CN115482803A

CN115482803A - Intelligent system parameter calibration method and system applied to in-vehicle road noise control

Info

Publication number: CN115482803A
Application number: CN202211041732.6A
Authority: CN
Inventors: 刘志恩; 程灿; 卢炽华; 李晓龙; 孙毅; 颜伏伍; 侯献军
Original assignee: Wuhan University of Technology WUT
Current assignee: Wuhan University of Technology WUT
Priority date: 2022-08-29
Filing date: 2022-08-29
Publication date: 2022-12-16

Abstract

The invention discloses a system parameter intelligent calibration method and system applied to in-vehicle road noise control.A sensor is arranged on a vehicle body to acquire vehicle noise data; recognizing a control area by using an in-vehicle vision sensor, and acquiring an in-vehicle road noise zone control mode according to the passenger sitting condition; based on a partition control mode and the noise reduction maximization of a control area, screening out the optimal reference signal combination of the control area by utilizing a multiple coherent analysis method; generating a secondary noise signal by adopting a VSS-DSAF algorithm in combination with an in-vehicle road noise partition control mode to achieve noise reduction of a specific area through interference cancellation; and performing adjustable parameter optimization design on the VSS-DSAF algorithm by utilizing a quadratic programming algorithm. The invention can solve the defects of high complexity, low convergence speed and poor stability of the active control calculation of the road noise in the vehicle at present. So that people sitting everywhere in the vehicle can obtain better noise reduction effect. Meanwhile, the problem that the low-medium frequency road noise of the electric automobile is obvious after the masking effect of the engine order noise is lost is solved.

Description

Intelligent system parameter calibration method and system applied to in-vehicle road noise control

Technical Field

The invention relates to the technical field of automobiles, in particular to a system parameter intelligent calibration method and system applied to in-vehicle main road noise control.

Background

With the rapid development of new energy vehicles represented by electric vehicles, vehicle power systems are developing from internal combustion engine driving to motor driving. The noise masking effect of the noise of the internal combustion engine is lost, and the road noise becomes a main source of the noise in the electric automobile. The effective reduction of road noise is an important way for the new energy automobile to improve the brand image and the comfort. The problems of difficult decoupling of automobile performance, poor control effect of low-frequency noise, complex structure of an optimization scheme, high production cost and the like can be caused by a Passive Noise Control (PNC) method such as designing or increasing a vibration absorber through an automobile body structure, and the weight of the automobile can be increased, which is contrary to the development trend of light weight of the automobile.

The main road noise control (ARNC) method utilizes the principle of interference cancellation to introduce a speaker (secondary sound source) in the sound field, and controls the speaker to emit cancellation noise with equal amplitude and opposite phase to the noise to be cancelled (primary noise), so as to form a mute zone in a specific area. The ARNC system not only can effectively reduce low-frequency noise which is difficult to control by a PNC method, but also can track the frequency spectrum change of the road noise in the vehicle to carry out self-adaptive control, and can realize the attenuation of the wide-band noise in the vehicle under the condition of not changing the structure and the working performance of the vehicle. Therefore, the ARNC technology becomes a hot spot for NVH research in the automobile industry.

The ARNC technology faces two key problems in real vehicle application, namely accurate acquisition of a reference signal with good correlation with the road noise in the vehicle and realization of a low-computation complexity high-noise-reduction-quantity adaptive filtering algorithm. The reference signal provides prior information for the active control of the road noise in the vehicle, and the correlation of the reference signal and the road noise is directly related to the noise reduction amount. At present, the conventional filtering-x least mean square (FxLMS) algorithm is mostly adopted by an ARNC system, and the FxLMS algorithm has the defects of high computational complexity, low convergence speed and poor stability in the application of the ARNC. The invention adopts an active noise control algorithm of non-delay sub-band filtering (DSAF), can decompose the reference signal into sub-bands for down-sampling, and completes the self-adaptive adjustment of the control filter coefficient in each sub-band, so that the calculation complexity is reduced; since the signal is decomposed into sub-bands, the signal energy is also decomposed into sub-bands, and theoretically, the algorithm is more stable in dealing with the impact interference.

In addition, researches have found that the road noise in the vehicle is subjected to partition control according to the number and the positions of drivers and passengers, the noise reduction amount of a specific area can be improved to a certain extent, and the sound quality in the vehicle is improved. However, the actual industrialization is not applied much at present, and the matching and calibration of the in-vehicle regional ARNC hardware mainly under different driver and passenger scenes are unstable, so that the method cannot be popularized and applied. In particular, the low and medium frequency road noise of the electric vehicle after the masking effect of the engine order noise is lost is very serious.

Disclosure of Invention

The invention aims to provide a system parameter intelligent calibration method and a system applied to the control of the road noise in the vehicle, the defects of high calculation complexity, low convergence speed and poor stability of the active control method and system for the road noise in the vehicle are overcome.

Secondly, the invention aims to solve the problem that the noise reduction effect is not balanced among the drivers and passengers, so that people sitting everywhere in the vehicle can obtain better noise reduction effect.

Further, the air conditioner is characterized in that, the invention aims to solve the problem that the low-medium frequency road noise is prominent after the masking effect of the engine order noise of the electric automobile is lost.

