Method and system for judging chronic lumbar muscle degeneration state of passenger based on vehicle vibration modal analysis

Abstract

The invention discloses a passenger lumbar muscle chronic degeneration state discrimination method and system based on vehicle vibration modal analysis, and relates to the technical field of automobile passenger health monitoring, biomechanics and signal processing, and the method comprises the following steps: collecting a vibration signal transmitted to the waist of a passenger during vehicle driving; the vibration signals are preprocessed; modal analysis is carried out on the preprocessed vibration signals, modal parameters at least comprising the inherent frequency and the damping ratio are extracted, the modal parameters of a preset frequency band related to the lumbar vertebra biomechanical response are focused on, and feature vectors are extracted from the modal parameters; constructing a lumbar muscle degeneration state discrimination model; according to the method, a non-contact and non-inductive monitoring mode is adopted, only an existing vibration sensor or an added vibration sensor of the vehicle needs to be utilized, passengers do not need to wear or contact extra equipment, the comfort is good, long-term use is easy, objective lumbar muscle degeneration state indexes are provided based on physical signals and an algorithm model, and the influence of subjective factors is reduced.

Description

Technical Field

[0001] This invention relates to the fields of vehicle occupant health monitoring, biomechanics and signal processing technology, specifically to a method and system for identifying chronic degenerative states of occupant lumbar muscles based on vehicle vibration modal analysis. Background Technology

[0002] Prolonged driving is a major factor that causes and exacerbates lumbar muscle strain, chronic pain, and even intervertebral disc degeneration. Early identification and intervention are crucial for preventing serious lumbar diseases and improving driving comfort and safety.

[0003] Existing methods for assessing chronic degeneration of lumbar muscles in occupants have the following limitations:

[0004] (1) Subjectivity / Lag: Reliance on passenger self-reports (questionnaires, pain scores) has the problems of strong subjectivity and lag (usually reported only after symptoms become obvious).

[0005] (2) Invasive / High cost: Medical imaging examinations (such as MRI, CT) are accurate, but they are expensive, inconvenient, cannot be monitored in real time, and are usually used for diagnosis rather than routine screening.

[0006] (3) Limitations of contact sensors: Integrating pressure distribution sensors or surface electromyography (sEMG) devices into seats can provide some information, but may affect riding comfort. sEMG requires skin contact and the signal is easily interfered with, making it difficult to use in a long-term, non-invasive manner.

[0007] (4) Lack of dynamic correlation: Existing methods are difficult to directly and dynamically assess the relationship between vehicle vibration, a key inducing factor, and the biomechanical response of the occupant's lumbar muscles.

[0008] Furthermore, vehicle vibration modal analysis is mainly used for NVH (noise, vibration, and harshness) performance optimization, structural health monitoring, and suspension tuning. Although it is known that vibration affects human comfort and health, there is currently no mature technology to directly and quantitatively correlate and differentiate vehicle vibration modal characteristics with the chronic degenerative state of specific muscle groups (such as lumbar muscles) of occupants. Therefore, a method and system for differentiating the chronic degenerative state of occupant lumbar muscles based on vehicle vibration modal analysis is needed to address the aforementioned issues. Summary of the Invention

[0009] The purpose of this invention is to provide a method and system for identifying the chronic degeneration state of occupant lumbar muscles based on vehicle vibration modal analysis, so as to solve the problems existing in the prior art mentioned in the background.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A method for identifying chronic degenerative changes in occupant lumbar muscles based on vehicle vibration modal analysis includes the following steps:

[0012] S1: Collects vibration signals transmitted to the waist of the occupant during vehicle operation;

[0013] S2: Preprocess the vibration signal;

[0014] S3: Perform modal analysis on the preprocessed vibration signal and extract modal parameters including at least natural frequency and damping ratio, with a focus on modal parameters in a predetermined frequency band related to the biomechanical response of the lumbar spine, and extract feature vectors from the modal parameters;

[0015] S4: Construct a lumbar muscle degeneration state discrimination model, which is obtained by simulating the biomechanical correlation data between the modal parameters of vehicle vibration and the chronic degeneration state of the occupant's lumbar muscles;

[0016] The feature vector is input into a pre-trained lumbar muscle degeneration state discrimination model to obtain the lumbar muscle chronic degeneration state discrimination result output by the lumbar muscle degeneration state discrimination model, and a quantization level mapping table is established.

