An abnormal sound prediction method for an intelligent cockpit audio system based on a frequency domain substructure method

Abstract

The application discloses an intelligent cabin sound system abnormal sound prediction method based on a frequency domain substructure method, and belongs to the field of automobile NVH and intelligent cabin acoustic design; the intelligent cabin sound system is modularized and decoupled into a loudspeaker excitation, a door assembly, a roof lining, a seat assembly and a body-in-white subsystem, and a lead matrix of each subsystem is independently established; the lead matrix of the subsystem is progressively completed by using frequency domain substructure coupling; the loudspeaker excitation is decomposed into structural concentrated force excitation and sound field distributed equivalent excitation, and is uniformly represented as an excitation force vector in the frequency domain; a target component vibration acceleration response spectrum is obtained, abnormal sound risk determination, positioning and visual cloud map output are completed according to three-layer BSR evaluation criteria of an acceleration RMS threshold, a characteristic frequency peak value and a transfer function peak-valley ratio; the application breaks through the bottleneck of large traditional calculation amount, weak decoupling capacity and inability to modularize iteration, and significantly improves the acoustic quality and development efficiency of the intelligent cabin.

Description

Technical Field

[0001] This invention belongs to the field of automotive NVH (Noise, Vibration, Harshness) engineering technology, specifically relating to a method for predicting abnormal noises (BSR, Buzz, Squeak, Rattle) in intelligent cockpit audio systems of new energy vehicles based on the frequency domain substructure method. Background Technology

[0002] With the rapid development of new energy vehicle technology, the acoustic quality of the intelligent cockpit, as the core space for human-computer interaction, has become one of the key indicators for measuring the overall high-end level of a vehicle. Compared with traditional fuel vehicles, new energy vehicles significantly reduce background noise during driving because they eliminate the internal combustion engine. While this characteristic improves the quietness of passengers, it also makes the problem of abnormal structural noise (i.e., "abnormal noise," BSR) induced by the audio speakers more prominent and easier for the human ear to detect.

[0003] Currently, to achieve an immersive audio experience and multi-zone separation control, smart cockpits often feature numerous speaker units in various locations, including but not limited to those installed on door trim panels, A / B / C pillar trim panels, dashboard, headliner (roof liner), and seat headrest backrests. When these speakers play music or voice commands, not only do their own mounting structures vibrate due to driving forces, but they also transmit sound radiation energy to surrounding structural components through sound-vibration coupling effects. These components, such as door sheet metal and trim panel assemblies, headliner frame and liner, B-pillar trim, and seat frames, resonate within specific frequency ranges, leading to friction and collision-like abnormal noises that severely impact the occupants' driving experience and overall vehicle quality perception.

[0004] From an engineering practice perspective, existing methods for predicting and analyzing abnormal noises mainly fall into the following three categories of technical approaches and their inherent limitations: The first category is simulation methods based on the Full Finite Element (Full FE) model. This method treats the entire vehicle or cabin structure as a unified finite element model, applying speaker excitation loads to calculate the dynamic response of each structural node to assess BSR risk. However, full vehicle finite element models often contain millions or even tens of millions of degrees of freedom and require detailed characterization of nonlinear or weakly connected components such as interior trim, sealing strips, and fasteners. Model building is time-consuming and computationally expensive, making rapid iterative evaluation difficult in the early stages of product design (conceptual design, scheme design phase). Furthermore, when integrating acoustic cavities into the model, this method also needs to handle acoustic-structure interaction equations, further multiplying the computational complexity.

[0005] The second category is based on Transfer Path Analysis (TPA). Classical TPA methods measure or calculate the frequency response function (FRF) between each excitation source and the target response point, and combine this with the excitation force load to decompose the overall response into the sum of the contributions of each transfer path, thereby identifying the dominant transfer path. However, traditional TPA methods have the following inherent drawbacks when applied to acoustic noise problems: First, traditional TPA uses "excitation force" as input, while the excitation mechanism of the loudspeaker on the surrounding structure involves distributed pressure loads at the sound field-structure interface, making it difficult to accurately extract the equivalent concentrated force; second, traditional TPA requires a complete set of measurements to be performed again when the system is modified (such as structural alteration or component replacement), making it unsuitable for convenient "virtual prototype" analysis; third, classical TPA lacks subsystem decoupling capabilities, failing to independently characterize the dynamic characteristics of the loudspeaker acoustic subsystem and the cockpit structural subsystem, thus limiting the engineering guidance value of the analysis conclusions.

[0006] The third category is based on numerical simulation of vibro-acoustic coupling. This typically employs a hybrid strategy combining the finite element method (FEM) with the boundary element method (BEM) or statistical energy analysis (SEA) to model the coupling problem between the loudspeaker's acoustic radiation field and the cabin structure. While this method offers good physical integrity, its computational model is complex to construct, highly sensitive to material parameters and damping characteristics, and its prediction accuracy in the mid-to-high frequency range (above 500Hz) is difficult to guarantee, limiting its engineering applicability.

[0007] In summary, existing technologies for predicting abnormal noises in the intelligent cockpit audio systems of new energy vehicles suffer from varying degrees of problems, including large model size, low computational efficiency, insufficient independent representation capabilities of subsystems, and poor early usability in engineering design. There is an urgent need for a new noise prediction method that can significantly improve computational efficiency while ensuring prediction accuracy and supporting independent modeling and flexible combination of subsystems. Summary of the Invention

[0008] The purpose of this invention is to provide a method for predicting abnormal noises in intelligent cockpit audio systems based on the frequency domain substructure method, so as to solve the technical problems of existing technologies in predicting BSR in cockpit audio systems of new energy vehicles, such as large computational model size, insufficient subsystem coupling analysis capability, poor early-stage usability, and inability to efficiently support parametric design iteration.