In order to solve the technical problems, the invention adopts the following technical scheme:

a system parameter intelligent calibration method applied to vehicle road noise control is characterized by comprising the following steps:

s1: the sensor is arranged on the vehicle body, collecting noise data of the automobile by using a sensor;

s2: recognizing a control area by using an in-vehicle vision sensor, and acquiring an in-vehicle road noise zone control mode and the control area according to the passenger sitting condition;

s3: based on the determined partition control mode, screening reference signals by using a multiple coherent analysis method according to a noise reduction maximization principle of a control area, and screening out an optimal reference signal combination of the control area;

s4: based on the screened optimal reference signal combination, and in combination with an in-vehicle road noise partition control mode, generating a secondary noise signal by adopting a main road noise control algorithm of a variable-step-size delay-free sub-band filtering algorithm based on a symbol error to realize noise reduction of a specific area by interference cancellation; the secondary noise has the same amplitude and opposite phase with the noise signal to be counteracted or the primary noise in the vehicle;

s5: and performing parameter optimization design on a main road noise control algorithm of the variable-step-size delay-free subband filtering algorithm based on the symbol error by utilizing a quadratic programming algorithm.

In the technical scheme, the step S1 is used for respectively acquiring noise data under various steady-state test working conditions of constant-speed running and all-throttle acceleration as time-varying unsteady-state test working conditions, wherein the noise data comprises frequency spectrum noise signals of a vibration acceleration sensor at a connecting point of a vehicle suspension and a vehicle body or at the side of the vehicle body as reference signals and frequency spectrum noise signals to be offset (primary) sent by microphones at the side of head pillows at four positions in front and rear rows in the vehicle.

In the above technical solution, step S2 is to arrange and combine the different numbers and different seating positions of the drivers and passengers in the vehicle, and the road noise zone control mode in the vehicle is divided into: the vehicle-mounted passenger seat control system comprises a centralized type vehicle-mounted passenger full seat and four distributed types of vehicle-mounted passengers which are respectively seated in a single area, a second area and a third area, wherein under each control mode, the seated positions of the passengers are in a road noise control area.

In the above technical solution, the step S1 of collecting noise data is performed according to a partition control mode:

when the face recognition system recognizes that only one passenger in the vehicle, namely a driver seat, needs to realize noise control, the microphone at the position is electrified to realize the collection of noise spectrum signals in the vehicle, and the microphones at other positions are in a power-off state and do not need to collect signals;

when the face recognition system recognizes a second area, a third area and a full seat, the microphones at the corresponding positions are electrified and acquire primary noise spectrum signals in the area, and the vibration acceleration sensor is always in a working state when acquiring the spectrum noise signals; in subsequent control, the secondary sound source (loudspeaker) or microphone in the non-control area is in the off state, and the microphone in the road noise control area is in the on state, so that the interference cancellation of the secondary sound source is generated.

In the above technical solution, step S3 introduces the acquired noise spectrum signal into MATLAB simulation analysis software, and selects an optimal reference signal by using a multiple coherence analysis:

the input signal is a vibration noise frequency spectrum signal acquired by an acceleration sensor at the passive end of the connection part of the suspension and the vehicle body, and the output signal is a sound pressure signal acquired by arranging a microphone at a passenger headrest in the vehicle;

calculating multiple coherence coefficients of the reference signal and the (primary) signal to be counteracted, obtaining a target noise reduction amount or a theoretical maximum noise reduction amount based on the coherence coefficients, sequencing and screening different combined signals according to the principle of the theoretical maximum noise reduction amount or the maximum coherence of the reference signal, and recombining signals based on the screened reference points.

In the above technical solution, the sorting and screening process of the different combined signals according to the principle that the theoretical maximum noise reduction amount or the coherence of the reference signal is maximum in step S3 is as follows:

firstly, calculating multiple coherence coefficients of all reference signals and primary noise spectrum signals to be counteracted, taking the reference signals corresponding to the maximum value of the multiple coherence coefficients, sequencing the reference signals into 1, and storing and recording the reference signals into a new reference signal selection set;

in a second iteration cycle, removing the reference signals in the sequence 1, calculating the multiple coherence coefficients of the remaining N-1 reference signals and the primary noise spectrum signal to be counteracted, taking the reference signal corresponding to the maximum value of the multiple coherence coefficients, sequencing the reference signal into 2, and storing and recording the reference signals in a new reference signal selection set;

the above steps are circulated until the last reference signal is stored and recorded in the new reference signal selection set; and selecting the position where the reference signal in the reference signal selection is positioned in the front-ranked reference signal as the arrangement position of the vibration acceleration sensor, and taking the position as the reference signal.

In the above technical solution, step S4 adopts a symbol error based variable step size non-delay subband filtering algorithm VSS-DSAF, and decomposing the reference signal into sub-bands, independently setting iteration step length for each sub-band, and generating a secondary noise signal in a full frequency band.

