Automobile seat vibration testing device, testing method and system based on multi-dimensional force feedback

Abstract

The application relates to the technical field of automobile seat dynamic characteristic testing and structure health monitoring, in particular to an automobile seat vibration testing device, a testing method and a system based on multidimensional force feedback, which comprises the following steps: acquiring six-dimensional force / torque signals of a seat bottom and acceleration response signals of a seat framework in real time; aligning time stamps of the two and preprocessing; obtaining a dynamic stiffness matrix and an equivalent damping coefficient, taking a calculation result as a feedback quantity; generating an error correction signal; generating a driving correction instruction and outputting the driving correction instruction to a driving component; constructing a real-time force-displacement hysteresis loop, extracting geometric characteristic parameters of the hysteresis loop; comparing the geometric characteristic parameters with preset health degree thresholds; if the thresholds are exceeded, generating a warning signal and adjusting output power or stopping the machine. The application has the effects of improving the problems of test deviation caused by single-dimensional feedback and the lack of state monitoring in related technologies, improving the vibration control precision in testing and the structure early warning capability.

Description

Technical Field

[0001] This application relates to the technical field of dynamic characteristic testing and structural health monitoring of automotive seats, and in particular to an automotive seat vibration testing device, testing method and system based on multidimensional force feedback. Background Technology

[0002] As a crucial component of the vehicle's interior, the structural strength, comfort, and durability of automotive seats directly impact the safety and experience of passengers. During the automotive research and development and validation phases, vibration durability testing is typically required to assess the reliability of seats under long-term driving conditions. Current automotive seat vibration testing primarily relies on multi-axial vibration tables, which use hydraulic or electric vibrators to simulate road surface excitation transmitted to the seat's underside during vehicle operation. During testing, the seat is fixed to the vibration table and undergoes prolonged sinusoidal frequency sweep or random vibration tests based on standard road condition spectra or customer-customized spectra to examine whether the seat frame, adjustment mechanisms, and connecting components exhibit fatigue fractures, loosening, or functional failures.

[0003] In related technologies, test control methods generally employ open-loop control or single-dimensional closed-loop control strategies based on acceleration feedback. Control systems typically place accelerometers at specific locations on the vibration table or fixture to collect acceleration signals in the vertical or single direction, comparing them with a preset target power spectral density. Iterative algorithms are then used to adjust the drive signal to reduce errors. While some advanced equipment incorporates force sensors to monitor load conditions, these are often used only for overload protection or simple force limiting control, failing to deeply integrate multi-dimensional force / torque signals into the real-time correction loop of the vibration waveform. For evaluating the dynamic characteristics of seats, related technologies largely rely on static stiffness measurements or modal analysis before and after testing. However, during vibration testing, only acceleration response data is typically recorded, lacking online monitoring methods for the seat's real-time dynamic parameters and energy dissipation characteristics.

[0004] Regarding the aforementioned technologies, since a car seat is a multi-degree-of-freedom system with complex nonlinear characteristics, relying solely on single-dimensional acceleration feedback is insufficient to effectively decouple coupled vibrations between the vertical, lateral, and rotational directions. This leads to significant lateral parasitic vibrations during testing, causing a deviation between the actual excitation applied to the seat and the target spectrum, thus affecting the accuracy of the test results. Furthermore, these technologies cannot detect minute changes in the seat's structural state in real time during testing; problems are often only discovered after macroscopic structural damage occurs. The lack of a structural health early warning mechanism based on real-time mechanical response characteristics not only risks excessive damage to test samples or even equipment safety accidents but also fails to provide refined dynamic data support for optimized seat design. Summary of the Invention

[0005] To address the issues of test bias and lack of condition monitoring caused by single-dimensional feedback in related technologies, and to improve the vibration control accuracy and structural early warning capability during testing, this application provides an automotive seat vibration testing device, testing method, and system based on multi-dimensional force feedback.

[0006] In a first aspect, this application provides a vehicle seat vibration testing device based on multidimensional force feedback, employing the following technical solution: A vibration testing device for automobile seats based on multidimensional force feedback includes: The platform assembly for supporting the seat under test includes a base, a column and a work surface, wherein mounting holes are arranged in an array on the work surface; A drive assembly for generating vibration is mounted within the base to transmit vibration excitation to the worktable surface. A multi-dimensional sensing component for acquiring six-dimensional force / torque signals of the seat under test in real time, the multi-dimensional sensing component being arranged between the workbench and the mounting interface of the seat under test; A motion response component for acquiring vibration acceleration response signals of a seat under test, the motion response component being arranged in key structural parts of the seat under test; A closed-loop control unit is used to adjust the vibration output of the drive component in real time based on the acquired six-dimensional force / torque signal and acceleration signal. The closed-loop control unit is connected to the drive component, the multi-dimensional sensing component and the motion response component to form a closed-loop feedback control.

[0007] By adopting the above technical solution, multi-dimensional sensing components are arranged on the installation interface to achieve the acquisition of six-dimensional force / torque signals, eliminating signal delay caused by the transmission path. Motion response components are arranged in the seat structure, working in conjunction with the closed-loop control unit to construct a multi-signal source feedback loop. This layout allows the drive components to receive interface force data and structural response data, replacing the single-point monitoring mode of the platform. The coordinated arrangement of the multi-dimensional sensing components and motion response components provides multi-dimensional input data to the control unit, establishing a correspondence between vibration excitation output and the seat's force state. This hardware architecture provides a physical basis for dynamic parameter identification, solving the problem that single-dimensional sensors cannot reflect spatially coupled forces, and realizing multi-dimensional data acquisition and control command issuance.

[0008] Secondly, this application provides a method for testing the vibration of an automobile seat based on multidimensional force feedback, applicable to the automobile seat vibration testing device based on multidimensional force feedback as described in the first aspect, using the following technical solution: A method for testing the vibration of automotive seats based on multidimensional force feedback, comprising: Real-time acquisition of six-dimensional force / torque signals from the bottom of the seat, and simultaneous acquisition of acceleration response signals from the seat frame; The six-dimensional force / torque signal and the acceleration response signal are time-stamped and pre-processed. The preprocessed six-dimensional force / torque signal and acceleration response signal are input into the pre-constructed dynamic feature extraction model to calculate the dynamic stiffness matrix and equivalent damping coefficient characterizing the current state of the seat. The calculation results are used as feedback quantities. The feedback quantity is compared with the preset target vibration spectrum to perform error analysis and generate an error correction signal. Based on the error correction signal, the inverse system control strategy is invoked to generate a drive correction command, which is then output to the drive component to adjust the output waveform of the vibration table. During the test cycle, a real-time force-displacement hysteresis loop is constructed based on the six-dimensional force / torque signal and the displacement signal obtained by integration, and the geometric feature parameters of the hysteresis loop are extracted. The geometric feature parameters are compared with the preset health threshold. If the geometric characteristic parameters are determined to exceed the threshold range, a structural failure warning signal is generated and the drive components are controlled to adjust the output power or shut down.

[0009] By employing the above technical solution, a dynamic feature extraction model is constructed using six-dimensional force / torque signals and acceleration response signals. The dynamic stiffness matrix and equivalent damping coefficient are calculated, and the calculation results are introduced into the control loop for error correction. The inverse system control strategy generates drive correction commands based on the error signals to adjust the vibration table output waveform. During the test cycle, a real-time hysteresis loop is constructed using force and displacement integrals. Geometric feature parameters are extracted and compared with health thresholds to form an online structural state assessment mechanism. Mechanical response characteristics are transformed into control feedback quantities and health early warning indicators, enabling closed-loop regulation of vibration excitation and structural damage identification. This replaces traditional open-loop or single-dimensional iterative modes, providing a failure early warning mechanism and ensuring the safety of the testing process.

[0010] Optionally, the specific steps for inputting the preprocessed six-dimensional force / torque signal and acceleration response signal into a pre-constructed dynamic feature extraction model to calculate the dynamic stiffness matrix and equivalent damping coefficient characterizing the current state of the seat include: The six-dimensional force / torque signal is decomposed into vertical and lateral components; A multidimensional input-output relationship is constructed by combining the displacement components obtained from the integral of the acceleration response signal. Based on the multidimensional input-output relationship, the complex frequency response function between the input force and the output displacement is calculated using a system identification algorithm. Based on the characteristics of the real and imaginary parts of the complex frequency response function, the dynamic stiffness matrix and equivalent damping coefficient of the seat in the vertical, lateral and rotational degrees of freedom are separated and output.

[0011] By employing the above technical solution, the six-dimensional force / torque signal is decomposed into vertical and lateral components, and a multi-dimensional input-output relationship is constructed by combining acceleration, integral displacement, and other parameters. The complex frequency response function is calculated using a system identification algorithm, and the dynamic stiffness matrix and equivalent damping coefficient under different degrees of freedom are separated based on the characteristics of the real and imaginary parts. Time-domain vibration data is converted to the frequency domain for analysis, extracting the mechanical characteristics of the seat in the vertical, lateral, and rotational directions. Real-time calculation of dynamic parameters provides state variables for the control loop, giving the feedback quantities physical meaning. This replaces static stiffness measurement and post-test modal analysis, enabling continuous tracking of dynamic characteristics during testing and providing mathematical model support for multi-dimensional vibration control.

