A performance detection system for a head multi-claw massager

Abstract

The application relates to the technical field of automatic testing and performance detection, and particularly discloses a head multi-claw massager performance detection system, which aims to solve the problem that the prior art cannot automatically, highly accurately and integrally comprehensively detect a multi-claw massager. The system comprises an integrated detection platform, a multi-modal sensing array, a high-speed data acquisition and processing unit and a comprehensive performance evaluation unit. A bionic head model carries the massager to be detected, and simultaneously collects mechanical, kinematic and acoustic vibration signals during operation of the massager, and through multi-source information fusion and correlation analysis, quantitative evaluation and comprehensive diagnosis are realized on the single-claw performance, multi-claw collaboration and internal mechanical state of the massager.

Description

Technical Field

[0001] This invention belongs to the field of automated testing and performance detection technology, specifically relating to a performance testing system for a multi-claw head massager. Background Technology

[0002] As an increasingly popular personal health care product, the performance of head massagers directly affects user comfort and potential health benefits. Multi-claw head massagers, in particular, can simulate more complex massage techniques and cover a wider area of ​​the head and acupoints. Therefore, accurate and efficient testing of their overall performance is crucial for ensuring product quality, guiding product design optimization, and enhancing market competitiveness.

[0003] Currently, performance testing methods for such massagers still have significant limitations. Mainstream testing methods largely rely on subjective sensory evaluation by testing personnel. This method is not only inefficient and lacks repeatability, but also fails to objectively quantify key dynamic parameters, making it difficult to meet the stringent requirements of consistency and efficiency in modern mass production. Another common improvement method is to use a single type of sensor (such as a single-point pressure sensor) for measurement. However, this method can only acquire force information of the massage claws at static or local points, and cannot comprehensively and synchronously capture the three-dimensional motion trajectory, collaborative working characteristics, and time-varying stability of multiple massage claws during dynamic operation. More importantly, existing technologies generally lack the ability to synchronously collect and deeply analyze the mechanical vibration and operating noise signals generated during the operation of the massager. These two types of signals are important carriers for indirectly assessing the internal mechanical health conditions such as motor performance, gear transmission status, assembly accuracy, and structural wear. Therefore, existing technologies struggle to achieve integrated, automated, and high-precision comprehensive testing and in-depth diagnosis of the force, motion, sound, and vibration multi-dimensional dynamic performance of multi-claw head massagers.

[0004] Therefore, there is an urgent need to propose an innovative technical solution to address the problems of fragmentation, subjectivity, and inefficiency in existing testing methods, thereby providing a reliable technical means for the research and development, verification, production quality inspection, and performance improvement of multi-claw head massagers. Summary of the Invention

[0005] The purpose of this invention is to provide a performance testing system for a multi-claw head massager, so as to solve the technical contradiction in the prior art that relies on manual labor or single-point sensors and cannot perform automated, high-precision, and integrated comprehensive testing of the multi-dimensional dynamic performance of multi-claw massagers.

[0006] The technical solution of the present invention is a performance testing system for a multi-claw head massager, comprising: An integrated testing platform includes a rigid base, a bionic head model set on the rigid base, and a multi-dimensional motion bearing mechanism set between the rigid base and the bionic head model. The multi-dimensional motion bearing mechanism is used to drive the bionic head model to perform multi-degree-of-freedom precise positioning. A multimodal sensing array is arranged on and around a bionic head model to synchronously acquire multidimensional dynamic response signals of the massager under test during operation. The multimodal sensing array includes at least a three-dimensional force sensing network for measuring three-dimensional forces, a high-speed visual tracking module for capturing motion trajectories, and an acoustic-vibration synchronous acquisition module for acquiring noise and vibration signals. A high-speed data acquisition and processing unit is connected to a multimodal sensor array signal to receive and synchronously process multidimensional dynamic response signals. The high-speed data acquisition and processing unit includes at least a synchronous trigger circuit, a multi-channel analog-to-digital converter, and a preprocessing algorithm module. The comprehensive performance evaluation unit is connected to the high-speed data acquisition and processing unit and is used to perform performance quantification and comprehensive correlation diagnosis based on the processed signal. The comprehensive performance evaluation unit includes at least a performance index quantification model and a multi-dimensional correlation analysis engine.

[0007] Preferably, the bionic head model has a layered structure of gradient composite material, comprising, from the inside out: The rigid support liner, made of engineering plastic or lightweight metal, serves to provide a mounting base and a stable spatial reference for the internal sensor array. The damping buffer intermediate layer, which covers the rigid support inner liner, is made of a polymer composite material with specific viscoelasticity. Its damping loss factor is controlled between 0.1 and 0.3, and it is used to simulate the buffering and damping characteristics of head tissue. The elastic biomimetic outer skin layer, which covers the damping buffer intermediate layer, is made of medical-grade silicone material with a Shore hardness of A10 to A20, and is used to simulate the tactile and frictional characteristics of scalp skin.

[0008] Preferably, the three-dimensional force sensing network consists of multiple miniature three-dimensional force sensors arranged according to a preset grid node layout. The sensors are installed through standardized threaded interfaces pre-embedded in the bionic head model, and each sensor can simultaneously measure normal force and tangential force.

[0009] Preferably, the high-speed vision tracking module includes at least two jointly calibrated high frame rate industrial cameras. The cameras start acquisition by receiving the same trigger signal from the synchronous trigger circuit, which is used to obtain the three-dimensional motion trajectory of the end of the massage claw based on a binocular stereo vision reconstruction algorithm.

[0010] Preferably, the acoustic and vibration synchronous acquisition module includes a high-sensitivity microphone array arranged in the space surrounding the bionic head model and multiple triaxial accelerometers fixed inside key mechanical nodes of the bionic head model, for synchronously acquiring noise spectrum and vibration signals.

[0011] Preferably, the synchronization trigger circuit is driven by a high-stability temperature-compensated quartz crystal oscillator to generate and distribute synchronization pulse signals to each acquisition device of the multi-mode sensing array to achieve microsecond-level time synchronization.

