Self-adaptive in-vehicle sound field partition multi-target control method and system considering scattering effect

Abstract

The invention discloses a self-adaptive in-vehicle sound field partition multi-target control method and system considering a scattering effect, and the method comprises the steps: dividing a sound field control region based on the sound listening demands of people in a vehicle; respectively arranging microphones inside and outside the sound field control area, and respectively measuring transfer functions from the loudspeaker and the equivalent sound source to the inside and outside of the sound field control area; based on the transfer function, taking passengers in the vehicle as an equivalent sound source, and establishing a scattering model for the passengers in the vehicle; acquiring a transfer function matrix in the vehicle based on the scattering model; and realizing in-vehicle sound field partition multi-target control based on the in-vehicle transfer function matrix. According to the invention, the influence of moving passengers in the vehicle on the sound field partition control effect can be eliminated. Meanwhile, the sound field reproduction effect can be considered while sound field partitioning is achieved, and dynamic balance of the sound energy contrast ratio, the bright region reproduction error and the loudspeaker array efficiency is achieved.

Description

Technical Field

[0001] The present invention relates to the field of multi-objective control methods and noise control methods for in-vehicle sound fields, and in particular to an adaptive in-vehicle sound field partition multi-objective control method and system taking scattering effects into consideration. Background Art

[0002] In the past few decades, the automotive industry has mainly focused on improving power and handling stability, but with the development of new energy vehicles, more and more companies have begun to pay attention to the NVH (noise, vibration and harshness) performance of vehicles, and in-car comfort has also become an important criterion for measuring vehicle performance. At the same time, with the advancement of technology, the types of in-car audio signals are increasing, including navigation sounds, music and other sounds that exist at the same time and interfere with each other, giving rise to personalized audio needs. In this context, sound field zoning control technology came into being.

[0003] The core goal of acoustic field partitioning is to provide different users with independent private acoustic environments in the same space and avoid interference between each other. Acoustic field partitioning control technology uses acoustic field reproduction technology to reproduce the target sound field in the bright area and keep it relatively quiet in the dark area, so that different users can enjoy their own acoustic space and avoid cross-interference.

[0004] However, the current sound field zoning control still has some limitations: first, the control target is single, and most control methods only consider the reproduction effect of a single target and ignore the control effects of other aspects; second, the problem of the existence of listeners in the listening range is not considered. There will inevitably be passengers in the car. When the passengers enter the sound field, the sound waves will be scattered by the passengers, resulting in reduced sound field zoning and sound field reproduction effects. Summary of the invention

[0005] In view of the problems that the existing sound field zoning control methods have a single control target, do not comprehensively consider the effects of sound field zoning and sound field reproduction, and do not consider the existence of listeners in the listening range, but in fact when passengers enter the sound field, the sound waves will be scattered by the passengers, resulting in reduced effects of sound field zoning and sound field reproduction, the purpose of the present invention is to develop an adaptive in-vehicle sound field zoning multi-target control method considering the scattering effect, and through this method to achieve multi-target control of intelligent cockpit sound field zoning and eliminate the influence of moving passengers on the sound field zoning control effect.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An adaptive vehicle interior sound field partition multi-objective control method considering scattering effect, comprising the steps of:

[0008] Divide the sound field control area based on the listening needs of the people in the car;

[0009] Arranging microphones in and outside the sound field control area respectively, and measuring the transfer functions from the loudspeaker and the equivalent sound source to and outside the sound field control area respectively;

[0010] Based on the transfer function, the passengers in the car are regarded as equivalent sound sources, and a scattering model is established for the passengers in the car;

[0011] Based on the scattering model, obtaining a transfer function matrix in the vehicle;

[0012] Based on the transfer function matrix in the vehicle, multi-objective control of the vehicle interior sound field partitioning is achieved.

[0013] Preferably, M is arranged in the bright area and the dark area respectively. b and M d Microphones are placed outside the bright area and the dark area respectively. e Additional microphones; measure the transfer function G from the loudspeaker to the bright and dark area microphones o , the transfer function G from the equivalent sound source to the microphone in the bright and dark areas s , the transfer function G from the loudspeaker to the additional microphone outside the bright and dark area oe , the transfer function G from the equivalent sound source to the additional microphone outside the bright and dark areas se .