[0012] Optionally, after performing error analysis on the feedback quantity and the preset target vibration spectrum to generate an error correction signal, the step further includes: The error correction signal is input to the adaptive iterative regulator, which dynamically adjusts the correction step size factor according to the current error trend. A high-precision driving spectrum correction is generated by adjusting the step size factor and the error correction signal. The high-precision drive spectrum correction is superimposed on the current drive signal to form a closed-loop control circuit, eliminating lateral parasitic vibrations caused by dynamic response deviations.

[0013] By employing the above technical solution, the error correction signal is input into an adaptive iterative regulator, and the correction step size factor is adjusted according to the error change trend. The adjusted step size factor is used to generate a driving spectrum correction amount, which is then superimposed on the current driving signal to form a closed-loop control loop. This allows the iterative process to automatically converge based on the real-time error state, avoiding oscillation problems caused by a fixed step size. The superposition of the correction amount directly acts on the driving signal input, achieving directional suppression of lateral parasitic vibrations. Adaptive step size adjustment optimizes iterative efficiency, eliminates lateral interference components caused by multi-degree-of-freedom coupling, ensures the actual loaded waveform remains consistent with the target spectrum, and improves the stability of the vibration control loop.

[0014] Optionally, the specific steps for constructing a real-time force-displacement hysteresis loop based on the six-dimensional force / torque signal and the displacement signal obtained through integration include: Periodically and synchronously sample the vertical force component and the corresponding vertical displacement signal in the six-dimensional force / torque signal to generate a discrete force-displacement data point set; Using computer graphics algorithms, a set of data points is fitted into a closed planar trajectory. Calculate the enclosed area and tilt angle of the planar trajectory graph; The enclosing area and tilt angle are used as geometric characteristic parameters to characterize the energy dissipation properties of the seat structure.

[0015] By employing the above technical solution, the vertical force components and corresponding vertical displacement signals are periodically and synchronously sampled to generate a discrete data point set, and a closed-plane trajectory is fitted using a graphics algorithm. The area enclosed by the trajectory and the tilt angle are calculated and used as geometric feature parameters characterizing energy dissipation. This process transforms the vibration response into a hysteresis graph, quantifies the energy loss caused by internal friction and plastic deformation through area, and reflects the stiffness change trend through angle. The extraction of geometric parameters provides input for health assessment, replacing empirical judgment. Real-time quantification of energy dissipation characteristics is achieved, establishing a mapping relationship between mechanical response and structural damage, and providing monitoring indicators for subsequent threshold comparison.

[0016] Optionally, the preprocessed six-dimensional force / torque signal and acceleration response signal are input into a pre-constructed dynamic feature extraction model. The preceding steps also include: The preprocessed signal is input into the digital twin simulation environment to drive the virtual seat model corresponding to the physical test bench to move synchronously. In the virtual model, the stress distribution cloud map of the seat in the next testing phase is predicted based on the current dynamic stiffness matrix; The predicted stress distribution cloud map is compared with the preset safety envelope to generate the pre-simulation evaluation results; If the pre-test evaluation results show that there are predicted stress points that exceed the safety envelope, an optimized test strategy containing path correction instructions will be generated in advance before the physical test begins.

[0017] By employing the above technical solution, preprocessed signals are input into the digital twin simulation environment to drive the synchronous movement of the virtual seat model. Based on the current dynamic stiffness matrix, the stress distribution cloud map for the next testing phase is predicted and compared with the safety envelope to generate a pre-test evaluation result. Virtual working condition simulation is completed before physical excitation is applied, identifying stress concentration areas. When the predicted stress exceeds the safety range, the system automatically generates an optimized test strategy including path correction instructions, adjusting the loading scheme in advance. This transforms post-test detection into pre-test prediction, avoiding structural damage caused by stress exceeding limits during physical testing. The combination of virtual pre-testing and physical testing optimizes the load application path, reducing the risk of sample damage.

[0018] Optionally, the specific steps for time-stamp alignment and preprocessing of the six-dimensional force / torque signal and acceleration response signal include: A multidimensional decoupling algorithm is used to separate the components of a six-dimensional force / torque signal, extracting the pure vertical excitation component and the coupling interference component. The coupled interference component is input as a feedforward quantity to the active cancellation module to generate an anti-phase compensation signal; The anti-phase compensation signal is vector-superimposed with the original drive signal to output the corrected drive command, thereby eliminating lateral parasitic vibration caused by multi-axis coupling or installation eccentricity. The separated pure vertical excitation components and acceleration response signals are subjected to coherence analysis, and abnormal data segments with coherence below a preset threshold are removed.

[0019] By employing the above technical solution, a multi-dimensional decoupling algorithm is used to separate the six-dimensional force / torque signal, extracting the pure vertical excitation component and the coupled interference component. The coupled interference component is used as a feedforward input to the active cancellation module to generate an anti-phase compensation signal, which is then superimposed with the original drive signal vector to output a correction command. This feedforward compensation mechanism eliminates lateral parasitic vibration components during the signal input stage, blocking the transmission path of coupled vibrations. Subsequently, coherence analysis is performed on the pure vertical component and the acceleration response, eliminating abnormal data segments with coherence below the threshold. This method achieves decoupling processing and data cleaning of the excitation signal, eliminating the interference of multi-directional coupling on control accuracy. The synergistic effect of anti-phase compensation and coherence filtering ensures the reliability of feature extraction.

[0020] Thirdly, this application provides a multi-dimensional force feedback-based automotive seat vibration testing system, which adopts the following technical solution: A multi-dimensional force feedback-based automotive seat vibration testing system includes: The acquisition module is used to acquire the six-dimensional force / torque signal at the bottom of the seat and simultaneously acquire the acceleration response signal of the seat frame; A memory for storing programs for a multidimensional force feedback-based automotive seat vibration testing method, as described above. The processor and the program in the memory can be loaded and executed by the processor to implement the multidimensional force feedback-based automotive seat vibration testing method as described above.

[0021] By adopting the above technical solution, the acquisition module synchronously collects six-dimensional force / torque signals and acceleration response signals, the memory stores the test method program, and the processor executes the program to complete signal processing, model calculation, and control command generation. The multi-dimensional signal acquisition, dynamic feature extraction, error correction, and health warning processes are integrated into the computing platform. Through modularization, automated execution of data flow and command issuance is achieved, reducing the need for manual intervention. The processor's computing power supports the synchronous operation of real-time closed-loop control and digital twin pre-simulation. The system integration solution replaces distributed control equipment, ensuring the sequential execution of each step of the test method and data consistency, improving the integration and operational stability of the test system.

[0022] Fourthly, this application provides a smart terminal capable of storing corresponding programs, employing the following technical solution: A smart terminal stores a computer program that can be loaded by a processor and executed any of the above-mentioned methods for testing the vibration of car seats based on multidimensional force feedback.

[0023] Fifthly, this application provides a computer storage medium capable of storing corresponding programs, employing the following technical solution: A computer-readable storage medium storing a computer program that can be loaded by a processor and executed any of the above-described methods for testing the vibration of an automobile seat based on multidimensional force feedback.

[0024] In summary, this application includes at least one of the following beneficial technical effects: By acquiring six-dimensional force / torque signals from the bottom of the seat and the acceleration response of the frame, a multi-dimensional decoupling algorithm is used to separate the pure vertical excitation and coupled interference components, and an adaptive iterative regulator is used to dynamically adjust the correction step size. This mechanism uses the coupled interference as a feedforward quantity to generate an anti-phase compensation signal, which is superimposed on the drive command to actively cancel lateral parasitic vibrations, while eliminating low-coherence abnormal data. This solves the problem that traditional single-dimensional feedback cannot effectively decouple multi-degree-of-freedom coupled vibrations, reduces the impact of lateral parasitic vibrations on test accuracy, and ensures a high degree of consistency between the actual loaded waveform and the target spectrum. A dynamic feature extraction model is constructed to calculate the dynamic stiffness matrix and equivalent damping coefficient of the seat in real time, and these are used as feedback variables to introduce into the inverse system control strategy to achieve real-time correction of vibration output. Simultaneously, combined with a digital twin simulation environment, the stress distribution of the next stage is predicted based on the current dynamic parameters and compared with the safety envelope to generate optimized testing strategies in advance. This replaces the traditional static measurement and post-analysis mode, realizing a shift from open-loop / single-dimensional iteration to multi-dimensional dynamic closed-loop with pre-prediction, improving the physical accuracy of the control response and avoiding structural damage caused by stress exceeding limits through virtual simulation. During the test cycle, displacement signals are acquired through integration and simultaneously sampled with vertical force to construct a real-time force-displacement hysteresis loop. Geometric feature parameters such as the enclosing area and tilt angle are extracted to quantify energy dissipation characteristics. The real-time extracted parameters are compared with health thresholds; if they exceed the range, a structural failure warning is triggered, and power is automatically adjusted or the system is shut down. A mapping relationship between mechanical response and structural damage is established, identifying minute structural changes before macroscopic failure occurs. This overcomes the shortcomings of traditional methods that lack online health monitoring capabilities, effectively preventing excessive sample damage and equipment safety accidents. Attached Figure Description

[0025] Figure 1 This is a flowchart of a method for testing the vibration of an automobile seat based on multidimensional force feedback, according to an embodiment of this application.