[0012] Preferably, the performance index quantification model is configured to calculate the single-claw performance index and the system collaborative performance index; Among them, the performance indicators of a single claw include at least static pressure uniformity, dynamic trajectory tracking error, force control accuracy, and force control bandwidth; The system's collaborative performance indicators include at least the multi-claw synchronization error, the consistency score of the collaborative force application mode, and the overall operational stability index.

[0013] Preferably, the multi-dimensional correlation analysis engine is configured to perform the following correlation analyses in sequence: Mechanical and kinematic correlation analysis is used to spatiotemporally align the three-dimensional motion trajectory of the massage claw with three-dimensional force data, and analyze the matching degree between the force direction and the motion direction; The correlation analysis between acoustic and vibration signals and mechanical status is used to determine the health status of internal mechanical components of a massager based on a pre-established fault knowledge base and diagnostic model, according to the characteristics of noise and vibration spectrum. Multi-claw synergistic effect correlation analysis is used to quantify the coupling relationship between claws by analyzing the cross-correlation of force signals, motion phase difference and vibration coherence between any two massage claws.

[0014] Preferably, it also includes an automated test process control module, which is configured as follows: Pre-store and execute standard test programs for different massage modes. The programs define the expected motion trajectory parameters and target force curve of the massage claw. When the test starts, a start command is sent to the massager under test and a hardware trigger signal is sent to the high-speed data acquisition and processing unit at the same time; During the detection process, the trajectory and force data of the massage claw are received in real time and compared with the expected value. Based on whether the deviation exceeds the preset tolerance threshold, the corresponding anomaly handling strategy is executed.

[0015] Preferably, the comprehensive performance evaluation unit is also configured to automatically generate a structured test report based on the analysis results after the test process is completed, and output the report as a specified format file and / or push the key judgment results to an external production management system in the form of structured data.

[0016] In summary, this application includes at least one of the following beneficial technical effects: 1. This invention, by constructing an integrated detection platform combining a bionic head model and a multimodal sensor array, achieves for the first time synchronous, in-situ measurement of the multidimensional dynamic performance of a multi-claw head massager, including force, trajectory, sound, and vibration. The combination of a three-dimensional force sensing network and a high-speed vision system overcomes the limitations of single-point measurement, enabling precise quantification of the force characteristics and motion accuracy of each massage claw in three-dimensional space, providing a reliable data foundation for evaluating complex collaborative massage techniques.

[0017] 2. This invention creatively introduces a multi-dimensional correlation analysis engine, elevating traditional isolated parameter detection to a system-level correlation diagnosis level. Through multi-layered analysis, including correlations between mechanics and kinematics, acoustic vibration and mechanical state, and multi-claw synergistic effects, this engine profoundly reveals the underlying mechanisms behind performance phenomena. For example, it diagnoses internal wear through vibration spectrum analysis and evaluates the quality of control algorithms through synchronization analysis. This in-depth analytical capability makes this system not only a detection tool but also a powerful diagnostic tool for product development and quality improvement.

[0018] 3. This invention ensures high repeatability of the testing process and comparability of results through a standardized biomimetic head model, automated testing process control, and a unified spatiotemporal data synchronization mechanism. This completely eliminates the subjectivity and random errors of manual testing, meeting the stringent requirements of modern production lines for testing efficiency, accuracy, and consistency, and providing key technical support for the intelligent, standardized production and quality control of massager products. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall technical solution architecture of the head multi-claw massager performance testing system proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of the multi-dimensional correlation analysis engine in this invention; Figure 3 This is a logical flow diagram of the automated testing process control and multimodal data acquisition in this invention. Detailed Implementation

[0020] This invention provides a performance testing system for a multi-claw head massager. Its purpose is to construct a technical platform capable of high-precision, synchronized, and integrated comprehensive evaluation of the mechanical, kinematic, and acoustic characteristics of a multi-claw massager applied to the human head region under dynamic operating conditions. Through a highly integrated hardware architecture and intelligent data processing logic, this system achieves a closed-loop testing process from raw sensor signal acquisition to multi-dimensional performance index output.

[0021] The overall technical solution of the head multi-claw massager performance testing system proposed in this invention consists of four core components: an integrated testing platform, a multimodal sensor array, a high-speed data acquisition and processing unit, and a comprehensive performance evaluation unit. These four parts work together in strict physical connection and logical dependency relationship to jointly complete the comprehensive performance testing task of the head multi-claw massager under test.

[0022] The integrated testing platform forms the basic supporting structure of this system. Its main function is to provide a standardized, repeatable, and highly realistic testing environment for the multi-claw head massager under test. This platform mainly consists of the following three components: (1) Rigid base: Made of high-strength aluminum alloy or cast iron material, it has high structural rigidity and thermal stability. It is used to support and fix all other components in the system, ensuring that the measurement reference will not be shifted due to external vibration or temperature drift during long-term continuous operation.

[0023] (2) Bionic head model: It is installed on a rigid base and its geometric shape is strictly constructed according to the international standard human head three-dimensional scanning database, covering key head areas such as the forehead, temples, top, and occipital region.

[0024] Below these areas, a standardized sensor interface matrix is ​​pre-embedded in a grid pattern inside the model. Each interface uses an M3 or M4 threaded hole with a depth of 6-8 mm. The hole axis is perpendicular to the model surface and is distributed at equal intervals of 20 mm × 20 mm.

[0025] The interface matrix is ​​surrounded by guide slots and numbered markers to facilitate the rapid positioning and installation of modules such as three-dimensional force sensors and accelerometers, ensuring a consistent sensor array layout across different tests.

[0026] (3) Multidimensional motion bearing mechanism: set between the rigid base and the bionic head model, the mechanism contains three precision electric displacement stages arranged orthogonally to each other, corresponding to the translational degrees of freedom in the X, Y and Z axis directions respectively.

[0027] Each displacement stage is equipped with a closed-loop feedback system with a grating ruler, achieving a positioning resolution of 1 micrometer and a repeatability accuracy better than ±2 micrometers.