[0014] Preferably, the established scattering model includes:

[0015] P(n)=G o q(n)+G s q s (n)

[0016] q s (n) = S (n) q (n)

[0017] P(n)=(G o +G s S(n))q(n)

[0018] G(n)=G o +G s S(n)

[0019] P e (n)=(G oe +G se S(n))q(n)

[0020]

[0021] Where P(n) represents the sound pressure of the current light and dark area control point; n represents the discrete time index; q s It represents the source intensity of the equivalent sound source generated after the incident sound field is scattered; P e (n) represents the sound field outside Me The sound pressure at the additional microphones; represents the estimated value of the scattering matrix S(n); represents the estimated value of the current in-vehicle transfer function; q(n) represents the driving signal of the speaker array.

[0022] Preferably, based on the framework of the LMS algorithm, the transfer function matrix in the current vehicle is obtained by iteratively minimizing the cost function J(n) The expression of J(n) is:

[0023] J(n)=||e(n)|| 2

[0024] Where e(n) represents the actual sound pressure P(n) at the current control point in the vehicle and its estimated value The error between

[0025] The update of J(n) expression is:

[0026]

[0027] Where α represents the regularization parameter; I represents the identity matrix; and H represents the Hermitian transpose of the matrix.

[0028] Preferably, the method for realizing multi-objective control of the partitioning of the sound field in the vehicle includes: using the weighted sum method WSM to transform the multi-objective optimization problem into a single-objective optimization, and applying the genetic algorithm NSGA2 to indirectly process the nonlinear relationship between multiple objectives through a nonlinear weight selection method, and determine the optimal weights and the driving signals that make the partitioning of the sound field in the vehicle the best under the power constraint of the speaker array; finally, applying the AGW iterative method based on the inverse generalized Rayleigh quotient iteration to reduce the oscillation during the convergence process.

[0029] The present invention also provides an adaptive in-vehicle sound field partition multi-objective control system considering scattering effect, the system is used to implement the above method, including: a partition module, a measurement module, a construction module, an acquisition module and a control module;

[0030] The division module is used to divide the sound field control area based on the listening needs of the people in the car;

[0031] The measuring module is used to arrange microphones inside and outside the sound field control area, respectively, to measure the transfer functions of the loudspeaker and the equivalent sound source to the sound field control area and outside respectively;

[0032] The building module is used to establish a scattering model for the passengers in the car based on the transfer function and taking the passengers in the car as equivalent sound sources;

[0033] The acquisition module is used to acquire a transfer function matrix in the vehicle based on the scattering model;

[0034] The control module is used to realize multi-objective control of the vehicle interior sound field partition based on the transfer function matrix in the vehicle.

[0035] Preferably, the measuring modules are arranged in the bright area and the dark area respectively. b and M d Microphones are placed outside the bright area and the dark area respectively. e Additional microphones; measure the transfer function G from the loudspeaker to the bright and dark area microphones o , the transfer function G from the equivalent sound source to the microphone in the bright and dark areas s , the transfer function G from the loudspeaker to the additional microphone outside the bright and dark area oe , the transfer function G from the equivalent sound source to the additional microphone outside the bright and dark areas se .

[0036] Preferably, the established scattering model includes:

[0037] P(n)=G o q(n)+G s q s (n)

[0038] q s (n) = S (n) q (n)

[0039] P(n)=(G o +G s S(n))q(n)

[0040] G(n)=G o +G s S(n)

[0041] P e (n)=(G oe +G se S(n))q(n)

[0042]

[0043] Where P(n) represents the sound pressure of the current light and dark area control point; n represents the discrete time index; q s It represents the source intensity of the equivalent sound source generated after the incident sound field is scattered; P e (n) represents the sound field outside M e The sound pressure at the additional microphones; represents the estimated value of the scattering matrix S(n); represents the estimated value of the current in-vehicle transfer function; q(n) represents the driving signal of the speaker array.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention can eliminate the influence of moving passengers in the car on the sound field partition control effect. At the same time, the present invention can take into account the sound field reproduction effect while realizing the sound field partition, and realize the dynamic balance of sound energy contrast, bright area reproduction error and speaker array efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0047] Figure 1 A schematic diagram of a method flow chart of an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of the arrangement of light and dark areas and microphones proposed in an embodiment of the present invention;