[0026] Figure 2 This is a flowchart illustrating the specific steps in this application embodiment to input the preprocessed six-dimensional force / torque signal and acceleration response signal into a pre-constructed dynamic feature extraction model to calculate the dynamic stiffness matrix and equivalent damping coefficient characterizing the current state of the seat.

[0027] Figure 3 This is a flowchart of the steps following the steps in this application embodiment: performing error analysis between the feedback quantity and the preset target vibration spectrum to generate an error correction signal.

[0028] Figure 4 This is a flowchart illustrating the specific steps involved in constructing a real-time force-displacement hysteresis loop based on a six-dimensional force / torque signal and a displacement signal obtained through integration, as described in this application embodiment.

[0029] Figure 5 This is a flowchart illustrating the steps before the preprocessed six-dimensional force / torque signal and acceleration response signal are input into the pre-constructed dynamic feature extraction model, as described in this application embodiment.

[0030] Figure 6 This is a flowchart illustrating the specific steps involved in aligning and preprocessing the six-dimensional force / torque signal with the acceleration response signal according to an embodiment of this application.

[0031] Figure 7 This is a block diagram of a car seat vibration testing system based on multidimensional force feedback according to an embodiment of this application. Detailed Implementation

[0032] The present application will be further described in detail below with reference to the accompanying drawings. These specific embodiments are merely illustrative of the present application and are not intended to limit it. Those skilled in the art, after reading this specification, may make modifications to these embodiments without contributing any inventive step, but such modifications are protected by patent law as long as they fall within the scope of the claims of this application.

[0033] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the appendices in the embodiments of this application will be described below. Figures 1-7 The technical solutions in the embodiments of this application are clearly and completely described. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0034] This application discloses a multi-dimensional force feedback-based automotive seat vibration testing device. The device includes a platform assembly, a drive assembly, a multi-dimensional sensing assembly, a motion response assembly, and a closed-loop control unit. The platform assembly supports the seat under test and provides a stable mounting base. The drive assembly generates multi-degree-of-freedom vibration excitation, and the multi-dimensional sensing assembly acquires the mechanical response of the seat bottom in real time. The motion response assembly monitors the dynamic response of the seat frame, and the closed-loop control unit adjusts the drive assembly in real time based on the aforementioned signals to achieve high-precision closed-loop feedback control.

[0035] The platform assembly includes a base, columns, and a worktable. The base is the load-bearing foundation of the entire device. In this embodiment, the base is made of high-rigidity cast iron and has internal reinforcing ribs to suppress its own resonance. The columns are vertically fixed to the four corners of the base to support the worktable and isolate it from ground vibration interference.

[0036] The worktable is horizontally positioned at the top of the column, and its surface is arranged in an array of mounting holes. In this embodiment, the mounting holes are arranged in a grid pattern, with a center-to-center distance of 50mm between adjacent mounting holes, and the thread specification is M12. The high-density array of mounting holes in this embodiment adapts to the mounting hole spacing of seat rails in different vehicle models. Seat rails can be directly fixed to the worktable with bolts without the need for a separate transition plate, improving the device's versatility and clamping efficiency.

[0037] The base is equipped with an air-bearing vibration isolation system at the bottom, including multiple airbag support feet and a leveling mechanism. Before the test begins, the base is suspended by inflating the airbags, thereby cutting off the transmission path of low-frequency vibrations from the ground. This ensures that the force / torque signals collected by the multi-dimensional sensing components originate only from the excitation of the drive components and the dynamic response of the seat itself, eliminating the influence of environmental noise on measurement accuracy.

[0038] The drive assembly is installed in a receiving cavity inside the base, and its output end is connected to the bottom center of the worktable. The drive assembly includes a servo motor, a crankshaft connecting rod mechanism, and an elastic coupling. The servo motor is fixed to the inner wall of the base, and its output shaft is connected to the input end of the crankshaft connecting rod mechanism via a coupling.

[0039] The crankshaft connecting rod mechanism includes an eccentric wheel, connecting rod, and guide slider, which can convert the rotational motion of the servo motor into the reciprocating linear motion of the worktable in the vertical direction. In other variations of this embodiment, multiple servo motors can be used in conjunction with universal joints to drive the worktable to achieve six-degree-of-freedom vibration, in order to meet the needs of more complex working condition simulation. An elastic coupling element is located between the connecting rod and the worktable to buffer high-frequency impacts and compensate for installation errors, preventing stress concentration caused by rigid connections from damaging the sensor.

[0040] A multi-dimensional sensing assembly is positioned between the workbench and the mounting interface of the seat under test. The assembly includes a six-dimensional force / torque sensor and an adapter plate. Since the car seat is fixed to the vehicle body via a slide rail, and the dimensions of the six-dimensional force / torque sensor do not match the mounting interface of the seat slide rail, this embodiment uses a dedicated adapter plate. The lower surface of the adapter plate is fixed to the mounting holes on the workbench using high-strength bolts, and the upper surface of the adapter plate is machined with mounting grooves or threaded holes that match the bottom surface of the seat slide rail.

[0041] A six-dimensional force / torque sensor is embedded in the central slot of the adapter plate, ensuring that all vertical forces, lateral forces, and torques applied by the seat are completely transmitted to the sensor's sensitive element via the adapter plate. In this embodiment, the six-dimensional force / torque sensor is preferably a strain gauge-type high-precision sensor with a range covering ±5kN (force) and ±500N·m (torque), and a nonlinearity error of less than 0.5%FS. Before use, static calibration is required using standard weights and a torque wrench, and dynamic sensitivity calibration is performed by applying a dynamic load of known frequency and amplitude through a vibrator to eliminate zero-point drift caused by temperature drift or long-term loading.

[0042] The motion response component is used to acquire vibration acceleration response signals from key structural parts of the seat under test. The component includes several triaxial accelerometers, respectively positioned at the headrest mounting point, backrest center, seat cushion front end, and slide rail connection points on the seat frame. Other key nodes reflecting the overall modal characteristics and local resonance of the seat can also be arranged as needed. The accelerometers are fixed to the seat frame surface using magnetic bases to ensure they do not loosen or fall off under high-frequency vibration. The signal output terminals of each accelerometer are connected to the data acquisition module of the closed-loop control unit via shielded cables, providing rich spatial distribution data for subsequent dynamic feature extraction.

[0043] The closed-loop control unit is connected to the drive component, multi-dimensional sensing component, and motion response component via signals. In this embodiment, the closed-loop control unit includes an industrial computer, a data acquisition card, and a power amplifier. During operation, the multi-dimensional sensing component acquires six-dimensional force / torque signals from the bottom of the seat in real time, while the motion response component synchronously acquires acceleration signals from key points. The data acquisition card converts the analog signals into digital signals and transmits them to the industrial computer. The industrial computer internally runs a pre-built dynamic feature extraction model and an inverse system control algorithm.

[0044] The industrial control computer performs time-stamp alignment and preprocessing on the six-dimensional force / torque signals and acceleration signals to calculate the dynamic stiffness matrix and equivalent damping coefficient characterizing the current state of the seat. The calculation results are used as feedback variables to perform error analysis with a preset target vibration spectrum, generating an error correction signal. Based on the error correction signal, the inverse system control strategy is invoked to generate drive correction commands, which are sent to the power amplifier to adjust the output waveform of the servo motor. Through a closed-loop feedback mechanism, the device compensates in real time for response deviations caused by the nonlinear characteristics of the seat, ensuring that the vibration environment actually applied to the seat remains consistent with the target spectrum. Simultaneously, combined with subsequent health assessment algorithms, online monitoring and failure warning of the seat's structural state are achieved.

[0045] This application also provides a method for testing automotive seat vibration based on multidimensional force feedback, applied to the aforementioned automotive seat vibration testing device based on multidimensional force feedback. (Refer to...) Figure 1A multi-dimensional force feedback-based automotive seat vibration testing device includes: Step S100: Acquire the six-dimensional force / torque signal at the bottom of the seat in real time, and simultaneously acquire the acceleration response signal of the seat frame.

[0046] Among them, the six-dimensional force / torque signal refers to the data sequence of force components along the three orthogonal directions (X, Y, and Z) and torque components around these three axes, collected by sensors installed at the interface between the seat slide rail and the vibration table. The acceleration response signal refers to the vibration acceleration data sequence collected by three-axis accelerometers located at key nodes of the seat frame, such as the slide rail connection point, the upper crossbeam of the backrest frame, and the front crossbeam of the seat cushion.

[0047] The general process is described as follows: A six-dimensional force / torque sensor is embedded between the test bench's worktable and the mounting interface of the seat under test. A three-axis accelerometer is fixed to preset measurement points on the seat frame using magnetic attraction or adhesive bonding. The outputs of each sensor are connected to a multi-channel synchronous data acquisition card via shielded cables. The acquisition card performs analog-to-digital conversion on the analog electrical signals according to a preset sampling frequency, such as 2048Hz. The system uses hardware-level trigger signals to ensure strict alignment of the sampling clocks of the force and acceleration channels, eliminating phase delay between channels. The acquisition card encapsulates the converted digital signals into data frames with timestamps and channel identifiers, and transmits them in real-time to the control unit's memory buffer via a high-speed bus, completing the synchronous acquisition of multi-source vibration response signals.

[0048] Step S101: The six-dimensional force / torque signal and acceleration response signal are time-stamp aligned and preprocessed.