[0028] The three displacement stages work in coordination through a unified motion controller (such as one based on the EtherCAT bus). They can automatically perform multi-axis interpolation motion according to the spatial coordinate parameters sent by the host computer, so that the bionic head model is accurately aligned with the initial point of action of the massager before testing.

[0029] After each startup or sensor replacement, the system performs an automatic calibration process: first, it drives the displacement stage back to the mechanical zero point, and then, based on the reference target ball fixed on the base, it performs spatial position calibration through a laser tracker or vision system to ensure the long-term stability of the measurement benchmark.

[0030] A multimodal sensor array is distributed within and on the surface of the bionic head model to synchronously acquire multidimensional dynamic response signals of the massage claw during operation. This array consists of three subsystems: (1) Three-dimensional force sensing network: It consists of multiple miniature three-dimensional force sensors. These sensors are arranged according to a preset grid node, with the node spacing usually designed to be 15 to 25 mm, and preferentially covering the projection area of ​​key massage acupoints such as Baihui, Taiyang, and Fengchi on the bionic head model.

[0031] Below each node, a matching standard threaded interface (such as M3 specification) is pre-embedded inside the model. The sensor is directly screwed in and fixed through a threaded mounting post, ensuring installation rigidity and repeatable position.

[0032] Each sensor can independently and synchronously measure the normal force (Z-axis direction) applied by the massage claw and two mutually perpendicular tangential forces (X-axis and Y-axis directions).

[0033] For the dynamic force characteristics of the massage claw, the sensor selection should ensure that its force range covers 0 to 50 Newtons, the sampling frequency is not less than 10 kHz (to capture possible transient impacts), and the static nonlinearity error is less than 0.5% of the full scale.

[0034] (2) High-speed visual tracking module: It consists of at least two high frame rate industrial cameras, which are mounted on rigid supports on both sides of the model.

[0035] The camera's mounting angle and focal length are pre-calibrated to ensure that its field of view overlap area completely covers the model surface and the possible range of motion of the massage claw.

[0036] Considering that the movement speed of the massage claw tip may be relatively fast, the camera frame rate should be selected to be no less than 2000 frames per second, with a pixel resolution of at least 1280×1024, so that the motion trajectory can be clearly captured without motion blur in most working conditions.

[0037] Each camera receives the same TTL trigger signal from the system's synchronization trigger circuit via its image acquisition card, ensuring that image acquisition and force / sound vibration signals are started synchronously with microsecond-level precision.

[0038] (3) Acoustic and vibration synchronous acquisition module: This module is used to synchronously acquire the noise and vibration signals of the massager during operation.

[0039] High-sensitivity microphone array: It consists of 8 directional condenser microphones, which are arranged in a spherical space with a radius of about 0.5 meters centered on the bionic head model, and are roughly evenly distributed in the horizontal and vertical directions.

[0040] This arrangement facilitates the approximate location and separation of noise sources (such as different motors or gearboxes) through subsequent acoustic array processing technology.

[0041] Each microphone's frequency response needs to cover the range of 20 Hz to 20 kHz that is audible to the human ear.

[0042] Multiple triaxial accelerometers: At least six in number, fixed to specific locations within the rigid support liner of the bionic head model via miniature magnetic bases or adhesives. These locations are typically chosen at points corresponding to skeletal structures in the frontal, parietal, and occipital bones, as these points are key mechanical nodes in the vibration transmission path and can effectively reflect the characteristics of the massager's vibration transmitted through the model's structure. The typical sensitivity of the accelerometers is 100 mV / g, and the upper limit of the frequency response should be no less than 5 kHz to cover mid-to-high frequency vibration components that may be caused by mechanical parts such as motors and gears.

[0043] The high-speed data acquisition and processing unit acts as the core data hub of the entire system. Its main function is to receive all the raw signals from the multimodal sensor array and perform high-precision time synchronization, digital conversion, and preliminary feature extraction on these signals.

[0044] This unit consists of three core parts: a synchronous triggering circuit, a multi-channel analog-to-digital converter, and a preprocessing algorithm module. Its specific operation is as follows: (1) Synchronous trigger circuit: As the core of the time reference of the entire acquisition system, this circuit is driven by a high-stability temperature compensated quartz crystal oscillator (TCXO) to generate a clock signal with a frequency stability better than ±1ppm.

[0045] This master clock signal is passed through a clock buffer / distribution chip with low jitter characteristics to generate multiple synchronous pulses of the same origin (such as TTL level).

[0046] These pulses are transmitted via coaxial cables of equal length or dedicated synchronous buses (such as daisy chain or star topology) to the data acquisition units of each sensor in the three-dimensional force sensing network, the image acquisition card of the high-speed visual tracking module, and the multi-channel acquisition device of the acoustic vibration synchronous acquisition module.

[0047] To keep the time jitter of the entire system within ±1 microsecond, the circuit design needs to precisely match the delay of each transmission path, and enable the hardware trigger mode at the receiving end (especially the camera acquisition card) to ensure that all channels start sampling simultaneously within a very short time after receiving the synchronization pulse (usually within one master clock cycle).

[0048] For individual sensors that do not support hardware triggering, their acquisition devices need to have a high-precision timestamp recording function so that microsecond-level time alignment can be performed in post-processing.

[0049] (2) Multi-channel analog-to-digital converter: It is constructed using one or more high-speed analog-to-digital converter (ADC) boards with a resolution of not less than 16 bits.

[0050] The total number of input channels on the board needs to be determined based on the actual number of sensors deployed, with a margin of safety. For example, if a 3D force sensor outputs 3 signals (Fx, Fy, Fz), a triaxial accelerometer outputs 3 signals, and a microphone outputs 1 signal, then when the system includes 20 force sensors, 6 accelerometers, and 8 microphones, the total number of input channels required is ( Road simulation input channel.

[0051] Therefore, an ADC system should support at least 90-100 channels to meet typical configuration requirements. These channels are usually grouped, and each group can be configured with its own sampling rate. For example, force signals and vibration signals require higher sampling rates (e.g., above 10kHz), while some sound analyses may not have stringent requirements for ultra-high sampling rates, allowing for more flexible allocation of system resources.