[0049] Figure 3 A schematic diagram of the AGW-SA algorithm flow in an embodiment of the present invention;

[0050] Figure 4 1 is a control effect diagram of the method of an embodiment of the present invention; wherein (a) represents the control effect of AC obtained based on three transfer function matrices; (b) represents the control effect of SD obtained based on three transfer function matrices; (c) represents the control effect of AE obtained based on three transfer function matrices. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0053] Embodiment 1

[0054] like Figure 1 FIG. 1 is a schematic diagram of the method flow of this embodiment, and the steps include:

[0055] S1. Divide the sound field control area based on the listening needs of the people in the car.

[0056] According to the different listening needs of the passengers in the car, the car is divided into two sound field control areas: bright area and dark area. The specific division results are as follows: Figure 2 As shown, the main driver's seat is the bright area and the right side of the rear row is the dark area.

[0057] S2. Arrange microphones inside and outside the sound field control area, and measure the transfer functions from the loudspeaker and the equivalent sound source to the sound field control area and outside the sound field control area.

[0058] Arrange M in the bright area and dark area respectively. b and M d Microphones are placed outside the bright area and the dark area respectively. e Additional microphones; measure the transfer function G from the loudspeaker to the bright and dark area microphones o , the transfer function G from the equivalent sound source to the microphone in the bright and dark areas s , the transfer function G from the loudspeaker to the additional microphone outside the bright and dark area oe , the transfer function G from the equivalent sound source to the additional microphone outside the bright and dark areas se The specific steps include:

[0059] Step 2.1: Arrange M in the bright area and dark area respectively b and M d Microphones are placed outside the bright area and the dark area respectively. e An additional microphone.

[0060] Step 2.2: Drive the in-car speaker array and measure the time domain transfer function G from the speaker to the bright and dark area microphones o (t), the time domain transfer function G from the loudspeaker to the additional microphone outside the bright and dark areas oe (t).

[0061] Step 2.3: L on the left side of the rear row s An equivalent sound source is used to drive the equivalent sound source to measure the time domain transfer function G from the equivalent sound source to the microphone in the bright and dark areas. s (t), time domain transfer function G of the equivalent sound source to the additional microphone outside the bright and dark areas se (t).

[0062] Step 2.4: Perform a fast Fourier transform (FFT) on the time domain transfer function collected by each microphone to obtain the frequency domain transfer function G at the corresponding frequency. o , G s , G oe and G se .

[0063] S3. Based on the transfer function, the passengers in the car are regarded as equivalent sound sources and a scattering model is established for the passengers in the car.

[0064] The equivalent sound source method is used to approximate the passengers in the car as 6 equivalent sound sources, and a scattering model is established for the passengers in the car. The specific steps include:

[0065] Step 3.1: The sound pressure P(n) of the current light and dark area control point can be expressed as:

[0066] P(n)=G o q(n)+G s q s (n)

[0067] Where n represents the discrete time index; q s It represents the source intensity of the equivalent sound source generated by the incident sound field after scattering. According to the superposition principle, q s The size of can be expressed as:

[0068] q s (n) = S (n) q (n)

[0069] Where S represents the scattering matrix representing the scattering characteristics of the scatterer; q(n) represents the driving signal of the speaker array.

[0070] P(n)=(G o +G s S(n))q(n)

[0071] The current transfer function matrix is:

[0072] G(n)=G o +G s S(n)

[0073] Step 3.2: Outside the sound field e The sound pressure P at the additional microphone e (n) can be expressed as:

[0074] P e (n)=(G oe +G se S(n))q(n)

[0075] Step 3.3: Just get the estimated value of the scattering matrix S(n) through step 3.2 You can get the estimated value of the current in-vehicle transfer function

[0076]

[0077] S4. Based on the scattering model, obtain the transfer function matrix inside the vehicle.

