modulation spectrogram features and MFCCs for acoustic scene classification
CN105869652A (zh) 心理声学模型计算方法和装置
CN106782588A (zh) 一种基于听觉小波包的毫米波生物雷达语音增强方法
JP2003271166A (ja) 入力信号処理方法および入力信号処理装置
Zhou et al. Whisper intelligibility enhancement using a supervised learning approach
CN108205127B (zh) 一种基于稀疏表示的水声信号处理方法
JPH096391A (ja) 信号推定装置
CN113012710A (zh) 一种音频降噪方法及存储介质
Unoki et al. Development of the MTF-based speech dereverberation method using adaptive time-frequency division
Ganapathy et al. Auditory motivated front-end for noisy speech using spectro-temporal modulation filtering
CN113409813B (zh) 语音分离方法及装置
CN111009259B (zh) 一种音频处理方法和装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18862004

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18862004

Country of ref document: EP

Kind code of ref document: A1