[0052] The board's maximum sampling rate is no less than 1 megasamples per second (MS / s) to ensure that all channels can be digitized with high fidelity when needed.

[0053] (3) Preprocessing algorithm module: This module runs on an embedded high-performance processor (such as FPGA or multi-core DSP) and contains a series of real-time processing programs for different data streams.

[0054] For three-dimensional force data: the program first preprocesses the raw force signal.

[0055] The window length of the moving average filter is usually set according to the highest operating frequency of the massage claw. For example, if the highest operating frequency is 50Hz, the window time can be set to 1 / 5 to 1 / 10 of the corresponding period (i.e., 4 to 2 milliseconds) to effectively smooth out dynamic changes in high-frequency noise without significantly affecting them.

[0056] For specific frequency band mechanical resonance interference that may exist in the signal and is unrelated to the massage action, wavelet transform can be used for targeted filtering; at the same time, the algorithm will continuously monitor the output of the sensor under no-load conditions and perform real-time zero-point drift compensation.

[0057] For high-speed visual image sequences: the program calls a binocular stereo vision reconstruction algorithm. This algorithm first extracts features from each pair of synchronized images (for example, using SIFT or ORB feature point detection algorithms to locate specific marker points at the end of the massage claw), and then calculates the coordinates of the marker points in three-dimensional space based on the pre-calibrated camera intrinsic and extrinsic parameters using the principle of triangulation.

[0058] By differentiating the coordinates of consecutive frames, the motion velocity and acceleration can be obtained.

[0059] The accuracy of trajectory repetition is calculated by comparing the standard deviations between the sets of three-dimensional trajectory points obtained by repeating the same action multiple times.

[0060] For acoustic and vibration data: the program processes sound and vibration signals separately.

[0061] The Fast Fourier Transform (FFT) with windowing (such as Hanning window) is mainly used to convert the time domain signal to the frequency domain.

[0062] The analysis frequency bands are preset according to the characteristics of the massager. For example, vibration analysis focuses on the 0-2000Hz range to capture the main characteristic frequencies of the motor and gears; noise analysis may focus on a wider frequency band or specific frequencies that are sensitive to the human ear.

[0063] In the processing, in addition to calculating the total sound pressure level and the root mean square value of vibration acceleration (RMS), the peak detection algorithm will automatically identify and record the amplitude of motor fundamental frequency (e.g., 100Hz), gear meshing frequency (e.g., motor fundamental frequency multiplied by the number of teeth) and their harmonic components in the spectrum. These are the key input features for subsequent mechanical state correlation analysis.

[0064] The comprehensive performance evaluation unit establishes a high-speed communication connection with the high-speed data acquisition and processing unit through gigabit Ethernet or PCIe bus, receives various pre-processed feature data, and performs in-depth performance quantification and comprehensive correlation diagnosis based on this data.

[0065] This unit mainly contains two functional modules: a performance index quantification model and a multi-dimensional correlation analysis engine.

[0066] The performance metric quantification model, based on pre-defined evaluation criteria, transforms the input multidimensional feature data into a series of quantifiable and comparable performance metrics. These metrics are mainly divided into two categories: (1) Single claw performance indicators: used to evaluate the basic working performance of a single massage claw, mainly including: Static pressure uniformity: This is characterized by calculating the standard deviation of the normal force measured at multiple preset contact points (corresponding to different sensors) of the same massage claw. The smaller the standard deviation, the more uniform the pressure distribution.

[0067] Dynamic trajectory tracking error: defined as the root mean square (RMSE) value of the Euclidean distance between corresponding points on the actual motion trajectory of the massage claw tip and the preset theoretical trajectory. The "preset theoretical trajectory" comes from the standard path coordinate point sequence defined for the current test program in the automated test process control module. During calculation, the system aligns the actually reconstructed trajectory points with the theoretical trajectory points based on a unified timestamp, and then calculates the root mean square (RMSE) value of the distance between all corresponding points. This value reflects the accuracy of motion control.

[0068] Force control performance evaluation includes two aspects: force control accuracy and force control bandwidth.

[0069] Force control accuracy: During the execution of the test program, the system records the actual output force of the massage claw at a specific target point. The target force value is defined by the currently running test program. Force control accuracy is usually expressed as the mean absolute deviation (MAE) or root mean square error (RMSE) between the actual output force and the target set force during the entire steady-state holding phase, reflecting the accuracy of static force control.

[0070] Force control bandwidth: This is tested through a dedicated evaluation subroutine. This subroutine sends a series of sinusoidal force command signals with increasing frequency to the massager through the system's control interface (such as USB or CAN bus). The system synchronously records the command force and the actual response force of the massage claw. By analyzing the amplitude attenuation and phase lag of the actual force to the command force at different frequencies (for example, by plotting a gain-frequency curve), the frequency corresponding to the gain attenuation to -3dB is determined as the force control bandwidth, which is used to evaluate the system's ability to dynamically track rapidly changing force commands.

[0071] (2) System Coordination Performance Indicators: These are used to evaluate the overall performance of multiple massage claws working together, and mainly include: Multi-claw synchronization error: used to quantify the temporal consistency of all massage claw movements. In the test procedure, a specific target state to be achieved (such as a specified force value) is defined for each massage claw i. or location The system records the time it takes for each massage claw to move from the start of its action to the first time it reaches its target state. Synchronization error These are the times. Relative to its average value The smaller the standard deviation, the better the synchronization of each claw in completing its respective target action.

[0072] Consistency of Cooperative Force Application Mode: Principal component analysis was used for evaluation. First, a complete "massage cycle" was defined, typically referring to the time T during which the tested massager completes one standard massage mode (such as "kneading mode #1") from start to finish. The time series of the normal forces of all N massage claws within this cycle T were then organized to form an N-row × T-column matrix F. Principal component analysis was then performed on this matrix to extract the first principal component representing the most important cooperative mode. Next, the original force sequence of each claw is calculated and... Pearson correlation coefficient between sequences Finally, take all The average value is used as the consistency score R.