[0009] To address the aforementioned problems, this application provides a method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method, specifically including the following steps: Step S1: System decomposition step, dividing the intelligent cockpit audio system of new energy vehicles into a speaker excitation subsystem and at least one cockpit structure subsystem. The speaker excitation subsystem includes at least one of a speaker diaphragm, frame, magnetic circuit system and mounting bracket assembly. The cockpit structure subsystem includes at least one of a door assembly subsystem, a headliner subsystem and a seat assembly subsystem. The door assembly subsystem includes a door outer panel, anti-collision beam, door interior panel and buckle connection structure. The headliner subsystem includes a headliner frame, sound-absorbing cotton layer and inner lining skin structure. The seat assembly subsystem includes a seat frame, seat cushion foam, backrest frame and headrest support structure. Step S2: Frequency response function matrix establishment step, establish the frequency response function matrix at the interface degree of freedom and the target response degree of freedom for each subsystem respectively. The frequency response function matrix is ​​expressed in the form of admittance matrix and is obtained by inverting the dynamic stiffness matrix of each subsystem. Step S3: Interface Degrees of Freedom Identification Step. Identify the interface degrees of freedom corresponding to the physical connection points between each subsystem, and establish the connection constraint matrix L to ensure that the interface degrees of freedom of each subsystem satisfy the displacement compatibility condition. ;in, This is the summation vector of the degrees of freedom of the interfaces of each subsystem. It is a Boolean constraint matrix whose elements take only 0, 1 or -1, representing the constraint relationship between the degrees of freedom of the coupled interfaces; Step S4: Subsystem coupling step, based on the frequency domain substructure coupling formula. The admittance matrices of each subsystem are combined to obtain the overall admittance matrix of the coupled system. ;in, This is a block diagonal matrix of the admittance matrices of each subsystem. For the connection constraint matrix, and The sufficient condition for its existence is that the degrees of freedom of each interface are linearly independent; Step S5: Excitation load characterization step. The excitation of the cockpit structure by the loudspeaker during operation is divided into two categories: structural excitation and acoustic field excitation. These are characterized in the frequency domain as the excitation force spectrum at the corresponding subsystem interface degree of freedom. ; Step S6: Response calculation and risk assessment step, based on the overall admittance matrix of the coupled system established in step S4. and the loudspeaker excitation load spectrum determined in step S5. Calculate the vibration displacement response spectrum at each structural target point in the cockpit:

[0010] Based on the preset BSR multi-level assessment criteria, the risk of abnormal noise triggering in each component is quantitatively assessed and located.

[0011] Furthermore, the frequency response function matrix obtained in step S2 can be obtained through finite element simulation, experimental measurement, or a combination of both. When using finite element simulation, an independent finite element model is established for each subsystem, and the frequency response function of each degree of freedom is calculated using the modal superposition method. When using experimental measurement, an exciter or hammer excitation is applied to the physical subsystem, and frequency response function data is collected using an accelerometer or laser Doppler vibration meter.

[0012] Furthermore, the frequency domain substructure coupling operation described in step S4 is performed in a progressive manner. First, the loudspeaker excitation subsystem is coupled pairwise with adjacent structural subsystems to obtain an intermediate coupled subsystem. Then, according to the connection topology, the remaining subsystems are coupled to the intermediate coupled subsystem in sequence until the overall admittance matrix of the coupled system containing all subsystems is obtained. .

[0013] Furthermore, the acoustic field excitation mentioned in step S5 is characterized by the acoustic-structure coupling matrix S, and the equivalent excitation force is calculated using the following formula: ,in The sound pressure vectors at each node of the cabin acoustic cavity are calculated from the loudspeaker sound radiation model. The elements of the acoustic-structure coupling matrix S are determined by the inner product integral of the structural surface shape function and the acoustic cavity sound pressure distribution function.

[0014] Furthermore, the multi-level BSR evaluation in step S6 includes: root mean square acceleration evaluation, peak characteristic frequency evaluation, and peak-to-valley ratio evaluation of the transfer function; when the root mean square acceleration value of any target structural component exceeds the preset BSR trigger threshold... At that time, it was determined that the component posed a risk of triggering abnormal noise.

[0015] Furthermore, it also includes step S7: optimizing the iteration steps. For the high-risk target structural components and their corresponding subsystem regions identified in step S6, the admittance matrix data of the corresponding subsystems is modified. After modification, steps S4 to S6 are re-executed to calculate the modified abnormal noise triggering risk, so that the calculation time of a single modification-evaluation does not exceed 1 / 10 of the direct simulation calculation time of the whole vehicle finite element model. The iteration convergence criterion is: the change in the root mean square value of acceleration between two consecutive iterations is less than 5%, or the preset maximum number of iterations of 10 is reached.

[0016] Furthermore, the method also includes a BSR risk cloud map generation step: mapping the BSR risk assessment results of each target structural component in step S6 onto the three-dimensional digital model of the cockpit in the form of color gradient, intuitively presenting the distribution of abnormal noise risk levels in different areas of the cockpit, and generating a BSR risk cloud map.

[0017] Furthermore, the speaker excitation subsystem described in step S1 includes a speaker unit and its mounting bracket assembly. The speaker excitation subsystem is installed at at least one of the following locations: door interior panel, dashboard, A-pillar trim panel, B-pillar trim panel, C-pillar trim panel, headliner, seat headrest, and seat back. When there are multiple speaker excitation subsystems, the admittance matrix of each speaker excitation subsystem is assembled in a block diagonal form, and the excitation force spectrum of each speaker is synthesized into a total excitation force spectrum according to the principle of linear superposition.

[0018] The technical solution of the present invention has the following significant advantages over the prior art: (1) Significantly improved computational efficiency: By using the frequency domain substructure method, the problem of millions of degrees of freedom of the whole vehicle is decomposed into several subsystems (the degrees of freedom of each subsystem are reduced to hundreds to tens of thousands). The time of a single coupled calculation is only 1 / 10 to 1 / 50 of that of direct simulation of the whole vehicle FE. For a typical cockpit model containing 5 subsystems, the whole vehicle FE simulation takes 12 hours, while this method only takes 15-30 minutes.