[0017] Preferably, the specific steps of S1 are as follows:

[0018] S11: Collect lumbar spine curve parameters, including lumbar curvature, intervertebral distance, curvature between L1-5, passenger weight, height and age;

[0019] S12: The frequency acquisition in the lumbar region mainly focuses on the L3 position of the spine. An accelerometer is installed according to the position of the spine to capture the vibration frequency of the lumbar spine. At the same time, the dynamic output of the seat cushion acceleration is recorded simultaneously. It can quantify the frequency spectrum and amplitude of spinal flexion, extension, lateral flexion and rotation, and measure the vibration frequency in real time.

[0020] S13: Finally, collect seat data, including seat equivalent mass, seat damping coefficient, and seat stiffness coefficient.

[0021] Preferably, the specific steps of S2 are as follows:

[0022] S21: The lumbar frequency of the signal is filtered by bandpass filtering, and an anti-aliasing filter is used to prevent the frequency from being too high due to improper driving habits. The output signal at the lumbar muscles is used for preliminary judgment.

[0023] S22: Further analysis of the specific physical condition of muscle and lumbar spine vibration is then conducted using the transfer functions of the input and output signals.

[0024] Since the natural frequency of the intervertebral disc is between 5 and 10 Hz, the interaction between vehicle vibration and the body's natural frequency varies with different vehicle speeds. According to "human lumbar spine resonance experiment data", vibration in this frequency range is prone to cause a sudden increase in intervertebral disc pressure (30%-50% higher than static), which is a key cause of lumbar muscle degeneration. The human body is prone to resonance at 4 to 8 Hz, causing personal injury. As the resonance frequency decreases, the lumbar spine vibration output frequency also decreases. When the resonance frequency decreases to 4-8 Hz, the lumbar spine reaches the danger period of injury. At the same time, the change in the transfer function can also be used to judge the change in the driver's posture.

[0025] Preferably, the specific steps of S3 are as follows:

[0026] S31: Derivation of the nonlinear relationship between input and output: Modal analysis is performed through the data output of hardware and software to record the natural frequency, damping ratio and mode shape of vibration excitation, and the data is segmented according to the driving conditions.

[0027] S32: Signal and noise analysis: Analyze the specific vibration environment of the vehicle. First, perform EMD decomposition on the measured time-domain signal to decompose it into the main signal, secondary signal and residual term.

[0028] S33: Fourier Transform and Noise Removal of Main Signal: Fourier transform is performed on the main frequencies of the lumbar spine response obtained after EMD decomposition, and the initial signal is processed to remove noise signals, so as to obtain the frequency domain signal of the main frequencies of the lumbar spine response. The peak value of the frequency domain signal response, i.e. the resonance peak, is recorded, and parameters closely related to lumbar spine injury are extracted.

[0029] To reduce errors, the eigenvectors of the modal parameters were acquired through two types of data comparison. The first was the time-domain signal directly output by the lumbar accelerometer, which was mainly decomposed by EMD transformation to acquire the main vibration signals, and the resonance amplitude, relative baseline frequency offset, and damping ratio of the signals were recorded. The second was the frequency response function H(w) derived by the H1 estimation method from the self-spectral density of the input signal spectrum of the seat cushion and the cross-spectral density of the output spectrum. The data results were further optimized by comparing the frequency response function with the output signal.

[0030] Secondly, the calculated transfer function can provide a reference for the current chronic degeneration of the lumbar spine. The main reference data include the damping ratio, the principal resonant frequency, and the correlation function coefficient. The higher the transfer rate, the greater the force the lumbar spine has to bear, and the higher the risk of injury. For the vibration frequency, we mainly observe the frequency change from 4 to 8 Hz. This is because the human lumbar spine is more likely to resonate at this frequency, which will aggravate lumbar spine injury. The pitching vibration of 1 to 2 Hz will increase the muscle load. The modal damping ratio indicates the speed at which the vibration energy of the system is dissipated.