[0073] Overall operational stability index: Defined as the reciprocal of the root mean square value of vibration acceleration measured by all accelerometers placed on the bionic head model when the massager is running under standard test procedures. The larger the index, the smaller the vibration caused by the operation of the whole machine and the better the stability.

[0074] The multi-dimensional correlation analysis engine is the core intelligent analysis component in the comprehensive performance evaluation unit. It operates based on a unified spatiotemporal correlation database, indexed by precise timestamps. This database aligns and fuses feature data acquired from four sensor channels—force, motion trajectory, sound, and vibration—to form a unified multi-dimensional feature dataset. Based on this dataset, the engine executes a three-tiered progressive correlation analysis according to a pre-defined logic: Layer 1: Mechanical and kinematic correlation analysis. This layer mainly examines whether the "force" and "position" of the massage claw match. The system will correlate the three-dimensional motion trajectory of the massage claw captured by the high-speed camera with the three-dimensional forces (including the normal force of pressing down and the tangential force of scraping) measured by the force sensor, according to time.

[0075] In specific analysis: For example, for a preset "kneading" motion, theoretically, the massage claw should move along a spiral line, and the direction of force should be roughly perpendicular to the head surface (i.e., mainly normal force). The system will calculate the tangential direction (movement direction) at each point on the actual motion trajectory, and then check whether the force component (tangential force) in this direction is large. If an excessively large tangential force is measured in a motion segment that should mainly be "pressing", it indicates that there may be unwanted "sliding" or "scratching". A "force-trajectory direction matching degree" index can be defined, for example, calculating the cosine value of the angle between the actual force vector direction and the theoretical force direction (such as the surface normal at that point) at each trajectory point, and then averaging it over the entire motion cycle. The closer this value is to 1, the more the force direction matches the expectation.

[0076] In addition, the system will overlay the previously calculated spatial distribution map of "dynamic trajectory tracking error" with the "pressure distribution cloud map" of the force sensor network. This allows for a direct observation of whether the pressure becomes uneven in areas with large trajectory deviations, thus analyzing the actual impact of mechanical precision on the massage effect. These results can ultimately generate a color heat map that visually shows where the "force" and "path" do not match.

[0077] Layer 2: Correlation analysis between acoustic and vibration signals and mechanical state. This layer is equivalent to "listening to the heart" and "taking the pulse" of the massager, judging the condition of internal parts through sound and vibration.

[0078] The foundation of this function is a pre-built "fault knowledge base" and corresponding diagnostic models. Building this knowledge base requires collecting sound and vibration spectrum data from a batch of massagers (including those with normal operation and those with various known faults, such as gear wear, loose bearings, and motor imbalance) under standard testing conditions. For each device, its spectral characteristics are extracted, such as amplitude values, sideband characteristics, and overall vibration energy at motor rotation frequency, gear meshing frequency and its harmonics, forming a feature vector, which is then labeled with its health status.

[0079] Then, this labeled data is used to train a classification model, such as a support vector machine (SVM) or a random forest, and the model learns which failure mode corresponds to different combinations of spectral features.

[0080] In actual testing, the system processes the sound and vibration spectrum of the current massager into the same feature vector, and then feeds it into the trained model for judgment. The model will output a classification result, such as "healthy", "suspected gear wear (confidence 85%)" or "excessive motor vibration", and give a comprehensive health score (such as 0 to 100 points). In this way, the health status of the internal machinery can be read from the noisy sound and vibration.

[0081] Layer 3: Multi-claw synergy effect correlation analysis. This layer focuses on whether multiple claws will "fight" or "drag down" each other when working together.

[0082] During the analysis, the system will select data from any two claws (such as claw A and claw B) for comparison.

[0083] Observe the force coordination: Calculate the cross-correlation function between the force signals, set a reasonable time window (such as a complete massage cycle) and lag range (such as ±100 milliseconds). If the cross-correlation function has a high positive peak near zero lag, it means that the two claws are very synchronized in applying force. If a significant negative value appears, it may mean that one claw is pressing down while the other claw is lifting up, indicating interference. If the negative correlation coefficient is consistently below -0.6, it is considered that there is a significant force conflict.

[0084] Observe the synchronization of movements: calculate the phase difference of the movement trajectory in the same direction. For example, compare the time difference of arriving at corresponding points on the movement path. If the standard deviation of this time difference between all claws is very small, it indicates that the movements are synchronized.

[0085] To examine vibration transmission: calculate the coherence function between their vibration signals, and analyze how much of the vibration of one claw is "transmitted" to the other claw at the main vibration frequency (such as the motor frequency). The frequency point with the higher coherence value is likely to be the main path of vibration propagation.

[0086] By filling all these pairwise analysis results (cross-correlation peak, average phase difference, coherence value of main frequency band) into a matrix, we can obtain a "claw coupling relationship matrix". Through this matrix, we can quantitatively see which two claws have the strongest mechanical coupling (large vibration transmission) and which two may move out of sync (large phase difference), providing very specific clues for optimizing mechanical layout and control algorithm.

[0087] To accurately simulate the mechanical properties of the human head, the bionic head model of this system adopts a layered structure of gradient composite material. This structure consists of three layers from the inside out, each with different materials and mechanical properties, working together to simulate the tactile sensation and response of a real head. (1) Rigid support liner: As the innermost structure, it is mainly made of engineering plastics (such as ABS, PC) or lightweight metals (such as aluminum alloy), with a thickness between 5 mm and 8 mm.

[0088] The selection of these materials and control of their thickness range are primarily aimed at providing sufficiently high local stiffness while ensuring a moderate overall weight for the model. This ensures the stable mounting of the internal sensor array, preventing measurement errors caused by substrate deformation, and also guarantees that the model's overall geometry does not change significantly when subjected to massage forces, thus providing a stable spatial reference for all measurements.

[0089] (2) Damping buffer intermediate layer: The intermediate layer is made of a polymer composite material with specific viscoelasticity, such as a polyurethane elastomer or silicone rubber composite material with a specific formulation.

[0090] A key design parameter for this layer is the material's loss factor (damping coefficient), which typically needs to be controlled within the range of 0.1 to 0.3. This range is determined by benchmarking the biomechanical properties of human scalp and subcutaneous tissue (which have moderate energy dissipation capacity) and after testing and verification.