[0049] Timestamp alignment refers to eliminating data timing misalignment caused by differences in transmission paths or slight drift in sampling clocks between different sensor channels, ensuring that the data from each channel corresponds precisely on a unified time axis. Preprocessing refers to performing baseline drift correction, noise filtering, and outlier repair on the raw signal to improve the signal-to-noise ratio.

[0050] The general process is described as follows: The system reads the hardware timestamps carried in each data frame and uses a linear interpolation algorithm to resample the channel data with microsecond-level timing deviations, ensuring that all signals are strictly aligned to a unified time grid. Then, a moving average filter is used to smooth the aligned signal, removing high-frequency electromagnetic interference and mechanical resonance noise. A polynomial fitting algorithm is used to estimate the sensor's zero-point drift curve, and the DC bias component is subtracted from the original signal. Finally, a preset amplitude tolerance band is used to perform amplitude limiting verification on the signal, eliminating singularities caused by instantaneous impacts or line interruptions, and filling in the gaps with the average of adjacent valid sampling points, outputting a standardized preprocessed signal.

[0051] Step S102: Input the preprocessed six-dimensional force / torque signal and acceleration response signal into the pre-constructed dynamic feature extraction model to calculate the dynamic stiffness matrix and equivalent damping coefficient characterizing the current state of the seat, and use the calculation results as feedback quantities.

[0052] Among them, the dynamic feature extraction model refers to a mathematical mapping model built based on frequency domain system identification theory, used to convert time-domain mechanical inputs and motion outputs into frequency-domain dynamic characteristic parameters. The dynamic stiffness matrix and equivalent damping coefficient refer to a set of quantitative indicators reflecting the seat's ability to resist deformation and the energy dissipation characteristics of internal friction under multidimensional coupled excitation.

[0053] The general process is as follows: The system uses the preprocessed six-dimensional force / torque signal as the input excitation vector. The acceleration response signal is converted into a displacement response vector after quadratic integration and high-pass filtering, and both are input into the dynamic feature extraction model. The model internally calls the Fast Fourier Transform (FFT) algorithm to convert the time-domain data to the frequency domain and calculate the frequency response function matrix between each degree of freedom. Then, the least squares complex frequency domain method is used to fit the modal parameters of the FRF curve, separating the real and imaginary parts of the transfer function. The FRF curve is obtained by comparing the excitation force spectrum and the response displacement spectrum acquired by the system in real time in the frequency domain, characterizing the dynamic transfer characteristics of the seat at each test frequency, and subsequently serving as the input benchmark for modal parameter identification. Then, based on the inverse solution of the multi-degree-of-freedom vibration differential equation, the dynamic stiffness matrix elements and equivalent damping coefficients characterizing the current vertical, lateral, and rotational degrees of freedom of the seat are calculated in real time. The system encapsulates the calculation results into a state feedback vector and pushes it to the control algorithm module.

[0054] Step S103: Perform error analysis between the feedback quantity and the preset target vibration spectrum to generate an error correction signal.

[0055] The preset target vibration spectrum refers to the ideal vibration power spectral density (PSD) or time-domain target waveform that the seat is expected to achieve during the test, converted from industry testing standards or actual vehicle road spectrum data. Error analysis refers to comparing the deviation between the current actual dynamic response characteristics and the target spectrum, quantifying the amplitude distortion and phase lag.

[0056] The general process is described as follows: The system calls the target vibration spectrum data stored in the database and converts it into the same frequency domain representation as the current feedback quantity. The dynamic stiffness matrix and equivalent damping coefficient output in step S102 are substituted into the seat-vibration table coupling transfer model to calculate the actual response spectrum under the current excitation. A frequency-by-frequency comparison algorithm is used to calculate the amplitude and phase differences between the actual response spectrum and the target vibration spectrum in the key frequency bands, generating a multi-dimensional error vector. The system performs weighted fusion of the error vector, converting the amplitude deviation into a gain compensation coefficient and the phase deviation into a timing offset based on a preset error-correction mapping function. These two are then combined and encoded into an error correction signal, which is output to the decision module.

[0057] Step S104: Based on the error correction signal, the inverse system control strategy is invoked to generate a drive correction command, which is then output to the drive component to adjust the output waveform of the vibration table.

[0058] Among them, the inverse system control strategy refers to a feedforward-feedback composite control algorithm that uses the mathematical inverse model of the coupled system of the controlled object, i.e., the vibration table and the seat, to back-calculate the desired output response into the required input drive quantity. Drive correction commands refer to control pulses or analog voltage commands used to adjust the output current, frequency, or phase of the servo driver.

[0059] The general process is described as follows: The control system loads a pre-trained inverse dynamics model of the vibration table-seat system. The error correction signal is used as the input excitation of the inverse model. By solving the inverse state-space equations, the additional excitation amount required to compensate for the current response deviation is calculated at the drive end. The system vector-superimposes the additional excitation amount with the current basic drive signal to generate a drive correction command that includes amplitude gain adjustment and phase lead / lag compensation. This command, after digital-to-analog conversion and power amplification, is sent to the servo controller of the drive component. The servo controller adjusts the motor torque and speed in real time according to the correction command, changing the motion trajectory of the excitation mechanism and dynamically correcting the output waveform of the worktable.

[0060] Step S105: During the test cycle, a real-time force-displacement hysteresis loop is constructed based on the six-dimensional force / torque signal and the displacement signal obtained through integration, and the geometric feature parameters of the hysteresis loop are extracted.

[0061] Among them, the force-displacement hysteresis loop refers to the closed nonlinear trajectory curve formed between the vertical force and the corresponding vertical displacement of the seat structure under periodic vibration excitation, characterizing the viscoelasticity of the material and the frictional properties of the structure. Geometric characteristic parameters refer to indicators used to quantify the hysteresis loop shape, such as the enclosing area, principal axis tilt angle, and loop eccentricity.

[0062] The general process is described as follows: During each complete vibration cycle of the test run, the system synchronously captures the peak data of the vertical force component and the vertical displacement data after integral filtering. The discrete data points are mapped to a two-dimensional Cartesian coordinate system according to the time series. A B-spline curve fitting algorithm or a least-squares ellipse fitting algorithm is used to smoothly connect the discrete point set into a closed planar trajectory. Then, a numerical integration algorithm, such as Green's theorem, is called to calculate the geometric area enclosed by the closed curve. This enclosed area represents the structural damping energy dissipation within a single cycle. Simultaneously, principal component analysis is used to calculate the angle between the major axis of the loop and the horizontal axis, extracting the principal axis inclination angle. The enclosed area and the principal axis inclination angle are used as core geometric feature parameters and stored in the feature cache.

[0063] Step S106: Compare the geometric feature parameters with the preset health threshold.

[0064] Among them, the health threshold refers to the allowable fluctuation range of the hysteresis loop geometric parameters that characterize the seat structure in normal working condition, based on the factory calibration data of the new seat or the results of finite element simulation. It includes the area reference range and the tilt angle tolerance zone.

[0065] The general process is described as follows: The system reads the area enclosed by the hysteresis loop and the principal axis tilt angle extracted in the current cycle from the feature cache. It then calls the health threshold configuration file stored in the database to obtain the area benchmark interval and tilt angle standard value for the corresponding test conditions. An interval matching algorithm is used to perform an inclusion check between the real-time area value and the area benchmark interval. Simultaneously, the absolute deviation between the real-time tilt angle and the standard value is calculated. If the area value consistently exceeds the upper limit or falls below the lower limit, or if the tilt angle deviation exceeds the tolerance zone, the system marks it as an abnormal state. If all parameters fall within the threshold range, it is marked as a normal state. The comparison results generate logical flag bits and are transmitted to the early warning decision module.

[0066] Step S107: If the geometric feature parameters are determined to exceed the threshold range, a structural failure warning signal is generated and the drive component is controlled to adjust the output power or stop.

[0067] Among them, the structural failure warning signal refers to the control command indicating that the seat frame, connectors or damping elements have suffered structural damage such as fatigue cracks, plastic deformation or stiffness degradation.

[0068] The general process is described as follows: When the comparison result is marked as abnormal, the early warning decision module immediately generates a structural failure early warning signal. The system executes a graded response based on the severity of the exceedance of the threshold. If the parameter deviates slightly from the threshold, the early warning signal triggers power derating logic, and the control system sends a ramp-down command to the drive component to reduce the output power of the vibration table with a preset gradient, thus slowing down the rate of damage propagation. If the parameter severely exceeds the limit or exceeds the limit for multiple consecutive cycles, the early warning signal triggers emergency shutdown logic, and the control system immediately cuts off the servo enable signal of the drive component and activates the braking mechanism to lock the worktable. At the same time, the system packages and archives the original data from multiple sources of sensors, hysteresis loop images, and characteristic parameters before and after the triggering of the early warning, generating a fault diagnosis log for subsequent failure analysis.

[0069] Reference Figure 2 The specific steps for inputting the preprocessed six-dimensional force / torque signal and acceleration response signal into the pre-constructed dynamic feature extraction model to calculate the dynamic stiffness matrix and equivalent damping coefficient characterizing the current state of the seat include: Step S200: Decompose the six-dimensional force / torque signal into vertical and lateral components.

