Virtual sensing method based on Kriging interpolation applied to active noise control

Abstract

The invention discloses a virtual sensing method based on Kriging interpolation, which is applied to active noise control. Comprising the following steps: (1) measuring and recording the distance between a virtual microphone and a physical microphone; (2) calculating the weighted weight of the physical microphone by adopting a Kriging interpolation method; (3) acquiring a sound pressure signal by using the physical microphone, weighting the sound pressure signal of the physical microphone by using the calculated weight, and estimating to obtain a sound pressure signal at the virtual microphone; and (4) updating by using an active noise control algorithm based on the estimated sound pressure signal at the virtual microphone. According to the method, the number and layout of the physical microphones are not required, and the sound pressure can be effectively estimated. The method can be applied to any active noise control algorithm, and the method is robust to the characteristic and azimuth change of the noise source position. If a secondary channel on-line modeling algorithm is combined, the method is also robust to the change of the secondary channel, such as the change of the position of a secondary sound source, the change of the position of a virtual microphone and the like.

Description

Technical Field

[0001] The invention belongs to the field of active noise control, and in particular relates to a virtual sensing method based on Kriging interpolation applied in active noise control. Background Art

[0002] Active noise control is a noise control method that makes the secondary sound source generate a signal that is inversely proportional to the original noise signal, thereby canceling each other out. In real life, due to the limitations of application scenarios, such as when used in active noise reduction headrests and ear clip headphones, it is inconvenient to place a physical microphone in the target noise reduction area. Virtual sensing technology is used to convert the target position of active noise control to obtain an effective noise reduction effect when there is no physical microphone at the target position. The target noise reduction position is called the virtual microphone position.

[0003] The current virtual sensing technology is mainly divided into two categories. One is based on additional filters. This type of method uses filters to model the sound pressure correspondence between physical microphones and virtual microphones, and then predicts the sound pressure at the virtual microphone by measuring the sound pressure with physical microphones. This type of method is generally sensitive to environmental changes, such as changes in noise source characteristics, positions, and acoustic transmission paths, which will cause the algorithm to diverge. The other is based on interpolation, represented by the forward difference method (CD Kestell, BS Cazzolato, CH Hansen, Active noise control in a free field with virtual sensors, J. Acoust. Soc. Am. 109 (2001) 232-243.), which assumes that the microphones are linearly distributed, and the order of interpolation is proportional to the number of microphones, which will lead to underfitting of the system when the number of microphones is low, and overfitting may occur when the number of microphones is large. Summary of the invention

[0004] The present invention proposes a virtual sensing method based on Kriging interpolation and applies it to active noise control. The method does not require pre-modeling of the sound pressure correspondence between a physical microphone and a virtual microphone and is insensitive to the direction and characteristics of the noise source. At the same time, the method has no requirements on the microphone placement method and number.

[0005] The technical solution adopted by the present invention is:

[0006] A virtual sensing method based on Kriging interpolation applied in active noise control comprises the following steps:

[0007] Step 1, measuring and recording the distance between the virtual microphone and the physical microphone;

[0008] Step 2, using Kriging interpolation method to calculate the weighted weights of the physical microphone;

[0009] Step 3, using a physical microphone to obtain a sound pressure signal, weighting the sound pressure signal of the physical microphone with the weighted weight calculated in step 2, and estimating the sound pressure signal at the virtual microphone;

[0010] Step 4: Based on the sound pressure signal at the virtual microphone estimated in step 3, the control filter coefficients are updated using an active noise control algorithm.

[0011] Furthermore, the step 2 is specifically as follows:

[0012] Assume that there are M physical microphones with positions x1(x1,y1,z1),…,x M (x M ,y M ,z M ), M physical microphones measure M sound pressures, denoted as e1, e2, …e M , estimate the sound pressure e0 at the virtual microphone x0(x0,y0,z0) based on the sound pressure of M microphones. Let the variation function be

[0013]

[0014] Where h is the distance between the physical microphone and the virtual microphone, and a is a set constant, called the correlation scale. ij =C(|x i -x j |), where |x i -x j | i.e. position x i and position x j The distance between them can be calculated by the matrix inversion method

[0015]

[0016] where λ1,…,λ M That is, the position is x1(x1,y1,z1),…,x M (x M ,y M ,z M ) is the weighted value of the sound pressure measured by the M physical microphones.

[0017] Furthermore, in step 4, any active noise control algorithm may be used, including any algorithm for online modeling of secondary channels.

[0018] The present invention proposes a new virtual sensing method, which is based on the Kriging interpolation method and has no requirements on the number and layout of physical microphones, and can effectively estimate the sound pressure. When applied to active noise control, the method is robust to the characteristics and azimuth changes of the noise source position. If combined with the secondary channel online modeling algorithm, the method is also robust to changes in the secondary channel, such as changes in the secondary sound source position, changes in the virtual microphone position, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is an algorithm block diagram of the control method of the present invention.

[0020] Figure 2 Schematic diagram of an experimental device in an embodiment of the present invention.

[0021] Figure 3 It is a comparison diagram between the effect of modeling the secondary channel using the method of the present invention (indicated by OMA, Overall Modelling Algorithm) and the actual secondary channel when the end of the experimental pipeline is a cover plate.

[0022] Figure 4 It is a comparison diagram between the effect of modeling the secondary channel using the method of the present invention (indicated by OMA, Overall Modelling Algorithm) and the actual secondary channel when the end of the experimental pipeline is open.