[0091] If the loss factor is too low (<0.1), the layer is too "elastic" and insufficient in buffering and absorbing the impact, which may not accurately reflect the gentle response of the head tissue to dynamic massage; if it is too high (>0.3), the layer is too "sticky" and may cause sluggish response, affecting the real-time performance of the impact measurement signal.

[0092] Therefore, choosing materials within this range can better simulate the buffering and damping characteristics of real head tissue in a wide frequency range (covering the main operating frequency of the massager), so that the dynamic force applied to the model can be transmitted and attenuated in a way that is closer to the real human body.

[0093] (3) Elastic bionic outer skin layer: The outermost layer is made of medical grade silicone material through casting or molding process. Its Shore hardness is set in the low range of A10 to A20. This hardness range is a typical choice to simulate the soft touch of human face and scalp skin.

[0094] Meanwhile, the silicone material itself has a moderate and stable coefficient of friction. At this hardness, it can provide sufficient (but not too much) friction between the massage claw and the massage claw, effectively preventing distortion of force or trajectory data caused by slippage during the test, and also avoiding interference with the normal movement of the massage claw itself due to excessive friction.

[0095] The thickness is controlled at 3 to 5 millimeters to simulate the feel of skin without excessively weakening the transmission of mechanical signals by the underlying damping and rigid layers.

[0096] These three layers work together to achieve a gradient change in mechanical properties from the outside in. The elastic outer skin layer is responsible for simulating the initial sensation and friction of skin contact; the damping layer below it is a key energy regulator, responsible for absorbing and dissipating the mid-to-high frequency impacts and vibrations brought by the massage claw movements, simulating the viscoelasticity of soft tissue; the innermost rigid layer serves as the stable foundation for all measurements.

[0097] This "soft-sticky-hard" gradient structure allows the entire bionic head model to comprehensively approximate the complex mechanical behavior of a real human head within a wide range of frequencies and force values ​​when it is massaged, including its dynamic force response, vibration transmission characteristics, and overall force-deformation relationship. This provides a highly simulated mechanical environment for the detection system.

[0098] As another preferred embodiment of the present invention, the calculation logic of the system collaborative performance index is as follows, which aims to quantitatively evaluate the effect of multiple massage claws working together: (1) Calculation of multi-claw synchronization error: This metric is used to quantify the temporal consistency of all massage claw movements. Before calculation, two key moments need to be clearly defined: the moment the movement begins and the moment the goal is achieved.

[0099] Action start time: The time stamp when the automated test process control module sends a unified start command to the massager is usually taken as the start time of all massage claw actions (t=0), and the system will record this time.

[0100] Target achievement time: For each massage claw i, the system continuously monitors its actual output force or position, and the target value is reached when its measured value first enters the preset target value. or When the claw reaches the target state within an allowable tolerance zone (e.g., ±2% of the target value) near the target value, this moment is recorded. .

[0101] Then, calculate the time to achieve all of these. arithmetic mean Synchronization error That is, each Compared to The standard deviation of is given by the following formula:

[0102] Calculated standard deviation The smaller the value, the more synchronized the movements of each massage claw are in time.

[0103] (2) Consistency assessment of collaborative force application mode: This indicator assesses how similar the force patterns are when each claw works together. Before calculation, it is necessary to determine which data segment to analyze.

[0104] Data selection: Select data within a complete "massage cycle". This cycle usually refers to the time T taken for the tested massager to automatically complete one standard massage mode (such as one complete "kneading mode #1") from beginning to end. The system will automatically identify the start and end points of this cycle.

[0105] The specific calculation steps are as follows: Data processing: Take out the sequence of normal force changes of all N massage claws over time within this period T, and organize it into a matrix F with N rows (representing N claws) and T columns (representing T time points).

[0106] Principal component analysis: Perform principal component analysis on this matrix F to find the first principal component that best represents the trend of changes in the combined force exerted by everyone. .

[0107] Correlation calculation and scoring: Calculate the force sequence for each claw individually and the common force sequence. Pearson correlation coefficient between sequences .

[0108] Consistency score: Finally, calculate the correlation coefficients of all claws. Taking the average value yields the consistency score R of the collaborative force application mode:

[0109] The closer the score R is to 1, the more consistent the force application pattern of each massage claw is with the dominant synergistic pattern.

[0110] (3) Calculation of overall operational stability index: This indicator reflects whether the massager vibrates or not during operation, and is used to evaluate its stability. During the standard test program, the vibration acceleration data measured by all M triaxial accelerometers installed on the bionic head model are retrieved, and the root mean square value of each accelerometer's data is calculated separately. .

[0111] The overall operational stability index S is calculated using all these root mean square values. It is defined as the reciprocal of the squared average (i.e., the total RMS value). The specific formula is:

[0112] The larger the value of the index S, the lower the overall vibration level measured, meaning the more smoothly the massager operates.

[0113] This system also includes an automated testing process control module, used to automatically control and monitor the testing process, ensuring standardization and consistency of testing. The workflow of this module mainly includes the following steps: (1) Test program pre-storage and definition: This module pre-stores standard test programs for different massage modes (such as kneading, tapping, vibration, etc.).

[0114] Each test program is stored in the form of an executable script or data structure, and the expected motion trajectory parameters (such as path coordinate point sequence), target force curve (force value and time relationship), working frequency, and duration of a single test are clearly defined for each massage claw during the test.

[0115] These programs can be edited, verified, and downloaded to the control module via the host computer software interface.

[0116] (2) Detection start and synchronous triggering: When a detection task is started, the automated test process control module (usually a real-time controller or industrial PC) executes a script with strict timing.

[0117] Command issuance: The controller first sends a start command to the massager under test via USB or CAN bus.

[0118] Synchronous triggering: Within the same control cycle (e.g., within 1 millisecond), the controller outputs a rising edge-triggered TTL pulse signal from its dedicated digital I / O board as a hardware synchronous trigger signal. This pulse signal is directly connected to the synchronous trigger circuit input port of the high-speed data acquisition and processing unit via a coaxial cable.