[0031] Preferably, the specific steps of S4 are as follows:

[0032] S41: Establish baseline status and have the driver perform quantitative posture calibration according to national standard items before the vehicle is actually driven;

[0033] S42: The lumbar spine modality obtained from the modal analysis is detected. The initial lumbar spine state is defined before the vehicle is driven. This is the initial state of the lumbar spine. Due to the difference in biomechanical values ​​caused by the difference in human weight, the weight coefficient is corrected according to the BMI of different members to obtain the actual human body weight.

[0034] S43: The establishment of a model for judging the state of lumbar muscle degeneration is based on the quantification of damage according to the degree of injury;

[0035] S44: Based on data from healthy in vitro specimens or patients with varying degrees of injury, and combined with MRI and X-ray diagnostics, verify the biomechanical rationality before outputting the model.

[0036] The acquisition of modal vibration modes reflects the dynamic coupling between the occupant and the seat by dynamically capturing the occupant's movements at different frequencies. This establishes a mapping relationship between the vibration frequency of the human body at different vehicle speeds and the state of the lumbar spine. Finally, a two-parameter data verification mechanism is adopted: EMD transformation: targeting the nonlinear characteristics of lumbar spine vibration, the resonance amplitude and frequency offset Δf are decomposed to reflect the muscle elasticity decay (degeneration leads to a decrease in natural frequency); H1 estimation method: the transmissibility is calculated through the frequency response function H(w) of the seat cushion and the lumbar spine. When the transmissibility is >3.5, it indicates "abnormal driving posture leads to excessive lumbar load". The results are cross-validated with EMD results to reduce errors.

[0037] Preferably, the vibration signal in S1 is collected by multiple vibration accelerometers installed at L1-5 and C1 of the lumbar vertebrae.

[0038] Preferably, the predetermined frequency band in S3 includes a vertical vibration frequency band of 4Hz to 8Hz.

[0039] Preferably, the feature vector extraction in S3 includes: extracting the natural frequency value, damping ratio value, and the mapping relationship between the natural frequency value and the vehicle speed of the dominant mode in the predetermined frequency band.

[0040] Preferably, the discrimination result in S4 includes a classification of the level of lumbar muscle degeneration or a score of the degree of degeneration, and the discrimination result is output through an in-vehicle display device, a mobile terminal or a cloud platform.

[0041] A system for identifying chronic degeneration of occupant lumbar muscles based on vehicle vibration modal analysis includes:

[0042] The vibration sensor module is used to collect vibration signals transmitted to the waist of the occupants during vehicle operation;

[0043] The vehicle dynamic parameter acquisition module and the occupant information input module are used to collect vehicle dynamic parameters and occupant status information;

[0044] A signal preprocessing module is used to preprocess the vibration signal;

[0045] The modal parameter identification module is used to perform modal analysis on the preprocessed vibration signal and extract modal parameters including at least natural frequency and damping ratio, with a focus on modal parameters in a predetermined frequency band related to the biomechanical response of the lumbar spine.

[0046] The feature extraction module is used to extract feature vectors from the modal parameters, and also to fuse vehicle dynamic parameters and occupant basic information into the feature vectors;

[0047] The lumbar muscle degeneration state discrimination model module stores and runs a pre-trained discrimination model, which is used to receive the feature vector and output the discrimination result of the chronic degeneration state of the lumbar muscles. The lumbar muscle degeneration state discrimination model is trained based on the biomechanical correlation data between vehicle vibration modal parameters and the chronic degeneration state of the occupant's lumbar muscles.

[0048] The output module is used to output the discrimination result.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] 1. This invention adopts a non-contact, non-sensory monitoring method, which only requires the use of existing or added vibration sensors in the vehicle. There is no need for passengers to wear or touch additional equipment, which is comfortable and easy to use for a long time. Based on physical signals and algorithm models, it provides objective indicators of lumbar muscle degeneration status and reduces the influence of subjective factors.

[0051] 2. By monitoring subtle changes in modal parameters, this invention can easily identify the risk of decreased or early degeneration of lumbar muscle function before passengers perceive obvious discomfort or pain. It also directly establishes the correlation between vehicle vibration excitation and the biomechanical response of the occupant's lumbar muscles, which can more realistically reflect the lumbar load under the specific scenario of driving.

