Active noise reduction method, system and device for linear noise source

Abstract

The invention discloses an active noise reduction method, system and device for a linear noise source, and the method comprises the steps: arranging a plurality of microphones for the noise source, and forming a microphone array; noise signals of the noise source are collected through the microphone array; performing mixed-base FFT conversion on the noise signal to obtain a frequency domain signal, and analyzing the frequency domain signal to obtain a spectrum feature of the noise signal; generating reverse sound waves according to the spectrum characteristics; establishing a loudspeaker array for the noise source, transmitting the reverse sound waves to the loudspeaker array, playing the reverse sound waves to the noise source through the loudspeaker array, and eliminating the noise source; arranging error microphones between the loudspeaker array and the noise source to form an error microphone array; residual noise signals appearing after the noise source is eliminated by the reverse sound waves played by the loudspeaker array are collected through the error microphone array; and judging whether the equipment has a fault or not according to the residual noise signal by setting a threshold value.

Description

Technical Field

[0001] The present invention relates to the field of noise reduction, and particularly to a method, a system, and a device for active noise reduction of a linear noise source. Background Art

[0002] With the development of industrial technology, noise pollution has become one of the important factors affecting people's quality of life and work efficiency. Especially the linear noise sources (such as spinning machines, conveyor belts, etc.) generated during the manufacturing, construction, and operation of various mechanical equipment. These noises not only pose a threat to the health of on-site workers but also may interfere with other activities in the surrounding environment. Therefore, how to effectively reduce or eliminate these noises has become an important research topic.

[0003] Traditional passive noise reduction methods mainly rely on using sound-absorbing materials or sound insulation barriers to reduce noise propagation. However, this method often requires a large amount of physical materials and has limited effectiveness in dealing with low-frequency noises. In contrast, active noise reduction technology achieves noise cancellation by generating a sound signal (i.e., a reverse sound wave) with a phase opposite to the original noise, which has higher efficiency and a wider scope of application. However, in order to ensure that the reverse sound wave can accurately cancel the noise, it is necessary to accurately capture the spatial and temporal characteristics of the noise source.

[0004] Existing active noise reduction systems usually use a microphone array to collect noise signals and use a speaker array to play reverse sound waves to achieve the purpose of noise reduction. However, in practical applications, due to the complexity and variability of the noise source, how to reasonably arrange the microphone and speaker arrays to ensure that the distance between them is less than half of the noise wavelength to avoid frequency aliasing, and at the same time determine the appropriate number and density of microphones according to the length and complexity of the noise source, has become a key challenge in improving the active noise reduction effect. In addition, the monitoring of the residual noise after active noise reduction is not fine enough, and it is difficult to detect equipment failures in a timely manner, which is also a deficiency of the existing technology for early warning. Summary of the Invention

[0005] The purpose of the present invention is to provide a method, a system, and a device for active noise reduction of a linear noise source, which solve the above-mentioned technical problems pointed out in the existing technology.

[0006] The present invention provides a method for active noise reduction of a linear noise source, including the following operation steps:

[0007] Arrange a plurality of microphones for the noise source to form a microphone array; collect noise signals from the noise source through the microphone array;

[0008] The noise signal is converted into a frequency-domain signal by performing a mixed-radix FFT, and the frequency-domain signal is analyzed to obtain the spectral characteristics of the noise signal; a reverse acoustic wave is generated according to the spectral characteristics;

[0009] A loudspeaker array is established for the noise source, the reverse acoustic wave is transmitted to the loudspeaker array, and the reverse acoustic wave is played on the noise source through the loudspeaker array to eliminate the noise source.

[0010] Preferably, the distance between the microphones in the microphone array is less than half of the noise wavelength of the noise source.

[0011] Preferably, the distance between the loudspeakers in the loudspeaker array is the same as the distance between the microphones in the microphone array.

[0012] Preferably, the noise signal is converted into a frequency-domain signal by performing a mixed-radix FFT, the frequency-domain signal is analyzed to obtain the spectral characteristics of the noise signal; a reverse acoustic wave is generated according to the spectral characteristics, and the specific operation steps are as follows:

[0013] The noise signal captured by the microphone array is converted into a digital signal by using an analog-to-digital converter, and after quantization, a discrete time-domain signal xi(n) is obtained;

[0014] The time-domain signal is converted into a frequency-domain signal by performing a mixed-radix FFT; the spectral characteristics of the noise signal are analyzed according to the frequency-domain signal;

[0015] The spectral characteristics are filtered according to the LMS algorithm to obtain a reverse acoustic wave.

[0016] Preferably, the time-domain signal is converted into a frequency-domain signal by performing a mixed-radix FFT, and the specific operation steps are as follows:

[0017] The time-domain signal is decomposed into a plurality of relatively prime factors according to the length of the time-domain signal by using a mixed-radix FFT;

[0018] Each relatively prime factor is subjected to a step-by-step radix transformation to obtain the radix set data of the radix transformation of each relatively prime factor;

[0019] The radix set data of the radix transformation of each relatively prime factor is re-ordered, and the relatively prime factors of the re-ordered radix set data are combined to obtain a frequency-domain signal.

[0020] Preferably, the spectral characteristics of the noise signal are analyzed according to the frequency-domain signal, and the specific operation steps are as follows:

[0021] The amplitude spectrum and phase spectrum of the frequency-domain signal are analyzed;

[0022] Search for the maximum amplitude value in the frequency-domain signal through the amplitude spectrum to determine the main frequency in the frequency-domain signal;

[0023] Determine the low-frequency band and high-frequency band in the frequency-domain signal according to the main frequency in the frequency-domain signal;

[0024] Determine a specific frequency band through the low-frequency band and high-frequency band; calculate the squared amplitude for each frequency point in the specific frequency band and sum up the squared amplitudes to obtain the band energy;

[0025] Draw a spectrogram through the band energy and the amplitude spectrum; integrate the frequency bands in the frequency-domain signal through the spectrogram to obtain the frequency distribution;

[0026] Take the frequency distribution of the frequency-domain signal as the frequency feature.