[0019] (2) The excitation modeling is complete and accurate: the acoustic-structural coupling matrix is ​​innovatively introduced. It characterizes the distributed excitation of loudspeaker sound pressure on a large-area structural surface, making up for the shortcomings of the traditional FBS method which only considers the transmission of structural forces, and improves the prediction accuracy by 15% to 30% compared with the pure structural coupling method.

[0020] (3) Practicality of risk assessment quantification: The established three-level BSR assessment criteria are closely integrated with the characteristics of human hearing perception, and can directly output the risk level and risk location without the need for additional subjective evaluation experiments to guide design modifications.

[0021] (4) Supports rapid design iteration: The modified admittance matrix formed in step S7 is recoupled to the closed loop of re-evaluation. The calculation time for a single iteration is no more than 10 minutes, and the optimization from high risk to low risk can be completed within 10 iterations. This is more than 20 times more efficient than the traditional iteration process of modifying CAD, re-meshing, and then re-simulating.

[0022] (5) The risk is visualized intuitively. This solution uses the BSR risk cloud map to directly map to the three-dimensional CAD model. Design engineers can intuitively identify risk areas and carry out targeted structural reinforcement or damping treatment, which reduces the communication cost between NVH engineers and structural engineers. Attached Figure Description

[0023] Figure 1 This is an overall flowchart of the abnormal noise prediction method for the intelligent cockpit audio system in Example 1.

[0024] Figure 2This is an overall flowchart of the abnormal noise prediction method for a smart cockpit audio system in a new energy vehicle in Example 2. Detailed Implementation

[0025] The following detailed description illustrates the specific implementation method: Example 1: like Figure 1 As shown, the abnormal noise prediction method for intelligent cockpit audio system based on frequency domain substructure method in this embodiment can be divided into the following six stages according to the implementation steps. The stages are interconnected and logically progressive, together forming a complete prediction method system.

[0026] Step S1: System decomposition and subsystem definition: First, the intelligent cockpit audio system is decomposed into a system with clearly defined physical meaning. Let the entire coupled system S consist of N interconnected subsystems, denoted as . The subsystems are spatially coupled through several discrete connection interfaces.

[0027] In typical application scenarios of intelligent cockpits in new energy vehicles, the division of subsystems is based on the following criteria: Subsystem A (loudspeaker excitation subsystem): This includes the loudspeaker diaphragm, frame, magnetic circuit system, and mounting bracket assembly. This subsystem is the active excitation source of the system, and its core dynamic characteristics are characterized by the loudspeaker impedance characteristics, diaphragm surface velocity distribution, and output admittance at the mounting point.

[0028] Subsystem B (Door Assembly Subsystem): This includes the outer door panel, anti-collision beam, interior door panel, and their snap-fit ​​connection structure. This subsystem is coupled to subsystem A through the connection interface at the speaker mounting flange and is the main load-bearing component for abnormal noises in the door area.

[0029] Subsystem C (Headliner Subsystem): This includes the headliner frame, sound-absorbing cotton layer, and inner lining skin structure. This subsystem is primarily excited by loudspeakers through the acoustic cavity radiation path, and also transmits structural vibrations through the connection point between the headliner frame and the body-in-white (BIW).

[0030] Subsystem D (Seat Assembly Subsystem): This includes the seat frame, seat cushion foam, backrest frame, and headrest support structure. This subsystem primarily applies distributed excitation to the seat structure through the cabin sound field and is an important analytical object for evaluating cabin acoustic comfort.

[0031] Subsystem E (Body-in-White Structure Subsystem): As the parent structure connecting the above-mentioned local subsystems, the body-in-white transmits and couples the local vibration responses through multiple connection interfaces, and is the main carrier of global dynamic characteristics.

[0032] The above decomposition strategy ensures that the frequency domain dynamic characteristics of each subsystem can be obtained through independent finite element modeling or experimental measurement, while the coupling relationship between the subsystems is strictly defined through the interface degrees of freedom.

[0033] Step S2: Establishing the frequency response function matrix of each subsystem: The core mathematical tool of the frequency domain substructure method is the frequency response function matrix of each subsystem under its free or constrained boundary conditions. The dynamic characteristics of each subsystem are described using the admittance matrix (i.e., the displacement / force mode frequency response function matrix) based on the degree of freedom division.

[0034] For a single subsystem ( Its equation of motion in the frequency domain can be expressed as:

[0035] in, , , These are the mass matrix, damping matrix, and stiffness matrix of the subsystem r, respectively. It is a frequency domain shift vector; It is the external force vector in the frequency domain; It is the angular frequency; The imaginary unit. Define a subsystem. The dynamic stiffness matrix is:

[0036] Then the admittance matrix of subsystem r is the inverse of the dynamic stiffness matrix:

[0037] Admittance matrix It is Complex matrices, For subsystem The degrees of freedom, its elements Indicates in When a unit harmonic force is applied to the degrees of freedom The frequency domain displacement response of the degrees of freedom, i.e., the displacement-force mode frequency response function.

[0038] There is no single limitation on the method for obtaining the frequency response function matrix of each subsystem; any one of the following methods or a combination thereof may be used: (1) Finite Element Method (FEM): An independent finite element model is established for each subsystem, and the frequency response function matrix within the required degrees of freedom is calculated by modal superposition or direct frequency response analysis. This method is suitable for simulation prediction when there is no physical prototype at the design stage.

[0039] (2) Experimental Modal Analysis and Frequency Response Function Measurement (EMA / FRF Testing): For the subsystem mounted on the physical sample or test fixture, the frequency response function of each degree of freedom is collected by exciter or hammer impact excitation using an accelerometer or laser Doppler vibration meter (LDV) to construct the experimental frequency response function matrix. This method is suitable for the sample verification stage and can correct simulation errors.