[0052] 3. The core sensor of this invention is a vibration accelerometer, which is low in cost, can reuse existing vehicle sensors or is easy to install, and the algorithm can be implemented in vehicle hardware or in the cloud. It can also improve health management, provide an effective tool for driver personal health management and fleet occupational health and safety management, and help prevent occupational lumbar muscle strain and degeneration. Attached Figure Description

[0053] Figure 1 This is a flowchart of the method of the present invention.

[0054] Figure 2This is a schematic diagram showing the installation position of the accelerometer of the present invention.

[0055] Figure 3 This is a flowchart of the modal parameter extraction process of the present invention.

[0056] Figure 4 The process for establishing the lumbar muscle degeneration status discrimination model of the present invention.

[0057] Figure 5 This is a diagram showing the environmental noise categories of this invention. Detailed Implementation

[0058] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0059] Please see Figure 1-5 The present invention provides the following technical solutions:

[0060] A method for identifying chronic degenerative changes in occupant lumbar muscles based on vehicle vibration modal analysis includes the following steps:

[0061] S1: Collects vibration signals transmitted to the occupant's waist during vehicle operation.

[0062] First, regarding the sensor type, a highly sensitive Viter intelligent triaxial accelerometer is used. This sensor is wearable on the human body, with an operating current of less than 25mA, a power supply voltage of 5V, and a detection cycle between 1 and 100Hz, meeting the human body's vibration frequency range. The sensor is installed at five locations on the lumbar spine (L1-L5), a high-risk area for spinal injuries. An electromyography (EMG) collector is installed at the iliac crest to record signals in real time with the sensors at L1-L5, ensuring signal synchronization. Simultaneously, a vibration sensor is installed inside the seat to measure the seat's vibration frequency response, collecting data every 10 minutes for a total of 6 times. Furthermore, the posture of various body parts is detected to determine the human posture at different vibration frequencies, and the input signal is determined by the mapping between these two measurements.

[0063] S2: Preprocess the vibration signal.

[0064] First, abnormal data caused by sensor malfunctions and motion artifacts (such as sudden swaying due to emergency braking) are eliminated. Second, bandpass filtering (0.5-50Hz) is used to remove power frequency interference and baseline drift. The lower limit of 0.5Hz is chosen to remove ultra-low frequency drift and motion artifacts, and low-pass filtering (cutoff frequency 5-10Hz) is used to remove high-frequency jitter. The upper limit is set to 50Hz to remove high-frequency electronic noise and extremely high-frequency vibrations unrelated to the lumbar spine's biomechanical response, reducing the influence of electronic equipment on the experimental results. Then, median filtering is used to smooth spatial noise. Finally, time synchronization is performed to ensure that the accelerometer and barometer data samples are consistent in time, and the data is segmented according to driving conditions such as straight-line driving, turning, acceleration, and road bumps.

[0065] S3: Perform modal analysis on the preprocessed vibration signal and extract modal parameters including at least the natural frequency and damping ratio, with a focus on modal parameters in a predetermined frequency band related to the biomechanical response of the lumbar spine, and extract feature vectors from the modal parameters.

[0066] S31: Derivation of the Nonlinear Relationship between Input and Output: Modal analysis is performed using hardware and software data outputs to record the natural frequency, damping ratio, and mode shape of the vibration excitation. The data is then segmented according to driving conditions such as straight-line driving, turning, acceleration, and road surface roughness. Since the lumbar spine signal is nonlinear and irregular, the frequency response function is derived by using the time-domain functions of the seat accelerometer and the surface lumbar spine accelerometer to establish reference parameters for the measured data. The derivation process of the frequency response function derived from the accelerometer and the seat cushion accelerometer is as follows:

[0067]

[0068] Where H(f) is the frequency response function; a spine (f) is the lumbar spine acceleration response; F(f) is the input excitation force;

[0069]

[0070] T(f) is the vibration transfer function from the seat to the lumbar spine; a seat (f) Seat acceleration;

[0071]

[0072] Z seat (f) The mechanical impedance of the seat, in units of (N·s / m), finally yields the frequency response function:

[0073]

[0074] The simplified representation is as follows: in:

[0075]

[0076] m is the equivalent mass of the seat (kg); c is the seat damping coefficient (N·s / m); k is the seat stiffness coefficient (N / m); f is the frequency (Hz); and the human body coupling coefficient m 耦合 for:

[0077] m 耦合 =m seat +0.7×m 等效

[0078] And transfer function calculation

[0079]