[0119] Timing Guarantee: To ensure the synchronization accuracy between the start of the massager and the initiation of data acquisition, the controller's task cycle needs to be set short and stable enough (e.g., 1ms), and the sending of the start command and the generation of the synchronization pulse must be completed immediately within the same cycle. This, combined with the slight fixed delay in circuit transmission, controls the system-level time deviation between "command execution" and "data acquisition start" to the millisecond level, meeting the synchronization analysis requirements of most massage movements. The even higher precision (microsecond level) synchronization within the high-speed data acquisition unit is guaranteed by its own synchronization trigger circuit.

[0120] (3) Real-time data monitoring and closed-loop comparison: During the detection process, this module not only acts as a commander, but also as a "supervisor".

[0121] Data Reception and Alignment: The control module continuously receives real-time trajectory and force data of the massage claw from the front-end processing unit via high-speed Ethernet or PCIe bus. These data streams are all timestamped with high precision. The monitoring thread runs at a fixed frequency (e.g., 100Hz or higher). In each monitoring cycle, it obtains the "expected value" for that moment from the expected trajectory and force curves defined by the test program (which may be defined at a higher time resolution) through linear interpolation based on the current time, and then compares it with the received "actual value".

[0122] Specific comparison and fault-tolerance logic: The system presets tolerance thresholds for different key parameters. These thresholds are usually determined based on product design specifications, safety limits, or experience. For example, the trajectory deviation threshold is set to ±2 mm, and the force deviation threshold is set to ±10%.

[0123] The monitoring thread calculates deviations in real time. Once it detects that the deviation of a certain parameter (such as the Y-axis position of claw 3) continuously exceeds its tolerance threshold (for example, exceeding the limit for three consecutive monitoring cycles), the control module will determine that an anomaly has occurred and immediately execute the preset response strategy: Log exceptions: Record a detailed exception event in the system log, including the time of occurrence, which parameter of which claw exceeded the limit, and by how much.

[0124] Execution strategy: Based on the preset configuration, the system can react in different ways: Alarm prompt: A prominent warning message pops up on the operation interface and an audible alarm is sounded to alert the operator, but the test continues.

[0125] Automatic pause: If the abnormality is serious (such as excessive force that may damage the equipment or model), the control module will immediately send an emergency stop command to the massager and pause data acquisition. This is to prevent the massager from continuing to operate under obvious fault conditions, which could generate meaningless data or even cause damage.

[0126] Continue to observe: For minor or occasional exceedances, the system can also choose to record only without interrupting the test, and then analyze them uniformly after completion.

[0127] This closed-loop monitoring mechanism makes the testing process no longer a simple "play-record" process, but rather enables preliminary online quality assessment and self-protection capabilities.

[0128] After the testing process is completed, the comprehensive performance evaluation unit will automatically package and generate a complete test report based on the previous analysis results. This function does not simply list the data, but organically combines the numerical results, charts and diagnostic conclusions according to the preset logic to form a document that can be directly used for quality judgment or R&D improvement.

[0129] (1) The report content is generated in a structured manner.

[0130] At the heart of report generation is a pre-configured report template that defines the entire skeleton of the report, such as where to place the summary, data tables, and conclusions. The template contains specific "placeholders" or "data-binding fields."

[0131] When the system is working, it will populate the template according to the following process: Data extraction and formatting: After the performance index quantification model and multi-dimensional correlation analysis engine have finished running, they will organize various index values ​​(such as pressure uniformity and synchronization error), analysis results (such as the judgment signs of "trajectory left deviation" and "spectral anomaly") and related feature data (such as matrix data of error heatmap) into a structured intermediate data object.

[0132] Content Mapping and Filling: The report generation engine reads this intermediate data object and then fills the data into the corresponding positions in the template according to a set of mapping rules. For example, if there is a table in the template marked as "Single Claw Performance Indicators", the engine will extract the values ​​such as "Static Pressure Uniformity" and "Dynamic Trajectory Tracking Error" for each claw in the intermediate data object and fill them in according to the table format. For another example, if there is a conclusion description in the template that "Massage claw number {claw number} has a continuous {deviation type} in the {direction} direction trajectory", the engine will replace it with the specific number (e.g., 3), direction (e.g., Y), and deviation type (e.g., left deviation) according to the analysis results to form a complete diagnostic conclusion: "Massage claw number 3 has a continuous left deviation in the Y direction trajectory".

[0133] Chart generation: For charts that need to be displayed, such as radar charts, spectrum comparison charts, and error heatmaps, the system will call the built-in chart library (such as Python's Matplotlib or business chart control) to render the chart images in real time using the data in the intermediate data object (coordination index values, spectrum data, trajectory error matrix), and then insert these images into the specified position in the report template.

[0134] In this way, a complete report containing data tables, visualizations, and textual conclusions is automatically generated.

[0135] (2) Report output and system integration.

[0136] The generated report content will be converted into a format that is easy to use and exchange.

[0137] Formatted packaging: Typically, the system will package the final report into a PDF file because the PDF format is universal and not easily tampered with. An HTML web version of the report can also be generated as needed for online viewing.

[0138] Output Distribution: The packaged report file will be sent out through the network interface on the integrated performance evaluation unit. The specific destination can be determined based on the actual production line configuration. On-site viewing: It can be saved directly to the local hard drive, or displayed on the connected system monitor via HDMI / USB.

[0139] Network archiving: More commonly, data is automatically uploaded to the company's central file server or database server via Ethernet using FTP or SCP protocols for long-term preservation and traceability.

[0140] Integration with Production Management Systems: To achieve truly automated quality control, the system can be deeply integrated with the Production Management System (MES). A typical approach is for the system to not only upload a PDF file but also push key judgment results from the report (such as "pass / fail" flags and main performance indicator values) and the report file's storage path to the MES via an API interface provided by the MES (e.g., a RESTful API) in a structured data packet (e.g., JSON format). Upon receiving this data, the MES can automatically trigger subsequent processes, such as marking the product as passable for warehousing or transferring non-conforming products to the repair station, while simultaneously pushing the report link to the quality engineer. If design-related issues are detected, this structured data can also be packaged and sent to the R&D department's PLM or project management system as input for product iteration.