[0070] The six-dimensional force / torque signal refers to the original time-series data sequence acquired through multi-dimensional sensing components, containing three orthogonal force components (Fx, Fy, Fz) and three orthogonal torque components (Mx, My, Mz). The vertical and lateral components refer to the degree-of-freedom partitioning of the six-dimensional signal for multi-input multi-output system identification. The vertical component includes the force Fz along the vehicle's vertical direction (Z-axis) and the torques Mx and My around the lateral (X-axis) and longitudinal (Y-axis) directions; the lateral component includes the forces Fy and Fx along the vehicle's lateral (Y-axis) and longitudinal (X-axis) directions and the torque Mz around the vertical (Z-axis) direction. This partitioning is used to construct a cross-coupling transfer model between the vertical, lateral, and rotational degrees of freedom.

[0071] The general process is described as follows: Coordinate system calibration: A standard vehicle coordinate system is established with the vehicle's driving direction as the X-axis, the lateral direction as the Y-axis, and the vertical direction as the Z-axis. A pre-defined coordinate transformation matrix, such as a direction cosine matrix or an Euler angle rotation matrix, is used to map the sensor's physical coordinate system to the vehicle's coordinate system. Matrix projection: The pre-processed six-dimensional force / torque signal vector is left-leaning by the coordinate transformation matrix to perform a spatial projection transformation, eliminating cross-coupling errors caused by sensor installation angles or seat tilt. Component stripping: Based on the transformed data sequence, the Z-axis force Fz, X-axis torque Mx, and Y-axis torque My are extracted by dimension index and encapsulated as a vertical component data stream. Simultaneously, the X-axis force Fx, Y-axis force Fy, and Z-axis torque Mz are extracted and encapsulated as a lateral component data stream. Channel synchronization: Independent data buffers are allocated for the vertical and lateral components, and the original hardware timestamps are strictly preserved to ensure timing consistency during subsequent multi-dimensional alignment with the displacement signal.

[0072] Step S201: Construct a multidimensional input-output relationship by combining the displacement components obtained by integrating the acceleration response signal.

[0073] The displacement component refers to the time-domain displacement data sequence of each measuring point on the seat frame obtained after the acceleration response signal has undergone double time-domain integration and high-pass filtering to eliminate accumulated drift. The multidimensional input-output relationship refers to the mathematical model characterizing the dynamic mapping between input excitation (force / torque components) and output response (displacement components) in a multi-degree-of-freedom space, usually represented by a multiple-input multiple-output (MIMO) transfer function matrix or discrete state-space equations. The general process is as follows: First, numerical integration is performed on the triaxial acceleration response signal preprocessed in step S101. The trapezoidal integration method or Simpson integration method is used to complete the first integration to obtain the velocity component, and the second integration to obtain the original displacement component. Next, a Butterworth high-pass filter with a cutoff frequency of 0.5Hz is used to filter the original displacement component, eliminating low-frequency divergence errors caused by sensor zero-point drift and the double integration operation, outputting clean displacement component data. Subsequently, the displacement component is aligned with the vertical and lateral components extracted in step S200 using a unified timestamp to construct the input excitation vector U(t) and the output response vector Y(t). Finally, based on the theory of linear time-invariant systems, U(t) and Y(t) are mapped to the discrete time domain, and a multidimensional input-output relationship equation containing vertical, lateral and rotational degrees of freedom is established, providing a structured data foundation for subsequent frequency domain system identification.

[0074] Step S202: Based on the multidimensional input-output relationship, the complex frequency response function between the input force and the output displacement is calculated using a system identification algorithm.

[0075] The system identification algorithm refers to a calculation method that estimates the dynamic transfer characteristics of a linear system based on experimental input and output data through frequency domain or time domain mathematical processing. This embodiment adopts the H1 frequency response estimation method based on cross-spectral analysis. The complex frequency response function refers to the complex ratio of the output displacement response to the input force excitation at a specific frequency. Its magnitude characterizes the amplitude-frequency characteristics, and the phase angle characterizes the phase-frequency characteristics, fully reflecting the system's resonance peaks, anti-resonance points, and phase lag behavior.

[0076] The general process is as follows: A Hanning window function is applied to the constructed multidimensional input vector U(t) and output vector Y(t) to suppress the spectral leakage effect caused by signal truncation. The Fast Fourier Transform (FFT) algorithm is used to convert the time-domain data to the frequency domain, obtaining the input spectrum U(f) and output spectrum Y(f). Then, the self-power spectral density Gxx(f) of the input signal and the cross-power spectral density Gxy(f) between the input and output signals are calculated. The complex frequency response function is calculated point-by-point using the H1 frequency response estimation formula H(f) = Gxy(f) / Gxx(f). This estimator, assuming no noise in the input measurement and random interference at the output, can effectively suppress background noise in the response signal and improve the FRF estimation accuracy. Finally, the complex frequency response functions calculated by pairing each degree of freedom are matrix-assembled according to the excitation channel (row) and response channel (column) to generate a complete complex frequency response function matrix H(f), covering all dynamic coupling information within the target test frequency band.

[0077] Step S203: Based on the real and imaginary characteristics of the complex frequency response function, separate and output the dynamic stiffness matrix and equivalent damping coefficient of the seat in the vertical, lateral and rotational degrees of freedom.

[0078] The real and imaginary parts refer to the fact that the complex frequency response function H(ω) can be expressed as the sum of the real part Re[H(ω)] and the imaginary part Im[H(ω)]. The real part mainly reflects the superposition effect of the elastic restoring force and inertial force of the system, while the imaginary part mainly reflects the energy dissipation and phase lag effect during the vibration process of the system. The dynamic stiffness matrix and the equivalent damping coefficient refer to the fact that the dynamic stiffness matrix is ​​a quantitative expression of the coupling stiffness of each degree of freedom of the seat under frequency-varying excitation. The equivalent damping coefficient is a frequency domain parameter that characterizes the comprehensive energy dissipation characteristics of the seat structure, such as internal friction, material viscosity, and connecting component friction.

[0079] The general process is described as follows: First, the real part matrix Re[H(f)] and the imaginary part matrix Im[H(f)] are extracted from the complex frequency response function matrix H(f) at each frequency point. This is based on the frequency domain dynamic equation H(ω) of the multi-degree-of-freedom linear vibration system. -1 =K dyn(ω) -ω 2 M+jωC eq(ω) The least squares complex frequency domain (LSCF) fitting algorithm or orthogonal polynomial fitting method are used to identify the global parameters of the real and imaginary parts of the FRF curves. In the above equations, H(ω) is the complex frequency response function matrix of the system at angular frequency ω; ω is the angular frequency of the vibration excitation, in rad / s; K dyn(ω) The frequency-varying dynamic stiffness matrix characterizes the seat's resistance to deformation at the current frequency; M is the mass matrix of the seat structure, which is a constant matrix; j is the imaginary unit; C eq(ω)This is the equivalent damping coefficient matrix characterizing the internal friction and energy dissipation properties of the seat structure. The formula decouples the frequency domain response characteristics into three terms: stiffness, inertial force, and damping force, providing a mathematical benchmark for subsequent inverse solving of dynamic parameters using the real and imaginary parts.

[0080] During the fitting process, the frequency-varying dynamic stiffness term is separated using the inflection points and extrema of the real part frequency response curve, and the equivalent damping term is separated using the peak width and phase jump characteristics of the imaginary part frequency response curve. Then, the identified stiffness parameters and equivalent damping coefficients for each degree of freedom are reorganized into matrices according to the vertical (Z / Mx / My), lateral (X / Y), and rotational (Mz) degrees of freedom to construct the dynamic stiffness matrix K(ω) and the equivalent damping coefficient matrix C(ω) representing the current state of the seat. Finally, the calculated matrices and coefficients are packaged into a structured data package and pushed to the closed-loop control unit in real time as system state feedback for subsequent inverse model control and error correction.

[0081] Reference Figure 3 The step after performing error analysis on the feedback quantity and the preset target vibration spectrum to generate an error correction signal further includes: In step S300, the error correction signal is input into the adaptive iterative regulator, which dynamically adjusts the correction step size factor according to the current error trend.

[0082] The adaptive iterative regulator is an algorithm module built into the closed-loop control unit, used to automatically optimize the iterative convergence speed and stability based on the evolution of the system's real-time response error. The error trend refers to the comprehensive characterization of the amplitude change rate, fluctuation frequency, and convergence direction of the error correction signal over multiple consecutive test cycles. The correction step size factor is the proportional coefficient that controls the adjustment amplitude of each driving signal iteration, directly affecting the system's rate of approximation to the target spectrum and the overshoot.

[0083] The general process is described as follows: Signal input: The adaptive iterative regulator receives the error correction signal and caches it in a circular data queue, forming a time series containing error data for the current period and N historical periods. Trend determination: The regulator calls the sliding window difference algorithm to calculate the first derivative (rate of change) and second derivative (acceleration) of the error sequence. Combined with a preset convergence criterion, it determines whether the current system is in a fast convergence region, a critical stable region, or a divergent oscillation region. Step size dynamic update: An adaptive adjustment strategy is executed based on the determination result. If the error amplitude is monotonically decreasing and the rate of change is stable, the correction step size factor is amplified by a preset growth coefficient to accelerate iterative convergence. If the error exhibits alternating positive and negative oscillations or amplitude rebounds, the correction step size factor is reduced by a decay coefficient to suppress system overshoot and resonance; if the error remains within the allowable tolerance band, the step size factor remains constant. Output transmission: The updated correction step size factor is encapsulated as a control parameter and pushed to the next processing module to complete the real-time tuning of the step size factor.