[0080] PSD 互谱 (f) is the cross-spectral density; PSD 自谱 (f) is the autospectral density;

[0081] In the specific calculation of the cross-spectral function and the self-spectral function, the PSD needs to be segmented and windowed, where PSD is the power spectral density function. The calculation of PSD is evaluated using the Welch method, which divides the signal into K segments, each of length L, and windows the signal of each segment. The window signal is as follows:

[0082]

[0083] After normalizing the window function, we obtain the following equation:

[0084]

[0085] Finally, the power spectral density of a certain segment is obtained as follows:

[0086]

[0087] Where f s The sampling frequency;

[0088] Finally, after weighted calculation, the average periodicity chart of all segments is obtained as follows:

[0089] Coherence function verification:

[0090]

[0091] Where γ 2 (f)>0.8, f0 is the resonant frequency;

[0092] Then, through least squares fitting:

[0093]

[0094] The cross-spectral function PSD 互谱 (f):

[0095]

[0096] X k (f) = FFT(x) k w[n])

[0097] Y k (f) = FFT(y) k w[n])

[0098] in It is X k (f) complex conjugate; X k (f) is the input function; Y k (f) is the output function.

[0099] S32: Signal and Noise Analysis: Analyze the specific vibration environment of the vehicle; First, decompose the measured time-domain signal into a primary signal, a secondary signal, and a residual term. Since the basic signal transmitted by the seat vibration is too weak, the IMF signal after S34EMD decomposition contains a large error, and the target signal in the residual term is too weak. According to the source of the noise, the noise can be divided into three types.

[0100] S321: Motion trajectory, which is the main source of noise. The main reason for the noise generated by the motion trajectory is the unconscious micro-movement of the human body, which is unpredictable. In the case of coughing or uneven road surface, the instantaneous impact signal can reach above 100Hz. This part is too inconsistent with the measured value and can be eliminated by bandpass filtering. The low frequency shift caused by unintentional posture changes has a more significant impact on the lumbar spine output signal. The low frequency input of posture changes is generally within 0.5Hz, which mainly affects the low frequency signal output.

[0101] S322: Physiological noise, mainly including various physiological activities of the human body, such as heartbeat. During normal driving, the main physiological activities include respiratory tremors, heartbeat and muscle tremors. Among them, muscle tremors have the greatest impact on the judgment process. Under the stretch reflex, the muscle activation frequency is maintained at 8-12Hz, while in the normal low-frequency environment it fluctuates randomly and is roughly stable at 4-8Hz.

[0102] To eliminate random muscle fluctuations under normal low-frequency conditions, the baseline status of drivers was initialized and a baseline reference standard was established before the experiment. A reference point was also established on the skin 2-5 cm above the bony prominence of the iliac crest at the upper edge of the pelvis (avoiding areas of muscle aggregation). Electromyography (EMG) signals were collected at this reference point using SEMG electrodes and compared with the experimental data. Finally, principal component analysis (ICA) was used to process the experimental data, carefully and accurately identifying and removing noise factors representing EMG activity. Pathological muscle tremors occur at 4-6 Hz and generally have a certain periodicity, while the heartbeat cycle is 1-2 Hz, and respiratory tremors are 0.1-0.3 Hz. Therefore, muscle tremors have a relatively direct impact on injury assessment, while heartbeat and respiratory tremors mainly affect the direct detection of vibration signals at low frequencies.

[0103] S323: Environmental noise signals, mainly including vibrations from the vehicle and electronic interference from nearby equipment. Electronic equipment operates at a stable frequency, typically 50-60Hz, which can be removed by bandpass filtering. Taking a common tire size of 0.65m as an example, the relationship between vehicle speed and vibration frequency is calculated as shown in the table below:

[0104] Table 1: Relationship between vehicle speed and vibration frequency

[0105]

[0106]

[0107] The table above shows that for every 20 km / h increase, the vibration frequency increases by an average of approximately 2.721 Hz. When the vibration frequency approaches the vehicle's natural suspension frequency (typically in the 10-15 Hz range), the amplitude increases dramatically, leading to strong vibrations. For example, at speeds of 60-80 km / h (frequency 8-11 Hz), the vehicle may be approaching the resonance zone.