[0078] Based on the LMS algorithm, microphones are additionally arranged outside the sound field to estimate the current scattering characteristics of the moving passengers in the car in real time, thereby estimating the transfer function matrix in the current car. The specific steps include:

[0079] Step 4.1: Based on the framework of the LMS algorithm, the transfer function matrix in the current car It can be obtained by iteratively minimizing the cost function J(n), and the expression of J(n) is:

[0080]

[0081] Where e(n) represents the actual sound pressure P(n) at the current control point in the vehicle and its estimated value The error between them is expressed as follows:

[0082]

[0083] From step 3.2, we can know that:

[0084]

[0085] Since step 3.2 requires the calculation of G se The inverse matrix of G se If the matrix is ​​not full rank or is close to being singular, it will lead to numerical instability. Regularization can help control the update amplitude of the scattering matrix S(n) to prevent it from changing drastically when it is subject to noise or external interference, which will lead to system instability.

[0086] Step 4.2: The expression of S(n) after Tikhonov regularization is as follows:

[0087]

[0088] Where α represents the regularization parameter; I represents the identity matrix; and H represents the Hermitian transpose of the matrix.

[0089] Step 4.3: According to steps 4.1 and 4.2, the expression of J(n) can be updated as follows:

[0090]

[0091] Step 4.4: To write the formula in the standard form of the LMS algorithm, the estimated scattering matrix Rewritten as the scattering vector d(n)

[0092]

[0093] In the formula, s LsL represents the estimated scattering matrix The elements in; L represents the number of speakers; L s Indicates the number of equivalent sound sources.

[0094]

[0095] Where Q(n) represents a diagonal matrix:

[0096]

[0097] Step 4.5: The cost function is updated as:

[0098] J=||y(n)-X(n)d(n)|| 2

[0099]

[0100] X(n)=G s Q(n)

[0101] Where X(n) represents the actual output signal; y(n) represents the expected output signal.

[0102] Step 4.6: To avoid overfitting, add a regularization term to the equation:

[0103]

[0104] In the formula, β represents the regularization coefficient.

[0105] Step 4.7: The scattering vector d(n+1) can be obtained by iterating step 4.7:

[0106]

[0107] μ represents the step size of the LMS algorithm; ||·|| F represents the Frobenius norm of the matrix; X H Represents the Hermitian transpose of the actual output signal.

[0108] Recombining the scattered vector d(n+1) into Through step 4.4, the estimated value of the transfer function matrix in the current car can be obtained Transfer function G of the light and dark areas in the car b and G d can be get:

[0109]

[0110] S5. Based on the transfer function matrix inside the car, multi-objective control of the sound field partition inside the car is realized.

[0111] The weighted sum method (WSM) is used to transform the multi-objective optimization problem into a single-objective optimization problem. At the same time, the genetic algorithm NSGA2 is applied to indirectly deal with the nonlinear relationship between multiple objectives through a nonlinear weight selection method, and the optimal weight and the driving signal that achieves the best control effect of the in-vehicle sound field partition are determined under the power constraint of the speaker array.

[0112] In this embodiment, the above-mentioned multiple objectives include: the ratio AC of acoustic potential energy between the bright area and the dark area, the error SD between the reproduced sound field in the bright area and the target sound field, and the ratio AE of the sum of squares of the speaker array weight vectors to the energy of the reference speaker that generates the same sound energy in the bright area. Their expressions are as follows:

[0113]

[0114] in,(·) H represents the Hermitian transpose of a matrix; P des It indicates the expected sound field in the bright area; q r Represents the driving signal of the reference loudspeaker.

[0115] The ideal sound field partition control effect should be to make AC as large as possible, and SD and AE as small as possible. Therefore, the optimization target F(q) is determined as:

[0116]

[0117] Use the weighted sum method to transform the multi-objective optimization problem between AC, SD, and AE into a single-objective optimization problem

[0118]

[0119] In the formula, w i represents the expected weights of different optimization objectives; q is calculated by w1R1+w2R2+w3R3 and R b The eigenvector corresponding to the minimum generalized eigenvalue between is obtained.