[0084] Step S301: Generate a high-precision driving spectrum correction amount by adjusting the step size factor and the error correction signal.

[0085] Among them, the high-precision drive spectrum correction refers to the frequency domain or time domain correction data sequence used to directly compensate the vibration table drive waveform after step size factor weighting and spectrum shaping. It has the characteristics of high signal-to-noise ratio, no phase change and strong frequency band targeting.

[0086] The general process is as follows: First, the dynamic correction step size factor and the real-time error correction signal are multiplied point-by-point to obtain the basic correction vector, achieving proportional scaling of the error amplitude. Next, a Fast Fourier Transform (FFT) is performed on the basic correction vector to convert it to a frequency domain representation. Based on a preset frequency band weighting table, gain amplification is applied to the correction components of the target frequency band, such as the band containing the resonant peak of the seat structure, while attenuation filtering is applied to non-critical high-frequency noise bands, completing spectrum shaping. Then, an Inverse Fast Fourier Transform (IFFT) is called to restore the shaped frequency domain data to a time domain waveform, and a Hanning window smoothing process is applied to eliminate the Gibbs effect introduced by frequency domain truncation. Finally, amplitude limiting verification and phase alignment compensation are performed on the time domain waveform, outputting a continuous, abrupt, high-precision drive spectrum correction value for subsequent superposition modules.

[0087] Step S302: The high-precision drive spectrum correction is superimposed on the current drive signal to form a closed-loop control circuit, thereby eliminating lateral parasitic vibrations caused by dynamic response deviation.

[0088] Lateral parasitic vibration refers to the undesirable lateral and torsional vibration components caused by structural asymmetry, installation eccentricity, or control channel crosstalk under multidimensional coupled excitation. Closed-loop control refers to the real-time data flow and control execution path of "signal acquisition - error calculation - parameter identification - drive correction - physical excitation - response feedback".

[0089] The general process is described as follows: Vector superposition: The control unit reads the basic drive signal waveform of the current cycle and adds the high-precision drive spectrum correction amount to it on a sample-by-sample-point vector basis to generate a synthetic drive command that includes main excitation compensation and cross-axis cancellation components. Boundary constraint verification: The synthetic drive command is subjected to a physical limit protection algorithm to verify whether its amplitude, rate of change, and frequency components exceed the safe operating envelope of the drive components, such as servo motors or hydraulic actuators. If the limits are exceeded, peak clipping and amplitude limiting are performed according to a preset slope to prevent actuator saturation or mechanical overload. Command issuance and execution: The verified synthetic drive command is output to the execution end of the drive component after digital-to-analog conversion and power amplifier. The drive component adjusts the magnitude, phase, and spatial vector direction of the excitation force according to the corrected command to cancel the lateral parasitic vibration components generated by multi-degree-of-freedom coupling in real time. Loop closure: The adjusted physical excitation is applied to the seat under test. The new mechanical response and acceleration signals are captured in real time by the multi-dimensional sensing components and motion response components and input again into the dynamic feature extraction model to start the next round of error calculation and iterative correction. This forms a continuous, real-time closed-loop control circuit, ensuring a high degree of consistency between the actual loaded vibration spectrum and the target spectrum in multidimensional space.

[0090] Reference Figure 4 The specific steps for constructing a real-time force-displacement hysteresis loop based on a six-dimensional force / torque signal and a displacement signal obtained through integration include: Step S400: Periodically and synchronously sample the vertical force component and the corresponding vertical displacement signal in the six-dimensional force / torque signal to generate a discrete force-displacement data point set.

[0091] Among them, the vertical force component refers to the dynamic load time-series data along the vehicle's Z-axis direction collected by the multi-dimensional sensing components. The vertical displacement signal refers to the Z-axis displacement time-series data obtained by the acceleration signal collected by the motion response components after double integration and high-pass filtering. Periodic synchronous sampling refers to the synchronous interception and pairing of force and displacement signals at equal intervals based on the complete period of vibration excitation or a fixed time window.

[0092] The general process is described as follows: Period identification: The control system monitors the excitation frequency of the drive components in real time or identifies the start point of the vibration signal period through a zero-crossing detection algorithm, generating a synchronous trigger pulse. Data alignment: Using the trigger pulse as a reference, the instantaneous values ​​of the vertical force component and vertical displacement signal are read in parallel under the same sampling clock, eliminating the phase delay introduced by integral filtering. Point set encapsulation: The force and displacement values ​​at each sampling moment are combined in coordinate pairs. Data pairs are continuously acquired within a complete vibration cycle, arranged in chronological order to generate a discrete force-displacement data point set, and cached in the feature extraction buffer.

[0093] Step S401: Use a graphics algorithm to fit the data point set into a closed planar trajectory graph.

[0094] Among them, graphics algorithms refer to mathematical calculation methods used to process discrete geometric data and achieve curve smoothing and topological closure. This embodiment adopts a periodic B-spline curve fitting algorithm. A closed planar trajectory graph refers to a continuous nonlinear loop curve formed by connecting discrete data points in a force-displacement two-dimensional coordinate system, with the ends connected and no intersections.

[0095] The general process is as follows: Data sorting: Extract the data point set, rearrange it in ascending and descending order based on the phase change characteristics of the displacement signal, and distinguish the data sequences of the loading path and unloading path. Curve fitting: Call the periodic cubic B-spline interpolation algorithm, take the rearranged discrete points as control points as input, and generate a smooth parametric curve passing through the data points by solving the node vector and basis functions, suppressing trajectory spikes caused by sampling noise. Topological closure: Identify the starting and ending control points of the fitted curve, calculate the coordinate deviation between them, and apply periodic boundary constraints to force the curve to connect end to end. The final output is a continuous and closed planar trajectory graph in the force-displacement coordinate system.

[0096] Step S402: Calculate the enclosing area and tilt angle of the planar trajectory graphic.

[0097] The enclosed area refers to the size of the geometric region enclosed by the closed hysteresis loop, which is numerically equal to the energy consumed by the seat in one vibration cycle due to internal friction, material viscoelasticity, and slippage of the connecting parts. The tilt angle refers to the angle between the major axis of the hysteresis loop and the displacement abscissa axis, reflecting the equivalent stiffness ratio of the seat structure.

[0098] The general process is described as follows: Area calculation: Numerical integration is performed on the closed trajectory graph using Green's formula. All discrete sampling points are traversed counter-clockwise along the curve boundary, and the areas of the infinitesimal elements formed by adjacent points are accumulated. The absolute value of the enclosed area is obtained by summing these areas. Angle calculation: Principal component analysis is performed on the trajectory graph. The covariance matrix of all data points is calculated, and the eigenvector corresponding to the largest eigenvalue of the matrix is ​​found. This vector represents the major axis direction of the hysteresis loop. Angle calculation: The cosine of the angle between the major axis eigenvector and the unit vector of the displacement coordinate axis is calculated, and the tilt angle is obtained through inverse trigonometric functions. The calculated area and angle values ​​are output to the parameter parsing module.

[0099] Step S403: The enclosing area and tilt angle are used as geometric feature parameters to characterize the energy dissipation properties of the seat structure.

[0100] Among them, energy dissipation characteristics refer to the ability of the seat to convert mechanical vibration energy into heat energy through internal damping materials, frame connections, and vibration damping elements under alternating loads. Geometric characteristic parameters refer to standardized numerical indicators extracted from hysteresis loop morphology and used to quantitatively assess the dynamic performance and health status of the structure.

[0101] The general process is described as follows: Physical mapping: The system maps the enclosing area to a single-cycle damping energy dissipation index; an increase in area indicates increased internal friction or material yielding. The tilt angle is mapped to a dynamic stiffness attenuation index; a decrease in angle indicates loosening of connections or degradation of support stiffness. Parameter normalization: The real-time area and angle values ​​are divided by the new seat calibration benchmark value to calculate the relative rate of change, eliminating dimensional differences caused by different seat models and load levels. Feature output: The normalized area change rate and angle deviation value are encapsulated into a two-dimensional geometric feature parameter vector, and after adding a timestamp and operating condition identifier, it is pushed to the health assessment module for subsequent comparison with preset thresholds and failure warning determination.

[0102] Reference Figure 5 The preprocessed six-dimensional force / torque signal and acceleration response signal are input into a pre-constructed dynamic feature extraction model. Previous steps include: In step S500, the preprocessed signal is input into the digital twin simulation environment to drive the virtual seat model corresponding to the physical test bench to move synchronously.

[0103] The digital twin simulation environment refers to a high-fidelity virtual mapping platform built based on the geometry, boundary constraints, and dynamic parameters of a physical test bench, possessing real-time data-driven and physics engine solving capabilities. The virtual seat model refers to a parametric finite element model established in the simulation environment, including the skeleton, slide rails, connectors, and vibration damping components. Synchronous motion refers to inputting real-time excitation signals acquired from the physical system as boundary conditions into the virtual model, ensuring its motion state remains spatiotemporally consistent with the physical seat.