[0027] Preferably, the spectral feature is filtered according to the LMS algorithm to obtain a reverse sound wave. The specific operation steps are as follows:

[0028] Use a filter to assign complex weights to each frequency band of the spectral feature and initialize the filter coefficients according to the complex weights;

[0029] Filter the frequency signal of the spectral feature through the LMS algorithm and the filter coefficients; generate the spectral signal of the reverse sound wave;

[0030] Perform inverse transformation on the spectral signal of the reverse sound wave through the mixed-radix FFT to obtain the reverse sound wave of the time-domain signal.

[0031] Preferably, an error microphone is arranged between the speaker array and the noise source to form an error microphone array; collect the residual noise signal that appears after the reverse sound wave played by the speaker array eliminates the noise source through the error microphone array; judge whether the device has a fault through the residual noise signal by setting a threshold. The specific operation steps are as follows:

[0032] With the noise source as the center, perform a grid-like distribution between the noise source and the speaker array;

[0033] Arrange error microphones in the grid to form an error microphone array; the distance between each error microphone in the error microphone array and the nearest speaker is less than 1 / 4 of the wavelength of the noise signal;

[0034] Collect the residual noise signal eliminated between the reverse sound wave and the noise source through the error microphone array;

[0035] Analyze the residual noise signal through the hybrid radix FFT, extract the frequency-domain energy distribution in the residual noise signal, and calculate the total residual energy in the residual noise signal based on the frequency-domain energy in the residual noise signal;

[0036] Preset a residual noise threshold k, and determine whether the total residual energy is greater than the residual noise threshold k;

[0037] If not, it is determined that the device has no fault problem;

[0038] If so, it is determined that the device has a noise reduction fault and an alarm is triggered.

[0039] Correspondingly, the present application also provides a system for active noise reduction of a linear noise source, including: a collection module; a unit module; an elimination module; a warning module;

[0040] The collection module is used to arrange a plurality of microphones for the noise source to form a microphone array; collect the noise signal of the noise source through the microphone array;

[0041] The unit module is used to convert the noise signal into a frequency-domain signal through the hybrid radix FFT, analyze the frequency-domain signal to obtain the spectral characteristics of the noise signal; generate a reverse sound wave according to the spectral characteristics;

[0042] The elimination module is used to establish a speaker array for the noise source, transmit the reverse sound wave to the speaker array, play the reverse sound wave to the noise source through the speaker array, and eliminate the noise source;

[0043] The warning module is used to arrange error microphones between the speaker array and the noise source to form an error microphone array; collect the residual noise signal that appears after the speaker array plays the reverse sound wave to eliminate the noise source through the error microphone array; judge whether the device has a fault by setting a threshold for the residual noise signal.

[0044] Correspondingly, the present application also relates to a storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the steps of a method for active noise reduction of a linear noise source are implemented.

[0045] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages:

[0046] Analyzing the above-mentioned method, system, and device for active noise reduction of a linear noise source provided by the present invention, it can be seen that in specific applications, by reasonably arranging multiple microphones, acoustic wave signals from the noise source can be effectively captured, ensuring the comprehensiveness and accuracy of the signals; the distance between microphones is less than half of the noise wavelength, avoiding the phenomenon of frequency aliasing, thus ensuring the accuracy of signal acquisition and avoiding the distortion or mixing of high-frequency noise.

[0047] Furthermore, the noise signals captured by the microphone array are converted into digital signals using an analog-to-digital converter to achieve digital processing and generate discrete time-domain signals; the use of a mixed-radix FFT can efficiently process time-domain signals of non-standard lengths, avoiding the limitations of traditional FFT; by decomposing the signal length into multiple relatively prime factors, the computational amount is significantly reduced and the transformation speed is increased; frequency-domain analysis can extract the frequency distribution of the noise signal and identify the main noise frequency components; the result of the base transformation of each relatively prime factor is based on the frequency-domain components of the current factor; in order to adapt to the transformation requirements of the next-level factor, the base-group data must be rearranged so as to obtain an accurate frequency-domain signal; by calculating the amplitude spectrum and phase spectrum of the frequency-domain signal, the spectral characteristics of the frequency-domain signal can be comprehensively analyzed. The amplitude spectrum reflects the intensity of the signal at different frequencies, while the phase spectrum shows the phase relationship of each frequency component; by finding the maximum amplitude value in the amplitude spectrum, the main frequency of the signal can be identified; through the analysis of the main frequency, amplitude spectrum, and phase spectrum, key frequency features can be effectively extracted from the signal, ensuring the best noise cancellation effect.

[0048] Furthermore, the LMS algorithm is used to simultaneously adjust the amplitude and phase of the complex weights, enabling real-time processing of the noise signal and generating a reverse acoustic wave; the mixed-radix FFT (combining the radix-2 and radix-4 algorithms) is used to convert the spectral signal of the reverse acoustic wave into a time-domain signal; the reverse conversion is to reverse the frequency signal and convert it back into a time-domain signal; the role of the reverse acoustic wave in the time-domain signal is to interfere with the noise signal, reducing the noise through phase inversion. The reverse acoustic wave is transmitted through the speaker array, and the role of the speaker array is to accurately play these reverse acoustic waves to generate an interference effect near the noise source, thereby achieving active noise cancellation. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0050] Figure 1Flowchart of a method for active noise reduction of a linear noise source provided in Embodiment 1 of the present invention;

[0051] Figure 2 Flowchart of generating a reverse sound wave for a method for active noise reduction of a linear noise source provided in Embodiment 1 of the present invention;

[0052] Figure 3 Flowchart of extracting frequency features for a method for active noise reduction of a linear noise source provided in Embodiment 1 of the present invention;

[0053] Figure 4 Flowchart of inverse conversion of a frequency signal for a method for active noise reduction of a linear noise source provided in Embodiment 1 of the present invention;

[0054] Figure 5 Flowchart of a system for active noise reduction of a linear noise source provided in Embodiment 2 of the present invention;

[0055] Figure 6 Structural schematic diagram of a storage medium applying the above - mentioned method for active noise reduction of a linear noise source in Embodiment 3.