[0108] S33: Fourier Transform (FFT) of the Main Signal and Noise Removal: Fourier transform (FFT) is performed on the main frequencies of the lumbar spine response obtained after EMD decomposition. Following the processing in S32, noise is removed to obtain the frequency domain signal of the main lumbar spine response. The peak values ​​(i.e., resonance peaks) of the frequency domain signal response are recorded, and parameters closely related to lumbar spine injury are extracted. The table below shows the output results of the frequency domain and time domain signals and the injury correlation with the extracted feature parameters:

[0109] Table 2: Main parameters of the time-domain function graph

[0110]

[0111]

[0112] Table 3: Main parameters of the frequency domain function graph

[0113]

[0114]

[0115] Finally, by comparing the reverse-derived frequency response function with the main frequency after EMD decomposition, severely degenerated samples are removed, and the current lumbar spine condition is assessed. The following are the criteria for assessing the risk of lumbar spine degeneration.

[0116] f0 < 3.5 Hz and |H(f0)| > 3.5 Hz.

[0117] 3.5 ≤ f0 < 4.0 Hz and Δf > 0.8 Hz

[0118] f0≥4.0Hz

[0119] Where |H(f0)| is the resonance amplitude, and Δf is the relative baseline frequency offset.

[0120] S4: Construct a lumbar muscle degeneration state discrimination model, which is obtained by simulating the biomechanical correlation data between the modal parameters of vehicle vibration and the chronic degeneration state of the occupant's lumbar muscles; input the feature vector into the pre-trained lumbar muscle degeneration state discrimination model, obtain the lumbar muscle chronic degeneration state discrimination result output by the lumbar muscle degeneration state discrimination model, and establish a quantization level mapping table.

[0121] S41: Establish baseline status (health status). Before actual driving, have the driver perform quantitative posture calibration according to national standards. This means that while driving, the body should be upright and firmly seated around the steering wheel, with both hands gripping the left and right sides of the steering wheel rim respectively; head upright, shoulders level, eyes looking straight ahead, observing both near and far, paying attention to the sides and up and down; upper body slightly leaning back against the backrest, chest slightly lifted, abdomen slightly tucked in, knees naturally apart, left foot placed below (or to the left of) the clutch pedal, and right foot on the accelerator pedal, maintaining a state of alertness and concentration at all times. The seat position varies from person to person. Adjusting the hands should ensure the left hand is positioned between 9 and 10 o'clock on the steering wheel, and the right hand between 3 and 4 o'clock, but the left hand must be higher than the right. The angle between the upper arm and forearm should be approximately 100°-110°. There should be approximately 10cm of space between the thigh and the lower edge of the steering wheel. Knees should be naturally apart, with the angle between the thigh and lower leg maintained at 90°-110°. Simultaneously, the vibration frequency of the driver at this time was recorded, and the test subject's BMI was calculated.

[0122] S42: The module for constructing and applying the lumbar muscle degeneration status discrimination model mainly consists of a risk assessment module. Its main task is to detect the lumbar spine modality obtained from modal analysis and define the initial lumbar spine state before vehicle operation. This is the initial state of the lumbar spine. Due to differences in biomechanical values ​​caused by differences in human weight, a weight coefficient correction is performed based on the BMI of different individuals to obtain the actual calculated human weight. Where f 校正 The expression is as follows:

[0123]

[0124] m 等效 =m×f 校正

[0125] S43: The assessment model quantifies damage based on the degree of damage, starting with the calculation of the damage index:

[0126]

[0127] Di is the damage index.

[0128] w1-w4 are the feature weights, which are generally taken as (0.2-0.4), and w1+w2+w3+w4=1;

[0129] Δf r It is a shift in the principal resonant frequency;

[0130] WR is the aspect ratio of the resonance peak;

[0131] HDI is the distortion index;

[0132] Δφ is the phase lag angle;

[0133] In the DI formula, w1-w4(0.2-0.4) was obtained through cross-validation using 100 clinical samples (30 healthy individuals, 30 individuals with mild degeneration, and 40 individuals with moderate to severe degeneration). The term "principal resonant frequency shift (Δf)" is used to determine the frequency shift. r The parameter with the highest weight (0.4) is the one with the strongest correlation to decreased intervertebral disc stiffness (R0.4). 2 =0.82), while eliminating individual differences, an age factor (Age / 40) and baseline values ​​are introduced to address the issue of "differences in lumbar spine biomechanics between young and old people." For example, the baseline value calculation for a 60-year-old driver needs to be magnified by 1.5 times (60 / 40), in which the individualized calibration injury index DI is used. personal

[0134]

[0135] in:

[0136] DI personalIt is an individualized calibration of the damage index;

[0137] DI current This is the currently measured damage index;

[0138] DI baseline It is the individual's baseline damage index;

[0139] Age is the age of a person.