In the above technical solution, the step S4 of using the symbol error-based variable step size non-delay subband filtering algorithm VSS-DSAF specifically includes the following steps:

step S41, designing an analysis filter:

h _n (i)＝h ₀ (i)e ^-j2πi/N ，i＝0，2...，L _p -1，n＝0，2，...N-1；

wherein L is _p Analyzing the filtering length to determine the calculation complexity; n is the number of subbands, h ₀ (i) By being a Matlab function h ₀ (i)＝fir1(L _p -1,1/N);

step S42, sub-band decomposition of the reference signal and the error signal:

subband reference signal:

in the formula: x is the number of _n (t) is a reference signal sequence, h _n (i) Is an analysis filter bank;

subband error signal:

in the formula: e.g. of the type _n (t) is the error signal sequence, h ₀ (i) By means of Matlab function h ₀ (i)＝fir1(L _p -1,1/N);

step S43, calculating a subband filtering reference signal and a subband error signal:

in the formula: d decimation factor, r _n (t) is the self-filtering reference signal sequence, hs _n (k) Estimating for the sub-band secondary path, wherein K is the number of reference signals;

step S44, subband control filter coefficient iteration:

μ _i (n)＝λμ _i (n-D)+(1-λ)min{μ _i，opt (n)，μ _i (n-D)}；

wherein: r is _n (t)＝[r _n (t)，r _n (t-D)，......，r _n (t-M+D)]、w _n (t + D) is a subband control filter coefficient sequence; m is the length of the full-band control filter, and the convergence speed is determined; ^H which represents the transpose of the conjugate, ^* represents a complex conjugate; mu.s _i Determining the impact interference resistance for an iteration step sequence, wherein i is more than or equal to 0 and less than or equal to N/2; e is constant coefficient to prevent denominator from being zero, sgn (e) _n (t)) is a sign function;

step S45, converting the sub-band control filter coefficients into full-band coefficients, performing 2M-point Fourier transform on the control filter coefficients of each sub-band, stacking according to the following rules (a) - (c) to obtain the Fourier transform of the full-band weight coefficients, and performing 2M-point discrete Fourier transform to obtain the full-band coefficients:

determining optimal parameters of an analysis filter: length L of analysis filter _p Full band control filter length M, iterative step sequence mu _i 。

In the above technical solution, step S5 utilizes quadratic programming algorithm to pair parameter L _p 、M、μ _i The adjustable parameters are optimized to make the parameter L _p 、M、μ _i The optimal value is reached so that the generated synthetic noise immunity tracks the broadband noise to achieve the ideal active path noise control performance:

taking the minimization of the average noise reduction of the negative value as a performance index, and carrying out quadratic programming optimization with the target as follows:

wherein: f. of _U To the upper limit of the frequency of interest, f _L In order to take care of the lower limit of the frequency, the frequency range of the road noise which is concerned about because the road noise is middle and low frequency broadband random noise is 50H _z ～500H _z MNR is mean noise reduction minimization; noise reduction target NR (f) _i )＝10log ₁₀ {e _off (f _i )/e _on (f _i ) }; wherein e _off (f _i ) For the steady-state error signal spectrum when no symbol error-based variable-step non-delay subband filtering algorithm is applied, e _on (f _i ) A steady-state error signal spectrum when a variable step size non-delay sub-band filtering algorithm based on a symbol error is applied;

the constraint condition is as follows; k is more than or equal to 1 _M ≤5，1≤K _Lp ≤5，μ _L ≤μ _i ≤μ _U ；

Wherein: mu.s _L 、μ _U Respectively are the upper limit and the lower limit of the iteration step length, and the general value range is 0-1;

calling a quadratic programming algorithm function in an MATLAB optimization toolbox:

X＝quadprog(H，f，A，b，Aeq，beq，lb，ub)；

then the quadratic programming optimization objective is:

the constraint conditions are as follows:

in the formula: min [ f (K) _M ，K _Lp ，μ _i )]-an objective function, H-a quadratic objective term symmetric matrix; f-linear target term real vector, A, b-linear inequality constraint, aeq and beq-linear equality constraint, lb and ub-limit condition upper and lower limits.

In the above technical solution, hs of the secondary noise signal generated in S4 _n (k) An off-line identification mode is adopted for estimating the sub-band secondary path, and mainly an electro-acoustic path of a process that an electric signal output by a controller generates secondary noise through a secondary sound source and transmits the secondary noise to an error microphone comprises a D/A converter, a reconstruction filter, a power amplifier, a loudspeaker, a propagation path from the loudspeaker to the error microphone, a preamplifier, an anti-aliasing filter and an A/D converter.

An intelligent system parameter calibration system applied to vehicle road noise control is characterized in that a program for realizing the method is stored.

Therefore, the method and the system adopt a partition control mode based on face recognition, and complete the development of an intelligent calibration system of algorithm parameters according to the matching of the in-vehicle partition ARNC hardware and the adjustment method of the algorithm parameters under different driver and passenger scenes.

Meanwhile, the partition control and the global control have different requirements in the aspects of selection of an optimal reference signal and determination of parameters of an adaptive filtering algorithm. According to the method, through zone control, optimization of variable-step-length non-delay sub-band filtering algorithm VSS-DSAF parameters based on symbol errors can be completed only by applying a quadratic programming function in the aspects of calculation complexity, convergence speed and impact interference resistance of the algorithm, optimal noise reduction performance is achieved, and the defects that the existing algorithm is difficult to rely on trial and error decoupling, high in calculation complexity, low in convergence speed and poor in stability are overcome.

And secondly, the secondary signals are generated according to the number and the positions of drivers and passengers, the noise reduction effect of each position can be realized according to the design requirement, and the noise reduction effect is more targeted.

Finally, the invention utilizes the control to generate the secondary signal low-efficiency primary noise signal, and can really solve the problem that the low-and-medium frequency path noise of the electric automobile is obvious after the masking effect of the engine order noise is lost.