[0120] The speaker array is power constrained and the NSGA2 algorithm is used to iteratively select the optimal weight w. i ,By introducing the weight optimization process of nonlinear relationships, the overall optimization process can adapt to the nonlinear relationships among multiple objectives.

[0121]

[0122] Wherein, η represents the constraint weight, and 0<η≤1; V represents the rated input voltage of the speaker.

[0123] Finally, the AGW iterative method based on the inversion of the generalized Rayleigh quotient iteration is applied, and the q(n) is used to update q(n+1) to reduce the oscillation during the convergence process. It can also eliminate the influence of scattering effect on the sound field partition control effect while realizing the multi-objective control of the sound field partition in the car. The AGW-SA algorithm process is as follows: Figure 3 shown.

[0124] The specific steps are as follows:

[0125] (1) Construct a generalized matrix:

[0126]

[0127] In the formula, represents the estimated value of the multi-objective weighting matrix; K represents and The generalized matrix of ;

[0128]

[0129] (2) Calculate the Rayleigh quotient of a generalized matrix:

[0130]

[0131] (3) Update the drive signal:

[0132]

[0133] D(n)=q(n)+γZ(n)

[0134] In the formula, γ represents the update direction of the driving signal q(n+1), and the updating method of the driving signal is:

[0135]

[0136] D is the signal update matrix;

[0137] The final control effect of this embodiment is as follows Figure 4 shown.

[0138] Embodiment 2

[0139] This embodiment also provides an adaptive in-vehicle sound field partition multi-objective control system that takes into account the scattering effect, including: a partitioning module, a measurement module, a construction module, an acquisition module and a control module; the partitioning module is used to divide the sound field control area based on the listening needs of the people in the car; the measurement module is used to arrange microphones inside and outside the sound field control area, and respectively measure the transfer functions of the speakers and equivalent sound sources to the inside and outside of the sound field control area; the construction module is used to establish a scattering model for the passengers in the car based on the transfer function, taking the passengers in the car as equivalent sound sources; the acquisition module is used to obtain the transfer function matrix in the car based on the scattering model; the control module is used to realize the in-vehicle sound field partition multi-objective control based on the transfer function matrix in the car.

[0140] The following will describe in detail how the present invention solves technical problems in real life in conjunction with this embodiment.

[0141] First, the division module is used to divide the sound field control area based on the listening needs of the people in the car.

[0142] According to the different listening needs of the passengers in the car, the car is divided into two sound field control areas: bright area and dark area. The specific division results are as follows: Figure 2 As shown, the main driver's seat is the bright area and the right side of the rear row is the dark area.

[0143] Afterwards, the measurement module arranges microphones inside and outside the sound field control area to measure the transfer functions from the loudspeaker and the equivalent sound source to the sound field control area and outside the sound field control area.

[0144] Arrange M in the bright area and dark area respectively. b and M d Microphones are placed outside the bright area and the dark area respectively. e Additional microphones; measure the transfer function G from the loudspeaker to the bright and dark area microphones o , the transfer function G from the equivalent sound source to the microphone in the bright and dark areas s , the transfer function G from the loudspeaker to the additional microphone outside the bright and dark area oe , the transfer function G from the equivalent sound source to the additional microphone outside the bright and dark areas se The specific steps include:

[0145] Step 2.1: Arrange M in the bright area and dark area respectively b and M d Microphones are placed outside the bright area and the dark area respectively. e An additional microphone.

[0146] Step 2.2: Drive the in-car speaker array and measure the time domain transfer function G from the speaker to the bright and dark area microphones o (t), the time domain transfer function G from the loudspeaker to the additional microphone outside the bright and dark areas oe (t).

[0147] Step 2.3: L on the left side of the rear row s An equivalent sound source is used to drive the equivalent sound source to measure the time domain transfer function G from the equivalent sound source to the microphone in the bright and dark areas. s (t), time domain transfer function G of the equivalent sound source to the additional microphone outside the bright and dark areas se (t).

[0148] Step 2.4: Perform a fast Fourier transform (FFT) on the time domain transfer function collected by each microphone to obtain the frequency domain transfer function G at the corresponding frequency. o , G s , G oe and G se .

