阅读说明：本技术 基于动态力识别的振动噪声贡献量分析方法及车辆 (Vibration noise contribution analysis method based on dynamic force identification and vehicle ) 是由 张健 贾文宇 马东正 范大力 李沙 宫世超 贾小利 尹冬冬 于 2021-06-17 设计创作，主要内容包括：本发明公开了一种基于动态力识别的振动噪声贡献量分析方法,包括：S1.确定待测车辆的目标点,在目标点处布置传感器；S2.基于驾乘场景对待测车辆进行试验,并收集传感器采集的工况数据G-R；S3.将待测车辆的悬架拆卸,并将由金属材质制成的悬架替换件按照设定扭力值安装到原悬架所在位置处；S4.采用锤击法对拆卸掉悬架后的待测车辆进行测试,获得悬架安装点至目标点处的频响函数H-R；S5.根据工况数据G-R和频响函数H-R建立动态载荷目标函数,运用迭代法求解悬架接附点动态载荷F；S6.根据悬架接附点动态载荷F与频响函数H-R计算各路径点贡献量,以此进行主路径识别与目标点响应预测。本发明能规避矩阵求逆带来的病态性问题,能够提高工作效率,且通用性强,可操作性高。(The invention discloses a vibration noise contribution analysis method based on dynamic force identification, which comprises the following steps: s1, determining a target point of a vehicle to be detected, and arranging a sensor at the target point; s2, testing the vehicle to be tested based on the driving scene, and collecting working condition data G collected by the sensor R (ii) a S3, disassembling a suspension of the vehicle to be tested, and installing a suspension replacement part made of a metal material to the position of the original suspension according to a set torsion value; s4, testing the vehicle to be tested after the suspension is detached by adopting a hammering method to obtain a frequency response function H from the mounting point of the suspension to the target point R (ii) a S5, according to working condition data G R Sum frequency response function H R Establishing a dynamic load objective function, and solving a dynamic load F of a suspension attachment point by using an iterative method; s6, according to the dynamic load F of the suspension attachment point and a frequency response functionH R And calculating the contribution amount of each path point so as to identify the main path and predict the response of the target point. The method can avoid the ill-conditioned problem caused by matrix inversion, can improve the working efficiency, and has strong universality and high operability.)

1. A vibration noise contribution analysis method based on dynamic force identification is characterized in that: the method comprises the following steps:

s1, determining a target point of a vehicle to be detected, and arranging a sensor at the target point;

s2, testing the vehicle to be tested based on the driving scene, and collecting working condition data G collected by the sensor_R；

S3, disassembling a suspension of the vehicle to be tested, and installing a suspension replacement part made of a metal material to the position of the original suspension according to a set torsion value;

s4, testing the vehicle to be tested after the suspension is detached by adopting a hammering method to obtain a frequency response function H from the mounting point of the suspension to the target point_R；

S5, according to the working condition data G obtained in the step S2_RAnd the frequency response function H obtained in step S4_REstablishing a dynamic load objective function, and solving a dynamic load F of a suspension attachment point by using an iterative method;

s6, according to the dynamic load F of the suspension attachment point and the frequency response function H obtained in the step S4_RAnd calculating the contribution amount of each path point so as to identify the main path and predict the response of the target point.

2. The vibration noise contribution analysis method based on dynamic force identification according to claim 1, wherein: the sensor direction is arranged according to the axis direction of a whole vehicle coordinate system O-XYZ, the direction from the vehicle head to the vehicle tail is + X, the direction from the main driving to the auxiliary driving is + Y, and the vertical direction is + Z.

3. The vibration noise contribution amount analysis method based on dynamic force identification according to claim 1 or 2, characterized in that: if the target point needs to solve the noise problem, a microphone sensor is installed at the target point, and if the target point needs to solve the vibration problem, a three-way vibration sensor is arranged at the target point.

Technical Field

The invention belongs to the technical field of dynamic load force identification, and particularly relates to a vibration noise contribution analysis method based on dynamic force identification.

Background

The NVH (noise, vibration and harshness) performance of the automobile is not only an important index of riding comfort, but also a key factor for improving the texture of the automobile product and strengthening the brand force. Therefore, each large automobile manufacturer has invested a lot of manpower and material resources in vehicle model research and development.