[0040] (3) Hybrid approach: The frequency response function at the key connection interface degrees of freedom is obtained through experimental measurement, and the remaining internal degrees of freedom are supplemented by finite element simulation data. A hybrid frequency response function matrix is ​​constructed through data fusion strategy to balance accuracy and efficiency.

[0041] Step S3: Interface Degrees of Freedom Identification and Connection Relationship Definition: In the frequency domain substructure method, the coupling between subsystems is described by "interface degrees of freedom". The set of degrees of freedom of each subsystem is divided into two categories: internal degrees of freedom and interface degrees of freedom.

[0042] For subsystems Its total degrees of freedom vector It can be represented in blocks as:

[0043] in, Represents the internal degree of freedom vector. This represents the interface degree of freedom vector. Correspondingly, the subsystem... The admittance matrix can be represented in terms of interior-interface blocks as follows:

[0044] The physical meaning of interface degrees of freedom corresponds to the connection points between subsystems, such as the translational and rotational degrees of freedom at the speaker mounting flange bolt holes, and the degrees of freedom at the interior panel clip connection points. The interface degrees of freedom are systematically identified through the following steps: First, based on the physical connection relationships of each subsystem (fastener connections, snap-fit ​​connections, adhesive bonding, etc.), the physical connection points between each subsystem are identified. Second, in the model of each subsystem, the nodal degrees of freedom corresponding to the above physical connection points are marked as interface degrees of freedom. Finally, a correspondence table of interface degrees of freedom is established to clarify the one-to-one correspondence of interface degrees of freedom between mutually coupled subsystems, i.e., the coupling condition matrix B, such that: (Displacement compatibility conditions), where, This is the summation vector of the degrees of freedom of the interfaces of each subsystem. It is a Boolean constraint matrix whose elements can only take the values ​​0, 1 or -1, representing the constraint relationship between the degrees of freedom of the coupled interfaces.

[0045] Step S4: Subsystem coupling based on admittance matrix synthesis: The frequency domain substructure method is applied to the prediction of cabin audio noise. The theoretical principle of FBS is rigorously explained below.

[0046] Frequency domain substructure method (FBS) is a systematic theoretical construction in the development of system synthesis and substructure technology. Its core idea is to synthesize the admittance matrix of each subsystem in the frequency domain by using the displacement compatibility condition and force balance condition satisfied by each subsystem at the interface, so as to obtain the overall frequency response function matrix of the coupled system without having to establish a finite element model of the overall system.

[0047] Suppose a coupled system consists of subsystems A and B connected by an interface degree of freedom (this example illustrates a two-subsystem coupling; the same logic applies to multi-subsystem coupling). The two subsystems must simultaneously satisfy the following two conditions at the interface: Condition 1: Displacement compatibility condition—that is, at the connection interface, the displacements of the corresponding degrees of freedom of the two subsystems must be equal: .

[0048] Condition 2: Force equilibrium condition – that is, at the connection interface, the internal forces between the two subsystems are equal in magnitude and opposite in direction: ;in Let be the coupling force vector experienced by subsystem r at the interface.

[0049] Under the above constraints, the frequency domain equations of motion for each subsystem can be uniformly written in the following form: ;in, Let be the vector of external excitation forces acting on subsystem r. This is a Boolean positioning matrix used to represent the global constraint force vector. (Lagrange multiplier vectors, representing the connection forces at the interface) are mapped onto the interface degrees of freedom of each subsystem.

[0050] By summing the dynamic stiffness matrices of each subsystem, the above equations can be rewritten in a summarized form. Define the summarized dynamic stiffness matrix. and summative displacement vector ,but: ,in, , To summarize the constraint matrix, satisfying (Matrix form of displacement compatibility conditions).

[0051] By combining the equations of motion with the displacement compatibility constraints, the interface coupling force (Lagrange multipliers) can be obtained. Then, substituting back into the equations of motion, we obtain the displacement responses of each subsystem for each degree of freedom. To completely transform the solution process into algebraic operations of the frequency domain admittance matrix, the following derivation path is adopted: From the summative equations of motion, in the "free" state without constraints, the summative displacement is: ;in The summation admittance matrix is ​​a block diagonal matrix consisting of the admittance matrices of each subsystem arranged along the diagonal.

[0052] After introducing Lagrange multipliers, the displacement response of the coupled system satisfying the constraints is:

[0053] After simplification, the overall admittance matrix of the coupled system This can be expressed as:

[0054] The above formula is the core synthesis formula of the frequency domain substructure method, also known as the "admittance matrix synthesis formula". This formula has the following important characteristics: First, the overall admittance matrix after coupling... Completely derived from the admittance matrices of each subsystem First, the connection relationship matrix L determines the overall model of the coupled system, eliminating the need to establish a holistic model. Second, when the admittance matrix of a certain subsystem changes, it is only necessary to update the admittance matrix data of the corresponding subsystem and re-execute the matrix algebra operation, thus realizing a "plug-and-play" modular analysis. Third, the case of multiple coupled subsystems can be handled under the same mathematical framework through step-by-step progression or one-time summary, exhibiting good scalability.

[0055] For the coupling of multiple subsystems: speaker subsystem A, door assembly subsystem B, headliner subsystem C, seat assembly subsystem D, and body-in-white subsystem E, the following strategy is adopted: First, based on the connection topology, the subsystem coupling operation is performed hierarchically and step by step; taking the speaker subsystem and door assembly subsystem as an example, the coupling of A and B is performed first to obtain the intermediate coupled subsystem AB, and then the coupling operation of AB and body-in-white subsystem E is performed at the corresponding interface. This process is repeated until the overall admittance matrix of the coupled system, which includes all subsystems, is finally obtained. .