[0141] In summary, this invention successfully solves the technical challenge of conducting integrated multi-dimensional dynamic performance testing of multi-claw head massagers by constructing a highly integrated biomimetic testing platform, deploying a multimodal synchronous sensor network, and introducing advanced data fusion and correlation analysis technologies. The system not only possesses basic testing capabilities of high precision, high efficiency, and high repeatability, but also reveals the underlying mechanisms behind performance phenomena through deep correlation analysis, providing strong technical support for product development, quality control, and intelligent manufacturing.

[0142] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0143] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A performance testing system for a multi-claw head massager, characterized in that, include: An integrated testing platform includes a rigid base, a bionic head model set on the rigid base, and a multi-dimensional motion bearing mechanism set between the rigid base and the bionic head model. The multi-dimensional motion bearing mechanism is used to drive the bionic head model to perform multi-degree-of-freedom precise positioning. A multimodal sensing array is arranged on and around a bionic head model to synchronously acquire multidimensional dynamic response signals of the massager under test during operation. The multimodal sensing array includes at least a three-dimensional force sensing network for measuring three-dimensional forces, a high-speed visual tracking module for capturing motion trajectories, and an acoustic-vibration synchronous acquisition module for acquiring noise and vibration signals. A high-speed data acquisition and processing unit is connected to a multimodal sensor array signal to receive and synchronously process multidimensional dynamic response signals. The high-speed data acquisition and processing unit includes at least a synchronous trigger circuit, a multi-channel analog-to-digital converter, and a preprocessing algorithm module. The comprehensive performance evaluation unit is connected to the high-speed data acquisition and processing unit and is used to perform performance quantification and comprehensive correlation diagnosis based on the processed signal. The comprehensive performance evaluation unit includes at least a performance index quantification model and a multi-dimensional correlation analysis engine.

2. The head multi-claw massager performance testing system according to claim 1, characterized in that, The biomimetic head model has a layered structure of gradient composite material, consisting of, from the inside out: The rigid support liner, made of engineering plastic or lightweight metal, serves to provide a mounting base and a stable spatial reference for the internal sensor array. The damping buffer intermediate layer, which covers the rigid support inner liner, is made of a polymer composite material with specific viscoelasticity. Its damping loss factor is controlled between 0.1 and 0.3, and it is used to simulate the buffering and damping characteristics of head tissue. The elastic biomimetic outer skin layer, which covers the damping buffer intermediate layer, is made of medical-grade silicone material with a Shore hardness of A10 to A20, and is used to simulate the tactile and frictional characteristics of scalp skin.

3. The head multi-claw massager performance testing system according to claim 1, characterized in that, The three-dimensional force sensing network consists of multiple miniature three-dimensional force sensors arranged according to a preset grid node layout. The sensors are installed through standardized threaded interfaces pre-embedded in the bionic head model, and each sensor can simultaneously measure normal force and tangential force.

4. The head multi-claw massager performance testing system according to claim 1, characterized in that, The high-speed vision tracking module includes at least two jointly calibrated high frame rate industrial cameras. The cameras start acquisition by receiving the same trigger signal from the synchronous trigger circuit, which is used to obtain the three-dimensional motion trajectory of the end of the massage claw based on a binocular stereo vision reconstruction algorithm.

5. The head multi-claw massager performance testing system according to claim 1, characterized in that, The acoustic and vibration synchronous acquisition module includes a high-sensitivity microphone array arranged in the space around the bionic head model and multiple triaxial accelerometers fixed inside the bionic head model at key mechanical nodes, for synchronously acquiring noise spectrum and vibration signals.

6. The head multi-claw massager performance testing system according to claim 1, characterized in that, The synchronization trigger circuit is driven by a high-stability temperature-compensated quartz crystal oscillator to generate and distribute synchronization pulse signals to each acquisition device of the multi-mode sensing array to achieve microsecond-level time synchronization.

7. The head multi-claw massager performance testing system according to claim 1, characterized in that, The performance index quantification model is configured to calculate single-claw performance index and system collaborative performance index; Among them, the performance indicators of a single claw include at least static pressure uniformity, dynamic trajectory tracking error, force control accuracy, and force control bandwidth; The system's collaborative performance indicators include at least the multi-claw synchronization error, the consistency score of the collaborative force application mode, and the overall operational stability index.

8. The head multi-claw massager performance testing system according to claim 1, characterized in that, The multi-dimensional correlation analysis engine is configured to perform the following correlation analyses in sequence: Mechanical and kinematic correlation analysis is used to spatiotemporally align the three-dimensional motion trajectory of the massage claw with three-dimensional force data, and analyze the matching degree between the force direction and the motion direction; The correlation analysis between acoustic and vibration signals and mechanical status is used to determine the health status of internal mechanical components of a massager based on a pre-established fault knowledge base and diagnostic model, according to the characteristics of noise and vibration spectrum. Multi-claw synergistic effect correlation analysis is used to quantify the coupling relationship between claws by analyzing the cross-correlation of force signals, motion phase difference and vibration coherence between any two massage claws.

9. The head multi-claw massager performance testing system according to claim 1, characterized in that, It also includes an automated test process control module, which is configured as follows: Pre-store and execute standard test programs for different massage modes. The programs define the expected motion trajectory parameters and target force curve of the massage claw. When the test starts, a start command is sent to the massager under test and a hardware trigger signal is sent to the high-speed data acquisition and processing unit at the same time; During the detection process, the trajectory and force data of the massage claw are received in real time and compared with the expected value. Based on whether the deviation exceeds the preset tolerance threshold, the corresponding anomaly handling strategy is executed.

10. The head multi-claw massager performance testing system according to claim 9, characterized in that, The comprehensive performance evaluation unit is also configured to automatically generate a structured test report based on the analysis results after the testing process is completed, and output the report as a specified format file and / or push key judgment results to an external production management system in the form of structured data.

Images