[0056] Markings: Acquisition module 10; Unit module 20; Elimination module 30; Early - warning module 40; Processor 1110; Communication interface 1120; Memory 1130; Computer storage medium 1140. Detailed implementation manners

[0057] Next, the technical solutions of the present invention will be described clearly and completely in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] Next, the present invention will be further described in detail through specific embodiments in conjunction with the accompanying drawings.

[0059] Embodiment 1

[0060] As Figure 1 shown, the present invention proposes a method for active noise reduction of a linear noise source, including the following operation steps:

[0061] S1: Arrange a plurality of microphones for the noise source to form a microphone array; collect noise signals of the noise source through the microphone array;

[0062] The distance between the microphones of the microphone array is less than half of the noise wavelength of the noise source;

[0063] It should be noted that in order to effectively collect the sound signals emitted by the noise source, multiple microphones need to be arranged near the noise source to form a microphone array. According to the frequency characteristics of the noise source, the distance between microphones needs to be less than half of the noise wavelength. This is to avoid the phenomenon of frequency aliasing and ensure that the collected signals are accurate enough.

[0064] For a linear noise source (such as a spinning frame), the microphone array should be arranged in a straight line along the length direction of the noise source. Such an arrangement can ensure that the array can capture the sound information from different positions, which is beneficial to subsequent signal processing, such as noise source localization, noise identification, active noise reduction, etc. (that is, for a linear noise source (such as a spinning frame), the microphone array should be arranged along the length direction of the noise source to form a straight line (that is, applicable to detecting the sound source in a specific direction, and each microphone will output the time-domain signal at its location, and these signals reflect the change of the sound intensity at that location over time; according to the obtained sound information from different positions in the environment, this is very important basic data for subsequent signal processing (such as noise identification, locating the noise source, active noise reduction, etc.); the distance between microphones should be determined according to the frequency characteristics of the noise source; generally speaking, the distance should be less than half of the noise wavelength to avoid the phenomenon of frequency aliasing; the number of microphones should be determined according to the length and complexity of the noise source; usually, arranging a microphone every 0.5 meters can meet the basic requirements; if the noise source is more complex (such as multiple sound sources), the microphone density can be appropriately increased (such as arranging a microphone every 0.3 meters));

[0065] The number and density of microphones need to be determined according to the length and complexity of the noise source. If the noise source is relatively simple, usually a microphone can be arranged every 0.5 meters; if the noise source is more complex (such as having multiple sound sources), a higher microphone density is required.

[0066] The microphones should be at the same height as the noise source to ensure that the noise signals can be effectively captured. At the same time, the orientation of the microphones should be consistent with the sound wave propagation direction of the noise source. If the sound waves emitted by the noise source are diffusive, the angles of the microphones can be appropriately adjusted.

[0067] S2: Perform a mixed-radix FFT on the noise signal to convert it into a frequency-domain signal, analyze the frequency-domain signal, and obtain the spectral characteristics of the noise signal; generate a reverse sound wave according to the spectral characteristics;

[0068] It should be noted that the noise signals collected by the microphone array are time-domain signals, and these signals are converted into frequency-domain signals through the fast Fourier transform (FFT); this conversion enables the distribution of the noise signals in the frequency domain to be displayed, facilitating subsequent noise analysis;

[0069] By analyzing the frequency-domain signal, the spectral characteristics of the noise signal are extracted. Spectrum analysis can reveal the frequency components of the noise, thereby helping to identify the characteristics of the noise source, understand its energy distribution and frequency characteristics;

[0070] S3: Establish a speaker array for the noise source, transmit the reverse sound wave to the speaker array, play the reverse sound wave on the noise source through the speaker array, and eliminate the noise source;

[0071] The spacing between the speakers in the speaker array is the same as the spacing between the microphones in the microphone array.

[0072] It should be noted that based on the spectral characteristics of the noise signal obtained by spectrum analysis, a reverse sound wave with the opposite phase and the same frequency as the noise signal is generated; the role of this reverse sound wave is to interfere with the noise signal and reduce the noise through phase inversion;

[0073] The reverse sound wave is transmitted through the speaker array. The role of the speaker array is to accurately play these reverse sound waves so that they produce an interference effect near the noise source, thereby achieving active noise cancellation.

[0074] S4: Arrange error microphones between the speaker array and the noise source to form an error microphone array; collect the residual noise signal that appears after the speaker array plays the reverse sound wave to eliminate the noise source through the error microphone array; judge whether the device has a fault by setting a threshold for the residual noise signal.

[0075] It should be noted that through research, it is found that after the speaker array plays the reverse sound wave to eliminate the noise source, there will be some small residual noises, which are not easily detected. Therefore, it is necessary to arrange error microphones between the speaker array and the noise source to analyze the residual noise signal, so as to detect whether the device has a fault and reduce manual patrol monitoring;

[0076] Specifically, as Figure 2 shown, in step S2, the noise signal is converted into a frequency-domain signal by hybrid radix FFT, the frequency-domain signal is analyzed to obtain the spectral characteristics of the noise signal; a reverse sound wave is generated according to the spectral characteristics, and the specific operation steps are as follows:

[0077] S21: Use an analog-to-digital converter to convert the noise signal captured by the microphone array into a digital signal, and obtain a discrete time-domain signal xi(n) through quantization (that is, i in xi(n) represents the i-th microphone, and n represents a discrete time point; therefore, xi(n) represents the amplitude of the noise signal captured by the i-th microphone at the n-th sampling moment);

[0078] It should be noted that the noise signal captured by the microphone array is converted into a digital signal through an analog-to-digital converter (i.e., the analog-to-digital converter converts the collected noise signal into a digital signal to achieve digital processing), thereby realizing digital processing. Digital signals are convenient for subsequent calculations and analyses, and compared with analog signals, digital signals have stronger anti-interference capabilities. Through quantization, the amplitude of the noise signal is converted into discrete values, thereby generating a discrete time-domain signal xi(n), and the signals of each microphone are represented separately, facilitating the processing of each signal. Quantization not only ensures the discreteness of the signal but also allows for the precise representation of the signal amplitude during the calculation process.

[0079] In noise control, by discretizing the noise signals captured by multiple microphones, it is possible to process the signals of different microphone arrays in subsequent frequency-domain analyses. This is the first step in dealing with complex noise environments. By separating the data of different microphones, it lays the foundation for the subsequent generation of reverse sound waves.