[0149] The building module takes the passengers in the car as equivalent sound sources based on the transfer function and establishes a scattering model for the passengers in the car.

[0150] The equivalent sound source method is used to approximate the passengers in the car as 6 equivalent sound sources, and a scattering model is established for the passengers in the car. The specific steps include:

[0151] Step 3.1: The sound pressure P(n) of the current light and dark area control point can be expressed as:

[0152] P(n)=G o q(n)+G s q s (n)

[0153] Where n represents the discrete time index; q s It represents the source intensity of the equivalent sound source generated by the incident sound field after scattering. According to the superposition principle, q s The size of can be expressed as:

[0154] q s (n) = S (n) q (n)

[0155] Where S represents the scattering matrix representing the scattering characteristics of the scatterer; q(n) represents the driving signal of the speaker array.

[0156] P(n)=(G o +G s S(n))q(n)

[0157] The current transfer function matrix is:

[0158] G(n)=G o +G s S(n)

[0159] Step 3.2: Outside the sound field e The sound pressure P at the additional microphonee (n) can be expressed as:

[0160] P e (n)=(G oe +G se S(n))q(n)

[0161] Step 3.3: Just get the estimated value of the scattering matrix S(n) through step 3.2 You can get the estimated value of the current in-vehicle transfer function

[0162]

[0163] The acquisition module obtains the transfer function matrix in the vehicle based on the scattering model.

[0164] Based on the LMS algorithm, microphones are additionally arranged outside the sound field to estimate the current scattering characteristics of the moving passengers in the car in real time, thereby estimating the transfer function matrix in the current car. The specific steps include:

[0165] Step 4.1: Based on the framework of the LMS algorithm, the transfer function matrix in the current car It can be obtained by iteratively minimizing the cost function J(n), and the expression of J(n) is:

[0166] J(n)=||e(n)|| 2

[0167] Where e(n) represents the actual sound pressure P(n) at the current control point in the vehicle and its estimated value The error between them is expressed as follows:

[0168]

[0169] From step 3.2, we can know that:

[0170]

[0171] Since step 3.2 requires the calculation of G se The inverse matrix of G se If the matrix is ​​not full rank or is close to being singular, it will lead to numerical instability. Regularization can help control the update amplitude of the scattering matrix S(n) to prevent it from changing drastically when it is subject to noise or external interference, which will lead to system instability.

[0172] Step 4.2: The expression of S(n) after Tikhonov regularization is as follows:

[0173]

[0174] Where α represents the regularization parameter; I represents the identity matrix; and H represents the Hermitian transpose of the matrix.

[0175] Step 4.3: According to steps 4.1 and 4.2, the expression of J(n) can be updated as follows:

[0176]

[0177] Step 4.4: To write the formula in the standard form of the LMS algorithm, the estimated scattering matrix Rewritten as the scattering vector d(n)

[0178] d(n)=[s 11 ,s 12 ,…,s 1L ,s 21 ,…,s 2L ,s Ls1 ,…,s LsL ] T

[0179] In the formula, s LsL represents the estimated scattering matrix The elements in; L represents the number of speakers; L s Indicates the number of equivalent sound sources.

[0180]

[0181] Where Q(n) represents a diagonal matrix:

[0182]

[0183] Step 4.5: The cost function is updated as:

[0184] J=||y(n)-X(n)d(n)|| 2

[0185]

[0186] X(n)=G s Q(n)

[0187] Where X(n) represents the actual output signal; y(n) represents the expected output signal.

[0188] Step 4.6: To avoid overfitting, add a regularization term to the equation:

[0189]

[0190] In the formula, β represents the regularization coefficient.

[0191] Step 4.7: The scattering vector d(n+1) can be obtained by iterating step 4.7:

[0192]

[0193] μ represents the step size of the LMS algorithm; ||·|| F represents the Frobenius norm of the matrix; X H Represents the Hermitian transpose of the actual output signal.

[0194] Recombining the scattered vector d(n+1) into Through step 4.4, the estimated value of the transfer function matrix in the current car can be obtained Transfer function G of the light and dark areas in the car b and G d can be get:

[0195]

[0196] The control module realizes multi-objective control of the sound field partition in the car based on the transfer function matrix in the car.