[0056] Step S5: Frequency domain characterization of loudspeaker excitation load: The excitation exerted by the loudspeaker on the cockpit structure during playback can be described by the following two mechanisms: Mechanism 1: Structural Excitation. The speaker's drive mechanism (voice coil-magnetic circuit system) generates reciprocating force during operation, acting directly on the speaker frame and transmitted via the mounting flange to the door panel or other mounting structures, manifesting as a concentrated force. This type of excitation can be measured in the frequency domain by measuring the interfacial force spectrum at the speaker mounting point. In terms of characterization, its magnitude is closely related to the speaker power and the impedance of the mounting structure.

[0057] Mechanism Two: Acoustic Excitation. The vibration of the loudspeaker diaphragm does work on the air medium inside the cabin, creating a sound pressure distribution field within the sealed acoustic cavity. This sound pressure field applies distributed pressure loads to various structural surfaces, converting sound field energy into structural vibrations through the acoustic-structural coupling interface. This type of excitation manifests in the frequency domain through the cabin's acoustic pressure spectrum. Characterized by the integral of the structural area, it can be equivalently represented as the equivalent surface load vector at each structural node. .

[0058] The two types of excitations are treated uniformly, and corresponding excitation vectors are applied within the admittance matrix framework of each subsystem. For structural excitations, the interface force at the loudspeaker mounting point is directly used as the excitation input at the corresponding subsystem interface degree of freedom; for acoustic field excitations, the acoustic-structure coupling matrix S is used to represent the acoustic field as an equivalent force acting on each structural node.

[0059] in, S is the sound pressure vector of each acoustic node in the cabin acoustic cavity, which can be calculated from the loudspeaker sound radiation model, including the analytical dipole sound source model or the radiated sound field obtained based on the diaphragm FEM-BEM analysis; S is the sound-structure coupling matrix, whose elements are determined by the inner product integral of the structural surface shape function and the acoustic cavity sound pressure distribution function.

[0060] In practical engineering applications, the frequency domain characterization of the loudspeaker excitation signal is achieved through the following steps: First, the acceleration spectrum and near-field sound pressure spectrum at the installation point corresponding to the standard test signal played by the loudspeaker at its rated power are acquired. The standard test signal can be played as powder noise, a swept sine wave, or an actual music signal. Then, through inverse structure filtering or sound intensity measurement methods, the measured quantities are converted into an equivalent excitation force spectrum. Finally, the equivalent excitation force spectrum is used as the input load for frequency domain substructure coupling calculations. .

[0061] Step S6: Vibration Response Calculation and BSR Risk Assessment Based on the overall admittance matrix of the coupled system established in the fourth stage and the loudspeaker excitation load spectrum determined in the fifth stage. It can calculate the vibration displacement response spectrum at each structural target point in the cockpit:

[0062] Furthermore, the vibration acceleration response spectrum at each target point can be derived from the displacement response spectrum through a second-order differential (i.e., multiplied by) )get:

[0063] After obtaining the vibration acceleration spectrum of each target structural component, a noise risk assessment criterion based on the BSR sensitivity index is introduced. The following multi-level assessment index system is adopted: Level 1: Root Mean Square (RMS) Acceleration Evaluation. The acceleration spectrum at each target point is integrated over the audio operating frequency band, typically 20Hz to 20000Hz. In practical engineering, the focus is on the low-mid frequency range of 20Hz to 500Hz and the resonant frequencies caused by this band. The RMS acceleration value is then calculated. With respect to the preset BSR trigger threshold Comparison. When At that time, it was determined that there was a high risk of abnormal noise being triggered at the target point.

[0064] The second level: Characteristic frequency peak assessment. Identify the resonant frequencies appearing in the acceleration spectrum, determine whether they are located in the frequency range sensitive to human hearing, especially the mid-to-high frequency range of 800Hz to 4000Hz, and classify the risk according to the magnitude of the peak amplitude, which is divided into low risk, medium risk, and high risk.

[0065] The third level: peak-to-valley ratio evaluation of the transfer function. By analyzing the frequency response curve shape of the transfer function of the coupled system, frequency bands that may lead to modal coupling resonance amplification are identified, providing guidance for optimizing the dynamic stiffness of the structure.

[0066] Further, a BSR risk cloud map is established, which visualizes the BSR risk assessment results of each target structural component and maps them onto the three-dimensional digital model of the cockpit. The distribution of abnormal noise risk levels in different areas is presented intuitively in the form of color gradient, providing structural design engineers with directly usable optimization guidance information.

[0067] For identified high-risk areas, the following targeted optimization suggestion generation strategy is provided: by locally modifying the dynamic stiffness matrix of the corresponding subsystem, such as adding additional mass blocks to the target structural nodes or adjusting the equivalent stiffness value of the installation point, the admittance matrix synthesis calculation is re-executed to quickly evaluate the improvement effect of the modification scheme, thereby supporting a rapid iterative design process of "modification-evaluation-re-modification".

[0068] Example 2 This embodiment uses a 6.5-inch mid-bass speaker audio system in the left front door area of ​​a certain new energy pure electric vehicle as the prediction object. The speaker is mounted above the left front door trim panel, has a rated power of 30W, and a frequency response range of 50Hz to 20000Hz. The goal is to predict the risk of BSR (Browser Resonance) noise in the left front door trim panel, outer door panel, and inner door anti-collision beam when the speaker is operating, and to optimize the structure of high-risk areas.

[0069] like Figure 2 As shown, the specific method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method, and the optimization steps are as follows.

[0070] Step S1: System decomposition, dividing the left front door audio system into the following 4 subsystems; Four subsystems: (1) Loudspeaker excitation subsystem A: including loudspeaker diaphragm (diameter 165mm, equivalent mass 8g), frame (cast aluminum, mass 120g), magnetic circuit system (ferrite magnet, equivalent stiffness 2.5×10 5 The subsystem consists of a N / m diameter sheet metal plate and a mounting bracket assembly (stamped steel plate, 350g in mass, connected to the door interior panel by 3 bolts). The finite element model of this subsystem includes shell elements and mass elements, totaling 1200 degrees of freedom (DOF), of which the interface degrees of freedom are the 6 translational degrees of freedom of the 3 bolt connection points at the mounting point (each bolt connection point considers the x, y, and z directions).

