CN-121983014-A - Semi-adaptive active noise reduction method for range hood with self-checking function

CN121983014ACN 121983014 ACN121983014 ACN 121983014ACN-121983014-A

Abstract

The invention discloses a semi-adaptive active noise reduction method of a range hood with a self-checking function, which is applied to a semi-adaptive active noise reduction system of the range hood. In the pre-training stage, a device position self-checking vector expression library, a device oil stain self-checking vector expression library, a range hood sound field database, a secondary channel estimation function library, a controller coefficient library, a range hood sound field feature library and a range hood sound field vector expression library are respectively established, and in the actual noise reduction stage, firstly, the self-checking of the position and oil stain state of an electroacoustic device is completed through image acquisition equipment, and noise reduction is interrupted when abnormality occurs. Otherwise, judging the sound field environment type of the range hood audio signal frame in real time, and selecting corresponding parameters from the secondary channel estimation function library and the controller coefficient library to perform active noise reduction and parameter fine adjustment respectively, so that the problems of slow convergence and easy howling of a fully self-adaptive system are avoided, and the defect that a fixed parameter system is difficult to adapt to environment change is overcome.

Inventors

LI LIN
YUAN YANG
XU YONGKANG
LOU QI
YUAN ZHENG
FENG FAN
XIE MINJIE
ZENG MING
LI YANRAN

Assignees

中电海康集团有限公司

Dates

Publication Date: 20260505
Application Date: 20251225

Claims (10)