For example, patent document CN106644512A discloses a method and a system for analyzing noise based on a powertrain load, the method comprising the following steps: acquiring position information of a microphone and position information of a three-way acceleration sensor; performing road test on a vehicle to be tested based on different working conditions to obtain road test data of a microphone and a three-way acceleration sensor; after a power assembly of a vehicle to be tested is disassembled, the power assembly reformed by metal parts is mounted on the disassembled vehicle to be tested in a suspension manner, and a microphone and a three-way acceleration sensor are arranged; testing the disassembled vehicle to be tested by adopting a hammering method to obtain frequency response functions of the microphone and the three-way acceleration sensor; calculating the load of the power assembly according to the road test data and the frequency response function to obtain a load matrix of the power assembly; and predicting the noise in the vehicle to be measured under the actual road working condition according to the power assembly load matrix, thereby comprehensively considering the influence of each direction and a plurality of power assembly suspensions.

For another example, patent document CN106679990A discloses a method for testing six-degree-of-freedom wheel center force and calculating vibration noise contribution rate of an automobile, which includes: establishing a coordinate system, arranging measuring points and installing sensors, arranging excitation points and measuring coordinates of the excitation points, measuring a frequency response function, solving a frequency response function matrix from the hub center to the response points, solving a frequency response function vector from the hub center to the target point, performing a road test, solving an excitation power spectrum density matrix, verifying the identification effectiveness of the excitation power spectrum density matrix, and calculating the energy contribution rate of each degree of freedom excitation to the response of the target point.

In the patent documents, a frequency response function matrix inversion method is adopted for dynamic load identification, so that the problem that matrix inversion ill-conditioned behavior is the first problem is solved. The solving approaches of matrix inversion ill-conditioned problems are generally divided into two categories, namely, the number of three-way vibration sensors of reference points and a regularization method. The matrix inversion method is used for calculating dynamic loads, three-way vibration sensors are required to be arranged under each path, in order to avoid matrix ill-conditioned problems, the number of the sensors is required to be at least 2 times or more than the number of the identified exciting forces, the number of the sensors is large, and the channel requirements of a data acquisition system are high. Meanwhile, certain requirements are placed on the arrangement positions of the three-way vibration sensors, the positions of the three-way vibration sensors of the reference points at the same path point are suitable, repeated vibration information is easily acquired when the three-way vibration sensors of the reference points are too close to the path point, matrix ill-conditioned performance is caused, information acquired too far away is impure, and errors are introduced. Undoubtedly bring certain difficulties for the implementers, and the position and the space arrangement space are considered at the same time. In post-processing calculation, a regularization method is used for correcting matrix ill-condition, most of the current commercial software such as an LMS.

Therefore, it is necessary to develop a new vibration noise contribution amount analysis method based on dynamic force recognition and a vehicle.

Disclosure of Invention

The invention aims to provide a vibration noise contribution analysis method based on dynamic force identification, which can avoid the ill-conditioned problem caused by matrix inversion, save equipment resources, improve the working efficiency, and has strong universality and high operability.

The invention relates to a vibration noise contribution analysis method based on dynamic force identification, which comprises the following steps:

s1, determining a target point of a vehicle to be tested, arranging a sensor at the target point, if the noise problem of the target point needs to be solved, installing a microphone sensor at the target point, and if the vibration problem of the target point needs to be solved, arranging a three-way vibration sensor at the target point;

s2, testing the vehicle to be tested based on the driving scene, and collecting working condition data G collected by the sensor_R；

S3, disassembling a suspension of the vehicle to be tested, and installing a suspension replacement part made of a metal material to the position of the original suspension according to a set torsion value;

s4, testing the vehicle to be tested after the suspension is detached by adopting a hammering method to obtain a frequency response function H from the mounting point of the suspension to the target point_R；

S5, according to the working condition data G obtained in the step S2_RAnd the frequency response function H obtained in step S4_REstablishing a dynamic load objective function, and solving a dynamic load F of a suspension attachment point by using an iterative method;

s6, according to the dynamic load F of the suspension attachment point and the frequency response function H obtained in the step S4_RAnd calculating the contribution amount of each path point so as to identify the main path and predict the response of the target point.

Optionally, the sensor direction is arranged according to an axis direction of a finished automobile coordinate system O-XYZ, a direction from a head to a tail of the automobile is + X, a direction from a main driving direction to a secondary driving direction is + Y, and a vertical direction is + Z.

Alternatively, if the target point needs to solve the problem of the noise, a microphone sensor is installed at the target point, and if the target point needs to solve the problem of the vibration, a three-way vibration sensor is arranged at the target point.