[0080] Converting the noise signal into a digital signal is to utilize digital signal processing techniques for subsequent analyses and operations. This process provides the basic framework for noise signal processing, enabling subsequent frequency-domain analyses, filtering, and reverse sound wave generation to be effectively carried out on a computer.

[0081] S22: Convert the time-domain signal into a frequency-domain signal through a mixed-radix FFT; analyze the spectral characteristics of the noise signal based on the frequency-domain signal.

[0082] It should be noted that when the length of the time-domain signal is not a power of 2 in the mixed-radix FFT, the traditional FFT may not be able to process it efficiently. By decomposing the length of the time-domain signal into multiple relatively prime factors (e.g., 3, 5, 7, etc.), the mixed-radix FFT can perform the frequency-domain transformation more efficiently and process signals of non-standard lengths. This method can reduce the computational amount and improve the processing speed, especially suitable for complex noise signals. After converting the time-domain signal into a frequency-domain signal through the mixed-radix FFT transformation, the energy distribution of the signal at different frequencies can be analyzed. Through the spectral characteristics, the main frequency components and specific frequency bands of the noise signal can be identified, which is crucial for the identification of the noise source and subsequent noise cancellation.

[0083] In the frequency-domain analysis of the noise signal, the frequency range of the noise can be accurately identified by capturing the frequency distribution and amplitude characteristics of the noise signal. This provides the basis for the subsequent generation of reverse sound waves opposite to the noise, ensuring that the reverse sound waves can effectively cancel the noise.

[0084] Converting the time-domain signal to the frequency-domain signal is the core step of noise analysis. Through spectral feature extraction, the main frequency components of the noise signal can be identified, providing a scientific basis for the design of reverse acoustic wave generation and noise cancellation; the mixed-radix FFT provides an efficient frequency-domain conversion method to ensure that effective spectral information can be obtained quickly in a complex noise environment;

[0085] S23: Filter the spectral features according to the LMS algorithm to obtain the reverse acoustic wave;

[0086] It should be noted that the LMS algorithm is an adaptive filtering algorithm, usually used for real-time processing and noise cancellation; the LMS algorithm adjusts the parameters of the filter by minimizing the error between the target signal (noise signal) and the output of the filter. According to the spectral features, the LMS algorithm can calculate the optimal parameters of the reverse acoustic wave in real time, making it opposite to the noise spectrum, thus effectively canceling the noise;

[0087] By adjusting the spectral features of the reverse acoustic wave through the LMS algorithm, it can be ensured that the frequency components of the reverse acoustic wave are opposed to those of the noise signal, thus eliminating the noise. The amplitude and phase of the reverse acoustic wave must match those of the noise signal, but with opposite phases, to achieve effective noise cancellation;

[0088] The generation of the reverse acoustic wave ensures that the noise is eliminated in a specific frequency band. Through the analysis of the noise spectrum and the precise generation of the reverse acoustic wave, the effective reduction or elimination of noise can be achieved in a noise environment;

[0089] By optimizing the generation of the reverse acoustic wave through the LMS algorithm, it can be ensured that the best effect of the reverse acoustic wave is always maintained in a dynamic and changing noise environment; the self-adaptability of the LMS algorithm enables it to cope with the changes in the noise signal and continuously adjust the reverse acoustic wave to maximize the noise cancellation effect; through precise spectral matching, the reverse acoustic wave can effectively reduce or eliminate the noise and improve the environmental quality;

[0090] Specifically, as Figure 3 shown, in step S22, convert the time-domain signal into a frequency-domain signal by performing a mixed-radix FFT on the time-domain signal; analyze the spectral features of the noise signal according to the frequency-domain signal, and the specific operation steps are as follows:

[0091] S221: Use the mixed-radix FFT to decompose the time-domain signal into multiple relatively prime factors according to the length of the time-domain signal;

[0092] Perform successive base transformations (i.e., the mixed-radix FFT) on each relatively prime factor to obtain the base group data of the base transformation of each relatively prime factor;

[0093] It should be noted that by using the mixed - radix FFT to decompose the time - domain signal into multiple relatively prime factors, "relatively prime factors" means decomposing the signal length N into the product of several relatively prime numbers (i.e., the greatest common divisor is 1). For example, if the signal length N = 105, it can be decomposed into 3×5×7, and each factor (such as 3, 5, 7) corresponds to the basis of a "basis transformation"; decomposing the length N of the time - domain signal into multiple relatively prime factors (such as 3, 5, 7) and using these factors to perform successive mixed - radix FFT on the signal can greatly improve the calculation efficiency of the FFT; this is because the combination of relatively prime factors has good mathematical properties, which greatly reduces the amount of calculation in the decomposition process; conventional FFT processing usually requires the signal length to be a power of 2, while the FFT transformation method of relatively prime factors (mixed - radix FFT) does not require the signal length to be a specific power, so it can effectively process signals of non - standard lengths, such as audio signals, noise signals, etc., avoiding padding with zeros to meet the requirements of the standard FFT.

[0094] And perform small - scale mixed - radix FFT on each subsequence corresponding to the relatively prime factors respectively, that is, perform basis transformation on 3, 5, and 7 respectively; first, rearrange the time - domain signal according to the decomposition factors 3, 5, 7; then perform the mixed - radix FFT of base 3 on the first factor (such as 3) to obtain an intermediate result of 3 sampling points (i.e., x(0), x(3), x(6), …… x(1), x(4), x(7), …, and x(2), x(5), x(8),); then perform the mixed - radix FFT of base 5 on the second factor (such as 5) to obtain 5 sampling points and continue to update the result; finally, perform the mixed - radix FFT of base 7 on the third factor (such as 7) to obtain 7 sampling points to complete the frequency - domain representation; the result of each basis transformation is called "basis - set data", which is the frequency - domain component corresponding to the current factor, and these basis - set data are the components of the final frequency - domain representation.