1. The self-adaptive active noise reduction method for the range hood with the self-checking function is characterized by being applied to a self-adaptive active noise reduction system of the range hood, wherein the self-adaptive active noise reduction system of the range hood comprises image acquisition equipment, an electroacoustic device and a controller, the electroacoustic device comprises a reference microphone, a secondary loudspeaker and an error microphone, and the self-checking function of the range hood comprises the following steps: Pre-training stage: The method comprises the steps of respectively establishing a device position self-checking vector expression library, a device oil stain self-checking vector expression library, a range hood sound field database, a secondary channel estimation function library, a controller coefficient library, a range hood sound field feature library and a range hood sound field vector expression library, wherein the range hood sound field database comprises M-class sound field environments with normal electroacoustic device positions, the secondary channel estimation function library comprises M-class secondary channel estimation functions which are respectively in one-to-one correspondence with the M-class sound field environments, the controller coefficient library comprises M-class controller coefficients which are respectively in one-to-one correspondence with the M-class sound field environments, and the range hood sound field vector expression library comprises M-class vector expressions which are respectively in one-to-one correspondence with the M-class sound field environments; The actual noise reduction stage: at the moment of starting the range hood, acquiring current image frames which are acquired by the image acquisition equipment and contain all electroacoustic devices, judging the position type of each electroacoustic device in the current image frames according to the device position self-checking vector expression library, and interrupting the work of the semi-adaptive active noise reduction system of the range hood when the position of the electroacoustic device is abnormal; Judging the oil stain state of the electroacoustic device of the current image frame according to the self-checking vector expression library of the oil stain of the electroacoustic device when the position of the electroacoustic device is normal, and interrupting the work of the semi-adaptive active noise reduction system of the range hood when the oil stain state of the electroacoustic device is abnormal; When the oil stain state of the electroacoustic device is normal, acquiring a current range hood audio signal frame acquired by a reference microphone in real time, extracting features corresponding to the range hood audio signal frame according to a range hood sound field feature library, normalizing the extracted features, and splicing the normalized features to obtain spliced features; Inputting the splicing characteristics of the current range hood audio signal frame into a trained range hood noise classification model to obtain vector expression of the current range hood audio signal frame, comparing the vector expression of the current range hood audio signal frame with a range hood sound field vector expression library in similarity, and judging the sound field environment type of the current range hood audio signal frame by combining with the oil stain state of an electroacoustic device of the current image frame; Selecting corresponding secondary channel estimation functions and controller coefficients from a preset secondary channel estimation function library and a controller coefficient library respectively according to the sound field environment type of the current range hood audio signal frame; the controller adopts a self-adaptive algorithm, calculates an inverted sound wave signal with the phase opposite to that of the current range hood audio signal frame according to the selected secondary channel estimation function and the controller coefficient and the residual noise acquired by the reference microphone, and transmits the calculated inverted sound wave signal to the secondary loudspeaker for emission, so that noise reduction is realized; And monitoring the noise reduction effect of the semi-adaptive active noise reduction system based on the residual noise acquired by the error microphone.
2. The method for semi-adaptive active noise reduction of the range hood with the self-checking function according to claim 1, wherein in the process of establishing a device position self-checking vector expression library, an image acquisition device is utilized to acquire a plurality of first image frames which simultaneously contain all electroacoustic devices, and all the first image frames are divided into two categories of normal electroacoustic device positions and abnormal electroacoustic device positions according to whether the electroacoustic device positions are abnormal or not, each category contains W first image frames, and each first image frame is marked with labels and numbers of the electroacoustic device position categories; and carrying out vector expression on each first image frame in each class, and carrying out average value calculation on the vector expressions of the W first image frames in each class to obtain vector expressions of two classes of first image frames, wherein the vector expressions of the two classes of first image frames form a device position self-checking vector expression library.
3. The semi-adaptive active noise reduction method of the range hood with the self-checking function of claim 1, wherein the oil stain state of the electroacoustic device comprises an oil stain type and an attached state; In the process of establishing a device oil stain self-checking vector expression library, acquiring a plurality of second image frames containing all electroacoustic devices by using image acquisition equipment, dividing all the second image frames into H types according to oil stain types and attachment states, wherein each type of second image frames contains K Zhang Dier image frames, the second image frames of the same type have the same oil stain types and attachment states, and labeling each second image frame with a label and a number of the oil stain state of the electroacoustic device; And carrying out vector expression on each second image frame in each class, and carrying out average value calculation on the vector expression of the K Zhang Dier image frames in each class to obtain the vector expression of the H-class second image frames, wherein the vector expression of the H-class second image frames forms a device oil stain self-detection vector expression library.
4. The method for semi-adaptive active noise reduction of a range hood with a self-checking function according to claim 3, wherein each type of sound field environment in a sound field database of the range hood comprises J range hood data, a reference microphone is used for collecting continuous I-frame range hood audio signal frames as one range hood data, each range hood data is calibrated to obtain calibration information, and the calibration information comprises sound field environment types, oil stain states of electroacoustic devices, fan gears and serial numbers.
5. The method for semi-adaptive active noise reduction of a range hood with a self-checking function according to claim 4, wherein the step of establishing a secondary channel estimation function library and a controller coefficient library comprises the steps of: Aiming at the sound field environment corresponding to each range hood data, carrying out secondary channel identification by adopting a secondary channel identification algorithm to obtain a secondary channel estimation function, substituting the secondary channel estimation function into a semi-adaptive active noise reduction system, carrying out active noise reduction by adopting an adaptive algorithm, and recording the secondary channel estimation function and the controller coefficient corresponding to the current range hood data when the semi-adaptive active noise reduction system achieves the optimal noise reduction effect; Aiming at M-class sound field environments, summing and averaging secondary channel estimation functions of J pieces of range hood data in each class to obtain secondary channel estimation functions of various sound field environments, and forming a secondary channel estimation function library by the secondary channel estimation functions of the M-class range hood data; And aiming at M kinds of sound field environments, summing and averaging the controller coefficients of J pieces of range hood data in each kind of sound field environments to obtain the controller coefficients of all kinds of sound field environments, and forming a controller coefficient library by the controller coefficients of the M kinds of range hood data.