[0071] (2) Door assembly subsystem B: includes the outer door panel (0.7mm DC04 steel plate, thickness 0.7mm), anti-collision beam (1.2mm hot-formed steel, cross-section cap shape), door interior panel (PP+TPO injection molded part, thickness 3mm) and 12 snap-fit ​​connection structures. The finite element model of this subsystem is modeled using shell elements, with a total of 8500 DOFs. The interface degrees of freedom include: 6 bolt connection DOFs shared with subsystem A, 8 hinge connection DOFs shared with subsystem C (at the door hinge), and 4 sealing strip bonding interface DOFs shared with subsystem D (equivalent to 8 DOFs of normal and tangential degrees of freedom of 4 nodes).

[0072] (3) Partial subsystem C of the body-in-white: including the lower section of the left front A-pillar, the left front sill beam and the front section of the left front longitudinal beam, all of which are welded parts of 1.5mm hot-formed steel plates. The finite element model of this subsystem has a total of 15,000 DOFs, and the interface degrees of freedom are the 8 hinge connection DOFs shared with subsystem B and the 6 bolt connection DOFs shared with subsystem E.

[0073] (4) Left side support subsystem D: includes the left side frame of the seat (welded steel pipe), the left side cushion foam (polyurethane, density 40kg / m³, thickness 80mm) and the left side backrest frame. The finite element model of this subsystem has a total of 6800 DOF, and the interface degrees of freedom are the 8 sealing strip bonding interfaces shared with subsystem B.

[0074] Step S2: Establishing the frequency response function matrix Formula for the dynamic stiffness matrix of subsystem A Calculate the total mass. g, modal damping ratio Extract the top 50 modes, highest analysis frequency Hz. The equivalent stiffness at the interface degree of freedom is N / m (radial) and N / m (tangential). Admittance matrix All dimensions The actual extracted sub-matrices are the interface DOF (6 points) and the target response DOF (200 points near the door interior panel mounting points), with dimensions of [dimension not specified]. .

[0075] Subsystem B was modeled using ANSYS Mechanical with 12,500 shell elements and 18,200 nodes. The modal synthesis method was used to reduce the order of the DOF to 8,500 (retaining the first 200 modes). The dynamic stiffness matrix was extracted from the interface DOFs (22 nodes: 6 shared with A, 8 shared with C, and 8 shared with D) and the target response DOF (500 nodes on the inner surface of the door trim panel), with dimensions of [missing information]. .

[0076] Subsystem C extracts interface DOF (14) and target response DOF (300), with the submatrix having dimensions of [missing information]. Subsystem D extracts interface DOF (8 elements) and target response DOF (200 elements), with the submatrix having dimensions of [missing information]. .

[0077] The admittance matrices of each subsystem are assembled into the overall admittance matrix in a block diagonal configuration. The total dimension is the sum of the dimensions of each subsystem.

[0078] The frequency response function was obtained using a combination of finite element simulation and experimental measurement methods: the interface degrees of freedom between subsystems (bolted connections, hinges) were measured experimentally using a force hammer and accelerometer to acquire the frequency response function, with a frequency range of 20Hz to 1000Hz; the internal degrees of freedom of each subsystem were calculated using finite element simulation, with a frequency range of 1000Hz to 20000Hz. The two sets of data were then fused and stitched together at 1000Hz using modal superposition.

[0079] Step S3: Identify interface degrees of freedom and define connection relationships.

[0080] Establish the connection constraint matrix based on the degree of freedom number of each subsystem interface. Taking the connection between subsystems A and B at the bolted joint as an example: Assume the interface node numbers for subsystem A are 1-6 (corresponding to the x, y, and z directions of the three bolts), and the interface node numbers for subsystem B are 7-12. Displacement compatibility requirements. Write it in matrix form. The corresponding non-zero elements in each row are: +1 in column 1 and -1 in column 7 of row 1; +1 in column 2 and -1 in column 8 of row 2; and so on. The matrix has 18 non-zero rows (corresponding to 18 independent compatibility equations), and its rank is 18, satisfying the following conditions: A sufficient condition for reversibility is that the degrees of freedom of each interface are linearly independent.

[0081] Step S4: Subsystem coupling, performing coupling operations in a progressive manner.

[0082] Step 1: Couple A and B. and The submatrix involving interface degrees of freedom is extracted and compared with... (12×12, corresponding to the 12 DOFs shared by A and B: 6 on side A + 6 on side B) Substitute them together into the admittance matrix synthesis formula:

[0083] Obtain the admittance matrix of the coupled AB subsystem , dimension (Subtract 12 duplicate DOF interfaces).

[0084] Step 2: Couple C to AB. Extract. The eight DOFs shared by C and Construct the corresponding 8 DOFs. ( Substituting into the formula, we get , dimension .

[0085] Step 3: Couple D to ABC. Similarly, extract the 8 shared DOFs and construct... ( ), thus obtaining the final overall admittance matrix of the coupled system. , dimension .

[0086] Comparison of computation time: Direct simulation of the entire vehicle FE model takes 12.5 hours; the method of calculating the frequency response function of the four subsystems separately (parallel computation) takes 2.5 hours, coupled computation takes 8 minutes, totaling about 2.6 hours, which is about 4.8 times more efficient. The computation time of a single modification-evaluation does not exceed 1 / 10 of the computation time of direct simulation of the entire vehicle finite element model.

[0087] Step S5: Excitation load characterization; The excitation of the cockpit structure by the loudspeaker during operation is divided into two categories: structural excitation and acoustic field excitation, which are characterized in the frequency domain as the excitation force spectrum at the corresponding subsystem interface degree of freedom.