[0104] The general process is described as follows: Environment initialization: The 3D CAD model of the physical test bench is loaded into the simulation platform, and the assembly drawing of the seat to be tested is imported synchronously. Material properties such as elastic modulus, Poisson's ratio, and density are assigned to each component, and a parametric finite element mesh is generated. Signal mapping: The six-dimensional force / torque signals and acceleration response signals preprocessed in step S101 are transmitted in real time to the drive module of the digital twin environment through an industrial data interface protocol. Motion driving: The simulation engine converts the received force / torque signals into nodal loads on the virtual model's installation interface and integrates the acceleration signals to convert them into displacement boundary conditions. State synchronization: A real-time physical solver, such as an explicit dynamics solver, is invoked to update the displacement, velocity, and acceleration of each node of the virtual seat in millisecond-level steps, ensuring that the motion trajectory and attitude response of the virtual model are strictly synchronized with the actual excitation of the physical test bench, forming a basis for virtual-real mapping.

[0105] Step S501: In the virtual model, predict the stress distribution cloud map of the seat in the next testing phase based on the current dynamic stiffness matrix.

[0106] The current dynamic stiffness matrix refers to the set of frequency-varying parameters, identified in real-time in step S102, characterizing the coupled stiffness of each degree of freedom of the seat under the current test frequency and load conditions. The next test stage refers to the test condition with a preset load spectrum frequency band or increasing amplitude that will be executed after the current vibration cycle ends. The stress distribution cloud map refers to the graphic data that visualizes the spatial distribution of equivalent stress on the surface and internal nodes of the seat structure in the form of colored contour lines, based on the finite element calculation results.

[0107] The general process is described as follows: Parameter mapping: The system reads the dynamic stiffness matrix output in step S102 and converts it into the equivalent stiffness parameters of the corresponding connecting elements and spring dampers in the virtual model, updating the boundary stiffness properties of the finite element model. Loading: A preset test condition sequence is called, and the target excitation spectrum for the next test stage, such as the target frequency range and amplitude envelope, is extracted and applied to the installation interface nodes of the virtual model. Finite element solution: The updated stiffness matrix and target excitation spectrum are input into the transient dynamics solution module of the digital twin platform. The solver iteratively solves the nodal displacement field and strain field based on the updated stiffness matrix and load boundaries, and calculates the equivalent stress value of each mesh node according to the material constitutive relation. Contour rendering: The visualization rendering engine is called to map the calculated stress scalar data onto the three-dimensional geometric surface of the seat, and rainbow color interpolation is used to generate a stress distribution contour map containing the stress gradient distribution for subsequent safety assessment.

[0108] Step S502: Compare the predicted stress distribution cloud map with the preset safety envelope to generate the pre-simulation evaluation result.

[0109] The preset safety envelope refers to a three-dimensional spatial threshold surface or stress-frequency tolerance range that characterizes the maximum stress the structure can withstand, based on the yield strength, fatigue limit standard, and design safety factor of the seat material. The pre-simulation evaluation result refers to a structured evaluation report generated by comparing the virtual stress distribution with the safety threshold, which includes safety area markers, coordinates of out-of-limit nodes, and risk levels.

[0110] The general process is described as follows: Threshold loading: The system reads preset safety envelope data from the material database or design specifications, including the allowable stress upper limit and fatigue damage accumulation threshold of key components such as slide rails, backrest frames, and seat crossbeams. Spatial comparison: The stress distribution cloud map grid node data is matched node by node with the safety envelope. A spatial coordinate indexing algorithm is used to locate the three-dimensional position of each node in the cloud map and its corresponding calculated stress value. Exceedance judgment: If the calculated stress value of a node is less than or equal to the safety envelope threshold corresponding to that position, it is marked as a safe state. If it is greater than the threshold, it is marked as an exceeded state, and the coordinates of the node, the stress exceeding ratio, and the component number to which it belongs are recorded. Result generation: The comparison status of all nodes is summarized, the proportion of exceeded nodes and the maximum exceeded amplitude are calculated, a pre-assessment result including "safe / warning / dangerous" risk levels is generated, and structured data is output to the decision control module.

[0111] Step S503: If the pre-test evaluation results show that there are predicted stress points that exceed the safety envelope, an optimized test strategy containing path correction instructions is generated in advance before the physical test begins.

[0112] Among them, predicted stress points exceeding the safety envelope refer to the set of mesh nodes determined in the virtual simulation to have calculated stress values ​​exceeding the material's allowable limit or fatigue safety threshold. Path correction instructions refer to the sequence of control parameters used to adjust the loading path of the vibration table excitation spectrum, including amplitude attenuation ratio, frequency band skipping instructions, shortening dwell time, or reducing step slope. Optimized testing strategies refer to the set of comprehensive loading scheme adjustments that avoid structural damage risks while ensuring testing effectiveness.

[0113] The general process is described as follows: **Cause Analysis:** When the pre-test evaluation results indicate the presence of stress points exceeding limits, the system locates the spatial position of these stress points and analyzes the causes of stress concentration, such as excessive excitation in the resonant frequency band, sudden changes in local stiffness, or load eccentricity. **Constraint Solving:** The system calls a constraint optimization algorithm, using "not exceeding the safety envelope" as a hard constraint and "maintaining the test target spectrum coverage" as the optimization objective, to deduce the feasible domain of the vibration table driving parameters. **Instruction Generation:** The system automatically generates path correction instructions. For example, in sensitive frequency bands that induce stress concentration, the excitation amplitude is attenuated by a preset ratio; a gentle transition slope is inserted to reduce the impact load rate; or the phase difference of multi-axis excitation is adjusted to disperse local coupling stress. **Strategy Encapsulation and Distribution:** The optimized test strategy containing the correction instructions is encapsulated into a standard control file and distributed to the closed-loop control unit. When the physical test bench executes this strategy, it will run according to the corrected loading path, thereby proactively avoiding structural damage risks before the physical test begins, achieving safe and efficient adaptive testing.

[0114] Reference Figure 6 The specific steps for time-stamp alignment and preprocessing of the six-dimensional force / torque signal and acceleration response signal include: Step S600: The six-dimensional force / torque signal is separated into components using a multi-dimensional decoupling algorithm to extract the pure vertical excitation component and the coupling interference component.

[0115] The multidimensional decoupling algorithm refers to a mathematical mapping model based on the sensor sensitivity matrix and cross-coupling coefficient, used to eliminate channel crosstalk in the acquisition process of multidimensional force / torque signals. The pure vertical excitation component refers to the effective excitation force data acting along the seat's vertical direction (Z-axis) and independent of interference from other degrees of freedom. The coupled interference component refers to the lateral force and rotational torque interference data caused by structural asymmetry, off-center installation, or actuator nonlinearity.

[0116] The general process is described as follows: Calibration loading: A pre-stored six-dimensional sensor calibration matrix is ​​invoked. This matrix, obtained through calibration with standard weights and a torque stage, characterizes the cross-sensitivity coefficients between channels. Matrix inversion: The real-time acquired raw six-dimensional signal vector is multiplied on the left by the inverse of the calibration matrix, performing a linear decoupling transformation. Component stripping: Based on the transformed data sequence, the pure vertical excitation component data stream is extracted according to the Z-axis index. Simultaneously, the X / Y axis force and Mx / My / Mz torque data are extracted and merged into a coupled interference component data stream. Buffer alignment: The separated two component data streams are allocated to independent buffers according to their original timestamps to ensure timing synchronization of subsequent feedforward compensation and coherence analysis.

[0117] Step S601: The coupled interference component is input as a feedforward quantity to the active cancellation module to generate an anti-phase compensation signal.

[0118] The active cancellation module refers to the feedforward compensation unit built into the control algorithm, used to generate control commands in real time that are equal in amplitude but opposite in phase to the disturbance force / torque. The anti-phase compensation signal refers to the drive compensation data sequence used to cancel lateral parasitic vibrations after phase reversal and system delay compensation.

[0119] The general process is described as follows: Signal input: The active cancellation module reads the coupled interference component data stream and uses it as a feedforward reference signal input to the digital signal processing unit. Phase reversal and delay compensation: The inverse model of the actuator transfer function obtained by the system identification is called to perform phase lead compensation on the feedforward signal to cancel the hardware response delay. Then, a 180-degree phase reversal operation is performed to generate a reference inverted waveform. Adaptive gain adjustment: Combined with the least mean square algorithm, the residual interference error is monitored in real time, and the amplitude gain coefficient of the compensation signal is dynamically adjusted to avoid overcompensation causing system oscillation. Signal encapsulation: The adjusted inverted waveform is encoded according to the driving protocol format and output as an inverted phase compensation signal, which is pushed to the signal superposition module.

[0120] Step S602: The anti-phase compensation signal is vector-superimposed with the original drive signal to output the corrected drive command, so as to eliminate the lateral parasitic vibration caused by multi-axis coupling or installation eccentricity.

[0121] Vector superposition refers to the algebraic addition of the main driving waveform and the compensation waveform according to their spatial degrees of freedom in the time or frequency domain, thereby achieving the synthesis of force / torque vectors. Lateral parasitic vibration refers to the additional vibration component excited in the undesired direction of the test bench due to multi-axis coupled excitation, which manifests as lateral swaying or torsion of the seat.