Drawings

Fig. 1 is a schematic diagram of a main road noise control based on the method of the present invention.

FIG. 2 is a flow chart of the system parameter intelligent calibration method applied to the in-vehicle main road noise control of the invention

FIG. 3 is a flowchart of a screening procedure for optimal reference signal combinations according to the present invention.

FIG. 4 is a schematic diagram of a symbol error-based variable step size non-delay subband filtering algorithm (VSS-DSAF) of the present invention.

Detailed Description

The invention mainly solves the problem that the low-medium frequency road noise is obvious after the masking effect of the engine order noise of the electric automobile is lost, and mainly utilizes an active noise control technology to carry out partition control and intelligent optimized calibration of controller control parameters on the low-medium frequency noise of the automobile so as to realize the purpose of in-automobile noise reduction.

The invention provides an intelligent optimization calibration method and system for active control technical parameters of an in-vehicle road noise partition, which mainly comprises the steps of collecting noise data as a reference signal through an in-vehicle sensor of an automobile chassis, and identifying a control area by using the in-vehicle sensor to obtain the control area; based on the noise reduction maximization of the control area of the partition control mode, screening reference signals by utilizing a multiple coherent analysis method, and screening out the optimal reference signal combination of the control area; and the multi-channel main road noise control algorithm based on low computation complexity generates secondary noise with the same amplitude as a noise signal to be counteracted (primary noise in a vehicle) and opposite phase to realize interference cancellation of noise in a control area, and optimization design is carried out on parameters of the main road noise control algorithm by utilizing an optimization algorithm, wherein the main road noise control principle based on the method is shown in figure 1.

Vehicle driving conditions and working conditions: two new energy test vehicles respectively provided with a multi-connecting-rod independent suspension structure and a torsion Liang Feidu vertical suspension structure are used as test data acquisition objects, and vibration noise tests are performed on a rough asphalt pavement and a smooth asphalt pavement. The vehicle speed is regulated to be set to be in a constant speed cruise mode, and the steady-state test working condition is taken when the speed is increased by 10km/h from 30-120 km/h; the acceleration mode adopts full-throttle acceleration and takes the full-throttle acceleration as a time-varying unsteady test working condition. The test contents comprise: vibration reference signals at the connecting point (the side of the vehicle body) of the vehicle suspension and the vehicle body and sound pressure signals beside the headrest at four positions of the front row and the rear row in the vehicle are obtained, a data analysis sample library is obtained, and the test working conditions are shown in table 1.

The required instrumentation is shown in table 2:

serial number	Device name	Number of
			1	Lab test equipment	1 set of
2	PCB triaxial acceleration sensor	A plurality of
			3	Microphone (Sound pressure sensor)	A plurality of
4	Triaxial BNC plug cable	A plurality of
			5	Notebook computer	1 table
6	Magnetic base	A plurality of

The specific positions of the sensor arrangement are as follows: the front ends of left and right triangular arms of a front suspension and the passive end of a shock absorber connecting point are respectively provided with a vibration acceleration sensor, the rear ends of the left and right triangular arms of the front suspension and the passive end of the shock absorber connecting point are respectively provided with a vibration acceleration sensor, the passive end of the upper end of a left shock absorber damper and a right shock absorber damper of the front suspension is respectively provided with one vibration acceleration sensor, the passive end of the left and right front wheel center of the front suspension is respectively provided with one vibration acceleration sensor, the passive end of the middle part of a left and right front wheel center of a vehicle body and a chassis is respectively provided with one passive end, the passive end of the front end of a left and right torsion beam of a rear suspension is respectively provided with one passive end, the passive end of the rear end of the left and right torsion beam of the rear suspension is respectively provided with one passive end, the passive end of the front end of the left and right torsion beam of the rear suspension is respectively provided with one passive end, 20 vibration acceleration sensors are arranged at 20 collection points, and microphones (sound pressure sensors) are arranged on the left side (DR), on the driving right side (DR), the auxiliary driving left ear (PL), the rear row right side (RLR) and the left ear (RRL).

And (3) active control design of the road noise subareas in the vehicle: aiming at the arrangement and combination conditions of different numbers and different seating positions of drivers and passengers in the automobile, the face recognition technology is adopted to recognize the seating conditions of the drivers and the passengers in the automobile, and the invention particularly points out that the face recognition technology is mature, is only borrowed from the prior art and is mainly used for recognizing the seating conditions of the passengers on each seat in the automobile, namely distinguishing the number of the passengers in the automobile and the respective seating positions of the passengers; therefore, active control architecture forms such as angles of centralized and distributed types, weight ratio among error channels and the like are formed, and a partitioned main road noise control rule base is established. The integrated type is full of passengers in the vehicle, the distributed type is that the passengers in the vehicle are respectively seated in a single area, a second area and a third area, and the seated positions of the passengers can be clearly identified according to the face recognition system, namely the seated positions of the passengers are set as road noise control areas.

Reference signal screening mode: lab test equipment data acquisition module is used for acquiring a frequency spectrum noise signal of an acceleration sensor and a microphone frequency spectrum noise signal at the ear side, wherein the frequency spectrum noise signal acquired by the acceleration sensor is a reference signal, and the microphone frequency spectrum noise signal at the ear side is a signal to be offset, namely a primary signal. And introducing the acquired signals into MATLAB simulation software, screening an optimal reference signal combination by using a multiple coherence analysis method, wherein the multiple coherence analysis method is based on the principle of calculating the coherence relationship between frequency domain signals of multiple input and multiple output signals, sequencing and screening different combination signals according to the theoretical maximum noise reduction (the reference signal has the maximum coherence), and recombining the signals based on the screened reference points until a target effect is achieved.