[0095] By decomposing into relatively prime factors and performing successive basis transformations, the FFT can efficiently process signals of any length, especially suitable for non - standard - length audio signals (such as noise signals); through efficient frequency - domain analysis of the time - domain signal, the main noise sources and their corresponding frequency ranges can be more accurately identified, which can help more precisely capture the noise spectrum characteristics and provide more effective noise - cancellation solutions, especially suitable for noise control in complex environments; this information is used later to design filters to ensure that the generated reverse sound wave can effectively cancel the original noise; through the frequency - domain analysis of the signal by this method, it can accurately determine which frequency components contain noise, so as to design the corresponding reverse sound wave. The reverse sound wave cancels the noise frequency components to achieve noise cancellation.

[0096] S222: Re - order the basis - set data of the basis transformation of each relatively prime factor, and combine the relatively prime factors of the re - ordered basis - set data to obtain the frequency - domain signal;

[0097] It should be noted that the signal length N is decomposed into multiple relatively prime factors (such as 3×5×7). The result of the basis transformation for each relatively prime factor is the frequency-domain component based on the current factor. To adapt to the transformation requirements of the next-level factor, the basis set data must be rearranged. This is because after each level of transformation, the organization of the basis set data changes according to the current factor, and the transformation of the next-level factor needs to process the data in a new organizational form in order to obtain an accurate frequency-domain signal.

[0098] The result of each level of basis transformation is for the frequency-domain components (i.e., the basis set data) of the current factor, and the order of these basis set data is usually different from the order of the original signal (i.e., the time-domain signal. As explained in the explanation in step S221, the relatively prime factors are rearranged after the time-domain signal decomposition, so it is different from the order of the time-domain signal). The transformation of the next-level factor needs to reorganize the basis set data according to the order of its own time-domain signal. Therefore, rearrangement must be performed to ensure the accuracy of the final frequency-domain signal. For example, the base 3 is divided into 5 groups of the base 5 mixed-radix FFT (i.e., x(0), x(5), x(10),…x(1), x(6), x(11) to x(4), x(9), x(14)).

[0099] The result of the transformation for each relatively prime factor needs to be multiplied by a rotation factor (also called the phase factor in the "butterfly operation") to compensate for the phase differences between different frequency components. Finally, the results of all basis transformations are combined in order to form a complete frequency-domain representation.

[0100] In noise processing, some noises may be concentrated in specific non-standard frequency bands (such as the combined frequency range of 3×5×7). The mixed-radix FFT can directly analyze these frequency bands without adjusting the signal length. This method can more accurately capture the spectral characteristics of the noise signal, especially in non-standard frequency bands. The result of the basis transformation for each relatively prime factor needs to be multiplied by a rotation factor (also called the phase factor in the "butterfly operation") to compensate for the phase differences between different frequency components. This ensures the phase consistency between the transformations of each relatively prime factor, thereby improving the accuracy of spectral analysis. Rearranging the basis set data and adding the rotation factor can ensure that each component in the frequency-domain signal is correctly combined, avoiding phase confusion caused by different transformation orders. The rotation factor (the phase factor in the butterfly operation) is used to compensate for the phase differences between frequency components and ensure the phase consistency during the synthesis of the entire frequency-domain signal. The use of rearrangement and rotation factor ensures the accurate synthesis of the frequency-domain signal, avoids incorrect results caused by inconsistent frequency-domain data, and improves the accuracy and reliability of the entire frequency-domain analysis.

[0101] S223: Analyze the amplitude spectrum and phase spectrum of the frequency-domain signal;

[0102] Find the maximum amplitude value in the frequency-domain signal through the amplitude spectrum, and determine the main frequency in the frequency-domain signal;

[0103] Determine the low-frequency band and high-frequency band in the frequency-domain signal according to the main frequency in the frequency-domain signal;

[0104] Determine a specific frequency band through the low-frequency band and high-frequency band; Calculate the squared amplitude for each frequency point in the specific frequency band, and sum the squared amplitudes to obtain the band energy;

[0105] Draw a spectrogram through the band energy and the amplitude spectrum; Integrate the frequency bands in the frequency-domain signal through the spectrogram to obtain the frequency distribution (i.e., the distribution of the band energy);

[0106] Take the frequency distribution of the frequency-domain signal as the frequency feature;

[0107] It should be noted that by calculating the amplitude spectrum and phase spectrum of the frequency-domain signal, the spectral characteristics of the signal can be comprehensively analyzed, and the main frequency components (main frequency) of the signal can be obtained; The amplitude spectrum reflects the intensity of the signal at different frequencies, while the phase spectrum shows the phase relationship of each frequency component; By finding the maximum amplitude value through the amplitude spectrum, the main frequency of the signal can be identified. The main frequency helps to determine the main components of the signal, and by analyzing the frequency bands (low-frequency band and high-frequency band), the frequency range and characteristics of the noise can be further refined;

[0108] After determining the main frequency, noise suppression can be carried out for a specific frequency band. When designing the reverse sound wave, ensure that it is consistent with the main frequency and characteristic frequency of the noise, so as to achieve more effective noise cancellation;

[0109] Through the analysis of the amplitude spectrum and phase spectrum, the key frequency characteristics can be effectively extracted from the signal, providing an important basis for subsequent filter design and noise suppression; By accurately identifying the frequency characteristics of the noise source, the reverse sound wave can be designed more accurately to ensure the best noise cancellation effect;

[0110] By summing the squared amplitudes within a specific frequency band, the band energy can be obtained, which can quantify the energy intensity of the signal in this frequency band; Band energy analysis helps to understand the energy distribution of the signal, especially in noise control, it can locate which frequency bands the noise mainly concentrates on; By plotting the distribution of the band energy to obtain the frequency distribution, the energy situation of different frequency bands in the signal can be clearly shown; This is crucial for noise analysis and filter design;

[0111] Through frequency distribution, it is possible to accurately analyze which frequency bands the noise is concentrated in, and then design reverse sound waves corresponding to these frequency bands to ensure that the generated reverse sound waves can effectively cancel the noise.