The invention has the following advantages: and solving the dynamic load of the path point by using the frequency response function from the path point to the target point and the working condition data of the target point through an iterative method. Matrix inversion is not needed, so that the ill-conditioned problem caused by the matrix inversion is avoided. A reference point three-way vibration sensor does not need to be arranged near a path point, the requirement on a data acquisition front end channel is low, equipment resources are greatly saved, the workload of arranging a large number of three-way vibration sensors is saved, and the working efficiency is improved; the method has the advantages that the active and passive sides of the path points are removed during the static frequency response function test, the method is particularly suitable for noise vibration analysis matched with suspensions in different states under the same vehicle body state, the method can be repeated only by collecting working condition data of the sensor at the target point each time, and the method is strong in universality and high in operability.

Drawings

FIG. 1 is a flow chart of the present embodiment;

fig. 2 is a flowchart of the iterative algorithm in the present embodiment.

Detailed Description

The invention is described in the following with reference to the accompanying drawings.

As shown in fig. 1, in this embodiment, a vibration noise contribution analysis method based on dynamic force identification includes the following steps:

s1, arranging a target point sensor: determining a target point of a vehicle to be detected, arranging a sensor at the target point, installing a microphone sensor at the target point if the noise problem needs to be solved, and arranging a three-way vibration sensor at the target point if the vibration problem needs to be solved. Such as: if the problem of noise in the passenger compartment is solved, a microphone sensor can be arranged near the ears of the passenger compartment seat. If the problem of seat vibration is to be solved, a three-way vibration sensor can be arranged at the seat guide rail.

In this embodiment, the sensor directions are arranged according to the axial direction of a finished automobile coordinate system O-XYZ (the direction from the head to the tail of the automobile is + X, the direction from the main driving to the auxiliary driving is + Y, and the vertical direction is + Z).

S2, collecting working condition data: vehicle to be tested is tested based on driving scene, and working condition data G collected by sensor is collected_R。

S3, disassembling the suspension, and installing a suspension replacement part: in order to eliminate the structural coupling problem, before the static frequency response function test is carried out, the suspension of the vehicle to be tested is disassembled, and a suspension replacement part (such as a metal sleeve) made of a metal material is installed at the position of the original suspension according to a set torsion value.

S4, testing the vehicle to be tested after the suspension is detached by adopting a hammering method to obtain a frequency response function H from the mounting point of the suspension to the target point_R。

S5, according to the working condition data G obtained in the step S2_RAnd the frequency response function H obtained in step S4_RAnd establishing a dynamic load objective function, and solving the dynamic load F of the attachment point of the suspension by using an iterative method.

S6, according to the dynamic load F of the suspension attachment point and the frequency response function H obtained in the step S4_RAnd calculating the contribution amount of each path point so as to identify the main path and predict the response of the target point.

In this embodiment, the step S5 is described with an example of solving the noise:

according to the sound transmission characteristics, the in-vehicle sound is obtained by the following formula:

G_R＝H_R·F；

for solving the dynamic load F (i.e. the optimal value of the solved dynamic load), i.e. finding F_dMeets the target min G_R-H_R·F_dI, |, corresponding to F_dSatisfy min (G)_R-H_R·F)²Wherein F is_dActual dynamic load calculated from data measured by the sensors; let U (F) be (G)_R-H_R·F)²And then obtaining:

wherein U (F) is RⁿA function having a first continuous partial derivative of RⁿThe error value actually generated by each iteration is optimized by a gradient descent method, and an appropriate initial value F is selected⁽⁰⁾And continuously iterating, updating the value of the F, and minimizing the target function until convergence.

As shown in fig. 2, toFor search direction, k is the number of iterations, λ_kFor the step size of the kth iteration, determined by a one-dimensional search, i.e. λ_kSuch that:

the iterative algorithm flow is as follows:

inputting: objective function U (F), search directionSetting the precision epsilon;

and (3) outputting: minimum point F of U (F)^*。

(5.1) taking an initial value F⁽⁰⁾∈RⁿSetting k to be 0;

(5.2) calculation of U (F)^(k))；

(5.3) calculating the gradientWhen in useThen stop iteration and let F^*＝F^(k)(ii) a Otherwise, withTo search for a direction, make

(5.4) placingCalculating U (F)^(k+1)) When | | | U (F)^(k+1))-U(F^(k)) I < epsilon or F^(k+1)-F^(k)If | < epsilon, stop iteration, let F^*＝F^(k+1)I.e. F ═ F^*(ii) a Otherwise, k is set to k +1, and (5.3) is switched in.

According to the dynamic load F of the suspension attachment point and the static test frequency response function H_RAnd calculating the contribution amount from each path point to the interior of the vehicle so as to identify the main path and provide reference for problem rectification and optimization. In addition, the dynamic load F and the static test frequency response function H of the suspension attachment point can be further tested_RAnd modulating to predict target point response.