The adoption of the noise reduction algorithm: and based on the screened optimal reference signal combination, generating a secondary offset signal by combining with the in-vehicle road noise partition control and adopting an active road noise control algorithm of a variable step length non-delay subband filtering algorithm (VSS-DSAF) based on a symbol error, and realizing noise reduction of a specific area through a secondary sound source.

Intelligent optimization of algorithm parameters: the cancellation signal (secondary signal) emitted by the secondary sound source affects the noise control performance, which is related to the strong coherence of the reference signal and the algorithm processing capacity, wherein the algorithm parameters mainly affect the calculation complexity and the convergence speed, and the parameter optimization of the symbol error-based variable-step-size delay-free subband filtering algorithm (VSS-DSAF) is mainly optimized by using a quadratic programming algorithm.

The flow chart of the intelligent calibration method for the system parameters applied to the in-vehicle main road noise control is shown in fig. 2. The method comprises the following steps:

two new energy test vehicles respectively provided with a multi-link independent suspension and a torsion Liang Feidu vertical suspension chassis structure are used as test data acquisition objects, and due to the difference of the structural forms of the multi-link independent suspension and the torsion Liang Feidu vertical suspension chassis, the arrangement positions of sensors are different, but the sensors are arranged at the passive end of the structural attachment point of a vehicle body and the chassis so as to acquire vehicle road noise signals more comprehensively. The invention takes the sensor arrangement position of the torsion Liang Feidu vertical suspension chassis structure as an example: the vibration acceleration sensor is characterized in that a vibration acceleration sensor is respectively arranged at the front end of a left triangular arm and a right triangular arm of a front suspension and the passive end of a connecting point of a shock absorber, a vibration acceleration sensor is respectively arranged at the rear end of the left triangular arm and the right triangular arm of the front suspension and the passive end of the connecting point of the shock absorber, one vibration acceleration sensor is respectively arranged at the passive end of the upper end of a left shock absorber and a right triangular arm of the front suspension, one vibration acceleration sensor is respectively arranged at the passive end of a left front wheel center, one vibration acceleration sensor is respectively arranged at the rear end of a left torsion beam and a right front wheel center, one passive end of a left torsion beam and a right torsion beam of a rear suspension, one passive end of the rear torsion beam is respectively arranged at the left end and the right rear end of the torsion beam of the rear suspension, 20 vibration acceleration sensors are respectively arranged at 20 collecting points, and a microphone (RRL) is arranged at the left side (DR), a left ear (PL), a right ear (RLR) of a rear row and a right side (RRL) of a main drive, and a microphone (2 instrument.

The automobile running road surface conditions are a rough asphalt road surface and a smooth asphalt road surface. The vehicle speed is regulated to be set to be in a constant speed cruise mode, and the steady-state test working condition is taken when the speed is increased by 10km/h from 30-120 km/h; the acceleration mode adopts full throttle acceleration and takes the full throttle acceleration as a time-varying unsteady state test working condition, the data acquisition mode is that a LMS test.Lab test equipment data acquisition module is used for acquiring a frequency spectrum noise signal of an acceleration sensor and a microphone frequency spectrum noise signal at the side of an ear under each vehicle speed on each road surface, wherein the noise frequency spectrum signal acquired by the acceleration sensor is a reference signal, the microphone noise frequency spectrum signal at the side of the ear is a signal to be counteracted, namely a primary signal, and the test working condition is shown in table 1.

Since a plurality of seats of a vehicle need to reduce noise, and the noise transmission paths of the seats are different, when the seats are controlled as a whole, the noise reduction performance is influenced compared with the regional control, and if the seats are controlled according to the sitting condition of passengers, the noise reduction effect of a local region is improved, and the resources in the vehicle can be efficiently utilized, so that the regional road noise control is performed according to the sitting condition of the passengers.

Aiming at the arrangement and combination conditions of different numbers and different seating positions of drivers and passengers in the vehicle, the seating condition of the drivers and the passengers in the vehicle is identified by adopting a face identification technology, and the invention particularly points out that the face identification technology is mature, is only used for borrowing the prior art and is mainly used for identifying the seating condition of each seat of the passengers in the vehicle, namely distinguishing the number of the passengers in the vehicle and the respective seating positions of the passengers; therefore, active control architecture forms such as angles of centralized and distributed types, weight ratio among error channels and the like are formed, and a partitioned main road noise control rule base is established. The integrated type is full of passengers in the vehicle, the distributed type is that the passengers in the vehicle are respectively seated in a single area, a second area and a third area, and the seated positions of the passengers can be clearly identified according to the face recognition system, namely the seated positions of the passengers are set as road noise control areas.