[0088] Regarding structural excitation, the interfacial force spectrum at the loudspeaker mounting point was obtained through measurement: under a loudspeaker rated power of 30W and a 1kHz sinusoidal excitation, the equivalent force peak at the mounting point is 2.5N, and the spectrum is relatively flat in the range of 50Hz to 500Hz, with a lower frequency response above 500Hz. Attenuation. Structural excitation force vector. The equivalent force at the target response degrees of freedom is obtained with a dimension of 6 (corresponding to 6 mounting points DOF). Matrix mapping.

[0089] Acoustic field excitation through acoustic-structure coupling matrix Characterization, the formula for calculating the equivalent excitation force is as follows: ,in The sound pressure vectors at each node of the cockpit acoustic cavity are calculated from the loudspeaker sound radiation model. The elements are determined by the inner product integral of the structural surface shape function and the acoustic cavity sound pressure distribution function.

[0090] Sound pressure is calculated using an analytical dipole sound source model: normal velocity of the loudspeaker diaphragm. m / s (at 1 kHz and 30 W excitation), directivity factor air density kg / m³, speed of sound m / s, wavenumber rad / m. Taking a distance of 0.3m from the center of the diaphragm as an example, the sound pressure amplitude is approximately 122.5 Pa. Considering the effects of multiple reflections and sound-absorbing materials within the cabin, the actual effective sound pressure is taken as... Pa.

[0091] The acoustic cavity is discretized into Each acoustic unit has a structural coupling interface (inner surface of the door trim panel) discretized into... There are 1 node, therefore Matrix dimension is Equivalent excitation force of sound field Dimension is 1222 (and (The number of rows is consistent), indicating the equivalent distribution of sound pressure at all target response degrees of freedom.

[0092] Total incentive is If there is One speaker, then ,in For the first The admittance matrix of the loudspeaker excitation subsystem For the first The excitation force spectrum of a loudspeaker.

[0093] Step S6: Response calculation and risk assessment; based on the overall admittance matrix of the coupled system established in step S4. and the loudspeaker excitation load spectrum determined in step S5 Calculate the vibration displacement response spectrum at each structural target point in the cockpit:

[0094] The vibration acceleration response spectrum at each target point is obtained from the displacement response spectrum through second-order differentiation:

[0095] The calculation frequency range is 20Hz to 20000Hz, and sampling is performed at 1 / 3 octave bands, for a total of 120 frequency points.

[0096] Risk assessment is conducted based on the three-tiered BSR assessment criteria in Example 1: First-level evaluation: Calculate the root mean square value of acceleration. The left front door interior panel is used as the target component (selecting the area with the largest acceleration among 500 nodes on the inner surface). The door trim panels are made of PP+TPO plastic. m / s². Due to The risk level is determined to be high, and further assessment is stopped. The risk level output is high risk (red), and the risk location is the upper part of the left front door trim panel near the speaker mounting point (coordinates: x=320mm, y=680mm, z=150mm, with the door hinge as the origin).

[0097] To fully illustrate the three-tiered assessment process, let's assume the following subsequent assessment if the first-tier risk is not triggered: Second-level evaluation (characteristic frequency peak evaluation): Taking the outer door panel as an example, its Threshold, first level not triggered. The outer door panel exhibits a resonance peak at 1,250 Hz, peak acceleration. m / s². 1,250 Hz is located within the sensitive frequency range of the human ear, from 800 Hz to 4,000 Hz, and the peak acceleration exceeds the background noise floor value by approximately 33 dB, which is considered a medium risk.

[0098] Third-level evaluation (transfer function peak-to-trough ratio evaluation): Taking the left side of the seat cushion as an example. The first two levels were not triggered.

[0099] Calculate the peak-to-valley ratio of the transfer function from the excitation point (door trim panel mounting point) to the target response point (center point of the seat cushion surface). The transfer function has a peak at 350 Hz. There is a valley at 280Hz. ,but The risk level was determined to be low, but a modal coupling resonance amplification was indicated, and structural dynamic stiffness optimization was recommended.

[0100] Step S7: Design optimization iteration; For the high-risk target structural components and their corresponding subsystem regions identified in Step S6, by modifying the admittance matrix data of the corresponding subsystems, steps S4 to S6 are re-executed to calculate the modified abnormal noise triggering risk. The calculation time for a single modification-evaluation is 12 minutes (12.5 hours for the entire FE vehicle). The convergence criterion for iteration is: the change in the root mean square value of acceleration between two consecutive iterations is less than 5%, or the preset maximum number of iterations of 10 is reached.

[0101] First iteration (n=1): The risk area is the upper part of the interior trim panel near the speaker mounting point (3mm thick, insufficient stiffness). The optimization solution is to add a 2mm thick PP reinforcing rib (15mm wide, 80mm long) to this area, effectively increasing the local stiffness by 20%. Modify the admittance matrix: Adjust the stiffness matrix of subsystem B. Increase the equivalent stiffness at the nodes corresponding to the stiffeners. Reassessment If the value is greater than 0.35 m / s², it remains a high-risk condition. The change is... Continue iterating.

[0102] Second iteration (n=2): Based on the first stiffener, add another cross-direction stiffener, which effectively increases the local stiffness by a total of 35%. It is still greater than 0.35 m / s², still considered high-risk but close to the threshold. The change is... ,continue.

[0103] Third iteration (n=3): Based on the reinforcing ribs, a 0.5mm thick butyl rubber damping sheet (covering an area of ​​50cm²) is pasted on the back of the interior panel, effectively increasing the damping ratio. (Increased from 0.02 to 0.07). It is classified as low risk (green). The change is... However, since it has already fallen below the threshold, it can be stopped.

[0104] Summary of optimization results: After 3 iterations (total computation time 36 minutes), the RMS acceleration of the left front door interior panel decreased from 1.23 m / s² to 0.28 m / s², a reduction of 77.2%, successfully eliminating the BSR risk. If iterations continue, the fourth iteration will add an additional 0.3 mm damping layer to the existing damping sheet. Change ; 5th iteration If the change is less than 5%, the convergence criterion is met, and the iteration stops.