[0112] Specifically, as Figure 4 shown, in step S23, the spectral features are filtered according to the LMS algorithm to obtain reverse sound waves. The specific operation steps are as follows:

[0113] S231: Use a filter to assign complex weights to each frequency band of the spectral features, and initialize the filter coefficients according to the complex weights;

[0114] Filter the frequency signal of the spectral features through the LMS algorithm and the filter coefficients; generate the spectral signal of the reverse sound wave;

[0115] It should be noted that in step S223, it has been stated that the spectral signal contains not only amplitude information but also phase information. By assigning complex weights, it is possible to adjust both the amplitude and phase of the signal simultaneously, enabling the filter to be optimized according to the characteristics of each frequency band. This processing method avoids pure amplitude adjustment and can more precisely control the frequency-domain characteristics of the signal; at the same time, the obtained band energy is used to obtain the frequency distribution, that is, the frequency characteristics; and the frequency distribution reflects the distribution of band energy and indirectly reflects each frequency band in the frequency signal; assign complex weights to each frequency band; and initialize the filter coefficients according to the complex weights. The initialization of the filter coefficients can be zero or small random values to avoid introducing additional noise in the initial stage and ensure that the system starts to be optimized gradually from the "zero intervention" state (that is, initializing the filter coefficients to zero or small random values avoids noise interference in the initial state and ensures that the system starts to be optimized gradually from the "zero intervention" state and avoids the system being possibly affected by excessive initial interference; through gradual adjustment, such an initialization method helps the system enter the adaptive filtering process more smoothly);

[0116] Use the LMS algorithm to simultaneously adjust the amplitude and phase of the complex weights to ensure that the reverse sound wave matches the noise amplitude and has the opposite phase (phase difference of 180°) within the target frequency band; in this way, the frequency-domain (that is, the frequency signal) processing can independently optimize each frequency band and avoid the global delay problem of time-domain filtering, especially suitable for multi-band noise (such as containing both low-frequency rumbling and high-frequency whistling simultaneously);

[0117] In this step, the LMS (Least Mean Square) algorithm is used to continuously adjust the filter coefficients, enabling the filter output to gradually approximate the reverse signal of the noise, thereby achieving noise suppression. The LMS algorithm calculates the error and updates the filter weights based on the error, adjusting the frequency characteristics (including amplitude and phase) of the reverse sound wave to ensure that the reverse sound wave opposite to the noise signal can be precisely reversed within the target frequency band. Using the LMS algorithm enables the filter to simultaneously adjust the amplitude and phase of the signal, ensuring that the reverse sound wave and the noise have matching amplitudes and opposite phases within the target frequency band, achieving the effect of acoustic cancellation. Especially for multi-band noise, the LMS algorithm can independently optimize each frequency band, avoiding the global delay problem of time-domain filtering. For multi-band noises such as low-frequency rumbling and high-frequency whistling, the frequency-domain optimization of LMS is particularly advantageous because it can effectively suppress noises in different frequency bands without being restricted by global delay.

[0118] Frequency-domain processing can independently optimize the noise suppression effect of each frequency band without introducing global delay. Especially when facing multi-band noise, traditional time-domain filtering may introduce unnecessary delays, while frequency-domain filtering can perform noise suppression more precisely and efficiently. Therefore, the combination of complex weights and the LMS algorithm can not only adjust the amplitude of the signal but also control the phase, which is crucial for generating accurate reverse sound waves. The accurate generation of reverse sound waves can effectively suppress noise, thereby improving the ambient noise conditions.

[0119] S232: Perform inverse transformation on the spectral signal of the reverse sound wave through a mixed-radix FFT to obtain the reverse sound wave of the time-domain signal.

[0120] It should be noted that a mixed-radix FFT (combining radix-2 and radix-4 algorithms) is used to transform the spectral signal of the reverse sound wave into a time-domain signal. The inverse transformation is to reverse the frequency signal and re-convert it into a time-domain signal. This transformation process is the same as step S22 and can both accelerate the algorithm operation speed and improve the accuracy of the reverse sound wave. In the process of generating the reverse sound wave, converting the frequency-domain signal back to the time-domain signal is a key step. Through the mixed-radix FFT, the spectral signal of the reverse sound wave can be efficiently transformed to obtain the reverse sound wave signal in the time domain. Such a reverse signal can interfere with the noise signal when they meet, thereby achieving noise cancellation. In a noise cancellation system, the reverse sound wave needs to be generated within the coherence time of the noise signal and have the opposite phase to the noise signal. If the delay in generating the reverse sound wave is too long, the effect will be greatly reduced. The mixed-radix FFT reduces the computational complexity by optimizing the calculation process, enabling the reverse sound wave to be generated faster and meeting the requirements of real-time processing.

[0121] The mixed-radix FFT reduces the computational complexity and is suitable for real-time systems. The optimized IFFT frame processing delay can be controlled within 1 ms, meeting the timing requirements for acoustic cancellation. In terms of real-time requirements, the reverse sound wave must be generated within the noise coherence time (for example, the period of low-frequency noise is about 10 ms). The mixed-radix FFT balances speed and accuracy. The optimized FFT algorithm can control the frame processing delay within 1 ms, which is very important for real-time acoustic cancellation. Real-time requirements mean that the reverse sound wave must be generated within the noise coherence time. Therefore, the mixed-radix FFT balances speed and accuracy, ensuring the accuracy and timeliness of the reverse sound wave. By performing the inverse transform of the mixed-radix FFT on the frequency-domain signal to obtain the time-domain signal, reverse sound waves are finally generated. These reverse sound waves will interfere with the noise signal, effectively reducing the intensity of the noise and improving the quality of the noise environment.

[0122] Flexibly adapt to complex noise environments: According to the frequency characteristics and complexity of different noise sources, the system can adjust the parameters of the reverse sound wave in real time to adapt to complex noise environments and ensure the continuous stability of the noise cancellation effect.

[0123] Specifically, in step S4, error microphones are arranged between the speaker array and the noise source to form an error microphone array. The residual noise signal that appears after the reverse sound wave played by the speaker array cancels the noise source is collected through the error microphone array. The following are the specific operation steps for judging whether the device has a fault by setting a threshold for the residual noise signal:

[0124] S41: With the noise source as the center, a grid-like distribution is carried out between the noise source and the speaker array.