6. The method for semi-adaptive active noise reduction of a range hood with a self-checking function according to claim 5, wherein the step of establishing a range hood sound field feature library comprises the following steps: For each piece of range hood data in a range hood sound field database, respectively calculating fifteen features of zero-crossing rate, short-time amplitude, energy, absolute average value, root mean square value, variance, standard deviation, kurtosis, skewness, peak index, peak factor, margin coefficient, spectrum centroid, mean square frequency and frequency variance of each range hood audio signal frame in the range hood data by using sampling points, and obtaining fifteen features of which each feature length is an I frame; The variance of fifteen features with the length of the I frame is calculated respectively, the first ten features with the highest variance are selected from the fifteen features, and then the pearson correlation coefficient is adopted to calculate the correlation between any two features in the ten features: When the correlation between the two features is smaller than a preset value, the correlation between the two features is low, and the two features are reserved; When the correlation between the two features is larger than or equal to a preset value, and the correlation between the two features is high, selecting the features with large variances from the two features to be reserved; Finally, only five characteristics, namely five main characteristics, namely short-time amplitude, kurtosis, peak index, margin coefficient and spectrum centroid are reserved, and the five main characteristics with the length of I frames are reserved for the I frame range hood audio signal frames of each range hood data in a range hood sound field database; Respectively carrying out normalization operation on each frame of the five main features to obtain five main features with normalized length of an I frame; and forming a range hood sound field feature library by using five main features of which the normalized length is I frames corresponding to all range hood data in the range hood sound field database.
7. The method for semi-adaptive active noise reduction of a range hood with a self-checking function according to claim 6, wherein in a pre-training stage, training a range hood noise classification model by adopting all range hood sound field features in a range hood sound field feature library to obtain a trained range hood noise classification model, wherein the range hood noise classification model is a deep neural network; in the pre-training stage, for each piece of range hood data in each type of sound field environment, extracting five main features with the length of I frames after normalization corresponding to the current piece of range hood data in the current type of sound field environment from a range hood sound field feature library, splicing the five main features after normalization, and inputting the five main features into a trained range hood noise classification model, wherein at a feature layer of the trained range hood noise classification model, outputting vector expression of the current piece of range hood data in the current type of sound field environment, carrying out mean value calculation on J vector expressions of the current type of sound field environment to obtain vector expression of the current type of sound field environment, and forming a range hood sound field vector expression library by vector expression of M type of sound field environment.
8. The method for semi-adaptive active noise reduction of a range hood with a self-checking function according to claim 5, wherein the monitoring of the noise reduction effect of the semi-adaptive active noise reduction system based on the residual noise collected by the error microphone comprises the following steps: real-time acquisition of residual noise e (n) using error microphone, mean square value of residual noise e (n) The calculation formula is as follows: ; under the condition that the range hood does not have howling, when the mean square value of residual noise The continuous preset seconds are unchanged and smaller than the first threshold, the semi-adaptive active noise reduction system achieves the optimal noise reduction effect, Representing the nth number in the residual noise e (n); and when the amplitude of the residual noise exceeds a second threshold value, judging that the range hood is in howling and immediately interrupting noise reduction.
9. The method for semi-adaptive active noise reduction of range hood with self-checking function according to claim 4, wherein the process of judging the position type of each electroacoustic device in the current image frame according to the device position self-checking vector expression library comprises the steps of normalizing the current image frame, carrying out electroacoustic device position vector expression on the normalized current image frame, carrying out similarity comparison on electroacoustic device position vector expression of the current image frame and vector expression in the device position self-checking vector expression library, and taking the device position type corresponding to the vector expression with highest similarity in the device position self-checking vector expression library as the electroacoustic device position type of the current image frame; The process of judging the electroacoustic device oil stain state of the current image frame according to the device oil stain self-checking vector expression library comprises the steps of carrying out vector expression of the device oil stain state on the normalized current image frame, carrying out similarity comparison on the vector expression of the electroacoustic device oil stain state of the current image frame and the vector expression in the device oil stain self-checking vector expression library, and taking the oil stain state corresponding to the vector expression with the highest similarity in the device oil stain self-checking vector expression library as the electroacoustic device oil stain state of the current image frame; The process of judging the sound field environment type of the current range hood audio signal frame comprises the steps of firstly reducing the range hood sound field vector expression library based on the electroacoustic device oil stain state of the current image frame, then carrying out similarity comparison on the vector expression of the current range hood audio signal frame and the reduced range hood sound field vector expression library, obtaining the sound field environment corresponding to the vector expression with the highest similarity in the reduced range hood sound field vector expression library, and taking the sound field environment as the sound field environment type of the current range hood audio signal frame.
10. The method of self-adaptive active noise reduction of a range hood with a self-checking function according to claim 1, further comprising an actual noise reduction stage, wherein the current range hood audio signal frame and the previous range hood audio signal frame have the same sound field environment type, the corresponding secondary channel estimation function and the controller coefficient are not required to be selected again, the controller coefficient is adjusted in a fine adjustment mode according to the noise reduction result by adopting an adaptive algorithm, and the calculation of the inverted sound wave signal is continued, and if the range hood audio signal frame which is larger than or equal to the continuous Q frame is different from the current range hood audio signal frame, the corresponding secondary channel estimation function and the controller coefficient are required to be selected again.