[0105] Step S8: BSR Risk Cloud Map Generation; The BSR risk assessment results of each target structural component in Step S6 are mapped onto the three-dimensional digital model of the cockpit in the form of color gradient, intuitively presenting the distribution of abnormal noise risk levels in different areas of the cockpit, and generating a BSR risk cloud map.

[0106] It should be noted that those skilled in the art can make various modifications and improvements without departing from the structure of this invention, and these should also be considered within the scope of protection of this invention. These modifications and improvements will not affect the effectiveness of the invention or the practicality of the patent. The scope of protection claimed in this application shall be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method, characterized in that, Includes the following steps: Step S1: System decomposition step, dividing the intelligent cockpit audio system of new energy vehicles into a speaker excitation subsystem and at least one cockpit structure subsystem, wherein the cockpit structure subsystem includes at least one of the door assembly subsystem, the headliner subsystem, and the seat assembly subsystem; Step S2: Frequency response function matrix establishment step, establish the frequency response function matrix at the interface degree of freedom and the target response degree of freedom for each subsystem respectively. The frequency response function matrix is ​​expressed in the form of admittance matrix and is obtained by inverting the dynamic stiffness matrix of each subsystem. Step S3: Interface Degrees of Freedom Identification Step. Identify the interface degrees of freedom corresponding to the physical connection points between each subsystem, and establish the connection constraint matrix L to ensure that the interface degrees of freedom of each subsystem satisfy the displacement compatibility condition. ;in, This is the summation vector of the degrees of freedom of the interfaces of each subsystem. It is a Boolean constraint matrix whose elements take only 0, 1 or -1, representing the constraint relationship between the degrees of freedom of the coupled interfaces; Step S4: Subsystem coupling step, based on the frequency domain substructure coupling formula. The admittance matrices of each subsystem are combined to obtain the overall admittance matrix of the coupled system. ;in, This is a block diagonal matrix of the admittance matrices of each subsystem. For the connection constraint matrix, and The sufficient condition for its existence is that the degrees of freedom of each interface are linearly independent; Step S5: Excitation load characterization step. The excitation of the cockpit structure by the loudspeaker during operation is divided into two categories: structural excitation and acoustic field excitation. These are characterized in the frequency domain as the excitation force spectrum at the corresponding subsystem interface degree of freedom. ; Step S6: Response calculation and risk assessment step, based on the overall admittance matrix of the coupled system established in step S4. and the loudspeaker excitation load spectrum determined in step S5. Calculate the vibration displacement response spectrum at each structural target point in the cockpit: Based on the preset BSR multi-level assessment criteria, the risk of abnormal noise triggering in each component is quantitatively assessed and located.

2. The method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method according to claim 1, characterized in that, The frequency response function matrix mentioned in step S2 can be obtained by finite element simulation, experimental measurement, or a combination of both. When using finite element simulation, an independent finite element model is established for each subsystem, and the frequency response function of each degree of freedom is calculated by modal superposition. When using experimental measurement, an exciter or hammer excitation is applied to the physical subsystem, and the frequency response function data is collected by an accelerometer or laser Doppler vibration meter.

3. The method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method according to claim 1, characterized in that, The frequency domain substructure coupling operation described in step S4 is performed in a progressive manner. First, the loudspeaker excitation subsystem is coupled pairwise with adjacent structural subsystems to obtain an intermediate coupled subsystem. Then, according to the connection topology, the remaining subsystems are coupled to the intermediate coupled subsystem in sequence until the overall admittance matrix of the coupled system containing all subsystems is obtained. .

4. The method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method according to claim 1, characterized in that, The acoustic field excitation mentioned in step S5 is characterized by the acoustic-structure coupling matrix S, and the equivalent excitation force is calculated using the following formula: ,in The sound pressure vectors at each node of the cabin acoustic cavity are calculated from the loudspeaker sound radiation model. The elements of the acoustic-structure coupling matrix S are determined by the inner product integral of the structural surface shape function and the acoustic cavity sound pressure distribution function.

5. The method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method according to claim 1, characterized in that, The multi-level BSR evaluation in step S6 includes: root mean square acceleration evaluation, peak characteristic frequency evaluation, and peak-to-valley ratio evaluation of the transfer function; when the root mean square acceleration value of any target structural component exceeds the preset BSR trigger threshold... At that time, it was determined that the component posed a risk of triggering abnormal noise.

6. The method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method according to claim 1, characterized in that, It also includes step S7: optimizing the iteration steps. For the high-risk target structural components and their corresponding subsystem regions identified in step S6, the admittance matrix data of the corresponding subsystem is modified. After modification, steps S4 to S6 are re-executed to calculate the modified abnormal noise triggering risk, so that the calculation time of a single modification-evaluation does not exceed 1 / 10 of the direct simulation calculation time of the whole vehicle finite element model. The iteration convergence criterion is: the change in the root mean square value of acceleration between two consecutive iterations is less than 5%, or the preset maximum number of iterations of 10 is reached.

7. The method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method according to any one of claims 1 to 6, characterized in that, The method also includes a BSR risk cloud map generation step: mapping the BSR risk assessment results of each target structural component in step S6 onto the three-dimensional digital model of the cockpit in the form of color gradient, intuitively presenting the distribution of abnormal noise risk levels in different areas of the cockpit, and generating a BSR risk cloud map.

8. The method for predicting abnormal noises in an intelligent cockpit audio system based on the frequency domain substructure method according to claim 1, characterized in that, The speaker excitation subsystem described in step S1 includes a speaker unit and its mounting bracket assembly. The speaker excitation subsystem is installed at at least one of the following locations: door trim panel, dashboard, A-pillar trim panel, B-pillar trim panel, C-pillar trim panel, headliner, seat headrest, and seat back. When there are multiple speaker excitation subsystems, the admittance matrix of each speaker excitation subsystem is assembled in a block diagonal form, and the excitation force spectrum of each speaker is linearly superimposed to synthesize the total excitation force spectrum.

Images