[0125] Error microphones are arranged in the grid to form an error microphone array. The distance between each error microphone in the error microphone array and the nearest speaker is less than 1 / 4 of the wavelength of the noise signal (that is, 1 / 4 of the wavelength of the noise source. The wavelength of the noise signal is related to the frequency of the frequency-domain signal. The 1 / 4 wavelength of the noise signal is obtained through the low-frequency band and the high-frequency band. The frequency of the frequency-domain signal has been described in step S223 and will not be elaborated here).

[0126] It should be noted that the error microphone array must be placed in the sound field superposition area between the noise source and the speaker array, and the spacing needs to meet the spatial sampling wavelength distance (that is, the distance between each error microphone in the error microphone array and the nearest speaker is less than 1 / 4 of the wavelength of the noise signal) to avoid signal distortion caused by phase delay.

[0127] S42: The residual noise signal after the cancellation between the reverse sound wave and the noise source is collected through the error microphone array.

[0128] Analyze the residual noise signal through the hybrid radix FFT, extract the frequency-domain energy distribution in the residual noise signal, and calculate the total residual energy in the residual noise signal based on the frequency-domain energy in the residual noise signal;

[0129] It should be noted that the small residual noise signals after cancellation are collected by the error microphone array, and these small residual noise signals are analyzed by the hybrid radix FFT. This step can refer to the step of obtaining the band energy by expanding step S22 (that is, the expansion scheme of this step S42 is the same as that of step S22 and will not be elaborated here), so as to obtain the frequency-domain energy distribution. By calculating the sum of the distributed frequency-domain energy, the total residual energy is obtained;

[0130] S43: Preset a residual noise threshold k, and determine whether the total residual energy is greater than the residual noise threshold k;

[0131] If not, it is determined that the device has no fault problem;

[0132] If so, it is determined that the device has a noise reduction fault and an alarm is triggered.

[0133] It should be noted that by setting the residual noise threshold k, it is possible to accurately determine whether the residual noise signal after cancellation has been maximally eliminated; if it has not been maximally eliminated, it will be greater than the residual noise threshold k, triggering an alarm, thereby reminding the operator that the device has a fault and the noise has not been maximally eliminated;

[0134] Embodiment 2

[0135] As Figure 5 shown, correspondingly, the present invention also proposes a system for active noise reduction of a linear noise source, including: an acquisition module 10; a unit module 20; an elimination module 30; an early warning module 40;

[0136] The acquisition module 10 is used to arrange a plurality of microphones for the noise source to form a microphone array; the noise signal of the noise source is collected through the microphone array;

[0137] The unit module 20 is used to convert the noise signal into a frequency-domain signal through the hybrid radix FFT, analyze the frequency-domain signal to obtain the spectral characteristics of the noise signal; generate a reverse sound wave according to the spectral characteristics;

[0138] The elimination module 30 is used to establish a speaker array for the noise source, transmit the reverse sound wave to the speaker array, play the reverse sound wave to the noise source through the speaker array, and eliminate the noise source.

[0139] The warning module 40 is configured to arrange error microphones between the speaker array and the noise source to form an error microphone array; collect residual noise signals that appear after the reverse sound waves played by the speaker array eliminate the noise source through the error microphone array; and determine whether the device has a fault by setting a threshold for the residual noise signals.

[0140] Embodiment III

[0141] As Figure 6 shown, on the other hand, based on the method for active noise reduction of a linear noise source provided in Embodiment I of the invention, Embodiment III also provides a computer storage medium 1140 (abbreviated as storage medium). The following is a schematic structural framework diagram of the computer storage medium provided for Embodiment III of the present invention, which includes:

[0142] A memory 1130 for storing computer programs;

[0143] A communication interface 1120 for connecting the memory 1130 to the processor 1110;

[0144] A processor 1110 for executing computer programs to implement the method for active noise reduction of a linear noise source involved in Embodiment I disclosed in any combination of the above embodiments.

[0145] In summary, from the method, system, and device for active noise reduction of a linear noise source proposed in the embodiments of the present invention, it can be seen that by reasonably arranging multiple microphones, the acoustic wave signals from the noise source can be effectively captured, ensuring the comprehensiveness and accuracy of the signals; the distance between the microphones is less than half of the noise wavelength, avoiding the phenomenon of frequency aliasing, thus ensuring the accuracy of signal acquisition and avoiding the distortion or mixing of high-frequency noise.

[0146] Furthermore, the noise signal captured by the microphone array is converted into a digital signal using an analog-to-digital converter to achieve digital processing and generate a discrete time-domain signal; the use of a mixed-radix FFT can efficiently process time-domain signals of non-standard lengths, avoiding the limitations of traditional FFT; by decomposing the signal length into multiple relatively prime factors, the computational amount is significantly reduced and the transformation speed is increased; frequency-domain analysis can extract the frequency distribution of the noise signal and identify the main noise frequency components; the result of the base transformation of each relatively prime factor is the frequency-domain component based on the current factor; in order to adapt to the transformation requirements of the next-level factor, the base-group data must be rearranged so as to obtain an accurate frequency-domain signal; by calculating the amplitude spectrum and phase spectrum of the frequency-domain signal, the spectral characteristics of the frequency-domain signal can be comprehensively analyzed. The amplitude spectrum reflects the intensity of the signal at different frequencies, while the phase spectrum shows the phase relationship of each frequency component; by finding the maximum amplitude value through the amplitude spectrum, the main frequency of the signal can be identified; through the analysis of the main frequency as well as the amplitude spectrum and phase spectrum, the key frequency features can be effectively extracted from the signal to ensure the best noise cancellation effect.