[0197] The weighted sum method (WSM) is used to transform the multi-objective optimization problem into a single-objective optimization problem. At the same time, the genetic algorithm NSGA2 is applied to indirectly deal with the nonlinear relationship between multiple objectives through a nonlinear weight selection method, and the optimal weight and the driving signal that achieves the best control effect of the in-vehicle sound field partition are determined under the power constraint of the speaker array.

[0198] In this embodiment, the above-mentioned multiple objectives include: the ratio AC of acoustic potential energy between the bright area and the dark area, the error SD between the reproduced sound field in the bright area and the target sound field, and the ratio AE of the sum of squares of the speaker array weight vectors to the energy of the reference speaker that generates the same sound energy in the bright area. Their expressions are as follows:

[0199]

[0200] in,(·) H represents the Hermitian transpose of a matrix; P des Indicates the expected sound field in the bright area; q r Represents the driving signal of the reference loudspeaker.

[0201] The ideal sound field partition control effect should be to make AC as large as possible, and SD and AE as small as possible. Therefore, the optimization target F(q) is determined as:

[0202]

[0203] Use the weighted sum method to transform the multi-objective optimization problem between AC, SD, and AE into a single-objective optimization problem

[0204]

[0205] In the formula, w i represents the expected weights of different optimization objectives; q is calculated by w1R1+w2R2+w3R3 and R b The eigenvector corresponding to the minimum generalized eigenvalue between is obtained.

[0206] The speaker array is power constrained and the NSGA2 algorithm is used to iteratively select the optimal weight w. i ,By introducing the weight optimization process of nonlinear relationships, the overall optimization process can adapt to the nonlinear relationships among multiple objectives.

[0207]

[0208] Wherein, η represents the constraint weight, and 0<η≤1; V represents the rated input voltage of the speaker.

[0209] Finally, the AGW iterative method based on the inversion of the generalized Rayleigh quotient iteration is applied, and the q(n) is used to update q(n+1) to reduce the oscillation during the convergence process. It can also eliminate the influence of scattering effect on the sound field partition control effect while realizing the multi-objective control of the sound field partition in the car. The AGW-SA algorithm process is as follows: Figure 3 shown.

[0210] The specific steps are as follows:

[0211] (1) Construct a generalized matrix:

[0212]

[0213] In the formula, represents the estimated value of the multi-objective weighting matrix; K represents and The generalized matrix of ;

[0214]

[0215] (2) Calculate the Rayleigh quotient of a generalized matrix:

[0216]

[0217] (3) Update the drive signal:

[0218]

[0219] D(n)=q(n)+γZ(n)

[0220] In the formula, γ represents the update direction of the driving signal q(n+1), and the updating method of the driving signal is:

[0221]

[0222] D is the signal update matrix;

[0223] The final control effect of this embodiment is as follows Figure 4 shown.

[0224] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. An adaptive vehicle interior sound field partition multi-objective control method considering scattering effect, characterized in that the steps include: Divide the sound field control area based on the listening needs of the people in the car; Arranging microphones in and outside the sound field control area respectively, and measuring the transfer functions from the loudspeaker and the equivalent sound source to and outside the sound field control area respectively; Based on the transfer function, the passengers in the car are regarded as equivalent sound sources, and a scattering model is established for the passengers in the car; Based on the scattering model, obtaining a transfer function matrix in the vehicle; Based on the transfer function matrix in the vehicle, multi-objective control of the vehicle interior sound field partitioning is achieved.

2. The adaptive in-vehicle sound field partition multi-objective control method considering scattering effect according to claim 1 is characterized in that: Arrange M in the bright area and dark area respectively. b and M d Microphones are placed outside the bright area and the dark area respectively. e Additional microphones; measure the transfer function G from the loudspeaker to the bright and dark area microphones o , the transfer function G from the equivalent sound source to the microphone in the bright and dark areas s , the transfer function G from the loudspeaker to the additional microphone outside the bright and dark area oe , the transfer function G from the equivalent sound source to the additional microphone outside the bright and dark areas se .