[0023] Figure 5 This is a comparison of the noise reduction effects of the OMA and FXNLMS algorithms. 4 At , the end of the experimental pipeline changes from being capped to being open. DETAILED DESCRIPTION

[0024] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0025] like Figure 1 As shown, the present invention is a virtual sensing method based on Kriging interpolation applied to active noise control, which mainly includes the following parts:

[0026] 1. Measure and record the distance between the virtual microphone and the physical microphone

[0027] Initialize the control filter to a non-zero vector, which can usually be simply set to the unit impulse response. Usually initialize the remaining vectors to 0.

[0028] 2. Use the Kriging interpolation method to calculate the weighted weights of the physical microphone.

[0029] 3. Estimate the sound pressure at the virtual sensor

[0030] The sound pressure signal is obtained by using a physical microphone, and the sound pressure signal of the physical microphone is weighted by the weighted weight calculated by the Kriging interpolation method to estimate the sound pressure signal at the virtual microphone.

[0031] 4. Active noise control using the estimated sound pressure signal at the virtual sensor

[0032] The estimated sound pressure signal at the virtual sensor can be used for any active noise control method, including the control filter update process and the secondary channel online modeling process.

[0033] The technical solutions in the embodiments of the present invention are described clearly and completely below in conjunction with the accompanying drawings.

[0034] 1. Simulation environment

[0035] Simulation environment such as Figure 2 As shown in the figure. The transfer function is generated by IMAGE MODEL. The room size is 3m×3m×3m, the reverberation time is 0.2s, the sampling rate is 2000Hz, the microphone and the speaker are placed on a plane with a height of 1.5m, the number of physical microphones M=10, and they are evenly distributed in a circle with the center at (1.3m, 1m, 1.5m). The positions of the two virtual sensor microphones are (1.3m, 1m, 1.5m) and (1.35m, 1.06m, 1.5m), the positions of the two primary sound sources are (0.5m, 0.3m, 1.5m) and (0.8m, 0.2m, 1.5m), the positions of the secondary sound sources are (1m, 0.6m, 1.5m), and the order of the room impulse response is 256.

[0036] 2. Parameter settings

[0037] The length of the primary channel, secondary channel and control filter are all set to 256. The control filter is initialized to unit impulse response, and the secondary channel and primary channel, as well as other vectors are initialized to 0. The noise source emits random white noise from 0 to 1000 Hz.

[0038] 3. Specific simulation process

[0039] Set a in formula (5) to 0.3, obtain the distance between the physical microphone and the virtual microphone according to the simulation environment settings, and calculate the weights λ1,…,λ of the physical microphone according to formula (6) and formula (2): M .

[0040]

[0041] When only noise source 1 emits noise, the comparison between the sound pressure at the virtual microphone estimated by the 10 physical microphones and the real sound pressure is as follows: Figure 3As shown, the sound pressure estimation error is -28.2dB.

[0042] Assuming that the virtual secondary channel from the secondary sound source to the virtual microphone 1 is known, the leakage FxLMS (LeakyFiltered-x Least Mean Squared, LFxLMS) active noise control algorithm is used for noise control. The LFxLMS algorithm combined with the present invention is denoted as Kriging-LFxLMS. 5 At the sampling point, the noise source changes from noise source 1 to primary noise source 2, and the noise reduction effect is shown in the figure below. Figure 4 As shown, it can be seen that the present invention can track the change of the orientation of the noise source to regain the noise reduction effect, and the present invention is robust to the change of the orientation of the noise source.

[0043] If in 5×10 5 At the sampling point, the virtual microphone changes from virtual microphone 1 to virtual microphone 2, and the virtual secondary channel changes. At this time, the online modeling algorithm in active noise control can be used, such as the synchronous modeling and control method (patent name: A synchronous modeling and control active noise control method, patent number: CN201811371068.5). The online modeling algorithm combined with the present invention is recorded as Kriging-OMA. The noise reduction effect diagram is as follows Figure 5 As shown, it can be seen that the present invention can be combined with the online modeling algorithm to achieve the effect of tracking the changes of the secondary channel, that is, the present invention is robust to the changes of the secondary channel.

[0044] The simulation results show that the present invention is robust to the characteristics and azimuth changes of the noise source position. If combined with the secondary channel online modeling algorithm, the method is also robust to changes in the secondary channel, such as changes in the secondary sound source position, changes in the virtual microphone position, etc.

Claims

1. A virtual sensing method based on Kriging interpolation applied in active noise control, characterized in that: The following steps are involved: Step 1, measuring and recording the distance between the virtual microphone and the physical microphone; Step 2, using Kriging interpolation method to calculate the weighted weights of the physical microphone; Step 3, using a physical microphone to obtain a sound pressure signal, weighting the sound pressure signal of the physical microphone with the weighted weight calculated in step 2, and estimating the sound pressure signal at the virtual microphone; Step 4: Based on the sound pressure signal at the virtual microphone estimated in step 3, update the control filter coefficients using any active noise control algorithm.

2. The virtual sensing method based on Kriging interpolation applied to active noise control according to claim 1, characterized in that: The step 2 is specifically as follows: Assume that there are M physical microphones with positions x1(x1,y1,z1),…,x M (x M ,y M ,z M ), M physical microphones measure M sound pressures, denoted as e1, e2, …e M , estimate the sound pressure e0 at the virtual microphone x0(x0,y0,z0) based on the sound pressure of M microphones; let the variation function be where h is the distance between the physical microphone and the virtual microphone, and a is a constant, called the correlation scale. Let C ij =C(|x i -x j |), where |x i -x j | i.e. position x i and position x j The distance between them can be calculated by the matrix inversion method where λ1,…,λ M That is, the position is x1(x1,y1,z1),…,x M (x M ,y M ,z M ) is the weighted value of the sound pressure measured by the M physical microphones.

3. The virtual sensing method based on Kriging interpolation applied to active noise control according to claim 1, characterized in that: Any active noise control algorithm in step 4 includes any algorithm that models the secondary channel online.

Images