[0147] Furthermore, through the LMS algorithm, the complex weights can adjust both the amplitude and the phase simultaneously, enabling real-time processing of the noise signal and generating a reverse sound wave; the mixed-radix FFT (combining the radix-2 and radix-4 algorithms) is used to convert the spectral signal of the reverse sound wave into a time-domain signal; the inverse transformation is to reverse the frequency signal and re-convert it into a time-domain signal; the role of the reverse sound wave in the time-domain signal is to interfere with the noise signal, reducing the noise through phase inversion. The reverse sound wave is transmitted through the speaker array, and the role of the speaker array is to accurately play these reverse sound waves so that an interference effect is generated near the noise source, thereby achieving active noise cancellation.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; those of ordinary skill in the art can modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for active noise reduction of a linear noise source, characterized in that: The steps are as follows: Placing multiple microphones on the noise source to form a microphone array; collecting noise signals from the noise source through the microphone array; Converting the noise signal into a frequency domain signal by performing mixed basis FFT on the noise signal, and analyzing the frequency domain signal to obtain a frequency spectrum feature of the noise signal; Generate reverse sound waves according to the frequency spectrum characteristics; Establishing a speaker array for the noise source, transmitting the reverse sound wave to the speaker array, and playing the reverse sound wave to the noise source through the speaker array to eliminate the noise source; Arranging an error microphone between the speaker array and the noise source to form an error microphone array; The error microphone array is used to collect the residual noise signal that occurs after the speaker array plays the reverse sound wave to eliminate the noise source; and a threshold is set to determine whether the device has a fault based on the residual noise signal.

2. The method for active noise reduction of a linear noise source according to claim 1, characterized in that: The spacing between microphones of the microphone array is less than half of the noise wavelength of the noise source.

3. The method for active noise reduction of a linear noise source according to claim 2, characterized in that: The spacing between the loudspeakers in the loudspeaker array is the same as the spacing between the microphones in the microphone array.

4. The method for active noise reduction of a linear noise source according to claim 3, characterized in that: The noise signal is converted into a frequency domain signal by performing mixed basis FFT, and the frequency domain signal is analyzed to obtain the frequency spectrum characteristics of the noise signal; and a reverse sound wave is generated according to the frequency spectrum characteristics. The specific operation steps are as follows: The noise signal captured by the microphone array is converted into a digital signal by using an analog-to-digital converter, and a discrete time domain signal xi(n) is obtained by quantization; Convert the time domain signal into a frequency domain signal by performing mixed basis FFT; and analyze the frequency spectrum characteristics of the noise signal according to the frequency domain signal; The frequency spectrum characteristics are filtered according to the LMS algorithm to obtain the reverse sound wave.

5. The method for active noise reduction of a linear noise source according to claim 4, characterized in that: The time domain signal is converted into a frequency domain signal by performing mixed basis FFT, and the specific operation steps are as follows: Decomposing the time domain signal into a plurality of coprime factors according to the length of the time domain signal by using a mixed basis FFT; Performing a basis transformation on each coprime factor step by step to obtain basis set data of the basis transformation of each coprime factor; The basis set data of the basis transformation of each coprime factor are reordered, and the coprime factors of the reordered basis set data are combined to obtain a frequency domain signal.

6. A method for active noise reduction of a linear noise source according to claim 5, characterized in that: The frequency spectrum characteristics of the noise signal are analyzed according to the frequency domain signal. The specific operation steps are as follows: Analyze the amplitude spectrum and phase spectrum of the frequency domain signal; Finding the maximum amplitude value in the frequency domain signal through the amplitude spectrum to determine the main frequency in the frequency domain signal; Determine a low frequency band and a high frequency band in the frequency domain signal according to a main frequency in the frequency domain signal; Determine a specific frequency band through the low frequency band and the high frequency band; calculate the square of the amplitude of each frequency point in the specific frequency band, and sum the square of the amplitude to obtain the frequency band energy; Draw a spectrum diagram using the frequency band energy and the amplitude spectrum; integrate the frequency bands in the frequency domain signal using the spectrum diagram to obtain a frequency distribution; The frequency distribution of the frequency domain signal is used as a frequency feature.

7. A method for active noise reduction of a linear noise source according to claim 6, characterized in that: The frequency spectrum characteristics are filtered according to the LMS algorithm to obtain the reverse sound wave. The specific operation steps are as follows: Using a filter to assign a complex weight to each frequency band of the spectrum feature, and initializing a filter coefficient according to the complex weight; The frequency signal of the frequency spectrum characteristic is filtered by LMS algorithm and filter coefficient to generate the frequency spectrum signal of the reverse sound wave; The frequency spectrum signal of the reverse sound wave is inversely converted through mixed basis FFT to obtain the reverse sound wave of the time domain signal.

8. A method for active noise reduction of a linear noise source according to claim 7, characterized in that: An error microphone is arranged between the speaker array and the noise source to form an error microphone array; the error microphone array collects the residual noise signal after the speaker array plays the reverse sound wave to eliminate the noise source; and the residual noise signal is judged by setting a threshold value whether the device has a fault. The specific operation steps are as follows: Taking the noise source as the center, a grid-shaped distribution is performed between the noise source and the speaker array; Arranging error microphones in a grid to form an error microphone array; The distance between each error microphone in the error microphone array and the nearest speaker is less than 1 / 4 of the wavelength of the noise signal; The error microphone array is used to collect the residual noise signal eliminated between the reverse sound wave and the noise source; Analyzing the residual noise signal by using the mixed basis FFT, extracting the frequency domain energy distribution in the residual noise signal, and calculating the total residual energy in the residual noise signal by using the frequency domain energy in the residual noise signal; Preset a residual noise threshold k, and determine whether the total residual energy is greater than the residual noise threshold k; If not, it is determined that the device has no fault problem; If so, it is determined that the device has a noise reduction failure and an alarm is triggered.

9. A system for active noise reduction of linear noise sources, characterized in that: include: Acquisition module; Unit module; Elimination module; Early warning module; The acquisition module is used to place multiple microphones at the noise source to form a microphone array; Collecting noise signals from the noise source by using the microphone array; The unit module is used to convert the noise signal into a frequency domain signal by performing mixed basis FFT, analyze the frequency domain signal to obtain the frequency spectrum characteristics of the noise signal; and generate an inverse sound wave according to the frequency spectrum characteristics; The elimination module is used to establish a speaker array for the noise source, transmit the reverse sound wave to the speaker array, and play the reverse sound wave to the noise source through the speaker array to eliminate the noise source; The early warning module is used to arrange error microphones between the speaker array and the noise source to form an error microphone array; The error microphone array is used to collect the residual noise signal that occurs after the speaker array plays the reverse sound wave to eliminate the noise source; and a threshold is set to determine whether the device has a fault based on the residual noise signal.

10. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the method for active noise reduction of a linear noise source as described in any one of claims 1 to 7 are implemented.

Images