Description

Semi-adaptive active noise reduction method for range hood with self-checking function Technical Field The invention belongs to the technical field of noise reduction of range hoods, and particularly relates to a semi-self-adaptive active noise reduction method of a range hood with a self-checking function. Background The range hood is an indispensable household appliance in a modern family kitchen, is arranged above a cooking bench, and can quickly suck cooking fume and gas waste materials away, so that the damage to human bodies is avoided. However, the high-suction fan ensuring the smoking effect often generates strong noise, and the long-term noise environment can lead to hearing degradation and even deafness. Therefore, the mute range hood is certainly the research and development focus of various manufacturers of large range hoods. Aiming at the characteristic that the noise of the range hood is broadband noise below 1KHz, more and more range hood manufacturers begin to explore the active noise reduction technology. Active noise reduction has been widely used in the field of headphones. However, in the field of range hoods, the range hood is seriously affected by wind noise, oil stains and other factors, and a great room for improvement still exists; for example, in terms of system adaptation: The prior patent CN 108954442A relates to a range hood with a noise reduction device and a noise reduction method, and is provided with a range hood main body and a three-dimensional space sound field noise reduction device for active noise reduction. The noise reduction device adopts a fully self-adaptive active noise reduction algorithm, real-time autonomous iteration parameters are changed according to the sound field of the range hood, the calculated amount of the algorithm is large, the requirement on hardware configuration is extremely high, and the cost is also high. The method has the defects of difficult convergence and easy dispersion, and is likely to generate howling, thereby influencing the user experience; the prior patent CN116105191A fume exhauster, a noise reduction method, a system and a storage medium thereof acquire a target sampling rate and a target sound velocity, acquire an actual sound velocity corresponding to a current gear, determine the current sampling rate corresponding to the current gear according to the target sound velocity, the target sampling rate and the actual sound velocity, and reduce noise of the fume exhauster by utilizing preset corresponding noise reduction parameters. The method adopts a fixed coefficient mode, is simple in calculation and low in cost, and does not have the problem of system divergence. However, the actual sound field environment of the range hood is complex and changeable, and the sound field type of the range hood is judged only according to non-audio acquisition equipment, so that a good noise reduction effect is difficult to achieve. For example, in terms of device self-test: the prior patent CN117831493a discloses a scene self-adaptive active noise reduction method and system, which uses a reference microphone signal and a multi-mode sensor signal to train a scene classification model, if the output result of the scene classification model is a preset scene type and an influence factor of the scene, an ANC parameter table is searched according to the preset scene type and the influence factor of the scene, and an ANC parameter is determined to be used for the operation of an ANC control module. According to the method, various range hood scenes are considered, but the device self-checking function is not provided, all devices of the active noise reduction system are distributed in the inner cavity of the range hood just like a black box, and the system has good noise reduction effect on the premise that the positions of all devices are normal and the oil stain state of the devices does not influence the noise reduction effect; The prior patent CN108916944B relates to a range hood with noise reduction and visual detection functions and a noise reduction method, and is provided with a range hood main body, a three-dimensional space sound field noise reduction device for actively reducing noise and a visual detection system, wherein the visual detection system for monitoring smoke conditions is assembled on the range hood main body. The visual detection system is not arranged in the inner cavity of the range hood, is only used for detecting the smoke concentration of kitchen ware outside the range hood, and cannot monitor all devices of the active noise reduction system arranged in the inner cavity of the range hood; The prior patent CN116132900B discloses a self-checking method, a device and a medium of an active noise reduction system, which comprises at least one microphone and at least one loudspeaker, wherein the at least one loudspeaker plays corresponding preset test signals, at least one target audio signal is synchronously acquired by the