[0122] The general process is described as follows: Time-domain alignment: The control system aligns the anti-phase compensation signal with the original drive signal of the current cycle at a unified sampling clock within microseconds to ensure precise phase matching. Vector synthesis: In the multi-dimensional drive vector space, the components of each axis of the compensation signal are added point-by-point to the corresponding components of the original drive signal to generate a composite waveform containing the main excitation and cancellation components. Boundary limiting: Servo safety envelope verification is performed on the composite waveform. If the amplitude or rate of change exceeds the physical limit of the driver, clipping is performed according to a preset slope to prevent actuator saturation. Command output: The verified composite waveform is converted into control commands and sent to the servo power amplifier to drive the actuator to generate a cancellation torque, suppressing the lateral parasitic vibration of the test bench in real time.

[0123] Step S603: Perform coherence analysis on the separated pure vertical excitation component and acceleration response signal, and remove abnormal data segments with coherence below a preset threshold.

[0124] Coherence analysis refers to a frequency domain statistical method that assesses the degree of linear correlation between input excitation and output response by calculating the ratio of their cross-power spectrum to their self-power spectrum. The preset threshold is a lower limit of the coherence coefficient set based on the sensor noise floor and the system signal-to-noise ratio, used to determine data validity. Abnormal data segments refer to invalid sampling intervals where the linear relationship between input and output is broken due to nonlinear distortion, sensor interruption, or strong external interference.

[0125] The general process is described as follows: Spectrum estimation: A Hanning window is applied to the pure vertical excitation component and the synchronously acquired acceleration response signal, and a Fast Fourier Transform is performed to calculate the input power spectrum, output power spectrum, and cross-power spectrum, respectively. Coherence coefficient calculation: The amplitude squared coherence function formula is used to calculate the coherence coefficient sequence point by point. Threshold determination: The calculated coherence coefficients are compared with a preset threshold within an interval. If the coherence coefficient within a certain time window is consistently lower than the threshold, it is determined that there is signal distortion or strong interference during that period. Data removal and marking: Data segments determined to be abnormal are removed from the effective analysis queue and marked as null values ​​on the time axis. Then, linear interpolation of the preceding and following effective data is used to fill the gaps, ensuring that the input data for the subsequent dynamic feature extraction model has a high signal-to-noise ratio and physical consistency.

[0126] Reference Figure 7 Based on the same inventive concept, embodiments of this application provide a multi-dimensional force feedback-based automotive seat vibration testing system, comprising: The acquisition module is used to acquire the six-dimensional force / torque signal at the bottom of the seat and simultaneously acquire the acceleration response signal of the seat frame; The memory is used to store the program for the automobile seat vibration testing method based on multidimensional force feedback as described above; The processor and memory can load and execute the program to implement the multi-dimensional force feedback-based automotive seat vibration testing method described above.

[0127] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0128] This application provides a computer-readable storage medium storing a computer program that can be loaded and executed by a processor to perform a method for testing the vibration of an automobile seat based on multidimensional force feedback.

[0129] Computer storage media include, for example, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media that can store program code.

[0130] Based on the same inventive concept, embodiments of this application provide a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor to perform a method for testing the vibration of a car seat based on multidimensional force feedback.

[0131] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0132] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A vibration testing device for automobile seats based on multidimensional force feedback, characterized in that, include: The platform assembly for supporting the seat under test includes a base, a column and a work surface, wherein mounting holes are arranged in an array on the work surface; A drive assembly for generating vibration is mounted within the base to transmit vibration excitation to the worktable surface. A multi-dimensional sensing component for acquiring six-dimensional force / torque signals of the seat under test in real time, the multi-dimensional sensing component being arranged between the workbench and the mounting interface of the seat under test; A motion response component for acquiring vibration acceleration response signals of a seat under test, the motion response component being arranged in key structural parts of the seat under test; A closed-loop control unit is used to adjust the vibration output of the drive component in real time based on the acquired six-dimensional force / torque signal and acceleration signal. The closed-loop control unit is connected to the drive component, the multi-dimensional sensing component and the motion response component to form a closed-loop feedback control.

2. A method for testing automotive seat vibration based on multidimensional force feedback, applicable to the automotive seat vibration testing device based on multidimensional force feedback as described in claim 1, characterized in that, include: Real-time acquisition of six-dimensional force / torque signals from the bottom of the seat, and simultaneous acquisition of acceleration response signals from the seat frame; The six-dimensional force / torque signal and the acceleration response signal are time-stamped and pre-processed. The preprocessed six-dimensional force / torque signal and acceleration response signal are input into the pre-constructed dynamic feature extraction model to calculate the dynamic stiffness matrix and equivalent damping coefficient characterizing the current state of the seat. The calculation results are used as feedback quantities. The feedback quantity is compared with the preset target vibration spectrum to perform error analysis and generate an error correction signal. Based on the error correction signal, the inverse system control strategy is invoked to generate a drive correction command, which is then output to the drive component to adjust the output waveform of the vibration table. During the test cycle, a real-time force-displacement hysteresis loop is constructed based on the six-dimensional force / torque signal and the displacement signal obtained by integration, and the geometric feature parameters of the hysteresis loop are extracted. The geometric feature parameters are compared with the preset health threshold. If the geometric characteristic parameters are determined to exceed the threshold range, a structural failure warning signal is generated and the drive components are controlled to adjust the output power or shut down.

3. The method for testing automotive seat vibration based on multidimensional force feedback according to claim 2, characterized in that, The specific steps for inputting the preprocessed six-dimensional force / torque signal and acceleration response signal into the pre-constructed dynamic feature extraction model to calculate the dynamic stiffness matrix and equivalent damping coefficient characterizing the current state of the seat include: The six-dimensional force / torque signal is decomposed into vertical and lateral components; A multidimensional input-output relationship is constructed by combining the displacement components obtained from the integral of the acceleration response signal. Based on the multidimensional input-output relationship, the complex frequency response function between the input force and the output displacement is calculated using a system identification algorithm. Based on the characteristics of the real and imaginary parts of the complex frequency response function, the dynamic stiffness matrix and equivalent damping coefficient of the seat in the vertical, lateral and rotational degrees of freedom are separated and output.

4. The method for testing automotive seat vibration based on multidimensional force feedback according to claim 2, characterized in that, The step after performing error analysis between the feedback quantity and the preset target vibration spectrum to generate an error correction signal further includes: The error correction signal is input to the adaptive iterative regulator, which dynamically adjusts the correction step size factor according to the current error trend. A high-precision driving spectrum correction is generated by adjusting the step size factor and the error correction signal. The high-precision drive spectrum correction is superimposed on the current drive signal to form a closed-loop control circuit, eliminating lateral parasitic vibrations caused by dynamic response deviations.

5. The method for testing automotive seat vibration based on multidimensional force feedback according to claim 2, characterized in that, The specific steps for constructing a real-time force-displacement hysteresis loop based on a six-dimensional force / torque signal and a displacement signal obtained through integration include: Periodically and synchronously sample the vertical force component and the corresponding vertical displacement signal in the six-dimensional force / torque signal to generate a discrete force-displacement data point set; Using computer graphics algorithms, a set of data points is fitted into a closed planar trajectory. Calculate the enclosed area and tilt angle of the planar trajectory graph; The enclosing area and tilt angle are used as geometric characteristic parameters to characterize the energy dissipation properties of the seat structure.

6. The method for testing automotive seat vibration based on multidimensional force feedback according to claim 2, characterized in that, The preprocessed six-dimensional force / torque signal and acceleration response signal are input into the pre-constructed dynamic feature extraction model. The preceding steps also include: The preprocessed signal is input into the digital twin simulation environment to drive the virtual seat model corresponding to the physical test bench to move synchronously. In the virtual model, the stress distribution cloud map of the seat in the next testing phase is predicted based on the current dynamic stiffness matrix; The predicted stress distribution cloud map is compared with the preset safety envelope to generate the pre-simulation evaluation results; If the pre-test evaluation results show that there are predicted stress points that exceed the safety envelope, an optimized test strategy containing path correction instructions will be generated in advance before the physical test begins.

7. The method for testing automotive seat vibration based on multidimensional force feedback according to claim 2, characterized in that, The specific steps for timestamp alignment and preprocessing of the six-dimensional force / torque signal and acceleration response signal include: A multidimensional decoupling algorithm is used to separate the components of a six-dimensional force / torque signal, extracting the pure vertical excitation component and the coupling interference component. The coupled interference component is input as a feedforward quantity to the active cancellation module to generate an anti-phase compensation signal; The anti-phase compensation signal is vector-superimposed with the original drive signal to output the corrected drive command, thereby eliminating lateral parasitic vibration caused by multi-axis coupling or installation eccentricity. The separated pure vertical excitation components and acceleration response signals are subjected to coherence analysis, and abnormal data segments with coherence below a preset threshold are removed.

8. A vibration testing system for automotive seats based on multidimensional force feedback, characterized in that, include: The acquisition module is used to acquire the six-dimensional force / torque signal at the bottom of the seat and simultaneously acquire the acceleration response signal of the seat frame; A memory for storing the program of the automobile seat vibration test method based on multidimensional force feedback as described in any one of claims 1 to 7; The processor and the program in the memory are capable of being loaded and executed by the processor and implementing the multidimensional force feedback-based automotive seat vibration testing method as described in any one of claims 1 to 7.

9. A smart terminal, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and executed according to any one of claims 1 to 7.

Images