3. The adaptive in-vehicle sound field partition multi-objective control method considering scattering effect according to claim 2 is characterized in that: The scattering model established includes: P(n)=G o q(n)+G s q s (n) q s (n)=S(n)q(n) P(n)=(G o +G s S(n))q(n) G(n)=G o +G s S(n) P e (n)=(G oe +G se S(n))q(n) Where P(n) represents the sound pressure of the current light and dark area control point; n represents the discrete time index; q s It represents the source intensity of the equivalent sound source generated after the incident sound field is scattered; P e (n) represents the sound field outside M e The sound pressure at the additional microphones; represents the estimated value of the scattering matrix S(n); represents the estimated value of the current in-vehicle transfer function; q(n) represents the driving signal of the speaker array.

4. The adaptive in-vehicle sound field partition multi-objective control method considering scattering effect according to claim 3 is characterized in that: Based on the framework of the LMS algorithm, the transfer function matrix in the current vehicle is obtained by iteratively minimizing the cost function J(n) The expression of J(n) is: J(n)=||e(n)|| 2 Where e(n) represents the actual sound pressure P(n) at the current control point in the vehicle and its estimated value The error between The update of J(n) expression is: Where α represents the regularization parameter; I represents the identity matrix; and H represents the Hermitian transpose of the matrix.

5. The adaptive in-vehicle sound field partition multi-objective control method considering scattering effect according to claim 4 is characterized in that: The method for realizing multi-objective control of in-vehicle sound field partition includes: using the weighted sum method WSM to transform the multi-objective optimization problem into a single-objective optimization, and applying the genetic algorithm NSGA2 to indirectly deal with the nonlinear relationship between multiple objectives through a nonlinear weight selection method, and determine the optimal weight under the power constraint of the speaker array, and the driving signal that makes the in-vehicle sound field partition control effect the best; finally, the AGW iterative method based on the inverse generalized Rayleigh quotient iteration is applied to reduce the oscillation during the convergence process.

6. An adaptive in-vehicle sound field partition multi-objective control system considering scattering effect, the system is used to implement the method according to any one of claims 1 to 5, characterized in that: include: Partition module, measurement module, construction module, acquisition module and control module; The division module is used to divide the sound field control area based on the listening needs of the people in the car; The measuring module is used to arrange microphones inside and outside the sound field control area, respectively, to measure the transfer functions of the loudspeaker and the equivalent sound source to the sound field control area and outside respectively; The building module is used to establish a scattering model for the passengers in the car based on the transfer function and taking the passengers in the car as equivalent sound sources; The acquisition module is used to acquire a transfer function matrix in the vehicle based on the scattering model; The control module is used to realize multi-objective control of the vehicle interior sound field partition based on the transfer function matrix in the vehicle.

7. The adaptive in-vehicle sound field partition multi-objective control system considering scattering effect according to claim 6 is characterized in that: The measuring modules are arranged in the bright area and the dark area respectively. b and M d Microphones are placed outside the bright area and the dark area respectively. e Additional microphones; measure the transfer function G from the loudspeaker to the bright and dark area microphones o , the transfer function G from the equivalent sound source to the microphone in the bright and dark areas s , the transfer function G from the loudspeaker to the additional microphone outside the bright and dark area oe , the transfer function G from the equivalent sound source to the additional microphone outside the bright and dark areas se .

8. The adaptive in-vehicle sound field partition multi-objective control system considering scattering effect according to claim 7 is characterized in that: The scattering model established includes: P(n)=G o q(n)+G s q s (n) q s (n)=S(n)q(n) P(n)=(G o +G s S(n))q(n) G(n)=G o +G s S(n) P e (n)=(G oe +G se S(n))q(n) Where P(n) represents the sound pressure of the current light and dark area control point; n represents the discrete time index; q s It represents the source intensity of the equivalent sound source generated after the incident sound field is scattered; P e (n) represents the sound field outside M e The sound pressure at the additional microphones; represents the estimated value of the scattering matrix S(n); represents the estimated value of the current in-vehicle transfer function; q(n) represents the driving signal of the speaker array.

Images