Lab test equipment data acquisition module collects spectrum noise signals of an acceleration sensor and spectrum noise signals of microphones beside ears according to the requirement of zone control, when a face recognition system recognizes the sitting condition of passengers, if the face recognition system recognizes that only one passenger in a vehicle, namely a driver seat, needs to realize noise control, the microphones at the position are powered on to realize the collection of the spectrum noise signals in the vehicle, and the microphones at other positions are in a power-off state and do not need to collect signals. Similarly, when the face recognition system recognizes a second area, a third area and a full seat, the microphone at the corresponding position is electrified and collects the in-vehicle noise spectrum signal of the area, and the vibration acceleration sensor is always in a working state when collecting the road noise. In subsequent control, the secondary sound sources (speakers) in the non-control areas are all in the off state, while the control areas are all in the on state to generate interference cancellation of the secondary sound sources.

And introducing the acquired signals into MATLAB simulation software, screening an optimal reference signal combination by using a multiple coherence analysis method, wherein the multiple coherence analysis method is based on the principle of calculating the coherence relationship between frequency domain signals of multiple input and multiple output signals, sequencing and screening different combination signals according to the theoretical maximum noise reduction (the reference signal has the maximum coherence), and recombining the signals based on the screened reference points until a target effect is achieved. The specific process of screening the optimal reference signal combination by the multiple coherence analysis method is as follows, and the screening procedure of the optimal reference signal combination is shown in fig. 3:

in the invention, the input signal is a vibration signal collected by 20 acceleration sensors at the passive end of the connection part of a suspension and a vehicle body, and the output signal is a sound pressure signal collected by a microphone arranged at a passenger headrest in the vehicle. Wherein, based on the coherence coefficient

Obtaining the target noise reduction amount according to the following formula:

in the formula (1), multiple coherence coefficient

The calculation formula is as follows:

in the formula: delta (f) is the theoretical maximum noise reduction,

for the purpose of the multiple coherence coefficients,

for an input signal x _n The resulting self-power spectrum of the output y (n); s _dd (f) Is the self-power spectrum of the primary noise signal d (n). S _xd Is x _n Cross-spectrum with d (n), superscript "T" denoting transposition, S _xx ＝X*X ^T Is x _n From the power spectrum, the superscript "-1" represents the generalized inverse of the matrix.

Firstly, calculating multiple coherence coefficients of all N reference signals and signals to be cancelled, taking the reference signal corresponding to the maximum value of the multiple coherence coefficients, ordering the reference signal into 1 and storing and recording the reference signal.

In the second iteration loop, the reference signals with the rank of 1 are removed, the multiple coherence coefficients of the rest N-1 reference signals and the signals to be cancelled are calculated, and the reference signal corresponding to the maximum value of the multiple coherence coefficients is taken and ranked as 2.

And the operation is circulated until the last reference signal. The position where the reference signal in the top ranking is located can be selected as the arrangement position of the vibration acceleration sensor of the road noise control system and used as the reference signal for system control.

The calculation formula for screening the reference signal by using the multiple coherence coefficient method is as follows:

in the formula: n is the number of input signals, from 1 to N, N is the total number of input signals or reference signals, f _u Is the highest frequency point, f _l Is the lowest frequency point.

In order to solve the problems of large calculation burden and low convergence speed of the FxLMS algorithm in broadband noise control, the invention adopts a variable step length non-delay sub-band filtering algorithm (VSS-DSAF) based on symbol errors. The reference signal is decomposed into sub-bands, each sub-band is independently provided with an iteration step length, the optimal noise reduction effect can be achieved in the full frequency band, and an algorithm schematic diagram is shown in fig. 4.

The algorithm is realized as follows:

step one, designing an analysis filter

h _n (i)＝h ₀ (i)e ^-j2πi/N ，i＝0，2...，L _p -1，n＝0，2，...N-1 (4)

L _p Is the length of the analysis filter, N is the number of subbands, h ₀ (i) By means of Matlab function h ₀ (i)＝fir1(L _p -1,1/N).

Step two, sub-band decomposition of the reference signal and the error signal:

1. subband reference signal:

in the formula: x is the number of _n (t) is a reference signal sequence, h _n (i) For analysing filter banks

2. Subband error signal:

in the formula: e.g. of the type _n (t) is the error signal sequence, h ₀ (i) By means of Matlab function h ₀ (i)＝fir1(L _p -1,1/N).

Step three, calculating a subband filtering reference signal and a subband error signal

In the formula: d decimation factor, r _n (t) is a self-filtering reference signal sequence, hs _n (k) For subband secondary path estimation, K is the number of reference signals.

Step four, controlling filter coefficient iteration by sub-band

μ _i (n)＝λμ _i (n-D)+(1-λ)min{μ _i，opt (n)，μ _i (n-D)} (11)

Wherein: r is _n (t)＝[r _n (t)，r _n (t-D)，......，r _n (t-M+D)]、w _n (t + D) is the subband control filter coefficient sequence, M is the full-band control filter length, ^H which represents the transpose of the conjugate, ^* denotes complex conjugation,. Mu. _i For the iteration step-size sequence, ∈ constant coefficient to prevent denominator from being zero, sgn (e) _n (t)) is a sign function.

And fifthly, converting the sub-band control filter coefficients into full-band coefficients (FFT-2 stacking method), performing 2M-point Fourier transform (FFT) on the control filter coefficients of each sub-band, stacking according to the following rule to obtain the Fourier transform (FFT) of the full-band weight coefficients, and performing 2M-point discrete Fourier transform (IFFT) to obtain the full-band coefficients.

Due to the algorithm parameter L _p 、M、μ _i The calculation complexity, the convergence rate and the anti-impact interference capability of the algorithm are determined, so that the algorithm parameters are simply tried and collected by an empirical method and obviously the optimal performance design cannot be met, and the algorithm parameters are optimized by quadratic programming.