[0140] S44: Model Validation and Optimization

[0141] Based on in vitro specimens or patient data of varying degrees of injury, combined with MRI / X-ray diagnostics, the biomechanical rationality is verified before model output:

[0142] The DI score was calculated based on S43, where the DI range (<0.2 / 0.2-0.5 / 0.5-0.8 / >0.8) corresponds to the clinical diagnosis of "no degeneration / minor wear of the annulus fibrosus / rupture of the annulus fibrosus / nucleus pulposus prolapse". The consistency of the results was confirmed by blind review by 30 orthopedic experts. The correspondence between the DI range and the degree of injury was then output based on the experts' assessments. The correspondence between the DI range and the degree of injury is shown below:

[0143] Table 4: Correspondence between DI range and damage degree

[0144] DI range Degree of damage Clinical correspondence DI < 0.2 healthy none 0.2 < DI < 0.5 minor injury Early degeneration of lumbar disc 0.5 < DI < 0.8 Moderate injury Fiber ring rupture DI > 0.8 severe injury Nucleus pulposus prolapse

[0145] In this invention, extreme road conditions (such as continuous bumps) may cause signal distortion, requiring correction based on gyroscope data. Due to differences in the internal structure of the human lumbar spine, it is impossible to distinguish between "lumbar muscle degeneration" and "lumbar osteoarthritis," fusion of bone mineral density parameters is necessary. Furthermore, the vibration frequency acquisition range varies across different vehicle models, necessitating further parameter normalization for each vehicle.

[0146] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for identifying chronic degenerative states of occupant lumbar muscles based on vehicle vibration modal analysis, characterized in that, Includes the following steps: S1: Collects vibration signals transmitted to the waist of the occupant during vehicle operation; S2: Preprocess the vibration signal; S3: Perform modal analysis on the preprocessed vibration signal and extract modal parameters including at least natural frequency and damping ratio, with a focus on modal parameters in a predetermined frequency band related to the biomechanical response of the lumbar spine, and extract feature vectors from the modal parameters; S4: Construct a lumbar muscle degeneration state discrimination model, which is obtained by simulating the biomechanical correlation data between the modal parameters of vehicle vibration and the chronic degeneration state of the occupant's lumbar muscles; The feature vector is input into a pre-trained lumbar muscle degeneration state discrimination model to obtain the lumbar muscle chronic degeneration state discrimination result output by the lumbar muscle degeneration state discrimination model, and a quantization level mapping table is established.

2. The method for identifying chronic degenerative states of occupant lumbar muscles based on vehicle vibration modal analysis according to claim 1, characterized in that, The specific steps of S1 are as follows: S11: Collect lumbar spine curve parameters, including lumbar curvature, intervertebral distance, curvature between L1 and L5, weight, height and age of passengers. S12: The frequency acquisition in the lumbar region mainly focuses on the L3 position of the spine. An accelerometer is installed according to the position of the spine to capture the vibration frequency of the lumbar spine. At the same time, the dynamic output of the seat cushion acceleration is recorded simultaneously. It can quantify the frequency spectrum and amplitude of spinal flexion, extension, lateral flexion and rotation, and measure the vibration frequency in real time. S13: Finally, collect seat data, including seat equivalent mass, seat damping coefficient, and seat stiffness coefficient.

3. The method for identifying chronic degenerative states of occupant lumbar muscles based on vehicle vibration modal analysis according to claim 1, characterized in that, The specific steps of S2 are as follows: S21: The lumbar frequency of the signal is filtered by bandpass filtering, and an anti-aliasing filter is used to prevent the frequency from being too high due to improper driving habits. The output signal at the lumbar muscles is used for preliminary judgment. S22: Further analysis of the specific physical condition of muscle and lumbar spine vibration is then conducted using the transfer functions of the input and output signals.