Effective elimination of wideband noise requires adjustment of some of the most influential parameters L _p 、M、μ _i (0 ≦ i ≦ N/2) associated with the non-delayed subband filtering algorithm based on the non-delayed architecture. The optimal length (M) of the adaptive filter is easily determined by theoretical calculations or by trial and error of a single channel main path noise control system, which is not the case for a multi-channel main path noise control system. For simplicity, it is assumed that the frequency resolution is determined as follows: n is a fixed value before each optimization instance, so there are 1+1+ (N/2+1) = N/2+3 design variables. In the invention, the adjustable parameters are optimized by utilizing a quadratic programming algorithm. The parameters are optimized so that the generated synthetic noise immunity tracks the broadband noise to realize the ideal active path noise control performance.

The parameter set to be adjusted is:

X＝[K _M ，K _Lp ，μ ₁ ，μ ₂ ，…μ _N/2 ] (14)

in the formula: k _M ＝log ₂ M-4 with a confidence interval of {1,2,3,4,5} and an M confidence interval of {32, 64, 128, 256, 512}; k _Lp ＝log ₂ (L _p and/N) +2 with confidence interval of 1,2,3,4,5.

Since the objective of the present invention is to achieve the objective of noise reduction, which is related to the noise reduction performance of the main road noise control in a specific frequency range, in order to obtain an optimal solution, the objective of noise reduction is:

NR(f _i )＝10log ₁₀ {e _off (f _i )/e _on (f _i )} (15)

wherein e _off (f _i ) For non-application of variable step size based on symbol error without delaySteady state error signal spectrum, e, at subband filtering algorithm (VSS-DSAF) _on (f _i ) The steady state error signal spectrum is obtained when applying a symbol error based variable step size non-delayed subband filtering algorithm (VSS-DSAF).

Taking the minimization of the average noise reduction of the negative value as a performance index, the optimization problem is expressed as:

optimizing the target:

wherein: f. of _U To take into account the upper limit of the frequency, f _L In order to care about the lower limit of the frequency, the road noise is middle-low frequency broadband random noise, so the frequency range of interest is 50H _z ～500H _z MNR is the average noise reduction minimization.

The constraint condition is as follows;

1≤K _M ≤5，1≤K _Lp ≤5，μ _L ≤μ _i ≤μ _U (17)

wherein: mu.s _L 、μ _U The upper and lower limits of the iteration step length are respectively, and the value range is generally 0-1.

Combining the optimization targets with constraint conditions, and calling a quadratic programming algorithm function X = quadprog (H, f, A, b, aeq, beq, lb and ub) in the MATLAB optimization toolbox according to an algorithm calculation result, wherein the above formula is rewritten into a quadratic programming form:

the constraint conditions are as follows:

in the formula: min [ f (K) _M ，K _Lp ，μ _i )]-an objective function, H-quadratic objective term symmetric matrix; f-the real vector of the linear target term, A, b-the linear inequality constraint,the invention can complete the optimization of algorithm parameters by only applying quadratic programming functions and realize the optimal noise reduction performance.

Secondary channel hs _n (k) The method adopts an off-line identification mode, which is mainly an electric-acoustic path in the process that an electric signal output by a controller generates secondary noise through a secondary sound source and is transmitted to an error microphone, and comprises a D/A converter, a reconstruction filter, a power amplifier, a loudspeaker, a propagation path from the loudspeaker to the error microphone, a preamplifier, an anti-aliasing filter and an A/D converter.

It will be appreciated by those skilled in the art that modifications and variations are possible in light of the above teachings, but that the invention is not limited to the particular embodiments disclosed. All alternatives with similar effects according to the principles and concepts of the present invention should be considered as the protection scope of the present invention.

Claims

1. A system parameter intelligent calibration method applied to vehicle road noise control is characterized by comprising the following steps:

s1: arranging a sensor on a vehicle body, and acquiring vehicle noise data by using the sensor;

s4: based on the screened optimal reference signal combination, and in combination with an in-vehicle road noise partition control mode, generating a secondary noise signal by adopting a main road noise control algorithm of a variable-step-size delay-free sub-band filtering algorithm based on a symbol error to realize noise reduction of a specific area by interference cancellation; the secondary noise has the same amplitude and opposite phase with the frequency spectrum signal of the primary noise to be counteracted;

2. The intelligent calibration method for system parameters applied to the noise control in the vehicle according to claim 1, wherein step S1 is performed under various steady-state test conditions of constant-speed driving and under a full-throttle acceleration condition as a time-varying unsteady-state test condition to collect noise data respectively, wherein the noise data comprises collecting a spectrum noise signal of a vibration acceleration sensor at a connection point between a vehicle suspension and a vehicle body or at a vehicle body side as a reference signal, and collecting spectrum noise signals to be cancelled or primary noise signals emitted by microphones at four positions on the side of a headrest at the front row and the rear row in the vehicle.

3. The method for intelligently calibrating system parameters applied to controlling the noise in the vehicle interior according to claim 1, wherein in step S2, for the situation of different numbers of drivers and passengers and different seating positions in the vehicle, the zone control mode of the noise in the vehicle interior is divided into: the vehicle-mounted passenger seat control system comprises a centralized mode in which the vehicle-mounted passengers are fully seated and a distributed mode in which the vehicle-mounted passengers are respectively seated in a single area, a second area and a third area, wherein in each control mode, the seating positions of the passengers are in a road noise control area.

4. The intelligent calibration method for the system parameters applied to the noise control in the vehicle according to claim 1, wherein the step S1 collects noise data and is performed according to a partition control mode:

when the face recognition system recognizes a second area, a third area and a full seat, the microphones at the corresponding positions are electrified and acquire primary noise spectrum signals in the area, and the vibration acceleration sensor is always in a working state when acquiring the spectrum noise signals; in subsequent control, the secondary sound source or the microphone of the non-control area is in the closed state, and the microphone of the road noise control area is in the open state, so that the interference cancellation of the secondary sound source is generated.

5. The intelligent calibration method for the system parameters applied to the in-vehicle road noise control according to claim 1, wherein the step S3 is to introduce the collected noise spectrum signals into MATLAB simulation analysis software to screen optimal reference signals by using a multiple coherence analysis method:

the system comprises a suspension, a vehicle body, an acceleration sensor, a microphone, a sound pressure sensor and a sound pressure sensor, wherein the input signal is a vibration noise frequency spectrum signal acquired by the acceleration sensor at the passive end of the connection part of the suspension and the vehicle body, and the output signal is a sound pressure signal acquired by the microphone arranged at the passenger headrest in the vehicle;

calculating multiple coherence coefficients of the reference signal and the primary noise spectrum signal to be counteracted, obtaining a target noise reduction amount or a theoretical maximum noise reduction amount based on the coherence coefficients, sequencing and screening different combined signals according to the principle that the theoretical maximum noise reduction amount or the coherence of the reference signal is maximum, and then combining the signals again based on the screened reference points.

6. The intelligent calibration method for the system parameters applied to the control of the road noise in the vehicle according to claim 1, wherein the step S3 is to perform the sorting and screening processes on different combined signals according to the principle that the theoretical maximum noise reduction amount or the coherence of the reference signal is maximum as follows:

in a second iteration cycle, removing the reference signals in the sequence 1, calculating the multiple coherence coefficients of the remaining N-1 reference signals and primary noise spectrum signals to be counteracted, taking the reference signals corresponding to the maximum value of the multiple coherence coefficients, sequencing the reference signals into 2, and storing and recording the reference signals in a new reference signal selection set;

the above steps are circulated until the last reference signal is stored and recorded in the new reference signal selection set; and selecting the position where the reference signal in the reference signal selection is positioned in the front as the arrangement position of the vibration acceleration sensor, and taking the position as the reference signal.

7. The intelligent calibration method for the system parameters applied to the in-vehicle road noise control according to claim 1, wherein step S4 adopts a variable step size non-delay subband filtering algorithm VSS-DSAF based on symbol errors to decompose the reference signal into subbands, each subband independently sets an iteration step size, and a secondary noise signal is generated in a full band;

step S4, adopting a symbol error-based variable step size non-delay subband filtering algorithm VSS-DSAF specifically comprises the following steps:

step S41, designing an analysis filter:

step S42, sub-band decomposition of the reference signal and the error signal:

subband reference signal:

subband error signal:

in the formula: d decimation factor, r _n (t) is the self-filtering reference signal sequence, hs _n (k) Estimating sub-band secondary paths, wherein K is the number of reference signals;

step S44, subband control filter coefficient iteration:

μ _i (n)＝λμ _i (n-D)+(1-λ)min{μ _i，opt (n)，μ _i (n-D)}；

wherein: r is _n (t)＝[r _n (t)，r _n (t-D)，......，r _n (t-M+D)]、w _n (t + D) is a subband control filter coefficient sequence; m is the length of the full-band control filter, and the convergence speed is determined; ^H which represents the transpose of the conjugate, ^* represents a complex conjugate; mu.s _i Determining the anti-impact interference capability for an iteration step length sequence, wherein i is more than or equal to 0 and less than or equal to N/2; e is constant coefficient to prevent denominator from being zero, sgn (e) _n (t)) is a sign function;

determining optimal parameters of the analysis filter: length L of analysis filter _p Full band control filter length M, iterative step sequence mu _i 。

8. The intelligent calibration method for system parameters applied to in-vehicle road noise control according to claim 1, wherein in step S5, a quadratic programming algorithm is used to perform quadratic programming on each adjustable parameter L of the analysis filter of the symbol error-based variable-step non-delay subband filtering algorithm VSS-DSAF _p 、M、μ _i Optimizing to make adjustable parameter L _p 、M、μ _i An optimum value is reached to generate a synthetic secondary anti-noise tracking wideband noise to achieve ideal active path noise control performance.

9. The intelligent calibration method for the system parameters applied to the in-vehicle road noise control according to claim 1, wherein the sub-band secondary path estimation for generating the secondary noise signal in S4 adopts an off-line identification mode, and mainly comprises an electrical-acoustic path in the process that the electrical signal output by the controller generates the secondary noise through the secondary sound source and transmits the secondary noise to the error microphone, wherein the electrical-acoustic path comprises a D/a converter, a reconstruction filter, a power amplifier, a loudspeaker, a propagation path from the loudspeaker to the error microphone, a preamplifier, an anti-aliasing filter and an a/D converter.

10. An intelligent system parameter calibration system applied to vehicle road noise control is characterized in that a program for realizing the method is stored.