4. The method for identifying chronic degeneration of occupant lumbar muscles based on vehicle vibration modal analysis according to claim 1, characterized in that, The specific steps of S3 are as follows: S31: Derivation of the nonlinear relationship between input and output: Modal analysis is performed through the data output of hardware and software to record the natural frequency, damping ratio and mode shape of vibration excitation, and the data is segmented according to the driving conditions. S32: Signal and noise analysis: Analyze the specific vibration environment of the vehicle. First, perform EMD decomposition on the measured time-domain signal to decompose it into the main signal, secondary signal and residual term. S33: Fourier Transform and Noise Removal of Main Signal: Fourier transform is performed on the main frequencies of the lumbar spine response obtained after EMD decomposition, and the initial signal is processed to remove noise signals, so as to obtain the frequency domain signal of the main frequencies of the lumbar spine response. The peak value of the frequency domain signal response, i.e. the resonance peak, is recorded, and parameters closely related to lumbar spine injury are extracted.

5. The method for identifying chronic degeneration of occupant lumbar muscles based on vehicle vibration modal analysis according to claim 1, characterized in that, The specific steps of S4 are as follows: S41: Establish baseline status and have the driver perform quantitative posture calibration according to national standard items before the vehicle is actually driven; S42: The lumbar spine modality obtained from the modal analysis is detected. The initial lumbar spine state is defined before the vehicle is driven. This is the initial state of the lumbar spine. Due to the difference in biomechanical values ​​caused by the difference in human weight, the weight coefficient is corrected according to the BMI of different members to obtain the actual human body weight. S43: The establishment of a model for judging the state of lumbar muscle degeneration is based on the quantification of damage according to the degree of injury; S44: Based on data from healthy in vitro specimens or patients with varying degrees of injury, and combined with MRI and X-ray diagnostics, verify the biomechanical rationality before outputting the model.

6. The method for determining the chronic degenerative state of occupant lumbar muscles based on vehicle vibration modal analysis according to claim 2, characterized in that, The vibration signal in S1 is collected by multiple vibration accelerometers installed in the lumbar vertebrae L1-L5 and C1.

7. The method for determining the chronic degeneration state of occupant lumbar muscles based on vehicle vibration modal analysis according to claim 4, characterized in that, The predetermined frequency band in S3 includes the vertical vibration frequency band from 4Hz to 8Hz.

8. The method for determining the chronic degeneration state of occupant lumbar muscles based on vehicle vibration modal analysis according to claim 4, characterized in that, The feature vector extraction in S3 includes: extracting the natural frequency value, damping ratio value, and the mapping relationship between the natural frequency value and the vehicle speed of the dominant mode in the predetermined frequency band.

9. The method for identifying chronic degenerative states of occupant lumbar muscles based on vehicle vibration modal analysis according to claim 1, characterized in that, The discrimination results in S4 include a classification of the level of lumbar muscle degeneration or a score of the degree of degeneration, and the discrimination results are output through an in-vehicle display device, a mobile terminal or a cloud platform.

10. A system for identifying chronic degeneration of occupant lumbar muscles based on vehicle vibration modal analysis, characterized in that, include: The vibration sensor module is used to collect vibration signals transmitted to the waist of the occupants during vehicle operation; The vehicle dynamic parameter acquisition module and the occupant information input module are used to collect vehicle dynamic parameters and occupant status information; A signal preprocessing module is used to preprocess the vibration signal; The modal parameter identification module is used to perform modal analysis on the preprocessed vibration signal and extract modal parameters including at least natural frequency and damping ratio, with a focus on modal parameters in a predetermined frequency band related to the biomechanical response of the lumbar spine. The feature extraction module is used to extract feature vectors from the modal parameters, and also to fuse vehicle dynamic parameters and occupant basic information into the feature vectors; The lumbar muscle degeneration state discrimination model module stores and runs a pre-trained discrimination model, which is used to receive the feature vector and output the discrimination result of the chronic degeneration state of the lumbar muscles. The lumbar muscle degeneration state discrimination model is trained based on the biomechanical correlation data between vehicle vibration modal parameters and the chronic degeneration state of the occupant's lumbar muscles. The output module is used to output the discrimination result.

Images